The Oxford Reference Guide to

ENGLISH MORPHOLOGY
The Oxford Reference Guide to

ENGLISH
MORPHOLOGY

Laurie Bauer, Rochelle Lieber,

and Ingo Plag

1
 1
Great Clarendon Street, Oxford, ox2 6dp,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Laurie Bauer, Rochelle Lieber, and Ingo Plag 2013
The moral rights of the authors have been asserted
First published 2013
First published in paperback 2015
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Cataloging in Publication Data
Data available
ISBN 978–0–19–957926–6 (Hbk.)
ISBN 978–0–19–874706–2 (Pbk.)
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
 CONTENTS

Detailed contents vii

List of tables xxi
Abbreviations and notation xxiii
Preface xxv

Part I Introduction

1 Aims and structures 3

2 Basic principles: terminology 7
3 Basic principles: methods 40
4 Orthography 46

Part II Inflection

5 Verb inflection 61
6 Adjective and adverb inflection 103
7 Noun inflection 121
8 Function words: pronouns, determiners, wh‑forms, deictics 150

Part III Derivation

9 Derivation: phonological considerations 161

10 Derived nouns: event, state, result 195
11 Derived nouns: personal and participant 216
12 Derived nouns: quality, collective, and other abstracts 245
13 Derived verbs 267
14 Derived adjectives 288
15 Derived adverbs 322
16 Locatives of time and space 333
17 Negatives 354
18 Size, quantity, and attitude 385
vi Contents

Part IV Compounding

19 Compounds: formal considerations 431

20 Compounds: semantic considerations 463

Part V Interaction

21 Combination of affixes 493

22 Affixation on compounds and phrases 509
23 Paradigmatic processes 518

Part VI Themes

24 Inflection versus derivation 533

25 The analysis and limits of conversion 545
26 Blocking, competition, and productivity 568
27 The nature of stratification 583
28 English morphology in a typological perspective 616
29 English morphology and theories of morphology 628

References 643
Index of affixes and other formatives 667
Index of names 674
Index of subjects 681
 DETAILED CONTENTS

List of Tables xxi

Abbreviations and notation xxiii
Preface xxv

Part I Introduction

1 Aims and structures  3

1.1 Aims 3
1.2 Structure of the book 4
2 Basic principles: terminology 7
2.1 Prospectus 7
2.2 ‘Word’ and related terms 7
2.2.1 Word-form and lexeme 8
2.2.2 Grammatical word 10
2.2.3 Orthographic word 10
2.2.4 Phonological word 11
2.2.5 Lexical item 12
2.2.6 Orthography and words 12
2.3 Word-elements 13
2.3.1 Morph 13
2.3.2 Morpheme and allomorph 14
2.3.3 Root, base, and affix 17
2.3.4 Vowel, consonant, and stress alternations 20
2.3.5 Suppletion 22
2.3.6 Compounds 24
2.3.7 Non-morphological word-elements 24
2.4 Word Classes 26
2.4.1 Defining word classes 26
2.4.2 Lexical versus grammatical, open versus closed classes 27
viii Detailed Contents

2.4.3 Transposition 27
2.4.4 Headedness 28
2.5 Inflection versus derivation 28
2.6 Lexicalization and productivity 29
2.6.1 The historical development of individual words 29
2.6.2 Availability 32
2.6.3 Profitability 32
2.6.4 Blocking 34
2.6.5 Notions of regularity and irregularity 34
2.7 Native and non-native 35
2.8 Lexical syntactico-semantic terminology 36
2.8.1 Argument, adjunct, and verbal diathesis 36
2.8.2 Referencing 38
2.9 Last word 39
3 Basic principles: methods  40
3.1 Prospectus 40
3.2 Sources of data 40
3.2.1 The corpora 40
3.2.2 Methods of obtaining and treating data from the corpora 42
3.2.3 Other sources of data 43
3.3 Citing data 44
3.4 Interpretation of data 45
4 Orthography 46
4.1 Prospectus 46
4.2 The representation of vowel sound length 47
4.2.1 Basics 47
4.2.2 Double consonants 48
4.2.3 Silent 49
4.2.4 Consonant doubling 49
4.2.5 Loss of silent <e> 51
4.3 The letter <i> replacing <y> 52
4.4 Velar softening 53
 Detailed Contents ix

4.5 Hyphenation 54
4.5.1 Hyphens in derivatives 54
4.5.2 Hyphens in compounds 55
4.5.3 Other lexical and grammatical hyphens 56

Part II Inflection

5 Verb inflection 61
5.1 Prospectus 61
5.2 Lexical versus auxiliary verbs 61
5.3 Lexical verbs 64
5.3.1 The verbal paradigm 64
5.3.2 Regular versus irregular verbs 66
5.3.3 Regular verbs 68
5.3.4 Irregular verbs I: the primary verb be, do, and have70
5.3.5 Irregular verbs II: all others 72
5.3.6 Defective paradigms 79
5.4 Auxiliary verbs 80
5.4.1 Core modal auxiliaries 80
5.4.2 Other auxiliaries 82
5.4.3 Auxiliary clitics and weak forms 84
5.4.4 Negation 86
5.5 Incorporation of infinitival to88
Appendix: Irregular verbs in English90
6 Adjective and adverb inflection 103
6.1 Prospectus 103
6.2 Semantic issues: degree and gradability 103
6.3 The affixal comparative and superlative 105
6.4 Double comparison 106
6.5 Adverbs 107
6.6 Adjectives 109
6.6.1 Irregular paradigms 109
6.6.2 Affixal versus periphrastic comparative and superlative 110
x Detailed Contents

7 Noun inflection 121

7.1 Prospectus 121
7.2 Plural 121
7.2.1 Semantics of pluralization 122
7.2.2 Regular marking of the plural 125
7.2.3 Regular suffix plus change in the base 129
7.2.4 Plurals with vowel alternation 131
7.2.5 Plurals in <n> 133
7.2.6 Unmarked plurals 134
7.2.7 Foreign plurals 136
7.2.8 Plurals of some complex lexical items 140
7.3 Possessive/genitive 141
7.3.1 Semantics 142
7.3.2 Spelling 144
7.3.3 Pronunciation 145
7.3.4 s-genitive versus of-genitive 146
8 Function words: pronouns, determiners, wh-forms, deictics 150
8.1 Prospectus 150
8.2 Pronouns 150
8.2.1 Personal 150
8.2.2 Possessive pronouns 151
8.2.3 Reflexive and emphatic pronouns 152
8.3 Determiners 153
8.4 Wh-words 153
8.5 Compound determinatives 154
8.6 Deictic pro-forms 155
8.7 Analysis of the forms 155
8.7.1 Initial <th>, <h>, and <wh> 156
8.7.2 Oblique <m> 157
8.7.3 Final <n>, <r>, <s>, and <t> 157
8.7.4 Thou and you 157
8.7.5 Analysability and morphemic status 158
 Detailed Contents xi

Part III Derivation

9 Derivation: phonological considerations  161

9.1 Prospectus 161
9.2 Phonology in derivational morphology 161
9.3 Allomorphy 164
9.3.1 Base allomorphy 165
9.3.2 Affix allomorphy 178
9.3.3 Extenders 181
9.4 Prosody 182
9.4.1 Auto-stressed affixes 183
9.4.2 Stress-shifting affixes 184
9.4.3 Stress preservation 187
9.5 Haplology 189
9.6 Prosodic morphology 190
9.6.1 Clippings 190
9.6.2 Hypocoristics in -ie191
9.6.3 Expletive insertion 191
9.7 Phonological selectional restrictions 192
9.8 Summary 193
10 Derived nouns: event, state, result  195
10.1 Prospectus 195
10.2 Formal considerations 195
10.2.1 Non-native affixation 196
10.2.2 Nominalizations with -ing202
10.2.3 Conversion 203
10.2.4 Nominalization by stress shift 204
10.2.5 Miscellaneous 206
10.3 Semantic considerations 206
10.3.1 Range of readings 207
10.3.2 Referencing of arguments or aspect 212
10.3.3 Predictability of nominal semantics 213
10.3.4 Count versus mass interpretation 214
xii Detailed Contents

11 Derived nouns: personal and participant 216

11.1 Prospectus 216
11.2 Formal considerations 217
11.2.1 Subject-referencing affixes: -er, -ant, -an, -st, -meister, -eer, -ster, -nik, -arian217
11.2.2 Object-referencing affixes: -ee226
11.2.3 Inhabitant and language: -ite, -ese, -ish, -i228
11.2.4 Gender: -ess, -ette, -trix230
11.2.5 Prefixes 230
11.3 Semantic considerations 230
11.3.1 Thematic domains 231
11.3.2 Athematic domains 241
12 Derived nouns: quality, collective, and other abstracts  245
12.1 Prospectus 245
12.2 Formal considerations 245
12.2.1 The suffixes -ness and -ity245
12.2.2 The suffixes -dom, -ship, and -hood248
12.2.3 The suffixes -ery, -age, -ana, and -ia250
12.2.4 The suffix -ism253
12.2.5 The suffix -y255
12.3 Semantic considerations 256
12.3.1 Abstracts 257
12.3.2 Collectives and location nouns 262
12.3.3 System of belief, action, or scientific study 266
13 Derived verbs  267
13.1 Prospectus 267
13.2 Formal considerations 267
13.2.1 Prefixation 267
13.2.2 Suffixation 269
13.2.3 Conversion 277
13.2.4 Back-formation and other processes 280
13.3 Semantic considerations 281
13.3.1 Prefixes 282
13.3.2 Suffixes 282
 Detailed Contents xiii

13.3.3 Conversion 285

13.3.4 Back-formation 286
13.4 Competition among verb-deriving processes 287
14 Derived adjectives  288
14.1 Prospectus 288
14.2 Formal considerations 289
14.2.1 Non-native affixes 291
14.2.2 Native affixes 303
14.2.3 Participial adjectives 306
14.3 Semantic considerations 306
14.3.1 Semantic content 307
14.3.2 Relational versus qualitative, gradable versus non-gradable 317
14.3.3 The suffixes -ic and -ical318
14.3.4 Multiple affixes and semantic interpretation 320
15 Derived adverbs 322
15.1 Prospectus 322
15.2 Are adverbs and adjectives the same category? 322
15.3 Suffixation 323
15.3.1 The suffix -ly323
15.3.2 The suffix -s328
15.3.3 The suffix -ward(s)328
15.3.4 The suffix -ways329
15.3.5 The suffix -wise329
15.4 The prefix A- 331
15.5 Conversion 332
16 Locative of time and space  333
16.1 Prospectus 333
16.2 Formal considerations 333
16.2.1 Non-native prefixes 334
16.2.2 Native prefixes 340
16.3 Semantic considerations 344
16.3.1 Core meanings 344
16.3.2 Native and non-native cohorts: a comparison 348
16.3.3 Argument structure effects 351
xiv Detailed Contents

17 Negatives 354
17.1 Prospectus 354
17.2 Formal considerations 354
17.2.1 Basic characteristics 354
17.2.2 Phonological and orthographic properties 359
17.2.3 Productivity 361
17.3 Semantic considerations 364
17.3.1 Types of negativity 364
17.3.2 Ranges of meaning 366
17.3.3 Redundancy 376
17.3.4 Rivals 377
17.3.5 Semantic restrictions on bases 381
18 Size, quantity, and attitude  385
18.1 Prospectus 385
18.2 Evaluation and size 385
18.2.1 Preliminaries 385
18.2.2 Diminution by affixation 386
18.2.3 Proper noun diminution: Hypocoristics 400
18.2.4 Clipping 402
18.2.5 Augmentatives 404
18.2.6 Reduplication 411
18.2.7 Expletive insertion 413
18.2.8 The prefixes pseudo- and quasi-414
18.3 Non-evaluative attitude: anti- and pro-417
18.4 Quantification and measure 418
18.4.1 Nominal -ful418
18.4.2 Verbal re-419
18.4.3 The suffix -fold420
18.4.4 The suffixes -some420
18.4.5 The suffix -ton421
18.4.6 The suffix -ish421
18.4.7 The prefix hypo-421
 Detailed Contents xv

18.5 Cardinal and ordinal numbers  422

18.5.1 The suffixes -teen and -ty422
18.5.2 More complex numbers: beyond 20 423
18.5.3 Ordinals and fractions 425
18.5.4 Number prefixes 426

Part IV Compounding

19 Compounds: formal considerations 431

19.1 Prospectus 431
19.2 Boundary issues: What is a compound? 431
19.2.1 Compounds: morphological or syntactic 431
19.2.2 Compounding versus prefixation and suffixation 440
19.2.3 Compounding and compounds: process versus product 442
19.3 Internal structure 443
19.3.1 Headedness 443
19.3.2 Constituency 443
19.3.3 Stress 444
19.3.4 Orthography 449
19.4 Types of compound 451
19.4.1 Nominal compounds 451
19.4.2 Verbal compounds 452
19.4.3 Adjectival compounds 453
19.4.4 Prepositional compounds 453
19.4.5 Neo-classical compounds 454
19.4.6 Phrasal compounds 456
19.4.7 Reduplicative compounds 457
19.4.8 Blends 458
20 Compounds: semantic considerations  463
20.1 Prospectus 463
20.1.1 General remarks 463
20.1.2 Classification 464
20.2 Argumental compounds 466
xvi Detailed Contents

20.2.1 Argumental compounds with deverbal heads 466

20.2.2 Argumental verbal compounds 471
20.2.3 Argumental compounds with relational nouns
and argument-taking adjectives as heads 472
20.2.4 Non-head is argument-taking 472
20.2.5 Exocentric examples 473
20.3 Non-argumental compounds 474
20.3.1 Attributive compounds 474
20.3.2 Coordinative compounds 479
20.4 The border between argumental and non-argumental compounds 482
20.5 The semantics of blends and neoclassical compounds 483
20.5.1 The semantics of blends 483
20.5.2 The semantics of neoclassical compounds 485
20.6 Miscellaneous compounds 487
20.6.1 Phrasal compounds 488
20.6.2 Reduplicative compounds 490

Part V Interaction

21 Combination of affixes 493

21.1 Prospectus 493
21.2 Suffix combinations and prefix combinations in derivation 493
21.2.1 Suffix combinations 493
21.2.2 Prefix combinations 497
21.2.3 Recursion 499
21.3 Combinations of prefixes with suffixes in derivation 500
21.3.1 Multiple hierarchical affixation 501
21.3.2 Parasynthetic affixation 502
21.4 Conversion and affixation 503
21.5 Inflection 505
21.5.1 Inflection in interaction with derivation 505
21.5.2 Inflection interacting with inflection 506
21.6 Derivatives with more than two affixes 507
 Detailed Contents xvii

22 Affixation on compounds and phrases  509

22.1 Prospectus 509
22.2 Data 509
22.2.1 Affixes on compound bases 509
22.2.2 Affixes on phrasal bases 513
22.3 Factors influencing affixation on compounds and phrases 514
22.4 The role of lexicalization 516
22.5 Conclusion 517
23 Paradigmatic processes 518
23.1 Prospectus 518
23.2 Terminological preliminaries 518
23.3 Inflection 521
23.4 Derivation 522
23.5 Compounding 524
23.6 Splinters 525
23.7 Conclusion 530

Part VI Themes

24 Inflection versus derivation 533

24.1 Prospectus 533
24.2 The basis of inflection and derivation 533
24.3 Some problem cases 535
24.3.1 The nominal plural 535
24.3.2 Adverbial marking with -ly536
24.3.3 Ordinal -th  536
24.3.4 Other numerical formatives: -teen and -ty  537
24.3.5 The participles 537
24.3.6 Summary 538
24.4 Clear-cut distinctions 538
24.5 Why might we need to know? 540
24.5.1 Productivity 540
24.5.2 Affix ordering 542
xviii Detailed Contents

24.5.3 Headedness 542

24.5.4 Conclusion 544
25 The analysis and limits of conversion  545
25.1 Prospectus 545
25.2 The issues 545
25.3 Cases for further consideration 549
25.3.1 Adjective to noun cases 549
25.3.2 Mention versus use: ‘but me no buts’  551
25.3.3 Formations related to prepositions: a down, to down, the down train  522
25.3.4 Minor phonological modification 552
25.3.5 Participles: his shooting, an interested party  555
25.3.6 Type coercion 557
25.3.7 Adverb formation real ale ◆ real good559
25.3.8 Compounds and phrases 561
25.4 Modelling conversion 562
25.4.1 Conversion, narrowly defined 562
25.4.2 Zero-derivation 563
25.4.3 Relisting 564
25.4.4 Underspecification/multifunctionality 564
25.4.5 The influence of pragmatics: contextuals 565
25.4.6 Conversion as inflectional 566
25.4.7 Conversion as metonymy 567
25.4.8 Various other nomenclatures 567
25.5 Summary 567
26 Blocking, competition, and productivity  568
26.1 Prospectus 568
26.2 New words and old 569
26.3 Competition 571
26.3.1 Competition in inflection 571
26.3.2 Competition in derivation 573
26.3.3 A diachronic view of competition 574
26.4 Blocking 575
 Detailed Contents xix

26.5 Productivity 578

26.6 Conclusion 581
27 The nature of stratification 583
27.1 Prospectus 583
27.2 Data 584
27.3 Patterns 610
27.4 Theoretical consequences 612
27.5 Conclusions 615
28 English morphology in a typological perspective 616
28.1 Prospectus 616
28.2 Broad classificatory schemes 617
28.2.1 The traditional Humboldtian classification 617
28.2.2 Head- versus dependent- marking 618
28.3 A finer-grained view 619
28.3.1 Inflection 619
28.3.2 Derivation 621
28.3.3 Conversion 622
28.3.4 Compounds 623
28.4 Conclusion 627
29 English morphology and theories of morphology 628
29.1 Prospectus 628
29.2 Broad theoretical models 628
29.2.1 Are there morphemes? IA, IP, and WP 629
29.2.2 Realization versus non-realization models 631
29.2.3 Analogical models and Construction Morphology 633
29.3 Specific claims 635
29.3.1 The Righthand Head Rule 635
29.3.2 The Unitary Base and Unitary Output Hypotheses 635
29.3.3 Blocking and the Elsewhere Condition 636
29.3.4 Lexical Phonology and Morphology: Level Ordering
and Bracket Erasure 637
29.3.5 The Monosuffix Constraint 638
xx Detailed Contents

29.3.6 The First Sister Principle and related proposals 638

29.3.7 The Lexical Integrity Hypothesis and related proposals 638
29.4 Implications 639

References 643
Index of affixes and other formatives 667
Index of names 674
Index of subjects 681
 LIST OF TABLES

Table 2.1 Native and non-native triplets 36

Table 4.1 Spelling of some noun-noun sequences according to different dictionaries 55
Table 5.1 Verbal inflectional paradigms 62
Table 5.2 Verbal paradigms 65
Table 5.3 Kinds of vowel alternations 74
Table 5.4 Syncretism of whole forms 74
Table 5.5 Syncretism of bases, verbs with suffixed forms 75
Table 9.1 Prosodic adjustments 167
Table 9.2 Vowel alternations 172
Table 9.3 Alternations between vowels 173
Table 9.4 Velar softening, spirantization, and palatalization 175
Table 11.1 Overlapping semantic domains 231
Table 13.1 The semantics of -ize and -ify283
Table 14.1 Formal characteristics of adjective-forming suffixes 290
Table 16.1 Formal characteristics of locative prefixes 334
Table 17.1 Formal properties of negative and private affixes
(native affixes are italicized). Parenthesized items are rare 355
Table 17.2 Distribution of readings 367
Table 19.1 Proportion of spellings in three corpora in percent 450
Table 19.2 Compound types by syntactic category 452
Table 22.1 Suffixes that frequently take compounds as bases 510
Table 22.2 Prefixes that frequently take compounds as bases 511
Table 22.3 Suffixes that only occasionally attach to compounds 512
Table 22.4 Prefixes that only occasionally attach to compounds 513
Table 22.5 Affixes on phrasal bases 513
Table 27.1 Native affixes and the bases they select 585
Table 27.2 Non-native suffixes and the bases they select 587
xxii List of Tables

Table 27.3 Non-native prefixes and the bases they select 589
Table 27.4 Non-native suffixes followed by the native suffixes 591
Table 27.5 Native suffixes followed by non-native suffixes 594
Table 27.6 Native suffixes followed by native suffixes 595
Table 27.7 Non-native suffixes followed by non-native suffixes 596
Table 27.8 Non-native prefixes preceding native prefixes 600
Table 27.9 Native prefixes preceding non-native prefixes 601
Table 27.10 Native prefixes preceding native prefixes 603
Table 27.11 Non-native prefixes preceding non-native prefixes 603
 ABBREVIATIONS AND NOTATION

The phonetic transcription system we use is basically that of Wells (2008).

{ } enclosing morphemes
/ / enclosing representations of pronunciation where phonetic detail is
not important
[ ] enclosing pronunciations; showing alterations to quotations
< > enclosing spellings
“ ” enclosing quotations
‘ ’ enclosing glosses; enclosing quotations within quotations
small caps new technical terms; lexemes; a representation of a variable by its
word class
Italics cited words and morphs; titles; data sources
Bold to mark matters of interest in examples and quotations.
* unacceptable
? of questionable acceptability
% used by some people, variable
◆ is related morphologically to
N not attested
~ varies with
ǁ divides British and American pronunciations or spellings
´ main stress
` secondary stress
. syllable boundary
A adjective
arch. archaic
Adv adverb
AusE Australian English
BNC British National Corpus
BrE British English
CanE Canadian English
COCA Corpus of Contemporary American English
dial. dialect(al)
IrE Irish English
N noun
xxiv Abbreviations and notation

NAmE North American English

NonSt non-standard
NZE New Zealand English
RP Received Pronunciation
SAf E South African English
ScE Scottish English
sl. slang
St standard
V verb
X a variable form or meaning, usually representing the meaning of a base
 PREFACE

This project began in 2007 over breakfast somewhere in Europe, we no longer remember
where. Laurie observed to Shelly that there was need for a comprehensive reference volume
covering the whole of contemporary English morphology. Although English is the most
extensively described of the world’s languages, no such volume existed; the closest to this
kind of work was Marchand (1969), and that was over 40 years old and covered only word-
formation. Shelly agreed that a comprehensive reference work would be useful, and we briefly
considered how it might get written. It could not be a handbook with multiple authors con‑
tributing individual chapters, as it needed to have a consistent authorial voice. But we agreed
that two people could not do it. Thus was Ingo recruited. We had no idea at the time how big
a project it would become, how exciting a collaboration it would be, and how much fun we
would have in the process. Nor did we realize how much new we would learn about a lan‑
guage whose morphology we thought we knew reasonably well.
This work is a true collaboration, with each of us contributing equally to the final product,
and hence the order of authors on the title page is alphabetical. Each of us participated in
some way in the drafting and revising of every chapter. The reader may be able to discern
different voices in the writing at different points, but we have done our best to make the
volume a seamless whole. We hope we have succeeded.
There are many individuals and groups we need to thank. Shelly wishes to acknowledge
the support of the Provost’s office of the University of New Hampshire for a Faculty
Fellowship in the early stages of our work. Thanks also to students Josh Albair and Heather
Froehlich for help with gathering data.
Laurie wishes to thank the Royal Society of New Zealand for a grant through its Marsden
Fund which supported the project. He would also like to thank Natalia Beliaeva and Liza
Tarasova for their work in support of the project.
Ingo wishes to thank the Volkswagen-Stiftung and the Thyssen-Stiftung for his Opus-
Magnum award, which allowed him to focus on this project for an entire year and to spend
parts of this year working with Laurie in Wellington. Ingo is also grateful to his colleagues at
his department in Siegen who supported him during his period of absence by keeping the
home fires burning.
Many thanks also go to Ingo’s student assistants at Siegen, who lent their support in vari‑
ous ways, for example by doing extensive corpus searches, proof-reading or providing other
kinds of valuable service: Julia Homann, Lena Hüsch, Jennifer Matthes, and Ute
Raffelsiefer.
xxvi Preface

All three authors wish to thank several people for providing comments on individual
chapters at various stages of the manuscript. These are Sabine Arndt-Lappe, Natalia Beliaeva,
Melanie Bell, Linda Hilkenbach, Julia Homann, Kristina Kösling, Gero Kunter, and Mareile
Schramm.
Finally, we wish to thank our spouses and families for suffering our yearly meetings. We
could not have completed this book without the support and hospitality of David, Yitzl, and
Raoul; Claudia, Jonas, Hannah, and Leo; Winifred and Ingrid.
 Pa rt I

Introduction
 chapter 1

Aims and structures

1.1 Aims
English is one of the best-described languages in the world: there are hundreds, if not thou-
sands of books and articles on various aspects of its structure and semantics, from descrip-
tive, historical, and theoretical perspectives. Nevertheless, there is no single-volume work
that focuses on the current state of English morphology. This volume is intended to fill that
gap by offering a comprehensive descriptive overview of English morphology and of the
theoretical problems associated with it.
Major aspects of English morphology are covered in recent textbooks (Carstairs-McCarthy
2002; Plag 2003; Harley 2006), but these are written for pedagogical purposes, and their
value as general reference works is limited. Marchand’s milestone monograph The Categories
and Types of Present-day English Word-formation (Marchand 1969) and Adams’s concise
Complex Words in English (Adams 2001) cover only English word-formation, but not inflec-
tion. Marchand (1969) is devoted to the historical development of English word-formation
and is largely based on dictionary data. Written over forty years ago, it is inevitably dated: we
would expect that there is much that is new to say about the state of English morphology in
the second decade of the twenty-first century. Adams (2001) is more recent, of course, but it
does not engage with theoretical issues. Finally Huddleston and Pullum’s (2002) Cambridge
Grammar of the English Language covers English morphology briefly, but as a small part of a
general grammar of English without developing the theoretical discussion, and without relat-
ing to disciplines other than traditional structural linguistics.
Morphological theory has also made great strides in recent years. Certainly, since
Marchand’s time there has been an explosion of interest in word-formation and inflection,
both in English and other languages, with the result that our understanding of the workings
of the mental lexicon is far more sophisticated than it was fifty years ago. And as will be dis-
cussed in detail in Chapter 3, our ability to find and manipulate data has improved consider-
ably with the development of vast computerized corpora.
In sum, we see a need for a comprehensive, up-to-date overview of contemporary English
morphology that is based on large collections of data that reflect the usage of words by con-
temporary speakers of English. In our view, such an overview should be as theory-neutral as
possible to allow access for the widest possible audience, but it should be theoretically
4 Aims and structures

informed, combining recent findings in corpus linguistics, theoretical linguistics, psycholin-

guistics and computational linguistics. Furthermore, it is our aim to include different varie-
ties of English. In general, differences between varieties seem to be rather minor, but there
are cases where they merit discussion.
The present book will offer a thorough, data-rich description of all phenomena of English
word-formation (derivation, compounding, conversion, and minor processes such as sub-
tractive morphology) and inflection. Our focus in the description will be on structure and
use, to the exclusion of various other interesting topics (e.g. processing or acquisition). The
descriptive account will, however, be complemented by a consideration of the theoretical
challenges that these phenomena present.

1.2 Structure of the book

As mentioned in Section 1.1, our goal is to offer a data-rich and relatively theory-neutral
description of English morphology, combined with a consideration of some of the theoreti-
cal issues raised by the phenomena we describe. We will cover not only areas of English
morphology that have been well-studied—nominalizations, agentive forms, comparatives,
root and synthetic compounds—but also areas that have been less well studied or hardly
studied at all: adjective-forming affixes like ‑al and ‑ic, abstract noun-forming suffixes like
‑hood, ‑dom, and ‑ship, neoclassical compounds, the morphology of numbers, and the status
of irregularly inflected forms in contemporary English.
There are a number of phenomena that we have chosen not to include, however. Among
these we have chosen not to discuss the formation of verb-particle constructions, on the
grounds that they are better treated as a matter of syntax than of morphology, although we
will discuss cases of conversion from verb-particle constructions (Chapter 19.2.3). We have
not included cases like forget-me-not or jack-in-the-box because we consider them to be lexi-
calized phrases (Chapter 19.2.1.2). Readers will also find that we have not discussed affixes or
formatives that are productive only in highly scientific or technical fields, for example, the
endings ‑ate (perchlorate) and ‑ose (glucose) that occur in chemical terms, or the suffix ‑mo
(twelvemo) that occurs only in the language of printers. Finally, in deciding which affixes to
treat in detail, we have largely limited ourselves to affixes that are available to native speakers
for coining new words or to affixes for which there are sufficient examples that they might
become available, for example, in jocular contexts. We will therefore exclude consideration of
historical remnants like ‑ric (bishopric), ‑ter (laughter), and ‑le (sparkle), which are clearly no
longer available to native speakers.
As might be imagined, there are many conceivable ways that a work of this sort might be
organized, and deciding which one most suited our aims or the demands of our readers was
a daunting task. We decided against listing individual affixes alphabetically and treating each
one separately (as Marchand 1969 does), on the grounds that doing so would preclude
 1.2 Structure of the book 5

making comparisons concerning the formal and semantic characteristics of various cohorts
of affixes and types of conversion and compounding. We decided to separate descriptive sec-
tions from more theoretical sections in the interest of making the book as useful as possible
to the widest range of potential readers, including theoretical linguists, psycholinguists, com-
putational linguists, English language teachers, and interested non-academics.
The book is divided into six parts. The remainder of this introductory part of the book is
devoted to setting out clearly and in detail the terminology we use (Chapter 2) and the meth-
ods we have applied in obtaining, analysing, and citing data (Chapter 3). Chapter 4 concerns
orthographic practices that overarch inflection and word-formation in English. The descrip-
tive chapters of the book (in Parts II–V) contain a detailed account of word-formation and
inflection in contemporary English, giving equal weight to formal considerations (phonology,
orthography, selectional properties, productivity) and to semantic considerations. Part II is
comprised of Chapters 5 to 8, which are devoted to the inflection of verbs, adjectives and
adverbs, nouns, and function words respectively. In Part III, we treat derivation, under which
we include affixation and conversion, but also, where relevant, minor processes like back-for-
mation, clipping, and the formation of acronyms. Chapter 9 gives a general overview of pho-
nological processes that recur in many processes of English derivation. Chapters 10 to 12
concern the derivation of nouns, Chapter 13 the derivation of verbs, Chapter 14 the derivation
of adjectives, and Chapter 15 the derivation of adverbs. Chapters 16–18 take up other sorts of
derivation, from locatives and negatives to evaluatives. Part IV is devoted to compounding,
one chapter (Chapter 19) to formal properties and another (Chapter 20) to matters of seman-
tic interpretation. In this part we include discussions of blends, as well as of neoclassical com-
pounds. Part V is the final descriptive section of the book, and is devoted to the ways in which
various morphological processes can interact with each other in English. Chapter 21 deals with
combinations of affixes, including issues of affix-ordering. Chapter 22 concerns the ways in
which complex forms (compounds and phrases) interact with affixation. Chapter 23 discusses
paradigmatic processes in English word-formation.
Part VI of this volume is devoted to some of the many areas of theoretical concern that
have been raised by the enormous amount of research on English morphology in recent
decades. We do not argue from the standpoint of any extant theoretical framework, but rather
survey the issues and show where the descriptive data we have set out in sections II–V bears
on them.
Although the organization of the book allows for a clear placement of most affixes, there
were nevertheless quite a few cases where we were forced to make assignments to one
chapter or another that might seem odd or arbitrary. As an example, we chose to put arch‑
in the chapter on Locatives of Time and Space, even though arch‑ (archfiend) might not
seem to be locative in the same sense as pre‑ or post‑. But since arch‑ is at least metaphori-
cally locational, and no other rubrics seemed obvious, we felt that that chapter was the best
we could do. We preferred this solution to having a chapter that we would be forced to call
‘Random Bits’.
6 Aims and structures

A final problem concerned the decision whether to treat particular forms as prefixes or as
first elements of compounds (either native or neoclassical); for example, were we to count
nano‑ (nanometric, nanostructure) or over‑ (overfly, overcoat) as prefixes or the lefthand ele-
ment of compounds? We made such decisions based on whether we could find signs of par-
ticularly affixal behaviour in such forms (as set out in Chapter 19.2.2) or where it seemed
useful to be able to draw parallels with obvious affixes. Again, some readers will inevitably
disagree with our judgements on given points.
This book is intended to be used as a reference work, of course, and readers can dip in and
look for what we have to say about one phenomenon or another. But those readers who have
the interest and stamina to read the book through systematically will sometimes be surprised,
as we were in gathering the data and writing the book, at the phenomenal exuberance of the
morphology of English and the range of resources speakers have at their disposal to create
and use new words.
 chapter 2

Basic principles: terminology

2.1 Prospectus
In this chapter we set forth the terminology that we will use in this book. Our intention
here is not to create a new theoretical framework or to endorse a particular existing theo-
retical framework, but rather to provide a relatively theory-neutral terminology that is
sufficient to fulfil the needs of this book, without doing violence to the core of commonly
accepted terms. We seek as much as possible to choose terms that are widely accepted
among morphologists of all stripes, and to avoid those that have developed specialized
meanings within a particular framework. We have aimed at clear, simple, and (as much as
possible) uncontroversial definitions of those terms. We start in Section 2.2 with a defini-
tion of ‘word’ and related terms, and of the various issues they raise. Subsequent sections
look at terms that concern word-internal elements, word classes, the distinction between
inflection and derivation, notions of lexicalization and productivity, native versus non-
native morphology, and terminology concerning syntactico-semantic concepts such as
‘argument’ and ‘adjunct’. In what follows we will use small capitals to mark terms for which
we provide definitions.

2.2 ‘Word’ and related terms

Like so many fundamental notions in linguistics, ‘word’ has resisted definition. Even a good
definition of ‘word’ in a particular language proves to be surprisingly elusive; a universal defi-
nition of the notion, in anything but the broadest terms, is apparently beyond the abilities of
current linguistics (see Bauer 2000; Dixon and Aikhenvald 2002). Part of the difficulty is
that, as an element of the English language, the word word can be used to denote things
which are conceptually very different from each other, and that we need a better classification
and a more precise terminology if we are to serve the needs of morphologists. Fortunately,
most of this terminology is widely accepted, although there are some terms which have vary-
ing usages in different theoretical frameworks.
8 Basic principles: terminology

2.2.1 Word-form and lexeme

In a sentence like (1), there are two answers to ‘How many different words are there in this
sentence?’

(1) Every time Fred sees violence on TV, he says seeing it destroys the soul.

In one answer, sees and seeing are forms of ‘the same word’, in the other, they are ‘different words’.
Where sees and seeing are ‘different words’, we will use the term word-form; where they are ‘the
same word’ we will use the term ‘lexeme’. In other words, ‘word-form’ refers to a phonological/
orthographic shape irrespective of meaning. Lexeme is an abstraction over one or more word-
forms that convey the same lexical meaning. Making this distinction then allows us to retain the
term word for occasions where the distinction between word-form and lexeme is not crucial to
the discussion (Bauer 1983). Word-forms are generally marked by the use of italics, while there is
less agreement on a notation for lexemes. We adopt here the notation of Lyons (1968) and use
small capitals. Notationally, therefore, we can say that sees and seeing are forms of see.
While this distinction has been well canvassed in the recent literature, it leaves various
questions unanswered. To begin with, it presupposes the distinction between inflection and
derivation, which is not necessarily as clear-cut as we might wish (see below, Section 2.5, and
Chapter 24). Derivational affixes create new lexemes from old lexemes, inflectional affixes
create new word-forms from old word-forms. To the extent that there are questions about the
distinction between inflection and derivation, the distinction between word-form and lex-
eme is equally in doubt.
Importantly, there is equivocation in the literature as to whether an item like but or the is
or is not a lexeme. For some, perhaps deriving from the French tradition, grammatical words
of this type (see Section 2.2.2 on the ambiguity of the term ‘grammatical word’) are not lex-
emes. For others, there is no reason why we should not see but as a lexeme invariably real-
ized by the word-form but. Here we take the second position, and by doing so make two
assumptions: first, that an item like but is part of the lexis of English and not simply part of
the grammar, and second that a lexeme can have a paradigm which includes only one form.
Taking this approach also allows us to avoid the issue of whether there is a lexeme this with
the two forms this and these, even though these are grammatical words in the sense used just
above: for us, there is a lexeme this.
It is important not to confuse the word-form that is used as the citation form of a
lexeme with the lexeme itself. We name the lexeme hit by using one of its word-forms
(although in principle we could name it in some more random way, e.g. as lexeme # 3,762). In
English, the form used as the citation form for a verb is the base form of the verb. In the case
of hit, because the plain non-past form, the preterite, and the past participle form are for-
mally identical, it is difficult to tell which form is being used except by means of the more
general parallel.
 2.2 ‘Word’ and related terms 9

In general terms, since lexemes are tied to particular inflectional paradigms (each lex-
eme is realized by a set of word-forms), we take the position that no noun and verb can
belong to the same lexeme, since they have different paradigms. This view is apparently
not shared by some dictionaries, which list, for example, butter (noun) and to butter (verb)
under the same headword. While we acknowledge the relationship between these two
words (which we attribute to conversion, see Chapter 25), we take it that they are distinct
lexemes.
While word-forms, by definition, include inflectional suffixes (if any), it is not clear
whether they include clitics. Is butcher’ll in The butcher’ll dice the meat for you a word-form of
butcher? We take this not to be the case, and rather treat forms like butcher’ll as consisting
of two separate word-forms; that is, host and clitic each constitute a word-form.
Word-forms are generally said to be definable in terms of various criteria. In particular,
they are supposed to be the smallest items which can stand alone as an utterance, be them-
selves movable, though the elements within them are not movable, and be uninterruptable.
Thus, given a form such as uncountable, we find the results set out in (2).

(2) a. What kind of noun is this? Uncountable.

b. There were uncountable reasons for this. The reasons were uncountable.
c. uncountable, *unablecount, *countableun, *countunable
d. *un totally countable, *uncount reason able

However useful these criteria may be, there are problems with them if they are interpreted as
necessary and sufficient conditions for word-form-hood. For example, apart from mentions
(for example, in the sentence “‘ran’ is the past tense of ‘run’”) past tense forms typically do
not occur alone. At the same time, (3) clearly does illustrate an acceptable utterance.

(3) Are you feeling cold?—Ish.

If ‑ish is not to be interpreted as a word-form, we need some mechanism for telling us pre-
cisely when the use in isolation criterion will or will not work.
The criterion of immovable elements is apparently broken by pairs such as armchair /
chair-arm, boathouse / houseboat, doghouse / housedog, but here the meaning is different, while
the criterion is really aimed at cases where there is no contrast of meaning. Extremely mar-
ginal cases are thus things like horse-zebra, which for some speakers is potentially synony-
mous with zebra-horse meaning ‘a cross between a zebra and a horse’, although for other
speakers, for example those who have special knowledge of interspecies breeding, the two
compounds might have distinct denotations. The interruptability criterion is apparently
broken by the marginal process of expletive insertion that gives us kanga-bloody-roo, Talla-
fucking-hassee.
10 Basic principles: terminology

2.2.2 Grammatical word

The term ‘grammatical word’ is used in two distinct ways in the literature (Bauer 2004b). The
first, for which we will retain the term grammatical word, contrasts with lexical word.
The former includes conjunctions, articles, prepositions, particles; the latter nouns, verbs
and adjectives.
For the second meaning, consider again (4).

(4) a. I will have to tread in your footsteps.

b. %It was strange to walk on the ground where he had tread.
c. The sheep is being shorn.
d. The sheep are being shorn.

In all these cases the items shown in bold are the same orthographic words, realizing a single
lexeme, but fulfilling different functions. The different functions are definable in terms of the
place the word-form holds in the relevant paradigm. A word defined in terms of its place in
the paradigm is sometimes termed a ‘grammatical word’, but, to keep the distinction clear, we
will prefer the alternative term, morphosyntactic word. Note that regular ‑ed forms of
English verbs will typically represent two morphosyntactic words: the past tense and past
participle forms. Not all identity of word-forms is due to a question of morphosyntactic
words. Bound, out of context, may be, among other things, the past participle of bind or a
noun meaning ‘jump’, but this is a matter of a single word-form belonging to two distinct
lexemes.
Another issue arises with respect to the form hit in the sentences in (5).

(5) a. I can never hit the bullseye.

b. I hit the bullseye last night for the first time.

Hit in (5a) has a different function from hit in (5b); in traditional terms, the first is the infini-
tive, the second is the past tense of hit. Although these two are the same word-form, they
constitute two distinct morphosyntactic words, as they realize two different sets of morpho-
syntactic features.

2.2.3 Orthographic word

An orthographic word is a word-form as defined by the spelling system. Thus, if we
define spaces and punctuation marks such as commas, colons, and so on as boundary mark-
ers in the English spelling system, an orthographic word is an element in a written text sur-
rounded by boundary markers. Next we have to ask whether apostrophes and hyphens are
 2.2 ‘Word’ and related terms 11

boundary markers or not. No matter which decision is taken, this decision does not affect the
problem of whether they are one or two lexemes, word-forms, and so on.
There are other places where there is inconsistency in English as to how to write individual
items: all right versus alright, in so far as versus insofar as. The standardization of English
spelling is so great that such examples are rare, but the fact that they persist is in itself a
noteworthy point.
The lay notion of ‘word’ is determined by the orthographic word; this is circular, since the
orthographic word is derived from some more fundamental appreciation of what a word is
(although, of course, there may be some confusion where historical change affects the basis
for taking a decision). It is an open question how far the orthographic word corresponds to
the more technical notion of ‘word-form’. Even linguists tend to assume that there is a good
match (perhaps with the exceptions mentioned above), but it might be possible to argue for
clitic status for items like the, a, an and even for subject pronouns like I, he, she, which would
make quite an important difference.

2.2.4 Phonological word

A number of scholars distinguish the notion of ‘phonological’ or ‘prosodic word’ from gram-
matical or orthographic word. The phonological word is the domain within which cer-
tain phonological or prosodic rules apply, for example, rules of syllabification or stress
placement. Phonological words may be smaller or larger than grammatical or orthographic
words. For example, Raffelsiefen (1999) has argued that complex words containing vowel-
initial suffixes like ‑ish consist of a single prosodic word, whereas complex words that contain
consonant-initial suffixes like ‑less contain two prosodic words. Words like fattish containing
the former suffix are syllabified as one unit with the base final consonant forming the onset
of the second syllable, as in (6a). In contrast, with a word like snackless, the base final conso-
nant forms the coda of the first syllable. The suffix ‑less constitutes its own phonological word
as in (6b):

(6) a. ((fæ)(tɪʃ ))P‑word

b. (snæk)P‑word(ləs)P‑word

Others define the phonological word slightly differently, for instance in terms of phonotac-
tics, so that unnaturally and brokenness are made up of more than one phonological word
each, since English words do not (otherwise) allow geminates within them (Goldsmith 1990:
122). Some scholars do not include prefixes within the phonological word (Vogel 2006: 532).
There is obviously a great deal more to be said about the distinction between the phonologi-
cal word and the grammatical word, and the justification for this distinction in English. We
will return to these matters in Chapter 9.
12 Basic principles: terminology

2.2.5 Lexical item

By lexical item we understand any item which, because of its lack of predictable semantics
or form, must be listed in the mental lexicon. We prefer the term ‘lexical item’ (used by
Carstairs-McCarthy 2002: 13) over Di Sciullo and Williams’s (1987: ch. 1) ‘listeme’. So lexical
items form a superset that includes lexemes as a subset. Lexical items also include:

• Items formed by the lexicalization of syntactic structure (e.g. forget-me-not, jack-in-the-

box, house of God, Australian Capital Territory).
• Idioms (e.g. by and large, red herring, take [ ]NP[+human] to the cleaners ‘ruin [NP] finan-
cially’, the man on the Clapham omnibus ‘the average man on the street’ (BrE), come the
raw prawn ‘attempt to deceive’ (AusE).
• Phrasal verbs (put up ‘accommodate’, put up with ‘tolerate’, run up ‘allow to accumulate’,
do in ‘murder’).
• Fixed figurative expressions (be like a dog with two tails ‘show extreme happiness’, be
between a rock and a hard place ‘have only two options, each of which is unpalatable’,
better the devil you know (than the devil you don’t know)).
• Proverbs (too many cooks spoil the broth, many hands make light work).

Many lexical items are items which can be found in standard printed dictionaries, not so
much because their meaning is totally unpredictable, but because they happen to be the way
the notion is expressed in English. For example, the Concise Oxford Dictionary (2008) lists
open marriage (not, for instance, free marriage), open sandwich (not, for instance, topless sand-
wich), open secret (not, for instance, public secret). So lexical items may be dictionary entries,
and as such are probably represented in a speaker’s mental lexicon.

2.2.6 Orthography and words

As mentioned in Section 2.2.3, the general assumption for English is that orthographic words
correspond to word-forms, and that these are ‘words’ par excellence. Moreover, thanks to high
levels of literacy in English, the lay notion of a word is determined by the orthographic word.
In most cases this does not give rise to great theoretical problems for the study of English
(though we will take up issues concerning the representation of compounds in Chapter 19).
However, from a theoretical standpoint, this is the wrong way around. As pointed out above,
we must assume that the orthographic word is based on the linguistic knowledge of speakers,
and is the output of that unconscious knowledge (based, in all probability, on criteria such as
uninterruptability). Thus it is that items are more likely to be written as single words the
more item-familiar (Meys 1975) they are or the more lexicalized they are. For example, the
OED gives the case of altogether written as two words in the earliest forms of English, but
being written as a single word from an early period (and there is a comment on confusion
 2.3 Word-elements 13

between altogether and all together in more modern usage). The change from all right to alright
is much more modern (first citation for alright in the OED is for 1893), but the process seems
the same. Thus it seems that orthography is influenced by lexicalization and frequency,
though the spelling seems to be a conservative representation. In principle, this could lead to
a mismatch between the linguist’s notion of a word and the lay notion as represented in the
orthography.

2.3 Word-elements
In this section we look at the elements that make up words. The elements we will consider
here are not purely phonological elements such as consonants, vowels, or syllables or their
orthographic equivalents, but elements which are capable of carrying meaning. We also con-
sider ways of forming words using processes which seem to be equivalent in function to the
use of the elements to be discussed here.
This is an area where there is a great deal of controversy (overt or covert) in the literature
on word-structure. The term ‘morpheme’ is one which has multiple definitions, many of
them incompatible, in writings about morphology. Our aim is not to produce yet another
definition or set of definitions, or to create a new way of considering morphological structure
in general. Rather we want to provide a terminology which can be used relatively transpar-
ently in discussing word structure, without leading to internal contradiction.

2.3.1 Morph
By a morph we understand any phonological (or graphemic) element of a word-form which
has function or meaning in the construction of a word. In (7) we illustrate some word-forms
analysed into morphs.

(7) a. snow-blind, sport-s-man, wind-mill

b. cat-s, hop-ing, lift-ed
c. pre-view, re-enter
d. Gandhi-ism, kill-er
e. dis-interest-ed-ness, un-contain-abil-ity

Clearly, there are different kinds of morph, and we will provide a more detailed classification
and nomenclature below.
Some morphs have the ability to stand alone as word-forms (whether or not they actually
do so in the word-forms in which we find them). In (7), for example, snow, cat, and contain
14 Basic principles: terminology

have this ability, while re‑, ‑ism, and ‑ic do not. A morph which has the ability to be a word-
form on its own is termed a potentially free morph (or free morph for short); a morph
that lacks this ability (except in cases of mention in linguistic discourse) is termed an obliga-
torily bound morph (or bound morph for short). When we mention obligatorily bound
morphs in isolation, we mark them with a hyphen which indicates where the rest of the
word-form would have to fall, as illustrated earlier in this paragraph.
Most of the potentially free morphs in (7) have lexical meaning, which we could look up
in a dictionary. With the obligatorily bound morphs, it may be clear in some instances that
there is a lexical meaning, while in others this is less clear. For instance, pre‑ has a meaning of
‘earlier in time’, but the ‘meaning’ of ‑ing is far less easy to express: rather it seems to have
some function (or set of functions) in creating word-forms of various lexemes.
Every word-form can be exhaustively analysed into a sequence of morphs. There are no
‘left over’ items in (7) which are not morphs. Note that, as elements of word-forms (or in the
case of potentially free morphs, potential word-forms), morphs are treated typographically
like word-forms, and written in italics when mentioned. Where appropriate, morphs may
also be presented in transcription.

2.3.2 Morpheme and allomorph

A morpheme is classically defined as the smallest meaningful unit of morphological analy-
sis; however, in order to be more precise and to make clear the relationship of the term
‘morpheme’ to that of ‘morph’, we need first to introduce a third term, ‘allomorph’. There
are many occasions on which morphs, though phonologically not identical, are function-
ally or semantically equivalent and are in complementary distribution. Consider a simple
example in (8).

(8) embark endanger

embed ensnare
embody entomb
emplane entrain

Bark, bed, body, plane, danger, snare, tomb, and train are potentially free morphs, and they are
preceded by an obligatorily bound morph which, in every case, has the same meaning, which
we might gloss roughly as ‘cause to be in’. In the examples in (8), sometimes this morph is
em‑, and sometimes it is en‑. The two variants are in complementary distribution, and we can
predict which of the two will occur in any given word-form: the version em‑ occurs before
bilabial consonants, and the version en‑ occurs before alveolar consonants (we avoid other
environments for ease of presentation). We say that these two morphs, em‑ and en‑ are allo-
morphs of the same ‘morpheme’. More generally, allomorphs are phonologically distinct
 2.3 Word-elements 15

variants that occur in complementary phonological environments. Groupings of allomorphs

may be termed morphemes.
The implication here (just as with the corresponding phonological terms of ‘allophone’
and ‘phoneme’) is that morphs are forms which occur in the flow of speech/writing, while
morphemes are abstract elements which derive from the linguist’s analysis. We speak of allo-
morphs where there are two or more morphs which belong to the same morpheme and
which are normally in complementary distribution. Each of the allomorphs is one possible
realization of the morpheme. Of course, in cases where there is only ever one morph to rep-
resent a particular morpheme, the same relationship holds, so that the morph dog is the only
morph to realize (or represent) the morpheme {dog}. Since allomorphs are morphs, they are
treated in the same way as morphs in terms of notation: they are italicized, sometimes
transcribed.
While the distinctions made above may seem relatively clear in the abstract, it has long
been noted that it is not always clear how to apply them in particular cases. There are a
number of recurrent cases where this causes disagreement. For example, we sometimes find
cases in which morphs are in complementary distribution, although they are not phonologi-
cally related to each other. The classic example of this is the allomorphy of plurals in English
(see Chapter 7). In regular cases, the allomorphs are /s/, /z/, and /ɪz/, where /ɪz/ occurs
after sibilants, /s/ occurs after other voiceless consonants, and /z/ occurs elsewhere, as illus-
trated in (9). This allomorphy is easily understood in phonological terms (assimilation and
epenthesis to break up illegal geminates), and is not controversial.

(9) horses /hɔːsɪz ‖ hɔːrsɪz/ judges /ʤʌʤɪz/

dogs /dɒɡz ‖ dɔːɡz/ flies /flaɪz/
cats /kæts/ strikes /straɪks/

However, what is controversial is the status of the ‑en that occurs—in standard varieties at
least—only with the lexeme ox. Is this another allomorph of the same morpheme, or is it an
allomorph of a synonymous morpheme altogether? As for our purposes nothing hinges on
the answer to this question, we leave this decision open.
Another difficult case is where there is form in common, but no meaning in common. The
classic example of this is English words borrowed from Latin, as illustrated in (10). We do not
need to know about the analyses of these words in Latin, but we do need to know whether
they are all analysable into two morphs in English.

(10) concur incur recur

deduce educe induce reduce
confer defer infer refer
commit demit emit remit
16 Basic principles: terminology

comport deport export import report

contend extend intend
convert evert invert revert

Those who argue for the one-morph solution claim that no element smaller than the entire
word here has a constant meaning (in English). That is, a recurrent meaning for the elements
in‑ and ‑fer that might be argued to occur in infer cannot be found except by an etymological
analysis. Those who argue for the two-morph analysis (e.g. specifically, Aronoff 1976) accept
this point, but claim that certain regularities of behaviour are lost here if these elements are
not recognized. For example, the ‑mit that occurs in these words always leads to a nominali-
zation in ‑mission. This particular argument goes to the heart of the definition of the mor-
pheme, which for most scholars demands a semantic unity. We will thus not accept elements
like ‑mit as morphs, but we will accept them as formatives, which we define as an overarch-
ing category that includes both morphemes and elements contributing to the construction of
words whose semantic unity or function is obscure or dubious.
A third vexed case is where the morphs have no associated meaning at all. These are some-
times termed empty morphs. An example is the <n> in the word platonic. This <n> has no
meaning; it is merely there to allow the morph Plato to be attached to the morph ‑ic. Items
like this can be analysed in a number of ways; it might be argued that the <n> is a matter of
base allomorphy (Platon‑ alongside Plato) or affix allomorphy (‑nic alongside ‑ic), or that it
is entirely separate from either base or affix. For the purposes of this volume where nothing
hinges on the specific analysis, we will take a rather agnostic stance here and refer to elements
of this sort simply as extenders. We allow extenders as morphs because we defined ‘morph’,
above, as having ‘a function or meaning’, and here it is the function which is paramount.
Another problematic case is where the contrastiveness of the morphemes is low. Just like
phonemes, morphemes are contrastive units. Part of the reason we recognize two morphemes
(and correspondingly, morphs) in smiles and smiled is that the two words contrast in meaning
(that is, in their temporal reference) and that this contrast can be located at the difference
between the <s> and the <d>, which must therefore belong to different morphemes.
However, there are places where the notion of contrast is less clear-cut, and where the iden-
tification of morphemes is more a matter of the preference of the analysing linguist.
As an example, consider the case of ‑er and ‑or (see Chapter 11). In most cases these ele-
ments are in complementary distribution (although the factors which determine the distribu-
tion may be complex), and are synonymous. However, there are sometimes said to be cases
like sailer and sailor (the first a ship, the second a person) where there appears to be a contrast.
Should such an example inevitably lead to setting up ‑er and ‑or as belonging to different mor-
phemes? Despite the contrast—the hallmark of the morpheme—most practitioners would
seem to believe that this is not sufficient, as it applies in a single isolated case and does not
occur in a larger group of words. Further, the contrast turns out to be spurious, as the OED
cites sailer as the older spelling, and the sailor spelling as a relatively recent development.
 2.3 Word-elements 17

Another related example comes from items like diner (‘location, e.g. place where one eats
on a train/particular kind of restaurant/person who dines’), driver (‘golf club/person who
drives’), sleeper (‘place on a train in which one can sleep/person who sleeps/brace laid to
support railway tracks’), smoker (‘place where one may smoke/person who smokes/device
which smokes food’). In this particular case, some might argue that we have homonymous
forms, but since they contrast in meaning, we might conclude that there are two (or more)
different morphs of the form ‑er belonging to two (or more) different morphemes. But most
practitioners resist any such conclusion, preferring to say that such cases involve polysemy
and that the semantics of the single morpheme is underspecified enough to cover all of these
examples (see Chapter 10).
The types of example illustrated above show why individual practitioners may disagree on
particular morphemic analyses without causing real damage to the notion of morpheme.
Some scholars, however, argue that the notion of morpheme is inherently flawed (see e.g.
Anderson 1992). They operate with the word-form as the basic unit, within which they iden-
tify morphs (defined rather less strictly than we have defined them here), but without the
link between morph and morpheme that has been drawn in this section, with the result that
they do not have a category of morpheme. While we accept that there are problematic
instances where the number of formal elements and the number of semantic elements does
not seem to match neatly (and a form like was in English might be just such a case, since the
meanings of ‘be’, ‘past tense’, and ‘singular’ are all wrapped up in what might be seen as a
single morph), we find the allomorphy relationship far too useful to discard out of hand, and
allomorphy implies morphemes.

2.3.3 Root, base, and affix

Three terms which are indispensible in analysing words are ‘root’, ‘base’, and ‘affix’,
although it is difficult to define them in a non-circular fashion; we therefore rely on exam-
ples to clarify these concepts. A root is the centre of a word, a lexically contentful morph,
either free or bound, which is not further analysable; it is what remains when all affixes
are removed. For example, in (11), the roots are contamin, member, warn, cut, align, smoke,
judge, pay, standard, impose, and nature. Affixes are obligatorily bound items which
attach to roots. In English there are two kinds of affix: prefixes and suffixes. Prefixes
are obligatorily bound morphs attached before (to the left of ) a root; suffixes are obliga-
torily bound morphs attached after (to the right of ) a root, as illustrated in (11) and (12).

Prefixes in English: de‑contaminate, dis‑member, fore‑warn, inter‑cut, mis‑align,

(11) 
non‑smoker, pre‑judge, re‑pay, sub‑standard, super‑impose, un‑natural

Suffixes in English: America‑n, fair‑ness, friend‑ship, girl‑hood, hospital‑ize, man‑ly,

(12) 
parent‑al, prefer‑s, pur‑ify, relate‑d, usher‑ette, watch‑er, wish‑ing
18 Basic principles: terminology

Because English frequently allows affixation to forms already containing affixes, we will
sometimes have reason to use the term base: a base is any morphological element to which
other elements are added in the creation of words. So a base may consist only of a root, as in
a word like friend‑ly, of more than one root, as in a compound like file cabinet, of a phrase like
old maidish, or of a root plus one or more affixes, such as when the suffix ‑ness attaches to
friend‑ly in friend‑li‑ness.
Although under normal circumstances affixes are obligatorily bound and roots are poten-
tially free, this is not always the case. Consider the words in (13):

(13) nomin‑ee, neuro‑jargon, robot‑icide, neuro‑logy

In all of these words, we have morphs that are bound. The morphs neuro‑, ‑(i)cide, and ‑(o)
logy do not occur independently, and they do recur in a number of words, as affixes do.
Similar to affixes, some of them typically occur in initial position (e.g. neuro‑) and others in
final position (e.g. ‑(i)cide). Still, most morphologists would hesitate to call them affixes. For
one thing, if we were to term them affixes, we would be left with the possibility of a word that
consists entirely of affixes (for example, neurology), which goes against the definition of affix
given above. For another, there are some items in this category which can occur either ini-
tially or finally (e.g. derm in dermatitis or endoderm), which is never the case for affixes in
English. We will therefore make a distinction between obligatorily bound roots (or
bound roots for short) and affixes. Obligatorily bound roots can serve as bases for affixes
or other obligatorily bound roots.
It now becomes necessary to clarify the difference between an obligatorily bound root and
an affix. The distinction is usually based on the type of semantic information that the morph
carries: bound roots are generally said to have more substantial lexical content than affixes
do, or as Bauer (1998b: 407) terms it, a higher degree of lexical density. The distinction makes
intuitive sense; the meaning of neuro‑ ‘pertaining to the nervous system or brain’ seems much
more substantial than the meaning of re‑ ‘again’. In practice, however, it is not always easy to
determine which morphs have sufficient lexical content to be considered bound roots, and
which fall below the threshhold. Most morphologists would agree that nomin‑, neuro‑, and
‑(i)cide are bound roots, and that in‑, pre‑, ‑ize, and ‑ness are affixes. But items like mini‑, mega‑,
and super‑ seem to fall on the line between affixes and bound roots. They convey notions of
size (and also evaluation) that are often in other languages conveyed by affixes. Inevitably the
amount of lexical content is a gradient matter, and different morphologists might draw the
dividing line between bound roots and affixes in a different place.
Bound roots of various sorts are very important in English morphology, perhaps particu-
larly so in neoclassical compounds like phytolith or neurogenic, and we will have occasion to
explore the properties of bound roots in many chapters of this work. We should therefore
make clear at the outset our assumption that while obligatorily bound roots are like free roots
in having lexical content, unlike free roots, they lack syntactic category (see also, for example,
 2.3 Word-elements 19

Giegerich 1999). Items like nomin‑ in nominate or nominee, or bapt‑ in baptize or baptism or
neuro‑ in neurology cannot be said to be nouns, adjectives, or verbs.
Another point we should clarify concerns cases like those in (14):

(14) a. She wore a mini. (viz. skirt/dress)

b. She drove a mini.
c. That’s super.
d. Isms and ologies (see e.g. Eskin 1995)
e. The concert was mega!
f. ‘Was it good?’ ‘Ish.’

In (14) we see a number of items that we might normally consider to be affixes apparently
being used in such a way that they are free—so clearly not obligatorily bound. We could, of
course, argue that items like ism have become lexemes (and for some of the items, like super,
which has a rather different meaning when it is used in isolation than when it is used as a
prefix in, for example, super-cooler, this would seem plausible). But on the whole speakers still
feel most of these elements to be prefixes being used as something else. In the terminology of
Hallidayan grammar (Halliday 1966: 111–18) they have undergone rankshift.
We will also make use of the term splinter on occasion. A splinter is a portion of a word
that is non-morphemic to begin with but has been split off and used recurrently on new bases
(free or bound). Formally it may be an affix plus some portion of its original base (e.g. ‑licious
as in bootielicious, ‑rific as in splatterific), part of an original compound (‑gate as in Monicagate,
‑scape as in deathscape), or merely the end of an originally mono-morphemic word (‑rama as
in cashorama). Semantically, splinters usually carry some of the semantic content of the origi-
nal word from which they split (‑licious from delicious, ‑scape from landscape, etc.), and there-
fore are more contentful than typical affixes (see Chapter 23 for a more detailed discussion of
splinters).
A term that we will not use in this work is stem, as it is used in at least two different ways
in the literature. The term is sometimes used in the sense that we have used obligatorily
bound root. But it is also used in the British tradition for the base to which an inflectional
affix can be added (see Section 2.5 on inflection). We try to avoid confusion here by eschew-
ing the term entirely, using instead the term obligatorily bound root as we have defined
it above, and when necessary referring to the base to which inflectional affixes attach as the
inflectional base.
Where we have need to refer to all words derived with the same affix or by the same mor-
phological process, we will use the term morphological category.
Finally we mention here two sorts of word-formation to which we will return in later chap-
ters. Both can be discussed in terms of affixes and bases but are not straightforward affixation.
20 Basic principles: terminology

Paradigmatic morphological processes are processes that are based on form‑meaning cor-
respondences between words, instead of the concatenation of formatives. For example, sets
of forms like baptism ◆ baptize ◆ baptist and agonize ◆ agonism ◆ agonist might be considered as
related in a paradigmatic fashion. In a word-based morphology all morphology might be
considered paradigmatic. In an approach that is agnostic as to the question of word-based
versus morpheme-based morphology, the term is mostly used to refer to processes which do
not lend themselves to a straightforward analysis in terms of a syntagmatic concatenation of
linguistic forms (see Chapter 23).
One sort of word-formation that might be considered paradigmatic is back-formation,
which is traditionally defined as the deletion of a suffix or prefix in analogy to pairs of base
and derivative that feature the affix in question (see Chapter 23 for discussion). A classic
example is the verb peddle, which is derived from peddler (originally spelled peddlar) on the
assumption the ‑er is the personal suffix occurring in writer, freighter, etc. Once back-formation
has occurred, it becomes invisible to speakers; linguistically naïve contemporary speakers
have no reason to think, for example, that peddle was derived from peddler, rather than the
other way around.

2.3.4 Vowel, consonant, and stress alternations

There are some instances in English morphology where changes which in most parts of the
system are marked with affixes are marked by changes to the phonological make-up of the
base. In such cases we find two phonologically related bases for the same lexeme. Such bases
may differ in vowels, in consonants, or in stress patterns. Where vowels are involved, this is
sometimes referred to in the literature as apophony; we will use the more transparent terms
vowel alternation, consonant alternation, and stress alternation (or stress
shift) here.
There are several areas in which we find vowel alternations. In one, singular and plural
nouns exhibit different vowels. The relevant pairs are shown in (15) (see also Chapter 7).
Historically known as umlaut, those changes were caused by assimilation of the vowel in
the root to a high vowel in a suffix. In every case, the suffix vowel has subsequently disap-
peared, so that the cause is not visible in current English.

(15) foot feet

goose geese
louse lice
man men
mouse mice
tooth teeth
woman women
 2.3 Word-elements 21

Vowel alternations also sometimes characterize the inflectional parts of irregular verbs, dis-
tinguish causative verbs from their non-causative counterparts, or link a few verbs with their
corresponding nouns. Historically such alternations have been known as ablaut. Some
examples of each type are provided in (16).

(16) a. vowel alternation in inflection

fall fell fallen
ride rode ridden
sing sang sung
b. vowel alternation in non-causative/causative pairs
fall fell
lie lay
rise raise
c. vowel alternation between verbs and corresponding nouns
abide abode
shoot shot
sing song

We also occasionally find consonant alternations distinguishing nouns and their correspond-
ing verbs or singular and plural in a few irregular nouns. Most, but not all, such cases involve
fricatives. Examples are presented in (17).

(17) a. noun/verb pairs

belief believe
defence defend
house (with final /s/) house (with final /z/)
portent portend
sheath sheathe
b. singular/plural pairs
house (with /s/) houses (with /z/) (but not always in ScE)
path (with /θ/) paths (with /ð/)
wife wives

There are innumerable cases of stress change brought about by morphological processes in
English, but the one that is of concern here is one that links words belonging to different
classes, as illustrated in (18) (for further details see Chapter 9).
22 Basic principles: terminology

(18) ábstractn, a abstráctv

íncreasen incréasev
ímportn impórtv
wálk‑outn walk óutv

2.3.5 Suppletion
In origin, the term suppletion is used for the situation where two (or more) lexical para-
digms are merged and forms from each of the lexemes are taken into a new paradigm (Boyé
2006): this is the situation we find with English go ◆ went. From there, the term has been
generalized to any case where stems in a paradigm are formally unrelated.
There are some cases in English which are uncontroversially viewed as suppletion, and
they are listed in (19). However, the definition given above is not sufficiently restrictive to
determine precisely where the boundaries of suppletion lie, what we must see as multiple
bases and what we must view as lack of morphological relatedness.

(19) bad/badly ◆ worse ◆ worst

be ◆ am ◆ is ◆ are ◆ was ◆ were
far ◆ further ◆ furthest
go ◆ went
good/well ◆ better ◆ best
little ◆ less ◆ least
much/many ◆ more ◆ most
one ◆ first
two ◆ second

It is quite clear that instances of allomorphy caused by regular phonological operations do

not count as suppletion. Thus alternations between the forms in persuade ◆ persuasive or
opaque ◆ opacify do not cause suppletion (see Chapter 9.3.1.5). It is less clear that vowel
apophony (whether umlaut or ablaut) does not: certainly sing and sung, mouse and mice are
usually perceived as different grammatical words in the paradigms of the relevant lexemes.
These are not normally considered suppletive, however. Even less are alternations such as do
◆ does /dʌz/, considered suppletive, despite the fact that they create unpredictably different
bases for morphological processes.
In (20) we present a series of potentially controversial instances (see also the discussion in
Dressler 1985). Each of the examples in (20) stands for an indefinitely large set of words in
English.
 2.3 Word-elements 23

(20) a. formally unrelated words

Indiana ◆ Hoosier ‘resident of Indiana’
Liverpool ◆ Liverpuddlian
dog ◆ canine
moon ◆ lunar
Manchester ◆ Mancunian
b. borrowing from different languages
hexagon ◆ sexennial
sheep ◆ mutton
c. competing affixes
intensify ◆ verbalize
d. alternative bases
Manchester ◆ Mancunian
mystify ◆ mystification
e. possibly suppletion
three ◆ third

The examples in (20) fall into distinguishable groups, some with borderline cases. There are
those where there is, by some fluke of the history of English, an unrelated word which fulfils
a function which, in some other instances, could be filled by a derivative (20a). We do not see
these as instances of suppletion at all, but as instances of vocabulary expansion by means
other than word-formation. These include the collateral adjectives like lunar and canine,
where the vocabulary expansion comes in the first instance through borrowing (see Koshiishi
2002). Liverpuddlian, originating as a joke about pools and puddles, is another instance of
this type. There are cases like (20b) where English happens to have borrowed synonymous
elements from different languages, but again there seems little reason to see these as supple-
tion rather than as prolific borrowing. As for (20c), even if the affixes ‑ify and ‑ize are synony-
mous and in complementary distribution (which is not entirely the case, though very nearly),
we view them as separate items rather than as unpredictable alternates in a paradigm. The
case of the intrusive ‑c‑ in mystification (20d), we treat as an extender (see Chapter 9.3.3 and
Chapter 10.2.1.3) rather than as an unpredictable base. This leaves only third (20e), and this is
the most suppletion-like example (especially since first and second were listed in (19)). The
reason it is perhaps less obvious than first and second is that historically it is derived from a
regular form, even if there are no other parallel formations today.
In short, we can (and will, in what follows) take quite a restrictive view of what involves
suppletion; others may take a less restrictive one, and include some of the types of relation
illustrated in (20).
24 Basic principles: terminology

2.3.6 Compounds
The definition of compound and the process of distinguishing compounds from phrases on
the one hand and derived words on the other is notoriously fraught with difficulty, and we
will discuss these issues in detail in Chapters 19 and 20. For our purposes here, we will pro-
vide a provisional and relatively uncontroversial definition of compound: we understand a
compound to be a lexeme that is composed of two or more bases, with the restriction that a
phrase may not occur in the righthand position.

2.3.7 Non-morphological word-elements

If by ‘morphology’ we mean ‘those bits of word structure which can be analysed into morphs
and morphemes’, there is some residue of elements that do not easily fit. Some of those are
considered here merely to draw attention to their existence and marginal status, as we will
have little or no occasion to refer to them in what follows.

2.3.7.1 Phonesthemes and sound symbolism

Phonesthemes are elements of a word which recur and where the sound of the element often
seems to reflect the meaning in some way, but which do not allow for an exhaustive analysis
of words into meaningful morphs (and morphemes) (Firth 1964: 184). Some examples are
provided in (21).

(21) a. glare, gleam, glimmer, glimpse, glint, glisten, glitter, glossy

b. bump, dump, hump, slump, thump
c. sleazy, slime, slippery, slosh, slovenly, sludge, sluggard, slum, sly

Although there may be some room for personal opinion as to which of the words belong
in particular lists, on the whole the words in (21a) have something to do with light (per-
haps particularly reflected light and small amounts of light); those in (21b) have something
to do with a dull sound; those in (21c) have to do with something unpleasant, especially
unpleasant to the touch. The effect is felt to be connected with the recurrent phonological
elements, /ɡl/, /ʌmp/, and /sl/. Note, however, that the remnant parts of the word are not
necessarily recurrent (‑int, ‑isten, ‑itter, and ‑ossy are not recurrent word-forming elements
with a constant meaning), and so an analysis of these elements as morphs would imply the
establishment of a large number of morphs which occur in only one word. Although there
is some evidence that phonesthemes show similarity to morphemes in terms of mental
processing (Bergen 2004), they will not be treated as a morphological phenomenon here.
As we will discuss in Chapter 8, we will take a similar position on items like th‑ence and
wh‑ence or h‑ither and th‑ither.
 2.3 Word-elements 25

Beyond phonesthemes there are some claims for sound symbolism in the English vocabu-
lary. Most of these claims are concerned with so-called onomatopoeic words, words that are
supposed to reflect directly the sounds they represent or whose production they represent,
or the sounds made by the entities they denote. Examples are given in (22).

(22) bang, bark, bow-wow, bump, coo, crash, cuckoo, groan, gurgle, hiss, miaow, murmur,
neigh, splash, squeak, squeal, twitter, whicker, whoosh

The degree of match with reality that is implied by such words is filtered through language
history and cultural expectations: other languages’ onomatopoeic words for the same sounds
or entities would not necessarily sound anything like these. Even if these sounds are taken to
represent real-world sound in some way, the words themselves are generally monomorphe-
mic. There are no elements in them whose function is to represent a particular sound or
sound-type.
The other main place where sound symbolism is sometimes claimed is with words denot-
ing size, including diminutives. It is sometimes claimed that words denoting smallness or
closeness are produced with a narrow passage between the front of the tongue and the hard
palate (so high front vowels and palatal consonants), while words denoting largeness or dis-
tance are produced with the mouth wider open (low vowels, possibly velar consonants). Here
has a higher vowel than there, and little has a higher vowel than large, illustrating the way in
which these correlations are supposed to work. Of course, distant and close, small and big have
the vowels precisely the wrong way round. In order to determine how far this supposed cor-
relation fits, some kind of statistical survey of all relevant English words would have to be
undertaken, although to our knowledge no such survey has been made. We note, though,
that diminutive suffixes in English (see further Chapter 18) such as ‑ling and ‑y do fit with the
general predictions to some extent.

2.3.7.2 Subtractive and prosodic morphology

English has a number of ways of forming new words that involve either the subtraction of
elements or the manipulation of the prosodic structure of words. Some of these processes are
more systematic than others, and it may be disputed whether they deserve to be treated as
morphological, as these ways of making new words do not readily allow for analysis into
morphs or morphemes and in some cases (initialisms and acronyms, not treated in this work)
are based on orthography rather than on morphology.
One type of word-formation that is based on the prosodic structure of words is clipping,
that is, the shortening of a word by deletion of some of its phonological material. Some sim-
ple examples are given in (23).

(23) a. brill ◆ brilliant, deli ◆ delicatessen, jumbo ◆ jumbo jet, lab ◆ laboratory, photo ◆
photograph
26 Basic principles: terminology

b. jamas ◆ pyjamas (BrE), phone ◆ telephone, blog ◆ weblog

c. tec ◆ detective (BrE, archaic), shrink ◆ headshrinker, flu ◆ influenza

As has been shown in Lappe (2007), there are several relatively systematic patterns that
determine the phonological shape of the output, with the resulting word conforming to par-
ticular requirements on syllable structure, stress, and its relation to the base word. Clippings
will be discussed in Chapter 18.
Another type of prosodic morphology is the process of expletive insertion (McMillan
1980). This looks like a form of infixation, except that what is inserted into the middle of a
word is not an affix, but an entire word. This process is illustrated in (24) and discussed in
more detail in Chapter 18.

(24) abso-bloody-lutely, guaran-frigging-tee, kanga-bloody-roo, propa-fucking-ganda

2.4 Word classes

Word classes, also traditionally called syntactic categories, form-classes, or parts of speech,
are classifications of the words of a language in accordance with their grammatical behaviour.
Because school-grammar inherited a set of word classes from classical grammar, it sometimes
comes as a surprise to the uninitiated to realize that these classes are creations of linguists,
and not part of nature. Accordingly, the precise criteria used to delimit a particular class and
the number of classes needed to classify the words found in any given language, are open
questions, and there may be many solutions to the problem. Having said that, we will mainly
be concerned with just four word classes here: nouns, verbs, adjectives, and adverbs. On the
whole these are the least controversial classes, and they are the most critical in a description
of the morphology of English. This is not to say that other word classes such as prepositions
and demonstratives never participate in processes of word formation in English, but merely
that this participation is infrequent and far more marginal than the inflection and derivation
of nouns, verbs, adjectives, and adverbs.

2.4.1 Defining word classes

Typically, word classes are defined in terms of a number of intersecting criteria, including
phonological, morphological, syntactic, and semantic criteria. As there is substantial discus-
sion of the criteria for determining word classes in both the theoretical and descriptive litera-
ture (Quirk et al. 1985; Huddleston and Pullum 2002), we will assume familiarity with the
basic criteria and simply refer the reader to the relevant sources (for example, Baker 2003 or
 2.4 Word classes 27

Croft 1991 for theoretical treatments and Quirk et al. 1985, and Huddleston and Pullum 2002
for descriptive treatments).
One issue that we should mention at least briefly, though, is whether we need to
assume the existence of distinct syntactic categories at all, given examples like those in
(25) and (26):

(25) a. The bottle is empty.

b. We should empty the bottles down the sink.
c. What should I do with the empty?

(26) a. His clothes were dripping by the time the rain stopped.
b. Her dripping nose was a real embarrassment to her.
c. They ate dripping on stale bread.

We will not provide the details here, but will discuss such examples in Chapter 25, as a
matter of conversion, that is, a change from one word class to another with no concomi-
tant change in form.

2.4.2 Lexical versus grammatical, open versus closed classes

The word classes noun, adjective, verb, and adverb are open word classes; that is they can
easily receive new members by the application of morphological processes. Ayto (1990), for
example, a dictionary of newly registered words from the last years of the 1980s, lists many
new adjectives, nouns, and verbs (e.g. aspirational, biodiversity, anonymize), a few adverbs
(e.g. futuristically) but no other types. Classes like prepositions, conjunctions, modals, deter-
miners, pronouns, and so on are so-called closed classes: any changes to them are rare, fre-
quently involving a change to a whole paradigm, and such changes do not usually arise
through the use of morphological processes. These closed classes provide grammatical infor-
mation, and are therefore termed ‘grammatical words’ (see Section 2.2.2.), while adjectives,
nouns, verbs, and adverbs are termed ‘lexical words’.

2.4.3 Transposition
The term transposition has been used in two ways in the morphological literature. One
use designates as transposition morphological processes that cause a change of word class
without concomitant semantic change other than what results from changing category. For
example, parent is a noun, but parental is an adjective that just means ‘pertaining to parents’;
amaze is a verb, but amazement is a noun that denotes the ‘process or result of amazing’. The
28 Basic principles: terminology

other use of the term designates a change such as the one illustrated in employ and employer
which involves both a category change from verb to noun, and also a specific semantic effect,
which we can loosely gloss as ‘one who verbs (someone)’. Because of the ambiguity in the
use of the term, we will avoid speaking of transposition, and where relevant, talk of category-
change and accompanying semantic effects.

2.4.4 Headedness
There is a tradition in morphological studies (Marchand 1969; Williams 1981; Lieber 1992) of
seeing the morpheme in a word which determines the word class of the word as the head
of the word. This is an extension of the syntactic notion of head, according to which the head
of the noun phrase big books is the noun books, the head of the prepositional phrase in the
woods is the preposition in, the head of the adjectival phrase extremely poor is the adjective
poor, and so on. The head of the phrase is seen as characterizing the phrase in terms of word
class and other grammatical and semantic properties. The difference between the syntactic
notion of headedness and the morphological notion in this tradition is that the set of criteria
used for determining headedness tends to be narrower in morphology (see Bauer 1990 for a
review, and Matthews 2007 for a wider critique of the way in which the notion of head has
been used). Moreover, it is not always clear how the notion of head is to be applied in mor-
phology, especially within inflectional morphology, a subject to which we now turn.

2.5 Inflection versus derivation

Appeal has already been made to the distinction between inflection and derivation: inflec-
tion creates word-forms from known lexemes, derivation creates new lexemes from other
lexemes. Thus, for example, the creation of a form employs from the lexeme employ would
almost universally be accepted as inflection, and the creation of a lexeme amazement from
amaze would almost universally be accepted as derivation. There are, however, other
instances which are far less clear-cut: both the creation of caterpillars from caterpillar and the
creation of usefully from useful have, by different authorities, been assigned to both categories
(see e.g. Beard 1982; Haspelmath 1996). Thus the notions of inflection and derivation must
be given some independent content if a suitable definition is to be provided.
This has been done in the recent literature by viewing inflection and derivation as
canonical categories, and by providing a number of criteria which, in the canonical cases,
distinguish the two, or tests which can be used to determine which kind of morphology
is being used in individual cases. There is a great deal of discussion about the distinction
in the literature (see for example, Scalise 1988; Dressler 1989; Anderson 1992; Plank 1994),
but here we will simply list some criteria from Plank (1994), referring the reader to
Chapter 24 for further discussion.
 2.6 Lexicalization and productivity 29

• Inflection is typically formally regular; derivation may not be.

• Inflection is typically semantically regular; derivation may not be.
• Inflection is typically fully productive; derivation typically shows (unmotivated) gaps
in productivity.
• Inflection does not add significantly to the meaning of the base, but contextualizes that
meaning; derivation adds to the meaning of the base.
• Inflection allows prediction that a form will exist to cover some notion; derivation does
not allow the same security of prediction.
• Inflection does not change the major category (noun, verb, etc.), but derivation may.
• Inflection is what is relevant to the syntax; derivation is not syntactically relevant.
• You cannot replace an inflected form with an uninflected form in a sentence, but you
can replace an inflectional base (which may have been created by derivational morphol-
ogy) with a morphologically simple form (that is, one belonging to a different
lexeme).
• Where both occur in the same word-form, derivational affixes are typically closer to the
root than inflectional ones.

Plank (1994) argues that criteria like these provide a clear-cut distinction between inflection
and derivation in English, and this is true for many affixes. But it is not true for all (see Beard
1982 on plurals; Haspelmath 1996 on ‑ly adverbs). In Chapter 24, we will return to the issue
of how to distinguish inflection from derivation in English. In the meantime, however, we
must make some sort of decision as to what to treat under inflection and what to treat under
derivation in the descriptive sections of this work. For presentational purposes, we here
stipulate that the following processes of English morphology are inflectional: plural and pos-
sessive marking on nouns, third-person singular marking on present-tense verbs, past tense
and past participle marking on verbs, present participle marking on verbs, comparative and
superlative marking on adjectives and adverbs. All other affixal processes in English are
derivational.

2.6 Lexicalization and productivity

2.6.1 The historical development of individual words
If we trace a word through its own individual history, we find a number of distinguishable
phases. Consider the word incentivize, listed by Ayto (1990) as originating in 1989. If we
accept that date seriously (though it may only represent the first attested date of the word in
print, not the first use of it), we can say that in 1988 there was no such word. However, there
was a pattern available to speakers of English in 1988, based on words like collectivize, compu-
terize, editorialize, fossilize, revolutionize, and dozens like them, where ‑ize is added to a noun
30 Basic principles: terminology

with any one of a relatively small number of meanings (see Plag 1999; Lieber 2004; and
Chapter 13 on the semantics of ‑ize). In 1988, therefore, incentivize was a possible or poten-
tial word, one which the system had the ability to create, but for which there was as yet no
perceived need.
We can rephrase that by saying that in 1988, incentivize was type-familiar to speakers of
English, but not item-familiar. Speakers were aware of the general pattern (this is what we
mean by the term type-familiar), but did not know this particular word formed on the
pattern. Not until the potentiality was taken up in 1989 did the word have the possibility of
becoming item-familiar (Meys 1975). In principle, the moment that incentivize was coined,
it moved from being a potential word to being an actual (or existing) word. Of course, the
moment of coining (or the written or spoken record of it) is rarely observable, and the transi-
tion between a potential word and an actual word is impossible to capture precisely. In many
cases, the same form may be coined several times before it becomes part of the usage of a
community, at which time we can say that the word has become institutionalized. For
that reason, the notion of actual words is notoriously fraught with difficulty.
Lexicographers like to distinguish between nonce words (new words which are used
but do not become institutionalized or widely established in community use) and neolo-
gisms which eventually become part of the community’s norm. From a morphological point
of view it is not clear that the two processes are distinct. Both deal with the creation of a new
word using the available resources of the language of the community; the difference is merely
a matter of whether speakers pick up the new word, something which is a sociological phe-
nomenon as much as a linguistic one. We will therefore confine ourselves to the term ‘neolo-
gism’ in this work, and avoid the use of the term ‘nonce’.
The moment a word is used any vagueness that there may be in its linguistic structure is
removed. For example, zinger (Ayto 1990) could in the abstract be a person, an animal, or an
object, and only after its first use does it become fixed as meaning ‘a riposte’. This is the start
of idiomatization. Typically it occurs with the passage of time. Idiomatized words do not
have a meaning that is compositionally retrievable from the elements (the morphemes)
which make up the word. For instance, the word driver, when it denotes a human, not only
denotes a person who makes a thing move forward, but denotes someone who drives a motor
vehicle on land. It is not clear how much of this information should be considered part of the
linguistic structure of the words concerned, and how much should be said to be part of usage;
it is clear, though, that extra information can be added to the perceived meaning of the word
that is not reflected in its morphological structure. Sometimes this extra information becomes
more important than the morphological information. The transmission in a car may indeed
transmit something from one part of the vehicle to another, but we think of it as a piece of
machinery rather than in terms of its transmitting function. We should also note that although
idiomatization typically occurs with the passage of time, it is nevertheless possible for words
to be coined with meanings that are idiomatic from their inception; for example, according
to the OED, the verb cannibalize was attested from the very beginning with the meaning ‘to
 2.6 Lexicalization and productivity 31

take parts from one machine to use in another’. It has never had the compositional meaning
‘to act like a cannibal’.
It is also the case that a word can lose its connections with its base because of changes in
the language or changes in society. To solicit can be a crime (‘to offer one’s services as a pros-
titute’), but a solicitor in the British legal system is a person of standing, who would not be
expected to be involved in soliciting of this kind. Solicit has gained new meanings since solici-
tor was coined. On the other hand, the world has changed so that fighter can denote a plane
as well as a person.
Words whose origins remain transparent change their phonological form in line with the
changing pronunciation of their elements. As the pronunciation of child has changed, so the
pronunciation of childhood has kept up. But where the semantic link is lost, a word may
change its form independent of the form of its elements. The history of English is full of such
cases: consider lord from loaf ward, hussy from housewife, health from heal (and ultimately
from whole) and forehead, which was once /fɒrɪd/ and changed its pronunciation as a conse-
quence of the spelling. Once phonological changes have occurred which divorce a morpho-
logically complex word from its base (and/or its affixes), a word loses its status as a
multi-morphemic unit, and becomes monomorphemic.
If the loss of motivation affects all the words formed with a particular affix or word-­
formation process, that process may be lost. At one stage in the history of English, there must
have been a process of vowel change that linked verbs to their causatives. We have only a few
such examples left (for example, fall ◆ fell, lie ◆ lay). Speakers today no longer perceive the
derivational relationship here, and the process is, to all intents and purposes, dead.
Similarly, there used to be a set of verbs beginning with the form to‑, just as we still have a
set of verbs beginning with with‑ (withdraw, withhold, withstand). We can no longer see any
constant meaning associated with the with‑ in these formations, but the to‑formations (some
listed in (27)) have vanished completely.

(27) to-hang ‘append’

to-hear ‘listen to’
to-lay ‘put forward’
to-set ‘affix’
to-stand ‘assist’

The vanishing of the individual words here has meant that the entire pattern is lost: we can
no longer make new words according to this pattern at all.
Having said that, it is often difficult to say whether a particular pattern has vanished or not.
The process which gives breadth, depth, length, warmth and others is often cited as an example
of a dead process, and it probably is. But coolth keeps getting reinvented (it has been around
since the sixteenth century). Wentworth (1941) is a lengthy illustration of the fact that the
32 Basic principles: terminology

suffix ‑dom, whose demise had been proclaimed repeatedly in the early twentieth century,
was (and still is) alive and well (see Chapter 12).

2.6.2 Availability
Sometimes the history of a word is a description of the process of lexicalization. Words
may move from being morphological complexes (sometimes, syntactic complexes),
through reducing levels of transparency, affected by both semantic and phonological fea-
tures, to the point of being unanalysable and monomorphemic or until they vanish com-
pletely. Processes which can still be exploited in the speech community for the creation of
new words are said to be available (or, in less specific terms, productive). The ablaut
giving rise to causative verbs and verb-formation with to‑ are no longer available to speak-
ers of current English. We cannot create a verb say meaning ‘to cause to sigh’ or a verb to-sit
meaning, perhaps, ‘to visit the sick’ and expect to be understood in the current state of the
language. On the other hand, incentivize is a relatively recent word which could be coined
and understood in the language of the late 1980s. The process of ‑ize suffixation is available
to speakers.
It is not always clear whether something is available or not. Bauer (2001: 9) suggests that
‑ment is no longer available in modern English, but we later (see Chapter 10) find some
evidence that it is still used occasionally. Nevertheless, in principle the question of whether
something is available is a yes/no decision. In general terms, when we speak of a process
being ‘unproductive’ or of ‘marginal productivity’ we mean that as far as we are able to
ascertain, the process cannot be freely used by members of the speech community of
English.

2.6.3 Profitability
Processes which are available are not all available to the same degree. Although ‑ness and ‑ity
are both available as suffixes in current English, we find that ‑ness is used to form neologisms
more often than ‑ity is. The degree to which something can be used to create new words is
termed the profitability of that process. A great deal of the variation in profitability can be
attributed to the number of domains in which a particular process can be used, but some of
it is a matter of competition between processes, some of it is a matter of the meaning associ-
ated with a particular process and the corresponding value of the process to speakers, and
some of it appears to be residual randomness.
By domain, we understand a linguistically defined set of bases that promote or permit
the use of a particular morphological process. Another way of looking at the same idea is
to say that there are constraints on when a particular process can be used. Consider some
examples:
 2.6 Lexicalization and productivity 33

The ‑ly suffix which creates adjectives like friendly, stately, weekly, womanly is, when it is
used to coin new words, used most notably on bases which denote humans (perhaps rational
beings). Examples from the British National Corpus (BNC) are given in (28).

(28) actorly, auntly, dancerly, designerly, headmasterly, readerly, ruffianly, spinsterly, writerly

Thus the domain (possibly one of the domains) of this suffix is bases which are [+ human].
The suffix ‑ity, despite a very few minor exceptions like oddity, is added to words which are
of non-native origin (see below, Section 2.7 and also Chapter 27). Typical examples include
those in (29):

(29) acidity, activity, chastity, circularity, fertility, feudality, mediocrity, obesity, orality,

scarcity, scrupulosity

One of the ways in which the non-native character of the word in the base can be guaranteed
is by having a non-native suffix. One of the suffixes which most obviously takes native words
and turns them into suitable bases for ‑ity suffixation is ‑able. So one of the domains for ‑ity
suffixation is words with the final suffix ‑able. Examples from Corpus of Contemporary
American English (COCA) are given in (30).

(30) absorbability, actability, ad‑lib‑ability, ageability, answerability, bendability, brushability,

climbability, cryability, fakeability, foldability, garageability, guessability

The nominal ‑al which derives nouns from verbs applies only to verbs which have stress on
the final syllable (some examples from BNC are given in (31)). The sole apparent exceptions
are the word burial (which historically was not derived with this suffix, though the etymo-
logical difference is no longer visible) and ‑al nouns which have subsequent prefixation which
changes the stress, for example mistrial. Note that this does not mean that all verbs which
have final stress take ‑al; it only means that to take ‑al, there must be final stress, or, to phrase
it differently, ‑al derivatives are stressed on the penult.

(31) accrual, acquittal, appraisal, approval, arousal, arrival, betrayal, betrothal, carousal,

committal, conferral, denial, dispersal, disposal

Finally, profitability is affected by competition between processes. Two processes compete

when they both have the potential to be used in the coining of new synonymous forms from
the same base. Usually, competition disappears over time, and processes become semanti-
cally or formally specialized. Thus, for instance, Plag (1999) argues that ‑ify and ‑ize are largely
in complementary distribution, depending on the form of the base, and thus compete only
with regard to a very small number of possible bases.
34 Basic principles: terminology

2.6.4 Blocking
Another factor which may limit the profitability of particular processes is blocking (Aronoff
1976). Blocking may be defined as the failure of a particular word to become established in a
community due to the fact that a word with the same base and the same meaning is already
established in that community. The classic example is the lack of the word gloriosity, ostensi-
bly blocked because there is an already established word glory. The basic argument is that the
existence of a word already bearing a meaning precludes the coining of another word with
the same meaning. We discuss the notion of blocking in detail in Chapter 26.

2.6.5 Notions of regularity and irregularity

The notion of productivity is tied in with the notion of regularity, but not in any direct man-
ner. Processes which are productive are often regular, but processes that are regular need not
be productive. In practice, the notion of regularity may mean any one of a number of things.
It may mean that something is rule-governed. The notion of rule-governedness itself is not
entirely obvious, but refers to a pattern which is frequent enough and predictable enough
that it seems feasible to give a set of statements which reflect the way in which words are
coined using a particular process. Rule-governedness would involve stating domains, for
example. At the other end of the scale, there would be cases that are clearly not rule-governed:
this is certainly the case for the plural form oxen, since this precise affixation process without
accompanying vowel change applies only to this word. This case must involve lexical listing,
rather than formation by rule, which would imply availability.
Regularity may also mean that something is frequent. However, if regularity is frequency,
it is not necessarily correlated with productivity. Something may be frequent in texts without
being productive at all. The example of the suffix ‑ment, cited above, is relevant here.
Conversely, rare morphological processes may be productive: it is difficult for step‑ to be
added to much, because in general it is added to names of close relatives, and there are not
many such potential bases. However, it is extended to words like step-son-in-law and step-
relations (BNC), which suggest a certain degree of profitability and availability given appro-
priate circumstances.
Regularity may also mean that a formation is morphotactically and semantically transpar-
ent. Highly profitable processes are always semantically transparent (though not necessarily
unambiguous), but the apparent continued availability of some vowel change processes for
irregular past tenses suggests that availability is not always the same as transparency.
Accordingly, it is preferable to be more specific than calling any process ‘regular’, though
we will talk of semantic regularity, meaning semantic constancy associated with a particular
process.
It should also be noted that because it is possible for there to be processes which are avail-
able but which are not equally profitable, the output sets (and correspondingly, the frequency
 2.7 Native and non-native 35

of the formation types in text) may differ considerably. It does not follow that only one (or
some other small number) of these sets is ‘regular’. Regularities may be statable over small
numbers of words as well as over large numbers.

2.7 Native and non-native

English is a language that has had a long history of contact with and borrowing from other
languages, such that both its vocabulary and the processes by which new words are formed
can be classed as native or non-native (although not always perfectly neatly). Other
terms used for the distinction are Germanic, for native, and Latinate or learned for
non-native. To understand the distinction it is necessary to know at least a bit of the his-
tory of English.
The English language is derived from a West Germanic language originally spoken along
the North Sea coast in modern Friesland. Various dialectally varied versions of this language
were brought to England with the Anglo-Saxon invaders, who overran the Celts (of whom
very few direct traces remain in the English vocabulary). With the Viking invasions, another
layer of Germanic, but North Germanic, was added to the West Germanic fundaments.
Words and word-formation processes from these Germanic origins are the ones that we call
native.
In 1066, the Norman invasion brought French to England. French was mostly used by
those with some kind of authority, not in everyday interaction in the villages, so that English
remained remarkably free of French influence on the grammatical system. However, a large
number of French words were added to the English vocabulary. With the Renaissance, in
order to improve English and make it a language fit for modern usage, a huge number of
Latin and Greek words were added to the Germanic and French ones. A large number of
pairs and triplets still exist in modern English, distinguished by their etymological origin.
Some of these are illustrated in Table 2.1.
Eventually, sufficient words containing particular affixes were borrowed so that a pattern
could be perceived in English, and new words were coined in English containing these non-
native affixes. At this point, those affixes become part of the English language. These patterns
are the ones that we refer to as non-native.
When undertaking an analysis of the non-native stratum of English word-formation, care
is always required. Some words have been borrowed in their entirety, for example, countess,
hostess, princess, and prioress; others such as adulteress, ambassadress, authoress, heiress, and
priestess are genuine English formations, in the sense that the affix is attached to a base on
which it could not be found in the source language. The latter cases are instances of English
word-formation; the former set are instances where a particular suffix can now be analysed,
but which were not instances of English word-formation. Of course, as far as native speakers
of the language are concerned, the two sets of examples cannot be distinguished.
36 Basic principles: terminology

Table 2.1 Native and non-native triplets

Germanic French Latin/Greek
abide endure tolerate
adder, snake serpent viper
ask enquire interrogate
blue azure cerulean
bold impudent audacious
breach infringement violation
choice preference predilection
dear precious valuable
dwell reside inhabit
foretell prophesy predict
hallowed sacred consecrated
hire payment remuneration
lithe pliant flexible

From Earle (1890)

It is clear from Table 2.1 that not only were new roots introduced, but also words that were
complex in their language of origin.

2.8 Lexical syntactico-semantic

terminology
A major goal of this work is to describe the semantic range and scope of the word-formation
processes of English. It is therefore necessary to have a vocabulary with which to discuss the
lexical semantic properties of these processes. Some terms that we make use of should be
familiar and will be used in the expected way. Among these are terms like: ‘count’ and ‘mass’;
‘gradable’ and ‘non-gradable’; ‘event’ and ‘state’; ‘agent’, ‘patient’, ‘theme’, ‘location’, ‘source’,
‘goal’, ‘instrument’, and ‘beneficiary’. What is especially problematic, however, is that there is
no established terminology for describing the kinds of quasi-syntactic relations that we find
in some types of derivation and compounding. By this we mean types of derivation or com-
pounding where the interpretation of a word or part of a word can be likened to that of, for
example, the subject or object of a sentence. We will therefore try to be especially clear about
the terminology that we will employ.

2.8.1 Argument, adjunct, and verbal diathesis

Some of the terminology we will make use of is familiar from syntactic theory. The terms
argument and adjunct are used in syntax to refer to phrases that stand in specific relations
to the verb. Arguments are phrases that are semantically required by the verb. Intransitive
 2.8 Lexical syntactico-semantic terminology 37

verbs are verbs that require one argument, typically the subject (although see below).
Transitive verbs require two arguments, subject plus either object or prepositional object.
Ditransitive verbs require two arguments in addition to the subject:

(32) Intransitive verbs: yawn, sneeze, arrive, fall

The chiropractor yawned.
The guests have arrived.

Transitive verbs:
(33)
a. With object: eat, cover, destroy, catch
Our dog eats insects.
The children destroyed the sandcastle.
b. With prepositional object: suffer, adhere
My mother-in-law suffers from insomnia.
We will adhere to your rules.

(34) Ditransitive verbs: show, put, give, tell

We showed the painting to the expert.
Someone put grease on the doorknob.

Note that although arguments are semantically necessary to the meaning of a verb, they may
not always be obligatory syntactically; a verb like eat can have an implicit object, as we find,
for example, in a sentence like I am eating. In referring to the range of arguments required by
a given verb, we use the term verbal diathesis.
One distinction that we will occasionally make reference to in this work is the distinction
between two types of intransitive verbs, those that are termed ‘unaccusative’, and those that
are termed ‘unergative’ (the terminology is admittedly awkward, but it is sufficiently standard
at this point that we will not try to improve on it). Unaccusative verbs are those intransi-
tive verbs whose single argument is interpreted as a theme or patient (for example, arrive or
fall). Unergative verbs are those intransitive verbs whose single argument is agentive, even
if involuntarily so (for example, yawn or sneeze). The former tend in English to be verbs of
change of state or of directed motion, the latter verbs that denote movements or actions of
the body, although there are some unaccusative and unergative verbs that do not belong to
either of these categories. Typical diagnostics are available in the syntactic literature for dis-
tinguishing unergative and unaccusative verbs; for example, unergatives, but not unaccusa-
tives, can be used in the so-called ‘reflexive plus result’ construction but not unaccusatives
(e.g. I yawned/*arrived myself tired).
Adjuncts are phrases that are not semantically necessary to the meaning of a verb. For
example, in the sentence in (35), the phrases in brackets express notions of location, time,
and beneficiary that add optional information:
38 Basic principles: terminology

(35) Matilda played the sonata [in the drawing room] [last night] [for the countess].

So whereas the verb play requires only a subject and an object, it may optionally occur with
phrases that express other semantic relations. The verb put, on the other hand, requires a
subject, object, and location as part of its diathesis.
While the distinction between argument and adjunct is relatively clear in theory, it is noto-
riously difficult to deploy in practice. For English there are a number of standard tests that
can be used to distinguish arguments from adjuncts (see, for example, Carnie 2006)—for
example, an adverb can be inserted before an adjunct but not before an argument; the order
of adjuncts can be changed, but not the order of arguments—but these tests sometimes give
conflicting results. We will not attempt to solve this problem here, but will try to stick to
cases where the distinction between argument and adjunct is relatively clear-cut.
Another issue with regard to arguments is the extent to which simplex items other than
verbs can be said to allow (or require) arguments. That some adjectives (e.g. fond) take argu-
ments is relatively uncontroversial. The ability of simple nouns to take arguments is some-
what more controversial. Here we will assume, following Löbner (1985), that nouns can
indeed be said to have arguments. Nouns like leg, mother, or author will be termed rela-
tional nouns here. These are nouns whose semantics presuppose another ‘participant’; a leg
must be a leg of something, a mother must be mother of someone, and so on. Nouns that do
not presuppose other phrases (e.g. book, wisdom) in this way can be termed sortal.

2.8.2 Referencing
It is very often the case in word-formation that a derived word that is based on a verb or
relational noun refers to one of the arguments of that verb or noun. For example, forms in
English derived with the suffix ‑er generally denote participants in an event that would be
expressed by the subject of a sentence, for example, agent (writer) or instrument (copier);
that is, a writer is someone who writes, a copier something that copies. Similarly, in a com-
pound like city employee, the first element of the compound denotes a participant that would
be expressed by the subject in a sentence, the second a participant that would be expressed
by an object; that is, the interpretation is that the city employs someone. We will have occa-
sion in Chapter 11 to go into the complex polysemy exhibited by personal affixes in very great
detail. We just point out briefly here that it would be inaccurate to refer to ‑er nouns as agent
and ‑ee nouns as patient nouns, as each of these affixes can express a range of thematic rela-
tions compatible with the functions of subject and object in a sentence. We could, of course,
call them ‘subjective’ and ‘objective’ affixes instead. However, the terms ‘subjective’ and
‘objective’ are already multiply polysemous and to use them in this specialized way is likely to
lead to unnecessary confusion.
We therefore propose to use the terms subject-referencing and object-referenc-
ing instead. When derived words or parts of compounds denote a participant that would
 2.9 Last word 39

normally be expressed as the subject of a sentence, we will call those words subject-­
referencing. Similarly, when derived words denote a participant that would normally be
expressed by the object or prepositional object of a sentence, we will call them object-
referencing and prepositional-object-referencing. A word like driver is subject-
referencing. The word truck in the compound truck driver is object-referencing, although
the compound as a whole is subject-referencing, as it is a hyponym of its head.

2.9 Last word

We have tried in this chapter to clarify the main terms that will recur throughout this work.
More specialized terms will inevitably be introduced in some chapters, and we will endeav-
our to define and exemplify them as we go along.
 chapter 3

Basic principles: methods

3.1 Prospectus
A central aim of this reference volume is to present as broadly as possible the contemporary
state of English morphology, comprising both inflection and word-formation. In the past,
such wide coverage of contemporary data was difficult if not impossible to achieve. Marchand’s
monumental Categories and Types of Present-Day English Word-Formation (Marchand 1969)
was mainly historical; his chief source of data was the Oxford English Dictionary (OED). Of
course, as Marchand’s intention was to trace the history of various word-formation processes,
the enormous richness of the OED was quite suitable for his aims. Our intent is not to revisit
Marchand’s territory, however, but rather to look closely at the current state of English word-
formation. In this chapter we discuss how we have generated the data on which this book is
based, the methods we have used in analysing it, and the conventions we have used for pre-
senting it.

3.2 Sources of data

3.2.1 The corpora
We take up this project at an auspicious time, as sources of data and the methods of analysis
have changed markedly in recent decades, allowing us to see the scope of English morpho­
logy in a way that would not have been possible just a decade ago. As in many sub-disciplines
of linguistics, the advent of digital corpora and lexical databases has had an enormous impact
on the study of morphology. The significant advances in both empirical and theoretical
approaches to morphology over the past decade are largely due to these methodological
tools. With the aid of corpus data, it is now possible to study English word-formation virtu-
ally in real time. Our most important electronic sources are listed below.

• The Corpus of Contemporary American English COCA (Davies 2008). This corpus, first
released in 2008, is the largest freely-available corpus of English, with 425 million words
at the time of writing. It is a balanced corpus with texts equally distributed over five
 3.2 Sources of data 41

genres: spoken language, fiction, popular magazines, newspapers, and academic jour-
nals, and it represents American English from 1990 through 2011 with roughly 20 million
words for each year.
• The British National Corpus BNC (e.g. Burnard 2007). The BNC represents a wide cross-
section of British English from the late twentieth century (1980–94), both spoken (c. 10
million words) and written (c. 90 million words). The written part includes, for example,
texts from newspapers, periodicals, and journals, academic books and popular fiction,
published and unpublished letters, and memoranda, school and university essays. The
spoken part consists of informal conversations and spoken language collected in various
contexts such as business or government meetings, radio shows, and phone-ins.
• CELEX (Baayen et al. 1995). CELEX is a lexical database which contains data from
German, British English, and Dutch, and has been extensively employed in linguistic
and psycholinguistic research. Apart from orthographic features, the CELEX database
comprises representations of the phonological, morphological, syntactic, and frequency
properties of words. We use the English part of CELEX, which has been compiled on
the basis of dictionary data and text corpus data. The dictionary data come from the
Oxford Advanced Learner’s Dictionary (1974, 41,000 entries) and from the Longman
Dictionary of Contemporary English (1978, 53,000 entries). The text corpus data come
from the COBUILD corpus (17.9 million words). Overall, CELEX contains lexical
information about 52,446 English lexemes, which are represented by 160,594 word
forms. Unlike the previously listed corpora, with CELEX all decisions concerning pho-
nological, morphological, or other aspects of the words concerned were taken by the
compilers of this database.
• Google Book Corpus. The Google Book Corpus is a corpus of 155 billion words taken from
1.3 million books written in American English from 1810 through 2009. This corpus went
on line in 2011 (Davies 2011).
• Other corpora such as The Corpus of Historical American English (COHA) (Davies
2010), The Wellington Corpora of Spoken and Written New Zealand English (WSC and
WWC, respectively, Bauer 1993; Holmes et al. 1994), the Morbocomp corpus (Scalise
and Bisetto 2012), and the Boston University Radio Speech Corpus (Ostendorf et al.
1996).
• Dictionaries of neologisms, reverse dictionaries, and other dictionaries, for example
Lehnert (1971); Mish (1983); Ayto (1990); Barnhart et al. (1990); Green (1991); Tulloch
(1991); Knowles (1997).

Of these databases, we have relied most heavily on COCA, as it is the largest available corpus
and the one that is updated most often. As much of what we can learn about contemporary
English morphology comes from looking for neologisms and encountering infrequently-
occurring forms, this corpus gives us the most promising material to work with. We have
made use of the other corpora to compare results, to fill in details and to look for possible
42 Basic principles: methods

cross-dialectal differences. COHA has allowed us to look for trends in word-formation over
the last two centuries.

3.2.2 Methods of obtaining and treating data from the corpora

To obtain data from the corpora, we searched for pertinent words or patterns by using string
searches. For example, to investigate words with a particular suffix, say, verb-forming‑ize as in
randomize, in COCA or the BNC, we would extract from the databases all orthographic
words that end in the strings <ize>, <izes>, <ized>, <izing>, <ise>, <ises>, <ised>, or <ising>.
The resulting lists of course still contained a large number of words that were either irrelevant
or problematic, and we would therefore subject them to standard data-cleaning procedures.
Obviously useless data—misspelled words, words with odd punctuation, words ending in
the string which obviously did not contain the affix (size, prize, bodywise)—were first elimi-
nated. This initial cleaning, however, inevitably left problematic forms that were more difficult
to deal with. The thorniest task was to decide whether particular words could be considered
to bear the affix in question.
There are a number of issues that arise with respect to such determinations. First, there is
semantic diversity. Semantically diverse affixes pose the question as to whether they really
represent a single morpheme or a number of different, homophonous ones. In those cases
where there was a sufficiently clear semantic relationship between different meanings or
readings, we opted for a polysemy approach, thus avoiding a proliferation of affixes. That is,
in some cases we made a judgement call about the degree of semantic relatedness that led us
to group certain affixes together as a single affix. Others might disagree with our decisions,
but usually we were inclined to go in the direction of lumping rather than splitting.
Second, there are derivatives that are not formed on potentially free bases but on obligato-
rily bound bases. Some theorists might argue that forms like baptize or propagate are mor-
phologically simplex, since they are not derived by the suffixation of ‑ize or ‑ate to an already
existing base word. However, we decided to adopt the position that independent of the theo-
retical approach one might take, there is an undeniable non-arbitrary connection of deriva-
tives like these two to other words featuring ‑ize and ‑ate, in that the two strings at least
indicate the verbal status of the word. Thus, even in these cases one could argue for the pres-
ence of a suffix which indicates the verbal category of the word, if nothing else. If such an
argument could be made, we did not eliminate these examples from consideration. In other
words, we took the position of being inclusive so as not to preclude data that might add to
our understanding.
The third problem is closely related to the previous one and concerns semantic and pho-
nological opacity. While the focus of this book is on productive morphology, which is over-
whelmingly transparent in nature, we inevitably encountered forms that were non-transparent,
either in terms of meaning or phonology. Opacity is a gradient notion and decisions about
morphological segmentability are difficult and often quite arbitrary. However, they in general
 3.2 Sources of data 43

concern only a rather small minority of words with particular affixes, such that the occasional
arbitrary decision normally does not unduly influence the overall picture. Indeed, decisions
on the inclusion or exclusion of words as members of a morphological category are most
important in quantitative analyses (e.g. of productivity). Since this book is not primarily
interested in quantification, but in patterns of structure and usage, we are confident that
occasional arbitrary decisions did not have a significant impact on our overall results.
In ­general, we have tried to document the potentially controversial data, as well as the
uncontroversial.
A word is also necessary about how we made judgements about the novelty of words. It is
obvious that claims about productivity are based on noticing the extent to which neologisms
formed by a particular process can be found. Our practice in determining the novelty of
forms was simply to check words against the OED On Line. Obviously this is an imperfect
method, but we reasoned that the OED On Line is updated constantly and there was no other
established source that would give us a better indication of whether words were established
or how established they were. Judgements about novelty are, like judgements about opacity,
difficult to make. Forms can be coined more than once, so we also used the OED to deter-
mine whether a word had a steady use over time or a use so intermittent (perhaps used only
one or twice several centuries ago) that an appearance of that word in a contemporary source
probably indicates a re-coinage.
For practical reasons, not all sources have been systematically used for the description of
each of the phenomena to be dealt with. For example, some chapters rely more on exempli-
fication with data taken from COCA, others more on the BNC, and yet others primarily
make use of data not collected from corpora at all. The choice of a particular source often
depended on whether we already had collections of cleaned data from one of the corpora, in
other cases it was a matter of personal interest or taste. In many cases, we simply went for the
largest corpus, that is, COCA, since new coinages of a particular morphological category are
often so rare that they cannot be found in sufficient numbers even in a corpus as large as the
BNC. In any case, the patterns described have been checked for their validity across varieties
and corpora whenever possible.

3.2.3 Other sources of data

All of the sources described so far are well-established tools of linguistic research. In addition
to these widely recognized sources we have also made use of other freely available resources
(both print and internet) that were not designed as research tools for linguists but neverthe-
less provide very interesting and pertinent data. Indeed, we freely admit to being opportunistic
scavengers, taking up whatever source of new words might be available. These non-scientifi-
cally oriented sources are especially useful in those cases where, due to the limitations of
electronic searches and the properties of the pertinent morphological categories, it is not pos-
sible to systematically search for and extract relevant forms from corpora. Among these cases
44 Basic principles: methods

are back-formations, truncations, blends, and even compounds of various sorts, as none of
these types can be accessed by the sort of string search that gives us data for affixes.
Two useful sources were Urban dictionary (http://www.urbandictionary.com/, last accessed
1 November 2012) and Wordspy (http://www.wordspy.com/, last accessed 1 November 2012),
which are websites that collect neologisms and provide relevant citations for these new
words. Other such sources were websites of specialized communities whose communicative
needs make the coinage, or collection and documentation, of derived or compounded words
necessary. For example, websites on dogbreeding (e.g. http://www.dogbreedinfo.com/
hybridmain.htm, last accessed September 2012) provide lists of names of mixed-breed dogs
that can be conveniently used to investigate patterns of blend-formation. Similarly, websites
designed to support genealogical research can be used to collect nicknames derived by trunca-
tion and/or ‑y‑ suffixation (as, for example, done by Lappe 2007).
Finally we have also made use of data collected by pure happenstance: items we encoun-
tered in newspaper, radio, TV, or materials collected by our students for classes. We have on
occasion used the internet to verify the existence of words or to examine popular usage. We
have also occasionally used material elicited in production experiments that were carried out
to test the properties of certain morphological categories for which corpus data are not easily
available for the above-mentioned reasons. We have, however, been cautious about asking
individual speakers for judgements on words. One thing that we have learned from this
project is that our intuitions about whether xyz could possibly be a word are often not very
good. We have frequently found novel words in our sources that seem completely unremark-
able in context, but which seem weird or improbable out of context, and might strike some
speakers as impossible. For this reason, our practice has been to be as non-judgemental as
possible. We have not excluded words from consideration because we felt that they were
somehow improbable or ‘wrong’. Others might dispute our non-prescriptive stance, but we
have chosen the path of inclusivity again in the hopes that such forms have something to tell
us about how speakers of English actually create new words and extend the meanings of
established words.

3.3 Citing data

Our intention in this volume is to make use of the data from the above-mentioned sources to
provide rich exemplification for all patterns discussed. We must therefore say something about
our conventions for listing examples and citing illustrative quotations from the data bases.
Examples are given in citation form, regardless of the inflectional form in which they were
attested in the corpus. A reader wanting to check the corpus for the citation of a particular
noun might therefore need to check the plural form as well as the singular to find the attesta-
tion. Similarly, to find the attestation of a verb, the reader might have to search the corpus for
the past tense or the 3rd person singular forms. Where lists of examples are given, they are
 3.4 Interpretation of data 45

alphabetized. Further, where affixes have two or more allomorphs, forms with extenders (for
example, ‑ation, ‑cation, ‑sion, ‑tion, ‑ion, ‑ution, and ‑ition or ‑able and ‑ible), we have chosen
to refer to the affix in discussion using a single form, rather than listing all possible forms at
each mention, except where specific allomorphs or forms with extenders are the direct topic
of discussion.
In addition to giving extensive lists of forms that illustrate particular affixes or other types
of word-formation, we attempt wherever possible to show how forms are used in context.
This is, of course, absolutely critical where we are discussing the semantic properties of vari-
ous word-formation processes. Where quotations are drawn from one of the corpora, we
begin the quotation with the name of the book, newspaper, magazine, TV programme, or
other source from which the quotation is drawn, followed by the year of publication or usage.
We treat examples drawn from other written sources (our own reading of novels or newspa-
pers, for example) similarly. This introductory material is always presented in italics. Where
examples are drawn from internet sources other than the corpora, we give the URL in paren-
theses next to the example.
Where we discuss phonological properties of affixes or other word-formation processes,
we transcribe forms using the IPA. Transcriptions are taken from the 3rd edition of the
Longman Pronunciation Dictionary (Wells 2008), with certain simplifications (e.g. eliminating
various diacritics) where fine phonetic detail is not needed. Where necessary we cite both
standard British English pronunciation and standard North American English in that order,
with the symbol ‖ separating the two.

3.4 Interpretation of data

There are innumerable places in the construction of such a comprehensive reference work as
this one where we found it necessary to rely on fine judgements. In addition to the cases
mentioned above concerning the polysemy of affixes, the treatment of opaque forms and the
like, we constantly found it necessary to make judgements about the category of the base in
a complex word or even the category of the output. With regard to the latter, we should
mention that although COCA and the BNC are parsed for part of speech, we did not neces-
sarily rely on the parsing given in the database, choosing rather to make our own judgements
as to part of speech in a given syntactic context.
With regard to the former, there are numerous cases where the base of a complex form
could theoretically be counted as either a noun or a verb for example, there being so many
forms in contemporary English that have been subject to conversion. Determinations of the
category of the base in cases such as these are notoriously tricky, and wherever we have
needed to make such determinations we have attempted to justify our reasoning. In some
cases we have left the decision open where we have not found sufficient evidence to support
a choice one way or the other. Readers may of course choose to dispute our decisions.
 chapter 4

Orthography

4.1 Prospectus
In this chapter we consider the way in which the morphological structure of words interacts
with their spelling. We will first have a look at how vowel length is represented in English
orthography, and the way in which this is affected by morphological processes of affixation.
We then consider some other orthographic processes which influence the spelling of mor-
phologically complex forms or the link between spelling and pronunciation.
Since it is of particular importance in this chapter, we will not talk of vowels and conso-
nants but of vowel sounds and consonant sounds and of vowel letters and consonant letters.
The vowel letters are <a, e, i, o, u> and sometimes <y> (though <y> is a consonant letter in
some positions, e.g. initially in the word yacht). The letter <w> may form part of a sequence
of letters that stand for a vowel sound (e.g. show).

4.2 The representation

of vowel sound length
4.2.1 Basics
We start with a note on terminology. English vowel sounds have traditionally been divided
into two groups, long and short, also analysed as tense and lax respectively. In this chapter,
where we concentrate on orthography, we will stick to the traditional terms as these are typi-
cally used in discussions of spelling, but in Chapter 9 where we focus on morpho-phonological
alternations, we use the terms ‘tense’ and ‘lax’, which have more currency among phonolo-
gists. Note also that much of what follows is informed by the discussion in Carney (1994).
Short vowel sounds differ from long vowel sounds phonologically in not occurring in
word-final position in a stressed syllable. So a monosyllabic word cannot end in /æ/ (a short
vowel), but can end in /u:/ (a long vowel). Monosyllabic words containing a short vowel
must end in a consonant. The English spelling system represents, not entirely consistently, a
much older stage of the language, and many vowels have changed from long to short or vice
versa since the spelling system was established. In native vocabulary, one of the ways of mark-
ing the length of vowel sounds was to use two vowel letters. As can be seen in (1), today this
 4.2 The representation of vowel sound length 47

is an untrustworthy guide. Some of the changes that lead to the forms illustrated in (1) give
rise to unpredictable irregularities in the spelling system. The difference between <oo> rep-
resenting /uː/, /ʊ/ or /ʌ/ (in, for example, food, wood, and blood respectively) is a matter of
historical changes operating at different periods of history. Originally, all these words would
have had the same vowel sound.

(1) Letters Short vowel sound Long vowel sound

<a> cat spa
<e> bed he
<i> bit I
<o> not fro
<u> fuss truth
<ai> plaid faint
<ea> head mead
<ei> foreign feint
<eo> leopard people
<ey> chimney prey
<oo> blood brood
<ou> touch foul
<ow> knowledge brow

From a phonetic point of view, long vowel sounds can be divided into monophthongs and
diphthongs, monophthongs being vowels of constant quality and diphthongs being vowels
which change their quality within their syllable. Because of the number of phonetic changes
that have affected the English vowel system since the spelling system was established, as well
as scribal practice, the spelling system no longer reliably indicates a distinction between
monophthongs and diphthongs. All that matters for the issues discussed in this chapter is
whether the vowel sound is long or short. The spelling system pairs the vowel sounds as long
and short sounds corresponding to the same vowel letter; this is not a phonetic correspond-
ence, and only sporadically a phonological one. It is basically a historical set of correspond-
ences. The correspondences are set out in (2).

(2) Vowel Short Example Long Example

letter pronunciation pronunciation
<a> æ cat eɪ name
<e> e bed iː me
<i> ɪ pit aɪ hide
<o> ɒ‖ɑ pot əʊ ‖ oʊ mode
<u> ʌ, ʊ but, put (j)uː mute
48 Orthography

It needs to be noted that the rules of English spelling are not the same for words of different
etymological origins. Only in words of Greek origin does <ph> spell /f/, for instance
(although the slang word phat extends this unpredictably). Only in words of French origin
does <eau> spell /əʊ ‖ oʊ/. While the English spelling system has ways of spelling native
words which reflect the distinctions between long and short vowel sounds, these same tech-
niques are not consistently applied to non-native words (on the native and non-native dis-
tinction, see Chapter 2.7 and Chapter 27). Where the rules on double consonants and silent
<e> fit words of French or classical origin, it must be taken that they do so either because the
spelling has been interpreted in that way, or by accident.

4.2.2 Double consonants

One general rule for spelling native words in English is that when a stressed vowel sound
is short, this is indicated in the spelling by having two consonant letters after the vowel
letter. This is true even when the two consonant letters are not the same, as in help, kiln,
swamp, whose spelling shows them to contain short vowel sounds (RP and related varie-
ties have a long vowel in ask, mast, etc., but this is a late development). The interesting
cases for our purposes are those where there is a double letter whose sole function is to
indicate that the preceding vowel sound must be read as being short. Some examples are
given in (3).

(3) batter, cotton, cuddle, marry, minnow, offal, pillow, posset, rabbit, tissue, udder

There are some things which need to be understood about double consonant letters. The
doubled versions of <c>, <ch>, and <ge> are <ck>, <tch>, and <dge> respectively. The letter
<v> is rarely doubled (since a double <v> would be indistinguishable from a <w> in hand-
writing), so that words like navvy are exceptional (and modern) rather than the rule. Doubling
does not occur with <x>, <sh>, or <th>.
We might expect double consonant letters to be used word-finally as well as word-­
medially, since long and short vowel sounds can be followed by a single consonant: cut with
a short vowel sound and coot with a long vowel sound both end with a single consonant.
While the two-vowel letter orthography helps here (and would once have helped more), it
has been shown above that it is no longer a trustworthy indication. Double consonants on
the ends of words are mainly restricted to <ck>, <dge>, <ff>, <ll>, <rr>, <ss>, <tch>, <tt>,
<zz>, and are not consistent. His and hiss, quiz and fizz, cut and butt illustrate some varia-
tion. There are some extra double letters which arise because there is a preference in English
for lexical words to be written with at least three letters, so add, ebb, egg, inn, odd provide
unusual double letters on this principle (most words affected by the short-word rule provide
double vowel letters rather than double consonants: bee, pee, see, wee—contrast he and we
which are grammatical).
 4.2 The representation of vowel sound length 49

4.2.3 Silent <e>

If a double consonant letter word-medially indicates a preceding short vowel pronunciation,
then a single medial consonant letter must indicate a long vowel sound. This is true, as shown
with monomorphemic words in (4). Note, however, that this rule works well only when the
single letter occurs at the end of a stressed syllable: elsewhere, because the vowels are often
reduced to /ə/ or /ɪ/, the number of consonants is not significant.

(4) baby, chafer, even, holy, ivy, open, spider, Viking

Since a final single consonant letter can follow the representation of what must be interpreted
as either a short vowel sound or a long vowel sound, some way is needed of distinguishing
between these two readings. The answer is the ‘silent <e>’, which makes the single consonant
letter orthographically intervocalic, and forces a long pronunciation. The history of this,
though, is just the opposite: the final <e> letters were once pronounced, so these single con-
sonant letters were intervocalic in phonological as well as in orthographic terms. This gives
rise to the examples in (5), where the final, ‘silent’ <e> forces a long interpretation of the
vowel sound.

(5) bide, dole, hate, mete, mule

4.2.4 Consonant doubling

The implications of this for morphology arise when a suffix with an initial vowel letter is
added to a base with a final stressed syllable containing a short vowel sound. If that base is
spelled as ending with a single consonant letter, that consonant letter will need to be doubled
to protect the short vowel sound of the base. Examples are given in (6).

(6) dim dimmed

hit hitting
man manned
sip sipped
stop stopping
sun sunny

There is a certain amount of consonant doubling where the previous syllable is not stressed.
In general North American English (NAmE) has the more regular form here, when it differs
from British English (BrE). Some examples are given in (7).
50 Orthography

(7) BrE NAmE

benefited (also benefiting) ✓
benefitted (also benefitting) ✓ ✓
biased (also biasing) ✓ ✓
biassed (also biassing) ✓
counsellor ✓
counselor ✓
marvellous (also marvelled, marvelling) ✓
marvelous (also marveled, marveling) ✓
traveler (also traveled, traveling) ✓
traveller (also travelled, travelling) ✓
worshiper (also worshiped, worshiping) ✓
worshipper (also worshipped, worshipping) ✓ ✓

Words which show similar variation between single and double letters in morphologically
complex words include: backlog, focus, grovel, handicap, hiccup, label, leapfrog, marshal, overlap,
pummel, signal, snivel, etc.
Note that only a few consonants double in this environment: we find <ll>, <pp>, <ss>,
<tt>, and occasionally <gg> in this kind of doubling, but it is not general.
In some words with final stress, the same confused doubling can also be found, where the
double letter in the BrE spelling is misleading after a stressed vowel, since it suggests a short
pronunciation, as shown in (8).

(8) BrE NAmE

control, controlled, controlling, controller ✓ ✓
control, controled, controling, controler ✓
woolen ✓
woollen ✓

Other words which show this variation between single and double letters in morphologically
complex words include: distil, appall, enthrall, extol, patrol.
Note that although woolen and wooly both exist, they are not treated in an entirely parallel
fashion, with woolly being preferred even in woolen areas.
Ironically, there are a few words which are spelled variably with a single or double final
consonant, usually following a short vowel sound, and here the double consonant tends to be
the NAmE spelling and the single consonant the BrE spelling.
 4.2 The representation of vowel sound length 51

(9) BrE NAmE

appal ✓
appall ✓ ✓
distil ✓
distill ✓ ✓
fulfil ✓
fulfill ✓ ✓
enrol ✓
enroll ✓ ✓
instal ✓
install ✓ ✓

While CanE mostly follows the NAmE pattern, and AusE, NZE, and SAf E mostly follow the
BrE pattern, there is variation in all of them; indeed, as illustrated above, BrE and NAmE are
not monolithic in regard to their spelling conventions here.

4.2.5 Loss of silent <e>

Since, in the current system of English, silent <e> can be seen to be necessary to represent
the long pronunciation of a vowel letter which appears before a single final consonant, its
function is lost when a vowel initial suffix is added. Thus, since the final <e> in hope is there
to ensure a long pronunciation for the <o>, there is no need for it in hoping, where the <i>
now fulfils the same purpose. This gives rise to pairs such as those in (10), where consonant
doubling maintains a short vowel sound.

(10) copping coping

fatted fated
hatter hater
hopping hoping
matting mating
mopping moping
rapping raping
ratting rating
ripper riper
sitting siting
tilling tiling
52 Orthography

The rule is general enough that it is the exceptions which need attention.

• The rule applies only to a silent <e>; a final <e> which is part of some other grapheme
is not affected. Thus see + ing gives <seeing> and not *<seing>.
• A final <e> can have other functions. One of these is to ensure the ‘soft’ pronuncia-
tions of <c> and <g> (/s/ and /ʤ/ respectively). Where the ‘soft’ pronunciation
would be threatened by the loss of <e>, the <e> is retained: charge + ‑able is
<chargeable>, peace + ‑able is <peaceable>, trace + ‑able is <traceable>. Despite this
rule, acknowledgment, fledgling, judgment, and knowledgable are standard alternatives
to acknowledgement, fledgeling, judgement, and knowledgeable, although sometimes
with regional variation.
• When an adjectival ‑y is added to a noun base the final <e> may be retained or deleted,
sometimes with variation between the two spellings. The spellings <stoney> and
<stony> are alternatives (with the latter the more common), but <dopey> seems to be
more common than <dopy>. Other relevant bases include bone, game, horse, joke, rope,
shake, tone (but the standard form from fire is fiery).

4.3 The letter <i> replacing <y>

A word-final <y> is replaced by <i> when a plural, 3rd person singular present tense, past tense,
past participle, comparative, or superlative affix is added, as illustrated in (11). The suffix in such
cases begins with an <e>, even in instances like skies (below), where this might not be expected.

(11) plural 3rd past past comparative superlative

sing tense participle
pry pries pried pried
sky skies
silly sillier silliest
empty empties empties emptied emptied emptier emptiest

The general rule normally does not apply:

• when the <y> is part of a digraph with another vowel letter: days, jockeys, toys, enjoyed,
obeyed, stays (but note irregular laid, paid);
• when the <y> is in by, as in lay-bys; in the abbreviation poly ◆ polytechnic (BrE) (pl.
polys);
• when the <y> is final in a proper name: the four Marys, the two Germanys.
 4.4 Velar softening 53

This rule also applies less regularly before the agentive/instrumental suffix ‑er in drier/dryer,
flier/flyer, but carrier, crier, supplier, and pryer.
The same <y> to <i> rule may apply before other suffixes as well, even where those suf-
fixes are consonant-initial. It does not apply if the change would lead to a sequence of <ii>.
It does not apply before a singular possessive <’s>. Some examples are given in (12), and
some counter-examples in (13).

(12) ally alliance

deny denial
hardy hardily
likely likelihood
marry marriage
merry merriment
pity pitiless
plenty plentiful
rely reliable

(13) baby babyhood, babyish, baby’s

busy busyness (this is a modern coinage, contrasting with
business, whose meaning is no longer transparent)
dry dryness
fairy fairydom, fairyhood
fly flying, flyable
lady ladyship, lady’s
party partyless

4.4 Velar softening

Typically spelling differences in complex words are accompanied by pronunciation changes.
In the case of forms like electric ◆ electricity, however, we find an apparent phonological alter-
nation between /k/ and /s/ with no accompanying difference in spelling. This particular
phenomenon has come to be called Velar softening. As this is a morpho-phonological
alternation that does not affect spelling, we will not look further at it in this chapter, but will
return to it in Chapter 9.
54 Orthography

4.5 Hyphenation
Hyphenation in English is subject to fashion and arbitrary prescription, some of which may
appear to give differences between national varieties. Accordingly, this section is less con-
cerned with giving rules than showing some of the ways in which hyphenation is used. We
have nothing to say in this section about hyphenation at line-breaks, where it is often a func-
tion of syllable structure, though we note that morphology can be used for identifying appro-
priate breaks, even here: for example a line-break at film-ing is preferred to the purely
phonological fil-ming. Nevertheless, we are concerned here with other hyphens which are
used to mark morphological structure in some way.

4.5.1 Hyphens in derivatives

On the whole, hyphens are rarely used to mark off suffixes but can be found marking off
prefixes. There are a number of cases where hyphenation can frequently be found:

• When the last letter of the prefix is the same as the first letter of the base. This is particu-
larly likely when the two identical letters are vowel letters: anti-inflammatory, co-ordinate,
de-energize. Counter-revolutionary and dis-satisfied are more likely to be found with no
hyphens. Even those cases with two identical vowel letters are not always hyphenated.
Also note the exceptional ski-ing (with the variant skiing), where a suffix may get a
hyphen.
• When the last letter of the prefix is a vowel letter and the first letter of the base is a dif-
ferent vowel letter. This is less common than the last case, but may be useful to prevent
sequences of vowel letters being perceived as digraphs: anti-emetic, psycho-active, semi-
automatic.
• When more than one prefix is used, including sequences where the same prefix is
repeated: re-discover, mega-mega-rich (but note hemisemidemiquaver (BrE) is usually not
hyphenated).
• Where the use of a hyphen allows the disambiguation of otherwise homographic words:
dis-ease (White 2002: 64) versus disease, re-cover versus recover, re-form versus reform,
re-mark versus remark, re-present versus represent, re-sign versus resign.
• Where the derivative has a phrasal base. In these cases, the entire word is hyphenated,
including the affix. An example is finer-than-thou-ness (Trevanian 1998: 201), or the estab-
lished (NAmE) stick-to-it-iveness.
• Where the prefix is added to a numeral or a word beginning with a capital letter or any
other typographically odd character: pre-1990, un-American, un-“allowable”.
• Where prefixes are coordinated: pre‑ and post-operatively, pro‑ and anti-government
forces.
 4.5 Hyphenation 55

• Where a prefix is felt to be particularly independent of its base. This usage is particularly
variable, partly because the notion of independence may be speaker-specific, and partly
because publishers may have conventions on these words. Prefixes such as anti-, ex-,
non‑, over‑, past‑, semi‑ seem particularly susceptible to this usage. The independence of
the prefix may simply mean that the writer is aware of creating a neologism, so that the
hyphen helps the reader parse an unfamiliar word.
• Where suffixes are felt to be unusual or unusually independent. An example is science
fiction-y (Vittachi 2008: 183). The suffixes ‑ish and ‑y seem particularly susceptible to this
use, perhaps because of their productivity.

Note that infixed expletives, though their status as affixes may be doubtful (see Chapter 18),
are separated off by hyphens: unbe-fucking-lievable, un-bloody-likely.

4.5.2 Hyphens in compounds

When it comes to writing sequences of noun + noun, English orthography is notoriously
variable. The two nouns may be written solid (with no space between them), they may be
hyphenated, or they may be written as two separate orthographic words, apparently at ran-
dom. Consider the data given in Table 4.1 where the spellings given in four different diction-
aries are tabulated.
Not only do the dictionaries not agree with each other, they are not obviously internally con-
sistent. While the four dictionaries given here represent different national varieties (Chambers
(2006), BrE; Webster’s Third New International (1993), NAmE; Macquarie (1997), AusE; OED
on-line, BrE), the differences between them seem to depend more on house style (for exam-
ple, a move to avoid the hyphen on the part of Webster’s) than on national variety. There are,

Table 4.1 Spelling of some noun–noun sequences according to different

dictionaries
Chambers Webster’s Macquarie OED
night + bird solid separate hyphen solid
night + rider solid separate solid hyphen
night + walker hyphen solid separate hyphen
night + watchman hyphen separate solid solid
nut + gall solid solid hyphen solid
nut + grass hyphen separate solid separate
56 Orthography

of course, many examples where the four works agree (needlefish and night owl are consistently
spelled); but the fact that there is so much disagreement indicates that there is no universally
accepted norm, and thus that the linguist cannot rely on the orthography of a given word to
provide linguistic information in these instances (see Chapter 19 for a more detailed discus-
sion of the variability of spaced versus non-spaced compound spellings).
Despite this inauspicious lack of systematicity, there are many other places where the use
of a hyphen is much more systematic. A hyphen is regularly (though not necessarily always)
used

• in compounds made of a letter and a word: e-mail, U-turn;

• in appositional compounds: singer-songwriter, builder-teacher-lawyers (Vittachi 2008:
88);
• in translative and co-participant coordinative compounds (see Chapter 20): the London-
Boston flight, a French-Greek dictionary, the Russian-Chinese talks. In these cases an
en-dash may be used rather than a hyphen, which may imply a distinct structure for
these constructions;
• in compounds made up of a derivative and a particle: a dressing-down, a passer-by, a
put-down, a runner-up, a walk-through;
• in phrasal compounds, where the phrasal first element of the compound is hyphenated:
The old manage-somehow-on-a-shoestring days were definitely gone (Meynell 1978: 125);
He sat there with a think-of-me-as-one-of-the-guys smile (Kernick 2003: 119);
• in compound verbs: freeze-dry, trickle-irrigate;
• in exocentric verb–noun compounds: know-all, scare-crow; some of these are written
solid: dreadnought, pickpocket;
• in compound adjectives, especially when used attributively (see also below): card-­
carrying member, fast-approaching train, ready-made clothing, sky-blue dress;
• when hyphenated compounds are coordinated: 1,000 short- and medium-range
missiles (BNC).

4.5.3 Other lexical and grammatical hyphens

Hyphens are also used within lexical items which are not formed by derivation or composi-
tion. For example, hyphens are used

• in noun + preposition + noun structures viewed as fixed lexical items: lady-in-waiting,

man-at-arms, mother-in-law. The same is true of place-names built on the same pattern:
Weston-super-Mare, Newcastle-upon-Tyne;
• in noun + adjective lexical items, whose structure is French: court-martial, governor-gen-
eral. These may also be written separately, and in some instances are more usually writ-
ten separately: lie direct, sum total;
 4.5 Hyphenation 57

• in numbers such as 37 or fractions when written out in full: thirty-seven, four-fifths

(see Chapter 18).

Hyphens may be used to resolve the ambiguity of written sentences by indicating constitu-
ency: one reading of They sold twenty dollar items can be indicated by the insertion of a
hyphen. In morphology this becomes relevant in any two-word construction which is used in
the structure ((X Y) N), where X and Y are variables over words of any word class, and may
be the same word class. In such constructions, the X and Y may be hyphenated to show their
constituency. Examples include before-tax profits, day-care facility, roll-neck sweater, small-
business operators, wrap-around skirt, yes-no question. Note that the hyphen would not neces-
sarily be used if these constructions were not used attributively. Whether (X Y) in this
structure should count as a compound or as a syntactic structure is controversial; see Chapter
19.2.1.2 for further discussion of this point. We find a similar use of hyphenation of compound
adjectives, such as able-bodied, fast-paced, never-ending, pitch-black, space-borne. In some cases
this hyphen resolves a potential ambiguity, as in 30-odd students versus 30 odd students.
 Pa rt II

Inflection
 chapter 5

Verb inflection

5.1 Prospectus
In this chapter we will be concerned with how the morphosyntactic categories of the verb are
encoded in English. After a look at the distinction between lexical and auxiliary verbs we will
deal with lexical verbs. Finally, we will turn to the inflection of the modal auxiliary verbs and
to the incorporation of infinitival to into certain verbs.

5.2 Lexical versus auxiliary verbs

English distinguishes between two kinds of verbs, lexical and auxiliary. The distinction is
based on a variety of diagnostic criteria, most of them syntactic in nature. Thus, auxiliary
verbs precede lexical verbs in the same clause, have certain types of complement, and behave
in a peculiar way in a number of syntactic constructions, such as negation, inversion, ellipsis,
and emphasis. The particulars of the syntactic behaviour of this class of words are well
described in the standard grammars and will not concern us here in great detail since we
focus on the morphological side of the matter. It should be noted, however, that with regard
to their syntactic behaviour, the auxiliaries do not show a completely uniform behaviour
across all syntactic constructions, and we will see that their morphological behaviour is also
not necessarily identical across different lexemes.
We will start our discussion of the morphological differences between the two major
classes of verb with the lexical verbs as a reference point. The number of different forms in
the inflectional paradigms of lexical verbs in standard varieties of English ranges from eight
for a single verb (be, with the forms be, am, are, is, was, were, being, been) to only three for
some 40 verbs (e.g. put with put, puts, putting). The different paradigms of individual verbs
show different kinds and numbers of syncretism, to be discussed more extensively in
Section 5.3.5.1. If we implement an analysis that recognizes morphosyntactic contrasts in
the system as soon as at least two verbs show this contrast (that is, we ignore some of the
categories that are found only with be), we can distinguish the six morphosyntactic
­categories given in Table 5.1, illustrated with three different verbs, with, five, four, and three
62 Verb inflection

Table 5.1 Verbal inflectional paradigms

Morphosyntactic category give talk put

past gave talked put

3rd sg gives talks puts
non-past
non-3rd sg give talk put
infinitive/subjunctive/imperative give talk put
present participle giving talking putting
past participle given talked put

different forms, respectively. We deal with be more fully in Section 5.3.4.1: it has a formal
person–number distinction within the past tense category (was versus were), one within
the non-3rd singular non-past category (am versus are), and one between the past tense
form and the counterfactual form for the 1st and 3rd person singular (I was, she was versus
I were, she were).
The details of this analysis will be discussed in Section 5.3. What is important for the dis-
cussion of the dichotomy between auxiliary verbs and lexical verbs is the fact that we can
safely posit six morphosyntactic categories for the English verb, with varying degrees of syn-
cretism across different verbs. We have a distinction between past forms and non-past forms,
there is a form that is used in infinitival, subjunctive, and imperative clauses, a present parti-
ciple form and the past participle form. The present participle occurs in progressive construc-
tions, the past participle in passive and perfect constructions. Both kinds of participle can
also be used as adjectives (as in a disturbing story, one of the invented alphabets, see Chapter 14
for discussion), and the present participle can serve as an event nominal (as in the teaching of
the arts, see Chapter 10).
There is variation in non-standard varieties of English as to the kinds of morphosyntactic
category, the number of such categories and the ways they are encoded. The present-day
standard system described here is historically derived from a much richer system of catego-
ries and realizations, and some contemporary conservative dialects of English have more
different verb forms than the standard, encoding more different categories. Some dialects of
English have different or fewer morphosyntactic category distinctions in their system.
Another parameter of variation is the distribution of the different forms across the different
categories, which can differ across dialects. For example, northern dialects of Britain are
known for their peculiar subject–verb agreement pattern (the ‘northern subject rule’, for
example, Pietsch 2005; Wales 1996: 45) where what is called the 3rd singular non-past form
in Table 5.1 is used variably in certain plural and non-3rd singular environments. Another
well-known example is the use of habitual be in African-American English, some varieties of
Caribbean English and IrE (e.g. Rickford 1986; Green 1998). A description of such dialectal
systems is not attempted here. However, we will document variation with forms that express
 5.2 Lexical versus auxiliary verbs 63

the same morphosyntactic category, such as preterite spelled ◆ spelt, or participle beat ◆ beaten
◆ bet (ScE, NZE, IrE).
Let us now turn to the auxiliaries. The class of auxiliary verbs that emerges from syntactic
analyses of verbal behaviour potentially has the members shown in (1).

(1) a. can, could, dare, may, might, must, need, ought, shall, should, will,
would
b. be, do, have
c. get, go, to, use
d. better

The items in (1a) are the traditionally acknowledged set of so-called modal auxiliary verbs.
Perhaps controversially, we list could, might, should, would as separate lexemes rather
than as past tenses of can, may, shall, or will, since their semantics does not correspond
simply to that of can, may, shall, and will with different time reference. Morphologically,
the modals can be distinguished from lexical verbs in important respects. Need and dare,
however, behave differently from the other modal verbs in (1a), and we will discuss their
behaviour in detail in Section 5.4. The remaining modals, which we will call ‘core modals’ for
ease of reference, have impoverished inflectional paradigms, lacking participles (*canning,
*maying, *shoulding) and 3rd person singular forms (*oughts, *shalls), and for some of them
even past tense forms (*musted, *oughted, *shalled, etc.). Core modals also share another
morphological property: they generally do not serve as bases for further word-formation.
The core modal will is also used as a lexical verb, with interesting morphological properties.
This will be discussed in Section 5.3.6.
The auxiliaries in (1b) are non-modal auxiliaries, and they share much of their syntactic
behaviour with the modal auxiliaries. They are, however, also used as lexical verbs, and
their special status has earned them the name of ‘primary verbs’ (e.g. Quirk et al. 1985). In
terms of morphology, they are characterized by the full paradigms that are typical of lexical
verbs, and will therefore be treated together with the lexical verbs, but in a distinct section
(Section 5.3.4).
The items in (1c) are controversial in their status. The aspect-marking verb use, as in She
used to follow him to work every day, partly behaves like an auxiliary in some varieties of
English, but in others it does not (see Section 5.4.2 for further discussion). Finally, the infini-
tival marker to is of unclear grammatical status, with analyses ranging from conjunction to
VP-subordinator (e.g. Huddleston et al. 2002: 1183–7) to non-finite modal auxiliary verb (e.g.
Pullum 1982). Given its (almost) invariable morphological behaviour we can afford to remain
agnostic as to its status and will consider it only in connection with its behaviour regarding
certain verbs preceding it, as in, for example, gonna, wanna (see Section 5.5).
64 Verb inflection

The lexical verb get also occurs in passive constructions in the same syntactic position as
be, which, together with its lexical morphology, can be taken as an argument for its belong-
ing to the set in (1b). Similarly, the verb go, in its present participle form plus a preceding
form of be plus a following to is in competition with will, and could therefore also be
included in the auxiliary verbs. There are, however, strong syntactic arguments not to include
them in the class of auxiliaries, as they behave like lexical verbs in crucial syntactic tests like
question formation and negation. Morphologically, nothing unexpected happens with these
verbs either. The future-oriented construction with go can be seen as an idiomatic use of the
present participle of this verb, and passive get enjoys the same full paradigm as in other
syntactic constructions. The two candidates get and go also lack a specific morphological
property that all finite auxiliary verb forms share and that distinguishes auxiliaries from lexi-
cal verbs, namely that they can act as a morphological host for the suffixed negator n’t.
Future-marking go and passive get cannot. The two verbs are thus syntactically and mor-
phologically not distinct from (other) lexical verbs and will not be treated as auxiliaries here.
Similar arguments would hold for some apparent auxiliaries as used in certain dialects of
English, such as ScE want (as in This car wants washed).
We are slightly controversial in adding the marginal example better, which, in some
varieties, occurs in sentences like I better do it, bettern’t I? The fact that it can take reduced
n’t shows its status, but it is rare in writing or formal speech where the older and fuller
form had better is preferred. We have little to say about this form beyond noting its
existence.
Based on this brief overview of the two classes of verbs we will now turn to the details of
the inflection of lexical verbs. Further intricacies of the morphology of auxiliary verbs will be
discussed in Section 5.4.

5.3 Lexical verbs

5.3.1 The verbal paradigm
In Table 5.1 we briefly introduced the paradigm of English lexical verbs with only minimal
justification and discussion. In this section we will take a closer look at the morphosyntactic
categories, their justification and realization across verbs. Table 5.2 recasts the paradigm, from
a different perspective, focusing on the forms and their names instead of the morphosyntac-
tic categories. We will treat the verb be separately in Section 5.3.4.
Table 5.2 has only five rows in the first column, as it recognizes the fact that the traditional
categories of infinitive, subjunctive, and imperative collapse formally with the non-3rd singu-
lar non-past tense form. We use the label plain form to express the fact that this form is
identical in form with the base form, that is the form to which affixes are attached. The differ-
ence between infinitive, subjunctive, and imperative is no longer morphologically marked in
 5.3 Lexical verbs 65

Table 5.2 Verbal paradigms: terminology, forms and morphosyntactic categories

Name of form Regular verb Irregular verb (Morpho-)syntactic category
realized by the form

plain form talk give infinitive, subjunctive,

base form imperative
non-3rd singular non-past
plain non-past form talk give tense
3rd singular non-past talks gives 3rd singular non-past tense
preterite talked gave past tense
present participle talking giving present participle
past participle talked given past participle

English, but is syntactically realized. That is, in certain syntactic constructions or sentence
types the plain form is used. There is thus only one morphosyntactic category involved,
which we could label ‘infinitive-subjunctive-imperative’ (cf. Huddleston 2002b: 83). We will
use the less clumsy, and well-established term ‘plain form’ instead. The reasons for recogniz-
ing the morphosyntactic category of non-3rd singular non-past tense in spite of its being also
realized by a plain form is that this form contrasts with the 3rd singular form, given the same
sentence type (I want some cake versus She wants some cake). We will use the shorthand
plain non-past form for this use of the base form. The base form, printed in small capital
letters, will also be used as the citation form of the respective lexeme.
The proposed analysis implies that the finite versus non-finite distinction is a syntactic
rather than a morphological one, as the non-finite infinitival form is identical to the finite
subjunctive, imperative, and plain non-past forms. This turns the distinction between these
identical forms into one of clause types.
As already pointed out in Section 5.2, the number and kinds of syncretisms that different
verbs show is variable and deserves further study. For example, from a comparison of the
two sample verbs talk and give we see that talk, like all regular verbs, has four different
forms, while give has five. However, if we also take into account the different vowel in the
preterite we might as well state that there is a base form syncretism with five of the six
forms, plus an additional base form (with a different vowel) as the preterite. The different
patterns of syncretisms will be discussed in more detail in the discussion of irregular verbs
in Section 5.3.5.1.
The semantics of the non-past and the past tense form require some comment. The non-
past is used not only to refer to events which habitually happen at a time including the present
(Mary leaves for work at eight o’clock) and for states which are currently in force (Wellington is
the capital of New Zealand, This coffee tastes wonderful), but also for eternal truths (The sun
rises in the east) and for ongoing events in the here-and-now, as in a running commentary
(Rooney scores for England!) or stage directions (Mrs Smith picks up her knitting) and
66 Verb inflection

performatives (I apologize for what I said). In academic discourse, it may be used to discuss
the content of any written text, whether modern or historical (Austen depicts her heroines with
great honesty). It is also used for scheduled future events (The plane arrives at noon tomorrow),
and in subordinate clauses more generally for future events (He will tell you when he arrives).
Despite our label of ‘non-past’ it may be used to narrate a presumed past event, typically in
colloquial language or in jokes (A horse walks into a bar, and the barman says, ‘Why the long
face?’).
The past tense is used as a narrative tense, to relate completed events in the past, but also
to mark attitudinal distance from an event, for instance for reasons of politeness (I wondered
whether you had time to see me) or to express a hypothesis or doubt (If you left now, you could
be in Edinburgh before dinner). For further details see, for example, Quirk et al. (1985) or
Huddleston (2002b).
We will now turn to the discussion of the distinction between regular and irregular verbs,
a distinction that we have already used in our discussion without proper justification.

5.3.2 Regular versus irregular verbs

As shown in Tables 5.1 and 5.2, we can distinguish between six morphosyntactic categories
that are morphologically encoded. Depending on how these categories are realized, gram-
marians have introduced the distinction between regular and irregular verbs. In spite of what
the terminology suggests, this distinction is not unproblematic, as will become clear as we go
along.
According to most analyses, regular verbs are those that form their past tense forms with
the predictable allmorphs of the so-called regular suffix ‑ed (as in showed, talked, ended), while
all other verbs are irregular. There is also some irregularity found in the marking of 3rd per-
son singular non-past, but this irregularity is restricted to the three verbs in standard varie-
ties: be, do, and say. The respective forms are is, does /dʌz/ and says /sez/ (~ regular /seɪz/
in some varieties), and do not provide a sensible basis for a general category distinction. The
recent verb text has, in some varieties, irregular third persons, either as texes or as textes,
with the /ɪz/ allomorph occurring in an unpredictable environment.
The irregular verbs as defined above may form their preterite and past participles by vari-
ous other means, for example by ablaut (sing ◆ sang ◆ sung), by suppletion (go ◆ went ◆ gone),
by partial suppletion and addition of an unexpected ‑t (bring ◆ brought ◆ brought), by having
the suffix ‑en in the participle form (show ◆ showed ◆ shown), or some combination of these
(give ◆ gave ◆ given). There is even a group where all three forms concerned are homophonous
(put ◆ put ◆ put). Thus, there seems to be a clear dichotomy of a nicely predictable set of
forms and a mixed bag of largely unpredictable miscellaneous forms.
One problem with such a view and the above definition of what is regular and irregular, is,
however, that it is not always clear which verbs should be treated as irregular. Sometimes the
right suffix allomorph is accompanied by some additional change in the base, as for example
 5.3 Lexical verbs 67

with have ◆ had, do ◆ did, teach ◆ taught, which makes the forms appear regular in terms of
suffixation, but irregular in terms of base allomorphy. Another problem is that some verbs
have variably regular and irregular past tense forms, for example betted ◆ bet, knowed ◆ knew,
learned ◆ learnt, which would make them belong to both classes. Traditional wisdom also
holds that irregular verb forms need to be stored in the mental lexicon, while the regular
suffix is generally applied to new verbs, and the regular verb forms need not be stored (and
mostly are not). However, as we will see in some of the discussions to follow, this is a rather
simplistic view of the matter that is unable to account for a number of empirical facts. Finally,
the terminology may suggest that the irregular verbs form their preterites and past participles
in rather arbitrary ways. As we will see, however, there are a number of clear sub-patterns
within the set of irregular verbs.
In spite of these problems we will use the established distinction here and treat all those
forms as regular whose past tense is fully predictable as being marked exclusively by the
expected allomorph of the ‑ed suffix, with no accompanying change of the base. All other
verbs are treated as irregular. Those verbs that have both regular and irregular variants of the
past tense will be treated with their irregular variants in the section on irregular verbs. The
theoretical and psycholinguistic status of the distinction between regular and irregular will
be discussed in more detail in Section 5.3.5.2.
One could of course ask the question whether regular verbs and irregular verbs might also
differ in other respects, that is, whether the difference in inflectional patterning is accompa-
nied by other, non-inflectional properties that distinguish the two kinds of verb. And indeed,
such differences can be found.
For example, it has long been observed that, on average, irregular verbs tend to be of higher
frequency than regular verbs. Empirical studies using corpus frequencies (e.g. Baayen and
Moscoso del Prado Martín 2005) have shown that the probability of a verb being irregular
dramatically increases with higher token frequencies. For illustration, we may look at the
frequency of verbs in the BNC lemmatized word list (Killgariff 2006), which contains the
1,281 most frequent verbs as found in this corpus. The top 13 verbs in this list are all irregular,
with the first regular verb, look, occupying rank 14. In contrast, there are only two irregular
verbs among the bottom 100 verbs, bet and bleed. High frequency protects word forms
from falling out of memory, and it therefore does not come as a surprise that the most irregu-
lar verb is also the most frequent one, be, which is in fact the second most frequent word of
English overall (after the), according to the BNC frequency list.
Apart from a frequency difference, there is also a difference between regulars and irregu-
lars that concerns their morphological families. A morphological family of a given word
consists of all morphologically complex words (i.e. types) in which that word occurs as a
constituent. For example, the morphological family of accept would contain accept, acceptable,
acceptably, acceptability, acceptance, accepter/acceptor, accepting, acceptive (based on the
attested forms in COCA). Verbs with irregular preterites tend to have morphological family
sizes that are significantly smaller than those of regular verbs, although the irregulars tend to
68 Verb inflection

have a higher token frequency and would therefore be expected to enter further derivation,
all other things being equal (Baayen and Moscoso del Prado Martín 2005).
Finally, there is some debate about the role of semantics in verbal inflectional patterning.
Contrary to standardly expressed views that the distinction between regular and irregular
verbs is one of form only (e.g. Kim et al. 1991), there is robust empirical evidence that regular
and irregular verbs may differ in their semantic properties. Evidence for semantic effects on
verb inflection has emerged from production experiments, as documented in Bybee and
Slobin (1982) and Ramscar (2002). Baayen and Moscoso del Prado Martín (2005) provided
large-scale distributional evidence on the role of semantics in inflection. Controlling for fre-
quency, the authors show that the number of synonyms is larger for irregular verbs.
Furthermore, sets of synonyms for irregular verbs are much more likely to contain other
irregular verbs than sets of synonyms of regular verbs are to contain other regular verbs. This
means that irregulars tend to live in a semantically more densely populated space, having
more semantic relations to other words and being more similar to each other than regulars.
Incidentally, this may have contributed to the survival of the many irregular forms in the
history of the language, as forms with stronger semantic ties to other words tend to have
stronger lexical representations in memory.
In sum, it seems that the difference in the inflectional behaviour of verbs goes together
with differences in other areas. We will return to some of these facts in the theoretical discus-
sion in Chapter 23.

5.3.3 Regular verbs

Regular verbs have only four different forms, one of them being the plain form, already dealt
with above. In the following we will discuss the three suffixed forms and some issues relating
to the orthography of the suffixed forms.

5.3.3.1 Preterite and past participle

Regular verbs are characterized by the fact that their preterite and past participle are fully
predictable and involve the suffixation of one of three allomorphs to the base form. The allo-
morph /ɪd/ or /əd/ occurs after base-final /d/ and /t/. All other bases take /d/ or /t/,
depending on the voicing of the base-final segment. Bases with voiced final segments take
/d/, bases with voiceless final segments take /t/. This is illustrated in (2). We transcribe the
syllabic allomorph as /ɪd/, ignoring the considerable variation between /ɪd/ and /əd/.

(2) /ɪd/ /d/ /t/

added /ædɪd/ played /pleɪd/ shipped /ʃɪpt/
faded /feɪdɪd/ dragged /dræɡd/ pushed /pʊʃt/
knitted /nɪtɪd/ teased /tiːzd/ packed /pækt/
 5.3 Lexical verbs 69

5.3.3.2 3rd singular non-past

The 3rd singular non-past form is derived from the base form by adding the pertinent allo-
morph of the suffix ‑s. This allomorphy is similar to the allomorphy of the plural ‑s (from
which it differs only by not being subject to haplology), and also similar to that of the past
tense suffix (in the general pattern of variation). Verbs ending in a sibilant (i.e. one of the
sounds /s, z, ʃ, ʒ, tʃ, dʒ/) take the allomorph /ɪz/ or /əz/, all other bases take either /z/ or
/s/, depending on the final segment of the base. If the base ends in a voiced segment the
voiced allomorph /z/ is chosen, if not, the unvoiced allomorph /s/ is chosen. Examples are
presented in (3). As with the past tense suffix, the variation between the two epenthetic
forms is ignored.

(3) /ɪz/ /z/ /s/

passes /pɑːsɪz ‖ pæsɪz/ plays /pleɪz/ ships /ʃɪps/
eases /iːzɪz/ drags /dræɡz/ puffs /pʌfs/
wishes /wɪʃɪz/ moves /muːvz/ packs /pæks/
judges /dʒʌdʒɪz/

5.3.3.3 Present participle

Apart from the plain forms, the present participle is the only form that is of the same struc-
tural make-up for all lexical verbs, no matter whether regular or irregular. It is derived by
adding the suffix ‑ing to the base form. The suffix has two allomorphs, /ɪŋ/ and /ɪn/, the lat-
ter being prescriptively viewed as a non-standard form, though it may be heard from other-
wise standard speakers. Variation between these two has been frequently subject to
sociolinguistic investigation (e.g. Trudgill 1974).

5.3.3.4 Orthographic issues

The default spelling of the suffixes is <ed>, <s>, and <ing>, but the attachment of these
strings to the orthographic representations of the verbal bases may bring with it some adjust-
ment in the orthographic representations of certain verb forms. These adjustments concern
the doubling of base-final consonant letters before <ed> and <ing>, the deletion of final
mute <e> before the same suffixes, and the change of <y> to or <ie> before <ed> and <s>,
respectively. These changes follow the general principles of morphology–orthography inter-
action in morphology that are laid out and illustrated in Chapter 4.
In addition to what follows from these general principles, the following peculiarities
should be noted with verbal inflected forms. First, the 3rd singular suffix ‑s has two spelling
variants, <s> and <es>, with the latter occurring after bases ending in <s> (<kisses>), <z>
(<buzzes>), <x> (<boxes>), <sh> (<pushes>), and <ch> (<matches>) and the relevant
words ending in <o> (e.g. <vetoes>, see Chapter 7). The suffix ‑ed has also two spelling vari-
ants, <ed> and <d>. The latter is used with words that end in <e> (baked, created, dyed, hoed),
70 Verb inflection

and after specific base allomorphs of certain verbs (e.g. did). The spellings <paid>, <laid>,
and <said> (and the spellings of pertinent derivatives, such as <mislaid>) are lexical excep-
tions to the rule that base-final <y> is preserved after vowel letters if it is part of a vowel
digraph (as in <employed> or <stays>). With regard to ‑ing, bases ending in <ie>, such as
<die, lie, tie, vie>, have <y> instead of <ie> before ‑ing.

5.3.4 Irregular verbs I: The primary verbs be, do, and have
5.3.4.1 be
Apart from its being a lexical verb, specifically, a copula, be is used as an auxiliary verb in
progressive aspect constructions and in passive sentences. The full paradigm of be is given in
(4). This verb is the only one in standard English that distinguishes the three person catego-
ries in the singular. Instead of a distinction between 3rd singular and non-3rd singular, be
encodes all three persons in the singular differently in the non-past (am ◆ are ◆ is). In the past
tense paradigm there is a distinction between 2nd singular and non-2nd singular (were ◆
was). Furthermore, (4) has a column for the counterfactual. As already mentioned, this cat-
egory is not formally marked with any other verb, and with all other verbs the preterite is
used in the respective clauses, which means that the counterfactual in English is a syntactic,
and not a morphosyntactic category. With be, however, we find a morphological distinction
between the counterfactual and the past tense encoded for the 1st and 3rd singular.

(4) non-past past counterfactual

singular 1st am was were
2nd are were were
3rd is was were
plural 1st are were were
2nd are were were
3rd are were were
imperative be
present participle being
past participle been

What is also striking with this verb is the degree of suppletion. One can differentiate between
six different bases that are phonologically quite distinct (am, is, are, was, were, be).
Although be is the most irregular verb of English with the highest number of different
forms and morphosyntactic distinctions, the distribution of its forms has things in common
with other irregular verbs and indeed with regular verbs. The non-finite forms are all derived
from a single base, be, as is the case in regular verbs and many irregular ones. Furthermore, in
standard varieties there is no person distinction in the plural forms, and the levelling of forms
 5.3 Lexical verbs 71

in the past tense goes beyond the levelling found in the non-past. Finally, the suffixes used for
the participles are also used by other verbs (‑ing by all verbs, ‑en by many irregular verbs).
The combination of negation and be brings about some interesting forms and issues,
which we will discuss in Section 5.4.4. There is also some variation in the forms of be used in
different dialects of English, as for example the number levelling of the preterite forms (I was
◆ we was), the use of been as a preterite in many non-standard dialects, and the use of invariant
be in African-American English.

5.3.4.2 do
Like be and have, do can be a lexical verb or an auxiliary. The auxiliary verb do is used in all
constructions requiring do-support, namely in negated sentences, interrogatives, negative
inversion, emphatic, and elliptical constructions. The paradigms of the lexical do and the
auxiliary verb do differ, in that the auxiliary does not have participle forms and the auxiliary
behaves differently in negative contraction. Negative contraction will be discussed with that
of the other auxiliaries in Section 5.4.4. In some varieties of English, the auxiliary and the lexi-
cal verb display further morphological differences. For instance, the regional English speak-
ers described in Cheshire (1982: 34–6) have a distinction between lexical dos /duːz/ and
auxiliary do, both used for all persons in the non-past. Furthermore, there are varieties that
have a paradigm do ◆ done ◆ done for the lexical verb, and do ◆ did for the auxiliary (Anderwald
2009: 126). The form done is also used as a perfective aspect marking auxiliary in Southern
U.S. English and African-American English (see Green 1998).
The paradigm for the lexical verb in standard English is given in (5):

(5) spelling phonology

plain form do /du/

base form
plain non-past form do /du/
3rd singular non-past does /dz/
preterite did /dd/
present participle doing /duŋ/
past participle done /dn/

Like regular verbs, the verb do has six forms in its paradigm, and it also shows the predictable
allomorphs of the 3rd singular and past tense suffixes, the same syncretism of the two plain
forms as regular verbs and the same suffix for the present participle. However, we find base
allomorphy with the 3rd singular (/duː/ ◆ /dʌ‑z/), with the preterite (/duː/ ◆ /dɪ‑d/), and
with the past participle (/duː/ ◆ /dʌ‑n/), and the past participle suffix is one that occurs with
irregular verbs, not with regular verbs.
72 Verb inflection

5.3.4.3 have
The primary verb have is similar to do in that its paradigm also closely resembles that of
regular verbs, with only some base allomorphy disturbing the otherwise regular picture.
Consider (6).

(6) plain form have

base form
plain non-past form have
3rd singular non-past has
preterite had
present participle having
past participle had

The 3rd singular, the preterite, and the past participle all show the predictable regular suffixes,
but do not have the base-final /v/. Disregarding this base allomorphy, the formal distinctions
and syncretisms are exactly the same as those of regular verbs.

5.3.5 Irregular verbs II: all others

In this section we take a closer look at the large number of remaining irregular verbs, all of
which show some deviation from the regular paradigm discussed above. Their exact
number is hard to determine, as it at least partially depends on how lexemes are treated
that are putatively derived from irregular verbs (e.g. misspell). In many cases, these are no
longer morphologically complex (e.g. become), but others certainly are (deepfreeze, outswim).
Another problem is of course the sampling procedure itself. Some lists we find in the litera-
ture contain verbs that other lists do not, irrespective of morphological complexity. Due to
these problems, the number of irregular verb lexemes given in the literature ranges, for
example, between 167 (Anderwald 2009: 3) and 470 (listed by englishpage.com). Palmer
et al. (2002: 1600–10) give 176, Quirk et al. (1985: 104) 261. Figures are of course lower
in those sources that define the term ‘irregular verb’ in a more restrictive way, for example
by reserving it for verbs that use vowel change to mark the past tense, or for verbs that do
not use a suffix.
What is clear is that the class of irregular verbs seems much less open than the class of
regular verbs. Claims that it is a closed class are certainly too strong, as new verbs can in
principle enter the irregular class by compounding, derivation, or back-formation. For exam-
ple, we found 126 such irregularly inflected forms at englishpage.com that were not listed by
Quirk et al. (e.g. hand-feed, interweave, outswim, overhang, preshrink, regrind,
telecast, test-drive, typewrite, underbid, unhang). There are extremely rare cases
where verbs move from the regular class into the irregular one. A well-known historical case
 5.3 Lexical verbs 73

involved teach, as a consequence of various sound changes that differentially affected differ-
ent forms.

5.3.5.1 The classification of irregular verbs

The kinds of deviations from the regular pattern which we find with irregular verbs are mani-
fold and have been the source of diverse attempts at categorizing the verbs into different
classes. All classifications refer exclusively to properties of and relations between three
forms—the plain form, the preterite, and the past participle—as the other two forms, the 3rd
singular and the present participle, are nicely predictable for all lexical verbs, with the excep-
tion of be, have, do, and say, as discussed above. Many irregular verbs have the suffix ‑en,
which has the allomorph /n/ after base-final vowels (and base-final /r/ in rhotic varieties),
and /ən/ elsewhere.
Turning now to existing descriptions and classifications of irregular verbs we can state that
the number and nature of the proposed classes differ from researcher to researcher, depend-
ing on which particular feature or features are chosen as the basis for the classification. Most
available classifications are based on the forms as spoken, which is an obvious road to take as
changes in vowel quality across the different morphosyntactic categories are very frequent
and important, and phonological vowel quality is not mapped straightforwardly onto the
spelling (consider plain form <read> /riːd/ ◆ preterite <read> /red/).
Quirk et al. (1985), to begin with, distinguish seven classes (sometimes with subclasses) on
the basis of the combinations in the distribution of three properties: whether there is a suffix
used, whether preterite and past participle is identical, and whether all three forms have the
same vowel. Palmer et al. (2002: 1600–9) set up four classes (with various subclasses) based
on four central criteria, that is whether there is some allomorph of the regular suffix discern-
ible, whether there is only a vowel alternation, and whether the past participle is formed with
the suffix ‑en. The fourth class is residual and contains ‘other formations’. The theoretical basis
of these two classifications is not very clear. Both taxonomies rely on formal similarities
between different sets of forms, with some of the criteria being syntagmatic in nature (i.e.
suffixation), some of them paradigmatic in nature (identity of vowels, or identity of certain
forms, across the paradigm). Why it is exactly these similarities and not others that form the
basis of the taxonomy seems to be partly arbitrary, partly a matter of frequency of particular
similarities, and partly even a reflection of the historical strong verb classes as they survive in
the modern system.
For further illustration of the problem of classification let us first look at a classification
based on the presence and quality of vowel change from plain form to the preterite. On the
basis of this feature, Baayen and Moscoso del Prado Martín (2005) end up with 32 classes
(one of them containing all verbs with no vowel change). In their sample, which is based
on CELEX, the number of verbs in each class ranges from only 1 to 38. Another classifica-
tion based on vowel alternations is also possible, focusing on whether forms have the same
or different vowels. We have classified the verbs from the Quirk et al. list accordingly and
74 Verb inflection

Table 5.3 Kinds of vowel alternations

Class Vowel pattern Example

A (N=39) a–b–c sing ◆ sang ◆ sung

B (N=134) a–b–b keep ◆ kept ◆ kept
C (N=42) a–b–a give ◆ gave ◆ given
D (N=3) a–a–b shear ◆ sheared ◆ shorn
E (N=75) a–a–a hit ◆ hit ◆ hit

ended up with the classification in Table 5.3, listing the kinds of vowel patterns using the
letters a, b, and c to denote different vowels, and using capital letters A to E to label the
resulting verb classes. Note that there are many lexemes among the 261 irregular verbs
given in Quirk et al. (1985) that have more than one irregular paradigm, which increases
the number of paradigms discussed from 261 to 293. We also give the frequency of the
pattern per class. We can see that the most frequent patterns are a–b–b and a–a–a, covering
more than two thirds of the verbs.
Anderwald (2009) sets up five classes on the basis of whole form syncretism. Given
three forms in the paradigm, and given that of these, none, two, or all three could be
identical, one arrives at five logically possible syncretisms, which in fact are all attested
and which constitute Anderwald’s five irregular verb classes (with various subclasses).
This is schematically represented in Table 5.4. All combinations occur, although with
varying frequency. The type frequency of each pattern, based on the Quirk et al. (1985)
list, is given in the first column, including more than one irregular pattern for a given verb
if listed and pertinent.
This is an insightful classification as it allows us to relate the irregular patterns to the regu-
lar pattern. Thus all regular verbs and a large number of irregular verbs distinguish between
the plain form and that for the preterite and the past participle (class 2, cf. regular talk and
irregular hold). One could therefore claim that this difference is a typical characteristic of
the English verbal system, as was done by early authors such as Lowth (1762: 85–6, cited in
Anderwald 2009: 98). The plain form and the preterite are the same for 46 verbs. Four of

Table 5.4 Syncretism of whole forms

Anderwald’s class Plain form Preterite Past Example
participle

1 (N=108) a b c give ◆ gave ◆ given

2 (N=129) a b b hold ◆ held ◆ held
3 (N=10) a b a come ◆ came ◆ come
4 (N=4) a a b beat ◆ beat ◆ beaten
5 (N=42) a a a hit ◆ hit ◆ hit
 5.3 Lexical verbs 75

these verbs have different past participles (e.g. bet with its past participle variant betted, and
beat with its past participle variant beaten), which makes them class 4 verbs, and 42 of them
also have a past participle of the same form (class 5, e.g. cost, cut, hit, hurt, put). There
are only ten verbs that have the same past participle form as the plain form but a different
past tense form (class 3).
Another classificational approach based on syncretism could focus on different kinds of
bases for one verb if there is a suffix present in at least one of the three forms. Thus, we could
posit that some verbs have more than one base form and that these different bases have dif-
ferent combinatorial properties. For example, give may be analysed as having one base form
give, which is used as a plain form, as the base for the suffix ‑s and, crucially, also for the past
participle suffix ‑en. This type of approach gives us another kind of syncretism, base syncre-
tism. Table 5.5 summarizes base syncretism patterns for suffixed irregular verbs, assigning
capital letters to the classes.
All of the five possible syncretism combinations are attested in the Quirk et al. list. Class A
shows that even suffixed verbs can have three different base forms (cf. drive ◆ drove ◆ driven).
Class B verbs take the same base for the participle as for the past tense (cf. also awake ◆ awoke
◆ awoken); in class C the plain form and participle bases are syncretic. The constellation of
bases labeled class D is rare and only two verbs show it: shear and swell (latter with the
forms swell ◆ swelled ◆ swollen). Class E shows an interesting implicational pattern. All verbs in
this class have an irregular participle suffix, but they differ in the marking of the preterite,
which can go either unmarked (beat ◆ beat ◆ beaten), or have a regular suffix (show ◆ showed ◆
shown), or an irregular suffix (spell ◆ spelt ◆ spelt).
Our overview has shown that there are very many conceivable properties and relations
between forms that could be employed as a basis for classification, some of which have
been used in the literature, some not. In (7) we list a number of features and their respec-
tive potential values. Most of these, sometimes in combination, have been used in the
literature to set up classes or subclasses. Different authors have used different kinds of
codings for the same kind of structural phenomenon, and we list the different codings
separately. We also indicate some of the sources that have used the respective criterion for
their classification.

Table 5.5 Syncretism of bases, verbs with suffixed forms

Class Plain form base Preterite base Past participle base Example

A (N=27) a b c ride ◆ rode ◆ ridden

B (N=55) a b b keep ◆ kept ◆ kept
C (N=33) a b a give ◆ gave ◆ given
D (N=2) a a b shear ◆ sheared ◆ shorn
E (N=34) a a a show ◆ showed ◆ shown
76 Verb inflection

(7) a. suffix
• presence of irregular suffix ‑en in the participle: yes, no (e.g. Palmer et al.
2002)
• presence of some form of ‑ed suffix: yes, no (e.g. Palmer et al. 2002)
• kind of suffix: regular ‑ed suffix, false allomorph of ‑ed, ‑en, none
• paradigm cell in which suffix occurs: only in preterite, only in participle, in both
preterite and participle
b. vowel
• change in base vowel (plain form versus preterite): none, many different changes
(e.g. Baayen and Moscoso del Prado Martín 2005)
• change in base vowel (plain form versus preterite): yes, no (e.g. Quirk et al.
1985)
• co-occurrence of vowel change and suffixation: yes, no (e.g. Palmer et al. 2002)
• number of ablaut stages: none, two-stage, three-stage (e.g. Quirk et al. 1985)
• kind of ablaut series: seven different ones (e.g. Katamba 1993)
c. syncretism
• kind of full form syncretism: five possible combinations (Anderwald 2009)
• kind of base syncretism: five possible combinations (see Table 5.5)
• vowel quality syncretism: five possible combinations (see Table 5.3)

Let us summarize our discussion of possible classifications of the similarities between differ-
ent sets of verbs. There is a long (and potentially open-ended) list of criteria and their com-
binations that can be employed to group forms into classes that share certain properties, and
the general absence of insightful pertinent methodological principles leads to classifications
that may be intuitively plausible, but mostly seem to lack any theoretical or practical relevance.
In particular, the whole question of productivity appears to have been ignored. This will be
discussed in the next section.

5.3.5.2 Productivity, variability, and irregularity

There is a widespread opinion that only the regular past tense formation is productive, while
irregular verb inflection is not. In order to see whether this is really the case we will first look
at the variation that many irregular verbs show and then revisit the potential evidence that
irregular past tense inflection is not totally unproductive.
Of the 261 verbs listed in Quirk et al. (1985), 72, that is, almost three out of ten, are listed
with variant forms (some of them regular, some irregular). If one included the great many
non-standard forms that can be found in colloquial usage (e.g. knowed with hundreds of
attestations in COCA), this figure is even higher. This means that the system is in fact much
 5.3 Lexical verbs 77

more variable than is allowed for in theoretical accounts which claim that irregular forms
block the formation of regular forms. Of the 72 verbs that have variable forms, 66 have only
two competing patterns, but four verbs have three competing patterns and two verbs have
four competitors in some of the cells of the paradigm. If we had widened the data set from
that provided by Quirk et al., we would have found even greater variability. The different sets
of verbs are illustrated in (8), (9), and (10).

(8) Verbs with two competing patterns (N=66)

abide ◆ abided ◆ abided
abide ◆ abode ◆ abode

(9) Verbs with three competing patterns (N=4)

a. bestride ◆ bestrode ◆ bestridden
bestride ◆ bestrode ◆ bestrid
bestride ◆ bestrode ◆ bestrode
b. chide ◆ chided ◆ chided
chide ◆ chid ◆ chidden
chide ◆ chid ◆ chid
c. cleave ◆ cleaved ◆ cleaved
cleave ◆ cleft ◆ cleft
cleave ◆ clove ◆ cloven
d. stride ◆ strode ◆ strid
stride ◆ strode ◆ stridden
stride ◆ strode ◆ strode

Verbs with four competing patterns (N=2) (The assumption is made that bade is pronounced
(10) 
/beɪd/; some speakers pronounce it /bæd/.)
a. bid ◆ bad ◆ bade
bid ◆ bade ◆ bade
bid ◆ bid ◆ bid
bid ◆ bade ◆ bidden
b. forbid ◆ forbad ◆ forbid
forbid ◆ forbade ◆ forbid
forbid ◆ forbad ◆ forbidden
forbid ◆ forbade ◆ forbidden
78 Verb inflection

What is remarkable with the verbs that have more than two competing patterns is that all
competing patterns are irregular ones, which raises the question of whether in the set of verbs
with only two competing forms it is always a regular pattern competing with an irregular one,
as suggested by the example in (8). This is not the case, however. Of the 66 verbs that have
two competing patterns, 19, that is almost one-third, have two irregular patterns competing
with each other.
A closer inspection of this variability reveals that all forms with an a–b–c pattern (in the
vowel or in the whole form) have a variant with an a–b–b pattern. Furthermore, those forms
whose bases end in /t/ or /d/ have a tendency to develop a–a–a variants, either instead of
a–a–b suffixed variants (e.g. bit ◆ bit ◆ bit ~ bitten), or instead of forms with different vowels
(shit ◆ shat ~ shit, or spit ◆ spit ◆ spit ~ spit ◆ spat ◆ spat). The recent usage of text as an invari-
able verb in some varieties of English illustrates this point perfectly. In sum, there are three
patterns that seem to be able to attract new verbs, the regular verb pattern, the two-stage
a–b–b ablaut pattern (in various instantiations) and the complete homophony pattern
a–a–a.
If we go beyond the set of words from Quirk et al., and also include, for example, the vari-
able irregular verbs that Anderwald (2009) investigates as ‘non-standard’ verb forms, this
picture is independently corroborated. Of these verbs, we find 16 attested in COCA with a
frequency of more than five regular preterite and past participle forms per lexeme (blow,
burst, bust, catch, come, dig, draw, give, grow, hear, know, run, see, shine, sing,
sink, teach, tell, throw). In addition, three verbs show up with irregular competing past
tenses, levelling the respective paradigms to an a–b–b ablaut pattern (see ◆ seen ◆ seen, sing ◆
sung ◆ sung, sink ◆ sunk ◆ sunk—and, indeed, all of the verbs which standardly have
/ɪ/‑/æ/‑/ʌ/ ablaut). There are even places where verbs borrow forms from other lexemes, as
with the confusion of bought and brought in several varieties of English.
Further evidence concerning the potential productivity of some of the irregular patterns
can be gleaned from studies involving past tense formation with nonce verbs, such as Bybee
and Moder (1983), Prasada and Pinker (1993), Marcus (1995, 1999), Xu and Pinker (1995),
Albright and Hayes (2003) or Wagner (2010). These experiments have also shown that speak-
ers create irregular past tense forms, and they do so especially if the nonce base form has
certain phonological properties. For example, generalizing across different studies, ablaut
forms (instead of regular preterites) seem most likely with nonce forms that are monosyl-
labic, have a complex, /s/-initial onset, have /ɪ/ as their base vowel, and end in a velar nasal
(e.g. Bybee and Moder 1983; Albright and Hayes 2003; Wagner 2010: ch. 3.2.2). Thus, the data
in these studies show that there are rather clear patterns, but these patterns cannot be gener-
ated by very general deterministic rules that produce a correct output from any given input
form. Instead, one needs either highly constrained rules that are specified to apply to very
small subsets of the data (as in Albright and Hayes 2003), or analogical models that can
­operate on the basis of similarity of forms to existing (cohorts of) forms in the lexicon
(e.g. Keuleers 2008).
 5.3 Lexical verbs 79

5.3.5.3 Orthographic issues

The regular spelling alternations, as discussed in Chapter 4 and in Section 5.3.3.4, apply, and
there are not many orthographic idiosyncrasies to be noted in addition. The irregular suffix
‑en has three spelling variants, <en>, <n>, and <ne>. The variant <n> occurs after bases that
end in <e>, <y>, <w>, <r> (<given>, <lain>, <shown>, <sworn>), the variant <ne> is
restricted to two verbs (<borne>, <done>), and <en> occurs elsewhere.

5.3.6 Defective paradigms

As mentioned in Section 5.2, the core modals are characterized by the fact that they have
defective paradigms, that is they lack certain forms. There are also, however, a few lexical
verbs that are, at least to some extent, defective. In most instances, this means that speakers
are unsure about what the relevant part of the verb might be, and avoid the issue whenever
possible. This gives rise to a gradient notion of defectiveness: some forms of some verbs are
less frequently used than might be expected, and in the limiting case are not used at all.
Instances are the verbs stride and beware, the former of which, for some apparently arbi-
trary reason, does not have a generally agreed past participle (strode, strid, stridden, strided).
The verb beware is even more impoverished and has only the plain form, with the conse-
quence that we only find it in infinitival, subjunctive, and imperative constructions.
There are also, however, less clear cases. repute and rumour, for example, have been said
to be defective (Ward et al. 2002: 1435), as they are assumed not to occur outside passive
constructions and therefore not to have the pertinent other forms. However, the example in
(11), from COCA, implies that this is really a preference rather than a strict absence of forms,
though it is less clear whether this is the case for rumour.

AB4 1991: Over his kindred he held a wary and chary care, which bountifully was
(11) 
expressed when occasion so required, reputing himself not only principal of the family
but a general father to them all

Ward et al. (2002: 1435–6) also mention that the verb say cannot occur in an active verb +
object + infinitive construction (*They say Kim to be a manic depressive). This looks like a
matter of verbal diathesis rather than defectiveness.
Another potential candidate for defectiveness is the lexical verb will ‘wish, want’, which
behaves syntactically like the modal verb and could therefore be classified as another primary
verb, together with be, do, and have. Lexical will has the full set of forms we know from
regular verbs (will ◆ wills ◆ willed ◆ willing), and even an alternate preterite form would, but
far-reaching restrictions on the usage of some of these forms make this verb appear to live on
the verge of defectiveness. The first peculiarity concerns the 3rd singular non-past form.
While willed is quite frequent as a preterite or past participle (with more than a hundred
attestations in the BNC) with no apparent restriction as to its subject (apart from sentience),
80 Verb inflection

the 3rd singular form wills seems highly restricted, especially in BrE. In the BNC we find only
12 attestations of wills, nine of which have God as a subject, and such usage appears to be very
formal. The same is apparently not true in NAmE. The form wills occurs in COCA more than
300 times with a wide variety of subjects. Some subjects denote power figures (God, Allah,
the president) but others are attested (proper names, a young Polish boy, pronouns that do
not refer to power figures). We frequently find wills followed by a reflexive (she wills herself ).
There is no particular formal flavour. Another peculiarity is the alternate preterite form,
would. Historically, this form originated as a preterite and subjunctive form, but is now largely
restricted to collocations with rather, as in (12).

FPK 1992: I would rather he’d stayed at home with us, but it was his choice, and he’s not
(12) 
a child any longer.

The analysis of potentially defective verbs suggests that defectiveness, outside the domain of
the core modal auxiliaries, is best seen as a very rare, and perhaps gradient, phenomenon.

5.4 Auxiliary verbs

In this section we will look more closely at the morphological behaviour of auxiliary verbs.
The auxiliary verbs be, do, and have were treated in Section 5.3.4, and we will include them
here only with regard to those aspects of auxiliary behaviour that were not discussed in that
section, namely weak forms, clitics, and negation.
Potential candidates for auxiliary status are the verbs in (13), where (13a) gives the core
modal auxiliaries and (13b) three auxiliaries whose morphological behaviour is, however,
quite different from that of the core modals.

(13) a. can, could, may, might, must, ought, shall, should, will, would
b. dare, need, use

We will deal with each group in turn in the following two subsections, and subsections 5.4.3 and
5.4.4 will then be devoted to clitics and weak forms, and to the negative forms of auxiliaries.

5.4.1 Core modal auxiliaries

5.4.1.1 The semantics of modals
The semantics of modals have been dealt with in many publications including Coates (1983),
Quirk et al. (1985), Palmer (1974, 1990, 2001), Huddleston (2002b). These studies indicate
that the whole question of the semantics of the English modals is far from simple and not
 5.4 Auxiliary verbs 81

easily summarized. We have no information to improve on the descriptions in such texts, and
the complex semantics associated with the modals has little direct relevance to the forms. We
should also note that there are many periphrastic expressions which carry modal meanings,
but which are not usually listed as ‘the modals’: forms such as be able to, be going to, be possible
that, be willing to, better (not), have (got) to, maybe, perhaps. We should also note that there is
a great deal of dialectal variability in the form and meaning of modals in English. This ranges
from the existence of ‘double modals’, such as might could in some varieties of English, to the
fact that haven’t got to may mean ‘there is no obligation to’ in some dialects and ‘must not’ in
others, to the fact that in some varieties shall is never or very seldom used at all. Overall
statements of precisely what the modals mean ‘in English’ are thus difficult. Rather than
spend a great deal of space on this topic, we present Figure 5.1 (adapted from Coates 1983: 26)
to illustrate the complexity of the meanings associated with the various modal verbs. While
many readings are presented in Figure 5.1, and the mapping between forms and meanings is
clearly complex (as Figure 5.1 illustrates), the meanings illustrated are closely related to one
another, to the extent that a given modal in context may be ambiguous between readings.

Modal verb Meaning/reading

must STRONG OBLIGATION

WEAK OBLIGATION

should

CONFIDENT INFERENCE

ought TENTATIVE INFERENCE

can ROOT POSSIBILITY

EPISTEMIC POSSIBILITY

may ABILITY

might PERMISSION

could VOLITION

would PREDICTION

will HYPOTHESIS

shall QUASI-SUBJUNCTIVE

Figure 5.1 Modal verbs and their meanings/readings. Bold lines represent primary use, thin lines
represent secondary or infrequent uses.
Adapted from Coates (1983: 26).
82 Verb inflection

5.4.1.2 The form of modals

As mentioned in Section 5.2, the core modals can be morphologically characterized by their
defective paradigms. They do not have non-finite forms that would allow them to occur in
the respective non-finite syntactic constructions, and they also lack 3rd person singular non-
past forms.
Since we have listed could, might, should, and would as separate modals on the basis of their
meaning, we might argue that the core modals do not even have a preterite form. However,
there are places where could (and the other forms listed here) can act as the preterite of can,
etc. under sequence of tense rules in indirect speech. This is illustrated in (14).

(14) a. He said, ‘I will come.’

b. He said he would come.
c. Mary says that she can come.
d. Mary said that she could come.

It thus seems that could and would have a dual function: one as a preterite form, and one as a
separate modal verb with their own meanings. Might and should are awkward as past tense
forms of may and shall, partly because of the restricted usage of may and shall, and partly
because the meanings of the erstwhile preterites have diverged semantically from the original
base to such a great extent.

(15) a. The teacher said, ‘Yes, you may leave early.’

b. The teacher said I might leave early.
c. My godmother says I shall go to the ball.
d. My godmother said I should go to the ball.

While (15b) is clearly possible, it sounds pedantic or old-fashioned, while (15d) does not
sound as though it is equivalent to (15c).
We might thus argue that we have modals like must and ought which have no preterite,
modals like can and will which have a preterite form that is homophonous with another
modal, and modals like may and shall which are transitional between the two.

5.4.2 Other auxiliaries

The three verbs given in (13b) are rather intricate cases, with need and dare sharing some
peculiarities. These two modals behave syntactically both as lexical verbs and as auxiliary
verbs. For example, they can take both bare infinitives (a sign of being an auxiliary) and
 5.4 Auxiliary verbs 83

to-infinitives (a sign of being a lexical verb). Interestingly, this distinction has other correlates,
syntactic and morphological. Huddleston (2002b: 111) implies for dare that the auxiliary use
goes together with a lack of 3rd person marking in the non-past. This is, however, not the
case, as the many attestations to the contrary show, two of which, from COCA, are shown
in (16).

Chicago Sun-Times 1999: And yet he dares think about it, even though he knows he has
(16) 
a bull’s-eye on his bumper
Symposium 1993: She is condemned because she dares reject the stereotype of the

submissive woman

Given that the morphological and syntactic facts do not clearly map onto each other, it is not
clear whether a lexical distinction between auxiliary dare and lexical dare is warranted at
all. Rather, we find variant forms expressing the same morphosyntactic property (e.g. 3rd
singular non-past).
With need, however, we do find a neat complementary distribution of structural proper-
ties. When need is used with a bare infinitive, that is as an auxiliary, its paradigm is defective
and has only one form, the non-past need, and no preterite, thus aligning with must and
ought. This is illustrated in (17a–c), from COCA.

(17) a. Sierra 1990: None is so poor that he need sit on a pumpkin.

b. Atlanta Journal Constitution 1996: He needs to write what makes his heart sing.
c. ABC_GMA 2007: He needed to win, he needed the black vote.

The verb use as an aspectual marker is also ambiguous in its status. For many speakers
it always behaves syntactically like a lexical verb, for others it has some of the syntactic
properties of auxiliaries. The construction is only used with anterior temporal refer-
ence, which has the morphological consequence that, independent of syntactic behav-
iour as a lexical or auxiliary verb, only the preterite and past participle forms occur.
This has led to some variation in the spelling of the form too. Thus one can easily find
attestations of <use to>, where standard orthography would have <used to>, as in (18),
from COCA.

Houston Chronicle 1999: he doesn’t spend nearly as much time as he use to on the
(18) 
practice range.

Furthermore, the construction is incompatible with the present participle (*She was using to
come by every day) and the verb thus ends up with a highly defective paradigm.
84 Verb inflection

5.4.3 Auxiliary clitics and weak forms

Being function words, auxiliaries occur most often unstressed, but must receive at least some
degree of stress in certain syntactic positions, for example before the gap in stranded position
(e.g. We’ll help you if we can, see, for example, Palmer et al. (2002: 1613) for further discussion).
Traditionally, we speak of ‘strong’ and ‘weak’ forms. In (19), we give these allomorphs for all
pertinent auxiliaries. In general, the weak forms involve a more centralized vowel and some-
times the loss of a consonant in addition. The non-strong realizations of are are treated fur-
ther below.

(19) spelling strong form weak form(s)

am /æm/ /əm/
be /biː/ /bi/
been /biːn ‖ bɪn/ /bɪn/
was /wɒz ǁ wʌz ~ wɑːz/ /wəz/
were /wɜː ~ weə ǁ wɝː/ /wə ǁ wər/
can /kæn/ /kən/
could /kʊd/ /kəd/
do /duː/ /dʊ, də/
does /dʌz/ /dəz, dz/
have /hæv/ /həv, əv/
had /hæd/ /həd, əd/
has /hæz/ /həz, əz/
must /mʌst/ /məst, məs/
shall /ʃæl/ /ʃəl, ʃə, ʃ/
should /ʃʊd/ /ʃəd, ʃd, ʃt/
will /wɪl/ /wəl, əl/
would /wʊd/ /wəd, əd/

Some of the verbs listed in (19) have yet other forms, namely clitic forms that attach to
the word immediately preceding them. There is also one verb, do, that has a proclitic,
that is /d/, for the forms do and did, which attaches to the following pronoun you, as in
non-past D’you need a hand? and past D’you have a good time? (20) gives the clitic
forms.
 5.4 Auxiliary verbs 85

(20) verb form (host +) clitic spelling host + clitic pronunciation

am I’m /aɪm/
are you’re /jɔː ~ jʊə ~ jə ǁ jʊər ~ jər/
we’re /wɪə ~ weə ~ wə ǁ wɪər ~ wər/
they’re
/ðeə ~ ðeɪə ǁ ðer ~ ðər/
do d’you /djuː, djʊ, djə, ʤuː, ʤə/
did d’you /djuː, djʊ, djə, ʤuː, ʤə/
’d /d/
have ’ve /v ~ f/
had ’d /d/
has ’s /z ~ s/
will ’ll /l/
would ’d /d/

The clitic forms are not simply to be seen as further reduced variants of weak forms for at
least three reasons. First, not all verbs that have weak forms have enclitic forms, second, the
syntactic distribution of the clitics is more restricted than that of the weak forms, and, third,
some pronoun–clitic combinations have weak and strong forms themselves (e.g. we’re and
did you illustrated in (20)). Note however that, depending on the kind of environment, the
distinction between some weak forms and the corresponding clitic is not always very clear.
For example, the orthographic representation <’ll> after a consonant letter, as in that’ll do it,
may be realized with only the clitic /l/ or with the weak form /əl/. Furthermore, some
sources list /ə ǁ ər/ as weak forms of are, and not, as we do, as clitics. We opt for a clitic analy-
sis of ’re on the grounds that a weak form analysis cannot account for the facts that the pro-
noun and the auxiliary form an indivisible phonological unit in /jɔː, jə, wə, ðeə, ðer, ðər/.
Furthermore, there are weak and strong forms of the pronoun clitic combinations themselves,
as shown in the last column of (20).
The distributional restrictions of the clitics are quite intricate and vary for different sets of
forms. In general, the clitics attach to preceding NPs, but there are some systematic excep-
tions to this. The least restricted forms are the clitic forms of has and is. They attach to the
widest range of preceding constituents, including NP subjects (Jane’s not at home), some
subordinators (the book that’s on the shelf ), interrogative words and phrases (How’s your
mother?; How bad’s your ear?) and a few more constructions (see Palmer et al. 2002: 1614–16
for more discussion).
More restricted is the encliticized realization of did, had, and would, /d/. It only attaches
to vowels and it occurs in fewer syntactic environments than the has/is clitic ’s. Which envi-
ronments are possible and which ones are not, is not entirely clear, however. Not all gram-
mars also list ’d as a clitic form of did, but the existing evidence strongly suggests its inclusion
86 Verb inflection

as such. Some examples from COCA are given in (21) (we have eliminated the spaces round
clitics used in the corpora for tagging purposes)

(21) a. American Scholar 2010: The phone rang. “How’d it go?” Lester asked. “With the
murderer? Did you talk to him?”
b. Sporting News 2009: Hey, how’d you like that? It was a pretty sweet catch, wasn’t it?

As with the clitic forms of has and is, the use of the ’d clitic may lead to ambiguities in specific
contexts. For example, I’d cut it can mean ‘I would cut it’ or ‘I had cut it’, or how’d they find us?
can mean ‘how did they find us’ or ‘how would they find us’.
The most restricted clitics are ’ll, ’m, ’re and ’ve, for which it has been claimed that they
attach only to NPs that are their own and sole subjects (e.g. Kaisse 1983: 98). This seems to be
true for ’ve and ’re (and trivially for ’m), as shown in (22) for ’ve, taken from Kaisse, but not
for ’ll, as many attested examples like those in (23) show (taken from COCA).

(22) a. I’ve got it.

b. *John and I’ve got it.

(23) a. Batman Returns 1992: Josh and Jen’ll put a spin on this.
b. Faulkner 2007: “Jenny and I’ll bring your TV,” said Martin.

One can also find spellings with two clitics attaching to one another, but in these cases the
second clitic, which is usually ’ve, is a weak form and realized with a schwa. Consider (24),
from COCA.

(24) Ind_Oprah 2004: I’d’ve been terrified if I thought anything like that

5.4.4 Negation
All of the auxiliaries, in contrast to all other verbs of English, have negative forms, in which
the negative element /nt/, spelled <n’t>, is suffixed to the auxiliary. In these cases, the nega-
tion has sentential scope. The negative form of the auxiliary is in free variation with construc-
tions that have the affirmative auxiliary form and the free sentential negator not, as shown in
(25) and (26), from the BNC. Examples like that in (26b) may be dialectal, being more com-
mon in northern British English.

(25) a. CD5 1992: there haven’t been any drugs here for ages.
b. Hansard extracts 1991–1992: there have not been any previous accidents of this
nature within the tunnel.
 5.4 Auxiliary verbs 87

(26) a. FSP 1989: Haven’t you been listening?

b. KCX 1992: Mm have you not been working then?

In contrast to the free form negation, negative auxiliary forms can never be used to express
VP-negation, that is a type of negation in which the negation has scope only over the VP, and
not over the whole clause. This leads to a potential meaning contrast between the two con-
structions, at least in writing, as shown in (27).

(27) a. I could not pay attention.

‘What I could do is not pay attention’ or ‘I was unable to pay attention’
b. I couldn’t pay attention.
‘I was unable to pay attention’

The negative suffix can be attached to each of the (non-clitic) forms of the respective verbs’
paradigms. The suffix integrates phonologically with the base, which, depending on its pho-
nological structure, may or may not lead to the addition of a syllable (cf. monosyllabic aren’t
with disyllabic couldn’t). If the addition of n’t leads to an increased number of syllables, the
nasal of the suffix is syllabic (e.g. /dɪdn̩t/). Bases with final /n/ do not have geminate nasals.
Six of the negative auxiliary forms show deviant allomorphies or spellings, or both. They are
listed in (28). Note that shan’t listed in (28) is not NAmE, and is rare or non-existent in some
other varieties as well.

(28) verb form negated verb form negated verb form

spelling pronunciation
can <can’t> /kɑːnt ǁ kæːnt/
do <don’t> /dəʊnt ǁ doʊnt/
must <mustn’t> /mʌsnt/
shall <shan’t> /ʃɑːnt/
used %
<usedn’t > ~ <usen’t> /juːsnt/
will <won’t> /wəʊnt ǁ woʊnt/

A number of additional peculiarities of auxiliary negation are noteworthy. First, the aspectual
verb use cannot take the negative suffix in all dialects or with all speakers. This is expected,
given its variable status as auxiliary or lexical verb discussed above. If the verb is lexical in
status, it does not have a negative form. A possible negative for used to is didn’t use(d) to, as
illustrated in (29) from the BNC.
88 Verb inflection

(29) Nottingham Oral History Project: They didn’t use to cut it in slices
Bacons College lesson: What do we have that we didn’t used to have?

Second, the verb can has an additional negative form, in which the free form negator is suf-
fixed to the form cannot. This form has initial stress and, just like the negative form can’t,
cannot be used to express VP-negation. The form cannot is more formal in character than the
form can’t, but the syntactic distributions of can’t and cannot seem to be equivalent (pace
Palmer et al. 2002: 1611). This is illustrated in (30), with examples from the BNC (although
Palmer et al. are right to the extent that the construction in (30a) is rare though found in both
BNC and in COCA, usually spelled <can not>).

(30) a. HHV 1992: Cannot we have a proper system of no-fault payments for medical
injuries?
b. G1D 1993: Can’t we have a vodka?
Dancy, Johnathan. 1992: Can we not then simply accept that justification continues
c. 
ad infinitum?

Third, not all of the negative auxiliary forms that can be found in the language are uniformly
considered part of the standard English system. Thus, ain’t occurs as an additional suppletive
negative form of am, are, is, have, and has in many varieties, and is described as being informal
or non-standard. Another such form is amn’t, which is especially frequent in IrE and ScE, but
seems not to be restricted to these areas. Another variant form for the 1st singular non-past
be is aren’t, which, again contrary to what Palmer et al. (2002: 1612) suggest, can be found in
both interrogative and affirmative clauses, especially in spoken registers, and more especially
in BrE, as shown by the examples from the BNC given in (31).

(31) a. KBW1992: Aren’t I hopeless?

b. KBD1992: I aren’t bothered for them.

5.5 Incorporation of infinitival to

The infinitive marker to combines with seven verbs to form morphologically complex units
that deserve special treatment. The verbs are want, go, have, use, got, ought, and sup-
pose. The spellings for the respective forms of want to and going to, <wanna> and
<gonna>, are established, for the other verbs we use <hafta>, <useta> , <gotta>, <oughta>
and <supposeta> /s(ə)pəʊstə ‖ s(ə)poʊstə/. The phenomenon is also known as to-contrac-
tion, a label that suggests that it is a phonological process of fusion. However, it can be argued
that we are not dealing with on-line phonological reduction due to fast speech but are faced
 5.5 Incorporation of infinitival to 89

with complex forms in their own right, which all show some peculiarities in phonological
shape and behaviour that are not readily described under the assumption of a syntactic and
phonological concatenation process of the two words involved. For ease of reference, we will
use the traditional term ‘contraction’ to refer to the phenomenon, but do so without commit-
ment to the underlying analytical assumptions. We will discuss the facts in detail for want
to, but we find the same patterns of usage with the other forms. We therefore discuss the
properties of the other verbs in more detail only if they differ from those of want to.
The form wanna can be used in all contexts in which the verb form want and the infinitival
marker to, introducing the complement of want, are syntactically adjacent. Notably, the
noun want adjacent to infinitival to does not contract (it virtually never occurs), nor do we
find contraction if the to following want introduces an adjunct, or a complement of another
verb. The examples of adjuncts in (32) are from Hudson (2006: 605).

(32) How much paint do you want to finish the house?

I’ll arrange for the books you want to be sent to your house.

The form wanna is only a possible realization of want to in those contexts where the plain
form or the plain non-past form of want, that is want, is possible, making incorporation with
wants, wanted, wanting impossible. The form is phonetically realized as /wɒnə ǁ wɑːnə ~
wɔːnə ~ wʌnə/, with an additional alternation in the final vowel in some varieties, where
schwa alternates with /ʊ/ if the following word starts with a vowel (see Hudson 2006: 606–8
for discussion). This variety-dependent alternation can also be found with the other verbs of
this group.
The future-oriented progressive form going to has a variant with incorporated infinitival
marker, gonna, which can be realized as /ɡənə, ɡɒnə, ɡʌnə ǁ ɡɔːnə, ɡɑːnə/. Contraction affects
go only in the pertinent idiomatic usage; the homo­phonous verb go cannot contract if fol-
lowed by to (*He’s gonna jail meaning ‘He is going to jail’).
The verb have in its idiomatic obligation-expressing usage with an immediately adjacent
to-infinitive can realize the two non-past forms have to and has to as /hæftə/ and /hæstə/.
Notably, these forms show regressive devoicing that is not found across the board (cf. li[v]e
to, do[v]etail, li[v]etrapping, fi[v]e times).
The aspect-marking construction with used to has already been described in Section 5.4.2.
The contracted form of this construction useta, pronounced /juːstə/, is phonologically differ-
ent from the non-aspect-marking form used to, as it does not normally alternate in stranded
positions. In contrast, non-aspectual used to, as in this is not what she is normally used to cannot
end in a schwa. In some dialects useta is used as a base form (as in She didn’t useta).
The idiomatic construction (have) got to expresses an obligation or necessity and, in its
contracted form is pronounced /ɡɒtə ǁ ɡɑːɾə/. In many varieties of English, including NAmE,
gotta does not rhyme with not to, which is an indication of the lexical status this form enjoys.
In some varieties of NAmE, and possible others as well, gotta can be a 3rd singular form. The
90 Verb inflection

form gotsta is found in NAmE and is used with all persons (contrary to what is suggested in
Pullum 1997: 89), as in (33) from COCA.

(33) Outdoor Life 1997: You gots to dig them deeper

The contracted variant for ought to is /ɔːtə ǁ ɔːɾə ~ ɑːɾə/. As pointed out by Pullum (1997),
oughta does not ryhme with thought to (as in you oughta be happy versus they are thought to be
happy), as the latter, at least in NAmE, cannot have a schwa. Thus, oughta, like the other verbs
discussed, has its own lexically specified phonological peculiarities.
Finally, the contracted form of supposed to, supposeta, shows the same kind of non-
alternation of schwa and /ʊ/ as the previously discussed verbs. Pullum (1997: 89) provides
the wonderful minimal pair of supposed to versus host to given in (34).

(34) a. You should pour the champagne, you’re sposta

b. You shouldn’t pour it, you should wait for the host [tʊ] / *host[ə]

At the syntactic level, supposeta co-occurs only with the clitic form of the copula, if there is a
choice (?You are supposeta take out the garbage).

Ap p e n dix

Irregular verbs in English

We provide below a list of 387 irregular verb forms in English derived from the various sources at
our disposal. We make no claim about the status of the forms we list beyond that we believe that
they are used. Specifically, some may be standard, others non-standard or dialectal, and we make
no attempt to mark such matters; we simply list the forms alphabetically in each category. The lack
of a form cannot definitely show that it does not exist, merely that we have not found it listed or
used. Under each fundamental verb, we list verbs created from it by word-formation, provided
that they too have irregular forms: such verbs are listed in italics. Grandstanded is not listed
because it is regular. Where there are alternative forms existing side-by-side there may be, but
need not be, a semantic difference between them. Forms which were originally past participles
but which are now only adjectives (forms such as drunken, sunken) are ignored. Forms which are
phonologically regular but orthographically irregular (laid, paid) are also ignored.

Base form Preterite Past participle Also regular Notes

abide abode abode Yes
be was were been
bear bore borne
 5 Appendix: Irregular verbs in English 91

Base form Preterite Past participle Also regular Notes

forbear forbore forborne
overbear overbore overborne
beat beat bet beat beaten bet
browbeat browbeat browbeat browbeaten
begin began begun begun
bend bent bent
unbend unbent unbent
bereave bereft bereft Yes
beseech besought besought Yes
bet bet bet Yes
bid bad bade bid bade bid bidden <bade> may be
pronounced /bæd/ or
/beɪd/
forbid forbad forbade forbid forbidden
outbid outbid outbid outbidden
overbid overbid overbid
rebid rebid rebid
underbid underbid underbid underbidden
bind bound bound
rebind rebound rebound
unbind unbound unbound
bite bit bit bitten
frostbite frostbit frostbitten
bleed bled bled
blow blew blown Yes
break broke broken
breed bred bred
crossbreed crossbred crossbred
inbreed inbred inbred
interbreed interbred interbred
outbreed outbred outbred
overbreed overbred overbred
bring brought brought
build built built
overbuild overbuilt overbuilt

(continued)
92 Verb inflection

Base form Preterite Past participle Also regular Notes

prebuild prebuilt prebuilt
rebuild rebuilt rebuilt
burn burnt burnt Yes
sunburn sunburnt sunburnt Yes
burst burst burst Yes
bust bust bust Yes
buy bought bought
overbuy overbought overbought
cast cast cast
broadcast broadcast broadcast Yes
forecast forecast forecast
miscast miscast miscast
overcast overcast overcast
rebroadcast rebroadcast rebroadcast Yes
recast recast recast
roughcast roughcast roughcast
sand-cast sand-cast sand-cast
telecast telecast telecast
typecast typecast typecast
catch caught caught Yes
chide chid chid chidden Yes
choose chose chosen
cleave cleft clove cleft cloven Yes
cling clung clung
clothe clad clad Yes
come came come come Yes
become became become
overcome overcame overcome
cost cost cost
creep crept crept Yes (regular in combination
creep out)
crow crew crew Yes
cut cut cut
recut recut recut
undercut undercut undercut
deal dealt dealt
 5 Appendix: Irregular verbs in English 93

Base form Preterite Past participle Also regular Notes

misdeal misdealt misdealt
redeal redealt redealt
dig dug dug Yes
dive dove dived dove Yes
do did done done
misdo misdid misdone
outdo outdid outdone
overdo overdid overdone
predo predid predone
redo redid redone
undo undid undone
drag drug drug Yes
draw drew drawn Yes
outdraw outdrew outdrawn
overdraw overdrew overdrawn
redraw redrew redrawn
withdraw withdrew withdrawn
dream dreamt dreamt Yes
daydream daydreamt daydreamt Yes
drink drank drunk drunk Yes
outdrink outdrank outdrunk
overdrink overdrank overdrunk
drive drove driven
outdrive outdrove outdriven
test-drive test-drove test-driven
dwell dwelt dwelt Yes
earn earnt earnt Yes rarely in writing
eat ate eat eaten
overeat overate overeaten
fall fell fallen
befall befell befallen
feed fed fed
hand-feed hand-fed hand-fed
overfeed overfed overfed
(continued)
94 Verb inflection

Base form Preterite Past participle Also regular Notes

spoon-feed spoon-fed spoon-fed
underfeed underfed underfed
feel felt felt
fight fought fought
outfight outfought outfought
find found found
fit fit fit Yes
refit refit refit Yes
retrofit retrofit retrofit Yes
flee fled fled
fling flung flung
fly flew flown
outfly outflew outflown
test-fly test-flew test-flown
forsake forsook forsaken
freeze froze frozen
deepfreeze deepfroze deepfrozen Yes
quick-freeze quick-froze quick-frozen
unfreeze unfroze unfrozen
get got got gotten
beget begot begotten
forget forgot forgot forgotten
give gave give given Yes
forgive forgave forgiven
misgive misgave misgiven
go went gone gone went Yes
forgo forwent forgone
undergo underwent undergone
grind ground ground
regrind reground reground
grow grew grown Yes
regrow regrew regrown
outgrow outgrew outgrown
 5 Appendix: Irregular verbs in English 95

Base form Preterite Past participle Also regular Notes

hang hung hung Yes The regular verb hang is
normatively taken to be a
hyponym of execute,
though colloquially the
irregular form is used here
as well.
overhang overhung overhung
unhang unhung unhung
rehang rehung rehung
have had had
hear heard heard Yes
mishear misheard misheard
overhear overheard overheard
rehear reheard reheard
heave hove hove Yes
hew hewed hewn Yes
hide hid hid hidden
unhide unhid unhidden
hit hit hit
hold held held
behold beheld beheld
uphold upheld upheld
withhold withheld withheld
hurt hurt hurt
keep kept kept Yes
ken kent kent Yes esp. ScE
kneel knelt knelt Yes
knit knit knit Yes
reknit reknit reknit Yes
unknit unknit unknit Yes
know knew known Yes
lead led led
mislead misled misled
lean leant leant Yes
leap leapt leapt Yes
(continued)
96 Verb inflection

Base form Preterite Past participle Also regular Notes

outleap outleapt outleapt Yes
learn learnt learnt Yes
mislearn mislearnt mislearnt Yes
relearn relearnt relearnt Yes
unlearn unlearnt unlearnt Yes
leave left left
lend lent lent
let let let
sublet sublet sublet
lie lay lain
underlie underlay underlain
light lit lit Yes
moonlight moonlit moonlit Yes
relight relit relit Yes
lose lost lost
make made made Yes
premake premade premade
remake remade remade
unmake unmade unmade
mean meant meant
meet met met
mow mowed mown Yes
plead pled pled Yes
prove proved proven Yes
disprove disproved disproven Yes
put put put
input input input Yes
output output output Yes
quit quit quit Yes
read read read
misread misread misread
proofread proofread proofread
reread reread reread
sight-read sight-read sight-read
rend rent rent
rid rid rid Yes
 5 Appendix: Irregular verbs in English 97

Base form Preterite Past participle Also regular Notes

ride rode ridden rode
outride outrode outridden
override overrode overriden
ring rang rung rung Yes The regular verb is
denominal
rise rose risen
arise arose arisen
run ran run run
outrun outran outrun
overrun overran overrun
rerun reran rerun
saw sawed sawn Yes
say said said
gainsay gainsaid gainsaid
see saw seen seen Yes
oversee oversaw overseen
seek sought sought
sell sellt sold sellt sold Yes
outsell outsold outsold
oversell oversold oversold
presell presold presold
resell resold resold
undersell undersold undersold
send sent sent
resend resent resent
set set set
beset beset beset
inset inset inset
misset misset misset
offset offset offset
preset preset preset
reset reset reset
typeset typeset typeset
upset upset upset

(continued)
98 Verb inflection

Base form Preterite Past participle Also regular Notes

sew sewed sewn Yes
oversew oversewed oversewn Yes
resew resewed resewn Yes
unsew unsewed unsewn Yes
shake shook shaken shook
shave shaved shaven Yes
shear shore shorn Yes
shed shed shed
shew shewed shewn Now old-fashioned. See
show.
shine shone shone Yes The transitive form is
more likely to be regular
outshine outshone outshone
shit shat shit shat shit Yes
shoe shod shod Yes
shoot shot shot
outshoot outshot outshot
overshoot overshot overshot
show showed shown Yes See also shew.
shred shred shred Yes
shrink shrank shrunk shrunk
preshrink preshrank preshrunk
shrive shrove shriven Yes
shut shut shut
sing sang sung sung
outsing outsang outsung
sink sank sunk sunk
sit sat sat
outsit outsat outsat
resit resat resat
slay slew slain Yes
sleep slept slept
outsleep outslept outslept
oversleep overslept overslept
slide slid slid
backslide backslid backslid backslidden
 5 Appendix: Irregular verbs in English 99

Base form Preterite Past participle Also regular Notes

sling slung slung
unsling unslung unslung
slink slunk slunk Yes
slit slit slit slitted is found, but is
denominal
smell smelt smelt Yes
outsmell outsmelt outsmelt Yes
smite smote smitten
sneak sneak snuck snuck Yes
sow sowed sown Yes
speak spoke spoken
misspeak misspoke misspoken
outspeak outspoke outspoken
overspeak overspoke overspoken
speed sped sped Yes
outspeed outsped outsped
spell spelt spelt Yes
misspell misspelt misspelt Yes
spend spent spent
misspend misspent misspent
outspend outspent outspent
overspend overspent overspent
underspend underspent underspent
spill spilt spilt Yes
overspill overspilt overspilt Yes
spin span spun spun
unspin unspun unspun
spit spat spit spat spit spitted is denominal
spoil spoilt spoilt Yes
spread spread spread
spring sprang sprung sprung
stand stood stood
misunderstand misunderstood misunderstood
understand understood understood
withstand withstood withstood
(continued)
100 Verb inflection

Base form Preterite Past participle Also regular Notes

steal stole stolen
stick stuck stuck
unstick unstuck unstuck
sting stung stung
stink stank stunk stunk
stove, stave stove stove Yes
strew strewed strewn Yes
stride strode strid stridden strode
bestride bestrode bestrid bestridden
bestrode
strike struck stricken struck
string strung strung
hamstring hamstrung hamstrung
restring restrung restrung
unstring unstrung unstrung
strive strove striven Yes
swear swore sworn
forswear forswore forsworn
outswear outswore outsworn
sweat sweat sweat Yes
sweep swept swept
swell swelled swole swollen Yes
swim swam swum swam swum
outswim outswam outswum
swing swung swung
take took taken
betake betook betaken
mistake mistook mistaken
overtake overtook overtaken
partake partook partaken
retake retook retaken
undertake undertook undertaken
teach taught taught Yes
misteach mistaught mistaught
reteach retaught retaught
tear tore torn
 5 Appendix: Irregular verbs in English 101

Base form Preterite Past participle Also regular Notes

retear retore retorn
tell tellt told tellt told Yes
foretell foretold foretold
retell retold retold
text text text Yes
think thought thought Thunk is sometimes used
jocularly as a past
participle.
outthink outthought outthought
overthink overthought overthought
rethink rethought rethought
thrive throve thriven Yes
throw threw thrown Yes
outthrow outthrew outthrown
overthrow overthrew overthrown
thrust thrust thrust
tread tread trod tread trod trodden Yes
retread retread retrod retread retrodden
wake woke woken Yes
awake awoke awoken Yes
reawake reawoke reawaken
rewake rewoke rewaken Yes
wear wore worn
rewear rewore reworn
weave wove woven Yes
interweave interwove interwoven Yes
reweave rewove rewoven Yes
unweave unwove unwoven Yes
wed wed wed Yes
rewed rewed rewed Yes
weep wept wept
wet wet wet Yes
rewet rewet rewet Yes
win won won
rewin rewon rewon

(continued)
102 Verb inflection

Base form Preterite Past participle Also regular Notes

wind wound wound
interwind interwound interwound
overwind overwound overwound
rewind rewound rewound
unwind unwound unwound
wring wrung wrung
write wrote written
handwrite handwrote handwritten
miswrite miswrote miswritten
outwrite outwrote outwritten
overwrite overwrote overwritten
rewrite rewrote rewritten
typewrite typewrote typewritten
underwrite underwrote underwritten
 chapter 6

Adjective and adverb

inflection

6.1 Prospectus
The inflection of adjectives and adverbs in English is confined to the marking of the morphosyn‑
tactic category ‘degree’ with positive, comparative, and superlative forms. This category also often
goes under the name of ‘comparison’ (e.g. Quirk et al. 1985), and we will use both terms inter‑
changeably. Following standard treatments we consider comparison in English a case of inherent
inflection (e.g. Haspelmath 2002: 67, Booij 2007: 101, see Section 24.2 for discussion). Similar to
the genitive, comparison of adjectives and adverbs involves two different structural realizations
for the comparative and the superlative, respectively. One option is suffixation by ‑er (compara‑
tive) or ‑est (superlative), the other a syntactic construction with more (comparative) or most
(superlative) preceding the adjective or adverb. We will refer to the suffixed realizations using the
terms affixal, synthetic, or morphological, and the terms periphrastic, analytic, or syntactic for
the constructions involving the degree markers more and most. Unlike Quirk et al. (1985) and
Pullum and Huddleston (2002) we will not restrict the term ‘inflectional’ to the affixal realiza‑
tion, since we recognize degree as an inflectional category that, similar to the inflectional category
of genitive (see Chapter 7), is either expressed morphologically or syntactically.
We will first discuss the semantic issues involved with comparison, then discuss the prop‑
erties of affixal degree-formation, before turning to a detailed description of the comparison
of adverbs and adjectives, including a detailed treatment of the determinants of the choice
between synthetic and analytic forms of comparative and superlative with adjectives. Due to
our focus on morphological issues, we will not delve into the properties of various syntactic
comparative constructions.

6.2 Semantic issues: Degree and gradability

The meaning conveyed by degree marking is one in which a comparison is made (implicitly
or explicitly) along a scale of values. The positive form denotes an unmarked given value (e.g.
wide), the comparative expresses a higher degree (wider), the superlative the highest degree
104 Adjective and adverb inflection

on the respective scale (widest). Note that the superlative form is sometimes used as an intensi‑
fier, in the sense of ‘very’, as in (1), from COCA. In attributive position this particular reading
of the superlative is often recognizable through the indefinite article, as shown in (1b).

(1) a. Rule Book 2011: “Most impressive,” said Poppins. “Much creative effort…”
b. Live event 8:00 PM EST 2010: Daniel showed extraordinary poise at a most difficult
moment.

Notably, a few ‑er suffixed forms may look like comparatives but do not express the meaning
‘more’ in these forms because they have taken on an extended meaning. They are listed in (2).

(2) earlier ‘before, previously’

later ‘subsequent(ly), afterwards’
lesser ‘insignificant’

In the appropriate context, these forms may, however, be interpreted as regular comparatives
(as in, for example, She arrived earlier than her sister).
It is uncontroversial, and follows from the meanings conveyed by degree marking, that
only those adjectives and adverbs that are gradable, or at least allow a gradable reading, can
have comparative and superlative forms. For an adjective or adverb to be gradable, we must
be able to conceive of it as denoting a quality that is not absolute, but rather present in greater
or lesser amounts along a scale (Rusiecki 1985; Bierwisch 1988a, 1988b, 1989; Pullum and
Huddleston 2002).
Adjectives (and the corresponding ‑ly adverbs) are also sometimes divided into those that
are qualitative and those that are relational, as exemplified in (3):

(3) a. qualitative: red, low, short, wide, fierce, intelligent, demanding, certain, private,
normal, quiet, aware
b. relational: Dutch, architectural, analytic, geographic, coastal, rhomboid, Palestinian

Most qualitative adjectives are gradable, with alleged exceptions being adjectives like dead,
pregnant, prime (as in numbers), male/female, unique. Relational adjectives are more fre‑
quently said to be non-gradable.
It seems, however, that nearly any adjective can be coerced into a gradable reading if it can
be construed as picking out a set of qualities that can be present or absent in various degrees.
So, although herbal typically means ‘made from herbs’ and seems solidly relational, under
appropriate circumstances we can conceive of this adjective as being gradable:

Jon Bonné SFC 2007: It’s subtle, showing a twinge of Pinot Gris ripeness but a more
(4) 
herbal nose: envision damp hay, ripe pear and celery.
 6.3 The affixal comparative and superlative 105

Even adjectives like dead or pregnant are often coerced into gradability, with something more
dead being further on in the process of dying or even more decomposed, and someone more
pregnant being further on in a pregnancy, as shown in (5).

Brimstone 2005: A sleeping man wouldn’t have his torso scorched and caved in upon
(5) 
itself like a burned log. She had seen many dead people during her childhood in
Colombia, and Mr. Jeremy looked deader than any of them.
Ind_Oprah 2003: The more pregnant that I got, the worse the beatings got.

As Pullum and Huddleston (2002) point out, prescriptivists often advise against comparative
and superlative uses of absolute adjectives (more unique, the most unique), but contemporary
English speakers generally find no difficulty in construing such adjectives as gradable.

6.3 The affixal comparative and superlative

As already mentioned, many adjectives and adverbs can form the comparative by affixation
of ‑er and the superlative by affixation of ‑est:

(6) red redder reddest

gentle gentler gentlest
hardy hardier hardiest
serene serener serenest

The comparative suffix is pronounced as /ə/ in non-rhotic dialects and /ər/ in rhotic dia‑
lects. The superlative suffix is pronounced /ɪst/ or /əst/. In order not to unnecessarily crowd
our phonetic transcriptions we will use only /ɪst/when transcribing pertinent forms.
In bases ending in syllabic [l̩], the /l/ sometimes but not always loses its syllabicity and
is resyllabified as the onset of the syllable formed by the affix. We find this loss of syllabicity
with gentle and simple, see (7a), but only optionally with little (7b). This elision and resyllabi‑
fication does not occur with bases ending in syllabic [r] or [n], see (7c). In non-rhotic dialects
the final [r] of the base is pronounced in intervocalic position, that is, when followed by the
suffix.

(7) a. gentle ◆ gentler ◆ gentlest

[ʤentl̩ ◆ ʤentlə ‖ ʤentlər ◆ ʤentlɪst]
simple ◆ simpler ◆ simplest
[sɪmpl̩ ◆ sɪmplə ‖ sɪmplər ◆ sɪmplɪst]
106 Adjective and adverb inflection

b. little ◆ littler ◆ littlest

[lɪtl̩ ◆ lɪtlə ~ lɪtl̩ə ◆ lɪtlɪst ~ lɪtl̩ɪst] (non-rhotic varieties)
[lɪtl̩ ◆ lɪtlər ~ lɪtl̩ər ◆ lɪtlɪst ~ lɪtl̩ɪst] (rhotic varieties)
c. meager ◆ meagerer ◆ meagerest
[mi:ɡə ‖ mi:ɡər ◆ mi:ɡərə ‖ mi:ɡərər ◆ mi:ɡərɪst]
often ◆ oftener ◆ oftenest
[ɒf(t)ən ‖ ɔːf(t)ən ◆ ɒf(t)ənə ‖ ɔːf(t)ənər ◆ ɒf(t)ənɪst ‖ ɔːf(t)ənɪst]

With some bases ending in /ŋ/ we find the addition of /ɡ/ before the suffixes, as shown
in (8):

(8) long ◆ longer ◆ longest

[ lɒŋ ‖ lɔːŋ ~ lɑːŋ ◆ lɒŋɡə ‖ lɑːŋɡər ◆ lɒŋɡɪst ‖ lɑːŋɡɪst]

Other bases that work this way are strong, young, and (for at least some speakers) wrong. For
speakers who allow superlatives in ‑ingest (winningest, sleepingest) (see below), there is no
addition of /ɡ/ before the superlative suffix.
The affixes are consistently spelled <er> and <est>, respectively, with the expected dou‑
bling of a final base consonant following a lax vowel (9a), replacement of final <y> by <i>
(9b), and elision of silent <e> before the suffixes (9c):

(9) a. thin ◆ thinner ◆ thinnest

fat ◆ fatter ◆ fattest
b. dainty ◆ daintier ◆ daintiest
bumpy ◆ bumpier ◆ bumpiest
c. white ◆ whiter ◆ whitest
fine ◆ finer ◆ finest

6.4 Double comparison

Although prescriptivists frown on it, double comparison is widespread in colloquial varieties
of English. In double comparison both the more/most and the ‑er/‑est markers are present, as
illustrated in the examples in (10) from COCA.

ABC_GMA 2006: Is it because pornographers are getting a little more craftier with
(10) 
the, with the, with the Internet
 6.5 Adverbs 107

CBS_Early 2007: if you want to put, like, an ornament in there, like a brooch or some‑
thing like that to make it look a little more dressier.
FantasySciFi 2009: Some of our more brighter engineers did this.
Rolling Stone 1996: I think when I’m most happiest is when I’m actually so tense, so
angry, so mad and disgusted, and I can relieve all that and be happy.

Interestingly, some adjectives are found with ‑er/‑est in this construction that would not
normally occur in the affixal construction. Some examples from COCA are provided in (11).
Such examples are rare, and it is not clear to what extent double comparison is subject to
different constraints from simple comparison.

(11) ABC_Nightline 1997: Who’s the most beautifulest baby in the world?
Associated Press News 1998: Most people probably would have thought of him as the
most wickedest man in town.

Double comparison seems always to be variable, and the difference, if any, between double
and simple comparison for those speakers who use both is not clear.

6.5 Adverbs
Not only adjectives, but also adverbs can undergo degree inflection. Since adverb degree
formation is different from adjectival degree formation in certain respects, we will treat it
separately in this section.
Before doing so, some further clarification is needed as to which kinds of non-adjectival
bases have comparative and superlative forms. Pullum and Huddleston (2002: 533), for
example, state that some determiners and prepositions may also instantiate the morphologi‑
cal category of degree. The pertinent forms are given in (12), with (12a) showing the deter‑
miners, and (12b) the prepositions:

(12) a. few ◆ fewer ◆ fewest

little ◆ less ◆ least
many/much ◆ more ◆ most
b. close ◆ closer ◆ closest
far ◆ farther ~ further ◆ farthest ~ furthest
near ◆ nearer ◆ nearest

The syntactic category of the words in (12) is, however, controversial, and most dictionaries
seem to list them as adjectives or adverbs or both. For the purposes of this chapter we will
108 Adjective and adverb inflection

also treat them as adverbs or adjectives. All non-adjectival forms that show degree formation
have in common that they meet the gradability requirement.
Adverbs in ‑ly differ from adjectives in that they generally do not take the synthetic degree
variants (see Chapter 15 for further discussion of ‑ly adverbs). Thus, apart from rare excep‑
tions, such as the archaic forms boldlier, quicklier, periphrastic comparative or superlative
forms are obligatory. Two cases are illustrated in (13) with examples from COCA:

(13) Our Lady of the Forest 2003: Tom moved more boldly into the doorway.
Time 2001: So it came along on a gold spoon, one from another, and I handed it to the

Queen, and then she tried to tip it out on the Prime Minister’s shoulder, most gently…

Some adverbs that are not formed from adjectives via ‑ly suffixation are attested with both
analytic and synthetic degree forms. Palmer et al. (2002: 1584) list the ones in (14a) as perti‑
nent, but even large corpora do not have the two variant forms for all of them. We found
analytic as well as synthetic variants only for fast, hard, and often in COCA. The less expected
variant of these words is exemplified in (14b), but it seems that in general variation in this
group is very limited.

(14) a. early, fast, hard, late, long, often, soon

b. BNC. The Daily Mirror: The way the English game is evolving it is becoming ever
more fast and physical. We were lacking in that department last year.
 CBS 48 Hours 10:00 PM EST 2010: And when you love that person, it makes it so
much more hard.
Fort Clay, Louisiana 2010: I wish I could come ashore oftener.

There are some ‑ly adverbs that also have unsuffixed variant forms, such as easy ~ easily, loud
~ loudly, quick ~ quickly, slow ~ slowly. Palmer et al. (2002: 1584–5) remark that while the posi‑
tive form of the unsuffixed variant (e.g. loud) seems less frequent than the positive of the
suffixed form (e.g. loudly), the comparative and the superlative seem to prefer the synthetic
variants based on the unsuffixed adverb (i.e. louder, loudest). This point can be nicely illus‑
trated with data from COCA, where we find, for example, only four attestations of talked loud
as against 19 of talked loudly, whereas for the comparative forms there are four instances of
talked louder and only one of talked more loudly. This pattern may also occasionally extend to
other adjectives and adverbs of high frequency, as evidenced in sentences such as That’s easier
said than done (instead of more easily), or Speak clearer! (instead of more clearly) (examples
from Quirk et al. 1985: 465).
A small number of adverbs, some of which we have already mentioned, have suppletive
forms for the comparative and the superlative. Notably, these suppletive forms are identical
to the comparative and superlative forms of the corresponding adjectives.
 6.6 Adjectives 109

(15) a. badly ◆ worse ◆ worse

well ◆ better ◆ best
little ◆ less ◆ least
far ◆ farther ~ further ◆ farthest ~ furthest
much ◆ more ◆ most

The difference between further and farther (and the corresponding superlatives) is not clear
(but see Quirk et al. 1985: 459 for a different view), and deserves further study.

6.6 Adjectives
6.6.1 Irregular paradigms
Among adjectives, there are several irregular paradigms, too. These involve unexpected vowel
alternations or suppletion:

(16) good better best

far farther/further farthest/furthest
bad worse worst
much, many more most
little less least

As indicated above, little has a regular paradigm in contemporary English as well. We also find
the archaic comparative and superlative forms of old, that is, elder and eldest, surviving in a
limited range of contexts. Thus, the form elder is most frequently used as a noun denoting a
senior person in charge of a tribe, society, or church, but in its form as an attributive adjective
it is most often used in designating familial relations (the elder daughter/uncle/grandchild,
etc.) and in the collocation elder statesman. Note that the noun elder also serves as a base for
adjectival derivation in elderly, which, given that inflectional suffixes do not typically occur
inside derivational ones, is an additional piece of evidence that the form with this reading has
lost its status as a comparative. The superlative eldest is almost always used in designating
familial relations.
The adjective bad shows some additional peculiarities. There is a mix of suppletion and
regular suffixation in the non-standard form worser, in which the comparative is thus doubly
marked. We also find the regular forms badder and baddest along with worse and worst, but
usually with a colloquial meaning of ‘impressive’ or ‘formidable’:

Washington Post 2002: Rather like the Sex Pistols, the YBAs get their special status by
(17) 
virtue of what they aren’t. There were proto-punks in the United States that the Pistols
110 Adjective and adverb inflection

took off from, just as there have been much badder bad boys in the American art
world than Damien Hirst or Chris Ofili.
Denver Post 2009: Of golf ’s four major championships, the U.S. Open has long been the

one most likely not only to puff out its chest, but, like the baddest kid on the playground,
actually go out looking for a fight.

There are also adjectives with defective paradigms. For example, although the adverb well has
a superlative form best, the adjective well ‘in good health’ is awkward under comparison:
weller is attested in COCA, we have not attested Nwellest, and best seems impossible.

6.6.2 Affixal versus periphrastic comparative and superlative

As mentioned at the beginning of this chapter, English provides two ways of forming the com‑
parative and superlative of adjectives. The matter has been widely discussed in the literature (e.g.
Quirk et al. 1985; Bauer 1994; Kytö and Romaine 1997; Huddleston and Pullum 2002; Dixon
2005; Hilpert 2008; Mondorf 2009a, to mention only a few), and it has been suggested that many
different factors may play a role. What has to be stated at the outset is that more recent corpus-
based studies have shown that the degree of variability is much greater than standardly assumed
in both descriptively (Quirk et al. 1985; Huddleston and Pullum 2002) and theoretically oriented
approaches (for example, Poser 1992; Kiparsky 2005; Dixon 2005; Embick 2007).
Much of the theoretical discussion of the comparative and superlative concerns the phe‑
nomenon of ‘blocking’, where the existence of the affixal forms of the comparative and super‑
lative is said to preclude the existence of the periphrastic forms. Our study of the corpora
(like others before) indicates, however, that while there may be a preference for one or the
other form, there in fact is no such thing as blocking. This is in line with other inflectional
categories we look at in this book (see Chapter 26 for discussion).
In the most extensive study of the phenomenon, Mondorf (2009a) provides large-scale
empirical evidence for the significant influence of phonological, morphological, lexical,
semantic, syntactic, and dialectal factors in the choice of comparatives. Multivariate studies
such as Szmrecsanyi (2005) and Hilpert (2008) have yielded results along the same lines. We
will discuss these factors in the following subsections. Unfortunately, no comparable studies
have yet been undertaken for the superlative, but it is more than likely that one would find
similar results. Peters (1999) shows for most adjectives in her investigation that comparative
and superlative have similar distributions of analytic and synthetic forms across the two cat‑
egories, but with a minority of adjectives, one finds opposite preferences for the two catego‑
ries involved, with the superlative preferring synthetic forms and the comparative preferring
analytic forms. For an explanation of these patterns a multivariate analysis of pertinent data
would be necessary. In view of this situation we will focus on the comparative, but wherever
possible we will complement the discussion of the comparative with some data and results of
our own, or from the literature, if available, concerning the realization of the superlative.
 6.6 Adjectives 111

6.6.2.1 Phonological factors

There are a number of phonological factors that have been shown to influence the choice of
analytic and synthetic degree variants. The one that is most often mentioned in grammar
books, and which is indeed also the factor emerging with the greatest predictive power in
multivariate analyses (e.g. Hilpert 2008), is the number of syllables. Other phonological fac‑
tors mentioned in the literature are the kinds of base-final segments, and the stress. We will
discuss each in turn.
According to received wisdom, monosyllabic adjectives typically take the affixal forms (as in
red ◆ redder ◆ reddest), disyllabic adjectives vary (e.g. happy ◆ happier ◆ happiest as against defunct
◆ more defunct ◆ most defunct), and trisyllabic and longer adjectives take only the periphrastic
form (e.g. horrible ◆ more horrible ◆ most horrible). While the length of an adjective counted in
number of syllables is generally a good predictor for the choice of the inflectional variant, it has
to be stated that monosyllables can still show variable behaviour. Notably, it is often the case that
both synthetic and analytic comparative and superlative forms are found with the same adjec‑
tive, which means that there is considerable within-type variability. In (18) we give some exam‑
ples that illustrate cases that go against the oft-cited generalization on monosyllabic adjectives.

Virginia Quarterly Review 2002: I worried you would find a girl in the jungle more brave
(18) 
than I, more fierce and valiant.
NPR_Morning 2002: I would say that this is bad, no question about it. But it’s only

slightly more bad than the usual bad situation for flu vaccination.
New York Times 2005: If these people were similarly cute with F.B.I. agents and the grand

jury, they’ve got an obstruction-of-justice problem possibly more grave than the hard-
to-prosecute original charge of knowingly outing a covert agent.
Analog Science Fiction & Fact 2003: She wasn’t ever going to get much more smart than

she already was, but she was so sweet!
San Francisco Chronicle 1999: Of her goofy television persona, she says, “I actually am

more weird than I am on the air, and I have to tone myself down.”

 irtually all one-syllable adjectives have affixal forms, but Palmer et al. (2002: 1583) note the
V
following as exceptions:

(19) cross, fake, ill, like, loath, prime, real, right, worth, wrong

However, all of the above (except worth and loath, both of which are syntactically anomalous
in a number of ways) are attested at least occasionally in the affixal comparative. COCA
yields the following examples:

Virginia Quarterly Review 2000: They bulge with home-canned green beans, Christmas
(20) 
sugar cookies, and a butcher-papered chunk of tenderloin, primest part of the deer.
112 Adjective and adverb inflection

Friends for Life 1995: “Do you suppose she’s all right? Or might she be iller than she

realizes? More important, is there anything we can do? I feel so helpless . . . so responsible
somehow.”
Geographical Review 1998: “That may not be entirely right, but it is turning out to be

righter than I would have thought.”
Atlantic Monthly 1994: “Sounds like shy to me,” my father said, brisk and certain. My father

was certain about everything. “Better come in.” “I’m not shy,” I said again, crosser.

There are more exceptions to the generalizations concerning monosyllabic and polysyllabic
adjectives, but these are morphologically determined and therefore treated in Section
6.6.2.2.
There is, however, another issue involved with the input-based generalization that trisyl‑
labic bases favour the periphrastic realizations of degree. As argued by Mondorf (2003,
2009a), what may really count is not the length of the base form of the adjective, but rather
the length of the inflected form, which may not exceed three syllables. Under this output-
oriented formulation it is possible to predict correctly that trisyllabic bases with final syllabic
consonants may take synthetic comparative and superlative suffixes if the final consonant is
resyllabified as part of the onset of the final syllable. This can account for forms such as
incredibler (COCA), and sensibler. However, it is not clear whether the output-oriented for‑
mulation is really more adequate, as it would predict the occurrence of many similar forms
such as Nprobabler or Nsuitabler. There is of course the additional complication that mor‑
phologically complex adjectives tend to avoid the synthetic variants anyway (see Section
6.6.2.2 for discussion) and that, due to the scarcity of pertinent monomorphemic adjectives,
the amount of data on which the output-based formulation could be tested against the tradi‑
tional input-based formulation is vanishingly small.
The form cartoonier (attested in COCA) may even suggest that what is important is not
the number of syllables but the prosodic shape, that is, that the base ends in a trochee, or, in
output-based parlance, that the synthetic output needs to end in a dactyl. This would elegantly
account for the fact that disyllabic iambs disprefer suffixal degree variants (see below), but,
like so many other proposed rules, would leak considerably. Furthermore, forms like absoluter
(COCA) do not conform to any of the rules we have considered so far.
Let us turn to the next phonological determinant, the final segments. Especially with
disyllabic adjectives, the influence of the final segments of an adjectival base can be of two
different kinds. There are effects that could be analysed as haplology effects, and there are
other effects of the final segment(s) whose motivation is not so clear.
Let us first consider haplology. Bases ending in <st> hardly ever take a synthetic superlative.
For example, none of the 122 instances of superlative honest in Mondorf ’s data has Nhonestest,
and similar results hold for just, moist, modest, robust, and unjust. Another haplology effect can
be observed especially, but not exclusively, in rhotic varieties of English with bases ending in
/r/, such as austere, bare, bitter, dire, mature, pure, sour, sure, slender. They avoid the identical
 6.6 Adjectives 113

onset and coda in a potential synthetic comparative by preferring the analytic variants to a
much greater extent than comparable bases with a phonologically different ending. In general
terms, adjectives ending in /l/ also prefer the analytic comparative, which could be interpreted
as another haplology effect, namely the avoidance of a liquid in both the onset and coda of the
final syllable of the inflected form.
Other phonological material at the end of the base may also have an influence on the
choice of synthetic or analytic variants. Thus, adjectives with final consonant clusters tend to
have a higher ratio of analytic comparatives than comparable adjectives without final clusters.
There have also been claims about adjectives with final high front vowels or /l/ plus a high
front vowel (Hilpert 2008). It seems, however, that these particular phonological sequences
are most often realizations of suffixes (i.e. ‑y and ‑ly, respectively), and we will therefore deal
with these in the section on morphological determinants.
Finally, stress has an influence on the choice of degree realization. Adjectives that end in a
stressed syllable are much more prone to analytic comparatives and superlatives than adjectives
ending in an unstressed syllable (e.g. Leech and Culpeper 1997; Hilpert 2008). Another puta‑
tive stress-related effect on the choice of degree variants is stress clash avoidance, as claimed by,
for example, Kuryłowicz (1964: 15), or Mondorf (2009a: Section 4.1). In this view, synthetic
variants are held to be overrepresented in environments with a stressed syllable following a
finally-stressed adjectival base. However, Mondorf’s own results are statistically inconclusive,
and in multivariate studies this factor did not survive as influential (e.g. Hilpert 2008: 406).

6.6.2.2 Morphological factors

In this subsection we look at the role of the morphological make-up of the base in degree
marking. There is a well-established general effect of morphological complexity at work (e.g.
Mondorf 2003; Hilpert 2008), such that morphologically complex adjectives prefer the peri‑
phrastic degree variants. However, there are some more things to be said about particular
kinds of morphological categories involved in this. We will discuss verbal participles, derived
adjectives, and compound adjectives in turn.
Perhaps the clearest morphologically related effect concerns verbal participles as adjec‑
tives, which generally do not form synthetic comparatives. Cases in point are, for example,
present participles like stunning or winning, and past participles such as blessed or worn. For
illustration of the general pattern consider stunning, which has 66 periphrastic attestations of
the comparative and 160 of the superlative in COCA, but no synthetic attestations at all. The
past participle worn also has no synthetically formed degree forms in COCA.
However, exceptions to these kinds of pattern can be found, as, for example, in the behav‑
iour of winning as attested in COCA. The synthetic superlative winningest has 272 attestations
as against only 28 periphrastic ones. In contrast, there is no synthetic comparative attested for
winning. Interestingly, the synthetic superlatives are all semantically restricted to sports con‑
texts (21a), while all but one periphrastic superlative instantiate the figurative meaning, as
shown in (21b). In (21c) we give an example of a periphrastic comparative.
114 Adjective and adverb inflection

(21) a. AP 2010: Schlossnagle is already winningest coach in school history

b. Arthas: the rise of the Lich King 2009: Arthas gave the girl his most winning smile.
c. Newsweek 1994: Clinton will have to try to project a more winning image.

Violations such as the one in (21a) of the strong general trend towards periphrastic degree
marking seem to be lexeme-specific. One of the more interesting points that emerges from a
study of participles in ‑ing in COCA is that they show a marked tendency towards allowing
the affixal forms in the superlative, at least in NAmE, but not in the comparative. This was
noted by Montgomery (1999) for Appalachian English, but seems not to be restricted to that
dialect of NAmE. Indeed, such forms are mentioned in Jespersen (1942: 353), and illustrated
with literary examples.
Another frequently observed morphological effect is that adjectives containing the prefix
un‑ prefer the synthetic degree forms, irrespective of how many syllables are involved, and
they thus often violate the generalization that polysyllabic adjectives prefer periphrastic
expression of degree. A particular striking example illustrating this fact is given in (22), from
COCA (note that this example also shows coercion of a relational adjective into a gradable
one, see Section 6.2 above):

Basic 2003: # STYLES # Then it’s unofficial. He takes a hit from an ASTHMA
(22) 
INHALER, as a ’71 PONTIAC GTO drives through the gate and pulls up. Hardy
emerges. # OSBORNE # (re: his appearance) Doesn’t get any unofficialer than
that . . . The two old friends embrace.

Going against the general trend of morphologically complex adjectives, those derived with
the suffixes ‑y or ‑ly, as in guilty, hungry, handy, risky, friendly, lively, manly have been found to
(slightly) favour the synthetic variants (e.g. Leech and Culpeper 1997; Hilpert 2008), but
there is massive type-dependent variation in the distributions. For example, in Mondorf ’s
(2009a: 34) data set, hungry has less than 20 per cent analytic comparatives while guilty has
80 per cent analytic comparatives. However, among the violators of the trisyllabic-plus rule
we find a preponderance of ‑y-suffixed adjectives. For finicky, lemony, orangey, slippery, watery
we find attested comparatives like finickier, lemonier, orangier, slipperier, waterier.
Adjectives formed by the suffixation of ornative adjectival ‑ed to multi-word bases, such as
broad-minded, full-flavoured, low-priced appear to be extremely variable across types with
some preferring the analytic, others the synthetic variants. For example, while full-flavoured
or high-priced strongly prefer the synthetic comparative in Mondorf ’s data set, broad-minded
or sure-footed show the opposite tendency (Mondorf 2009a: 48).
With other adjectival suffixes there is a clear trend towards periphrastic comparatives and
superlatives, with some variation in many of the categories. Using COCA, we tested the suf‑
fixes ‑al, ‑an, ‑ant, ‑ate, ‑esque, ‑ful, ‑ic, ‑ile, ‑ish, ‑ist, ‑ive, ‑less, ‑ous, ‑some with a pseudo-random
selection of pertinent adjectives listed in (23) and (24), excluding adjectives from the sample
 6.6 Adjectives 115

with very low frequency, in order to avoid adjectives with no degree-marked forms attested
at all. In (23) we list the frequencies of synthetic and analytic comparative variants separated
by a slash, while (24) gives the respective figures for the superlative variants of these adjec‑
tives. All variants that occurred less than 50 times were checked individually and non-pertinent
forms were manually weeded out. Variants with a frequency of 50 or more were not individu‑
ally checked, so that these figures, indicated by ‘>50’, ‘>100’ or ‘>1000’ in (23) and (24), may
slightly overestimate the frequencies of these variants.
With the data in (23) we can first observe that all adjectives sampled are attested with peri‑
phrastic comparatives. Many of the pertinent adjectives also have the occasional synthetic
comparatives in COCA, though of much less frequency than the corresponding periphrastic
variant (‑ant, ‑ful, ‑ish, ‑ous, ‑some). Some adjectives (for example, those in ‑ant, ‑ate, ‑esque,
‑ist, ‑ive, ‑less) are exclusively attested in COCA with periphrastic comparative forms, but this
does not necessarily mean that synthetic forms of these adjectives are generally impossible.

Comparative choice with selected complex adjectives (suffixal/analytic attestations in

(23) 
COCA)
a. global (0/ >100), mortal (0/4), racial (0/2), verbal (0/34)
b. spartan (0/14), Tuscan (0/2), human (0/ >100), Balkan (0/1), terran (0/1), urban
(0/ >100)
c. brilliant (0/ >50), fragrant (0/17), buoyant (0/25), fluent (1/22), decent (1/29),
stringent (0/ >100)
d. ornate (0/27), prostrate (0/1), private (0/ >100)
e. burlesque (0/4), grotesque (0/23)
f. careful (1/ >100), thankful (1/19), doubtful (0/25), needful (0/2), playful (0/ >50),
spiteful (0/4)
g. cosmic (0/14), ethnic (0/6), gothic (0/3), mystic (0/3), psychic (0/2), scenic
(0/30)
h. foolish (1/40), bullish (0/25), hawkish (0/ >50), purplish (0/1), sluggish (0/24),
youngish (0/1)
i. centrist (0/ >50), racist (0/12), sexist (0/3)
j. active (0/ >100), festive (0/38), fictive (0/1), massive (0/ >100), native (0/8),
pensive (0/11)
k. aimless (0/1), baseless (0/1), harmless (0/6), lifeless (0/1), pointless (0/1), wingless
(0/1)
l. famous (4/ >100), bulbous (0/6), gracious (0/ >50), heinous (0/24), spacious
(0/ >100), wondrous (0/16)
m. awesome (0/22), handsome (35/ >50), lonesome (1/11), fearsome (0/22), irksome
(0/6), loathsome (0/10)
116 Adjective and adverb inflection

Almost all adjectives in the sample are also attested with superlative forms. Among the
superlatives, synthetic superlatives are generally extremely rare, with only three of the
morphological categories having adjectives using it at all (‑an, ‑ous, ‑some). It is with the
last suffix that we find a synthetic degree form variant that outnumbers the analytic ones
(handsome) or that reaches at least the same frequency (lonesome).

Superlative choice with selected complex adjectives (suffixal/analytic attestations in COCA)

(24) 
a. global (0/10), mortal (0/4), racial (0/1), verbal (0/4)
b. spartan (0/2), Tuscan (0/0), human (1/49), Balkan (0/0), terran (0/0), urban (0/9)
c. brilliant (0/ >100), fragrant (0/23), buoyant (0/4), fluent (0/3), decent (0/43),
stringent (0/ >100)
d. ornate (0/15), prostrate (0/0), private (0/ >100)
e. burlesque (0/0), grotesque (0/16)
f. careful (0/>50), thankful (0/25), doubtful (0/9), needful (0/5), playful (0/12),
spiteful (0/0)
g. cosmic (0/2), ethnic (0/0), gothic (0/0), mystic (0/1), psychic (0/0), scenic
(0/>100)
h. foolish (0/18), bullish (0/6), hawkish (0/18), purplish (0/0), sluggish (0/8),
youngish (0/0)
i. centrist (0/5), racist (0/20), sexist (0/3)
j. active (0/ >100), festive (0/18), fictive (0/0), massive (0/ >100), native (0/1),
pensive (0/0)
k. aimless (0/0), baseless (0/1), harmless (0/7), lifeless (0/2), pointless (0/5), wingless
(0/0)
l. famous (1/ >1000), bulbous (0/0), gracious (0/ >50), heinous (0/ >50), spacious
(0/11), wondrous (0/30)
m. awesome (1/49), handsome (94/75), lonesome (4/4), fearsome (0/>50), irksome
(0/5), loathsome (0/11)

Let us now turn to compounds, which have been claimed to take only periphrastic degree
expression (e.g. Bauer 1988: 134). This claim is problematic, however. Different cases can be
distinguished, depending on the position of the adjective(s) involved. Adjectival compounds
with a present participle as head and an adjective in the left position (e.g. easy-going, fast-moving,
long-standing) can indeed take analytic degree forms on the adjective in first position (e.g. more
easy-going) or show affixal inflection on the initial adjective (easier-going). Which pattern is
preferred varies across compounds, with some compounds strongly preferring one variant, for
example analytic for easy-going, others preferring the other variant, for example synthetic for
fast-moving (see Mondorf 2009a: 45 for the distributions of individual adjectives).
 6.6 Adjectives 117

A similar situation can be found with other compounds that have other adjectival heads.
They seem to be able to take both kinds of variants, although our corpora attest this for only
a minority, which is not surprising given the general very low frequency of such compounds.
In (25a) we list some synthetic forms, in (25b) some periphrastic forms, all from COCA.

(25) a. bottom-heavier, bug-friendlier, butt-uglier, eye-easier, self-esteemier, sing-songier,

smart-assier, squeaky-cleaner, sugar-plummier
b. more accident-prone, more broad-based, more environment-friendly, more
sing-songy, more world-weary

There are also structures with adjectives in the first position and nouns in the second posi‑
tion that are sometimes analysed as compounds and which pose the question of how to grade
the adjective in these structures, for example hard-line, large-scale, left-wing, low-key, or short-
term. As shown by Mondorf (2006, 2009b), both types of variant are common in this group,
very often attested on the same base, as illustrated for hard-line in (26), from COCA:

(26) a. USA Today 1991: these changes of command might instead be signalling a shift to an
even harder-line administration
b. CNN_Q&A 2003: Can those settlers, the more hard-line settlers, remain where
they are in places like Nablus and Hebron?

The comparison of compounds presents a serious challenge to stratal approaches to the lexi‑
con in which compound-formation is ordered at a later derivational stage than affixation, and
in which the insertion of an affix into a compound, as in (26a) is predicted to be impossible.
Furthermore, with cases such as those in (25a) we are dealing with level ordering paradoxes,
as the semantic scope of the comparative suffix is the whole compound, while comparative
suffixation to such long polysyllabic adjectives should be impossible for prosodic reasons
(see Booij and Rubach 1984; Kiparsky 1982c; Selkirk 1982: 101; Booij 1993: 36–41 for other
level ordering paradoxes).

6.6.2.3 Lexicalization
The preceding discussion has already shown that preferences towards the respective variants
are frequently lexeme-specific, and there are often no clear explanations available why one
particular adjective would be more prone to periphrastic comparison, while another, quite
similar adjective would prefer synthetic variants. Nevertheless, it is possible to find some
patterning in this apparently confused picture, using the notion of lexicalization. The degree
of lexicalization can be conceptualized as the strength of the lexical representation of a par‑
ticular lexeme and its forms. Frequency of occurrence and spelling have been successfully
used as correlates of this notion of lexicalization (e.g. Sepp 2006; Plag et al. 2007, 2008 for
compounds, Mondorf 2000, 2009a, 2009b for the comparative).
118 Adjective and adverb inflection

Following up on earlier claims by many authors (e.g. Sweet 1891: 327; Bolinger 1968: 120;
Gnutzmann et al. 1973: 425, 434; Quirk et al. 1985: 463), Mondorf (2009a) and Hilpert (2008)
provide ample empirical evidence that frequency plays a role in the choice of the degree variants.
Adjectives with higher frequency tend to have higher proportions of synthetic variants, and the
same holds for adjectives with a higher ratio of comparative forms as against positive forms. Note
that Hilpert’s multivariate model proves that the observed frequency effect is not the conse‑
quence of the fact that more frequent adjectives also tend to be shorter, and hence more prone to
synthetic comparison, but that it is an independent effect alongside the length effect.
Spelling can be used as an indicator of lexicalization as more fused spellings tend to indicate
the writers’ perception of the structure in question as a single lexical item (see again the refer‑
ences cited above for discussion). Mondorf (2009a, 2009b) now shows that with decreasing
orthographic (i.e. lexical) fusion the proportion of synthetic comparatives increases for all
types of pertinent compounds. That is, with two-word spellings such as high risk, full flavoured
we find a higher proportion of the synthetic variant (higher risk, fuller flavoured) than with
hyphenated or one-word spellings (higher-risk, higherrisk, fuller-flavoured, fullerflavoured).

6.6.2.4 Semantic factors

Semantics also acts as a determinant in comparison realization. Mondorf (2009a) provides
data that show that with abstract readings of monosyllabic adjectives (as for clean in a clean
break), the chances of periphrastic comparative formation increase dramatically, overriding
the prosodic constraint against such forms.
Another semantic effect seems to be that adjectives that appear to be less easily construed
as gradable appear to be more prone to choose periphrastic comparatives. The comparison of
the adjective right is a case in point with 80 per cent of more right (as against 20 per cent of
righter) in Mondorf ’s data (2009a: 12).

6.6.2.5 Syntactic factors

It has also been shown that certain syntactic environments may favour particular degree vari‑
ants. Thus, the presence of to-infinitival complements, prepositional complements, premodi‑
fication of the adjective, and predicative (as against attributive) position all increase the
chances of a periphrastic comparative or superlative (e.g. Leech and Culpeper 1997; Lindquist
1998; Mondorf 2003; Hilpert 2008; Mondorf 2009a). Examples from COCA illustrate this
with some monosyllabic adjectives, which might be expected to prefer the synthetic variant.

(27) a. Parenting 2010: “Children are more apt to stay seated if the parent simulates a car
trip,” she says.
b. PBS_Newshour 2009: We couldnt [sic] be more proud of her as a family.
c. Devil red 2011: Leonard leaned over him. “Let me make it even more clear.”
d. NBC_MeetPress 2010: I have never been more proud.
 6.6 Adjectives 119

Mondorf (2009a) also finds some evidence that there is a tendency for sentence-final as
against sentence-medial positions to increase the likelihood of periphrastic comparison and
attributes this to the fact that, being one word longer, this tendency is to do with creating
end-weight in the pertinent sentences. It is unclear, however, how this effect relates to the
attributive/predicative position effect and whether it would survive in a multivariate analysis.
Finally, Szmrecsanyi (2005) has found that the use of the word more in the discourse pre‑
ceding the comparative choice context in question increases the chances of a periphrastic
comparison, similar to what we see happening in the genitive alternation (see Chapter 7). An
example from the BNC showing this effect is presented in (28), where more friendly is chosen
instead of the competing friendlier.

Broadcast news: the atmosphere was totally different, much more relaxed, much more
(28) 
friendly

6.6.2.6 Dialectal factors

There also seem to be differences in degree marking when it comes to different varieties of
English. Several studies (e.g. Kytö and Romaine 2000; Mondorf 2009a) have found, for
example, that NAmE is more prone to analytic comparison than BrE. The reasons for such
differences are largely unclear, and, to our knowledge, other varieties have not been investi‑
gated systematically.

6.6.2.7 Summary
To summarize, a large number of factors have been found to be influential in the choice
between analytic and synthetic forms of degree marking. Overall these factors generally do
not lead to categorical contexts but rather increase the chances of the occurrence of one vari‑
ant over the competing alternative realization. From a theoretical standpoint this is very
interesting as it means that the realization of an inflectional category is probabilistic, instead
of being grammatically or lexically strictly determined. While this situation is very similar to
the genitive alternation (see Chapter 7), it seems typologically a rather rare phenomenon.
Another theoretical issue is whether there may be any more general mechanism that unites
the different determinants. Applying Rohdenburg’s (1996) Complexity Principle, Mondorf
(2009a: 6) argues for such an overarching mechanism and states that ‘[i]n cognitively more
demanding environments which require an increased processing load, language users tend to
make up for the additional effort by resorting to the analytic (more) rather than the synthetic
(‑er) comparative.’ Although this seems to be an appealing explanation for a number of the
variables we have discussed, it also raises a number of problems that cast some doubt on the
generality of this approach, or on its rationale.
The underlying assumption for making the Complexity Principle work with degree forma‑
tion is that analytic forms are generally easier (or should we say more easy?) to process than
synthetic forms, all other things being equal. If this were generally the case, it is, for example,
120 Adjective and adverb inflection

unclear why certain rather complex adjectives (e.g. those prefixed by un-, see above) or cer‑
tain more complex constructions with adjectives, for example those involving a complement
introduced by than (see, for example, Hilpert 2008), would still prefer the synthetic variant.
So, as far as we can see, there is no independent psycholinguistic evidence for the assumption
that analytic degree structures are more easily processed than synthetic ones. And even if this
assumption were correct, why would the synthetic variants persist as a stable feature of the
language? A final problem is that it is unclear whether ‘processing complexity’ is the same in
production and perception. In principle, what could be advantageous in production may be
detrimental in perception, or the other way round. At present, there is only indirect support
for the Complexity Principle operating in degree formation.
 chapter 7

Noun inflection

7.1 Prospectus
In this chapter we discuss two morphological processes: the marking of plurals and the mark-
ing of possession. Plural marking shows a great deal of variation and irregularity, which con-
trasts with the relative regularity of possession marking. Some argue that possession marking
is not strictly inflectional, but whether it is or not, it interacts with plural marking in ways
which suggest that the two need to be dealt with together.

7.2 Plural
The English number system has two values, usually termed singular and plural, but more
accurately termed unmarked and plural. The label ‘unmarked’ is preferable because there are
several cases where the unmarked form is used even though there may be reference to several
entities either implicitly or explicitly. Consider the examples in (1).

(1) a. The tiger is a fierce animal.

b. Tiger skins can fetch a great deal of money.
c. More than one tiger was seen.
d. Every tiger is carnivorous.
e. %It is three mile long.

Typically the semantically unmarked form (the ‘singular’) is also morphologically unmarked,
while the plural form is indicated in some way. We will first discuss the semantics of plurali-
zation, then deal with the various formal ways of marking plural, starting with the regular
suffix ‑s (variously pronounced). Then other ways of marking plurals and the less regular
cases will be considered.
122 Noun inflection

7.2.1 Semantics of pluralization

It might be thought that the semantics of pluralization is simple: plural forms denote ‘more
than one’. But there is more to be said: on the one hand, we can mean a number of things by
‘more than one’, and, on the other, plurals do not always denote ‘more than one’ in a straight-
forward way.
English inflection being rather impoverished, the plural marking can be deployed to cover
a number of different concepts that in other languages might be distinguished morphologi-
cally. For example, a plural noun like poodles is interpreted as generic, as in (2a) or equally
well as collective or distributive in (2b):

(2) a. Poodles never shed.

b. The poodles ate the meat.

That is, (2b) can be interpreted as several poodles eating a piece of meat all together, or each
of several poodles eating its own piece of meat sequentially. English leaves it to context to
disambiguate such nuances.
We must also mention here several sorts of form–meaning mismatches that can be found
in English. The first is collective nouns such as people, police, folk, cattle, poultry, livestock, ver‑
min. There is psycholinguistic evidence (from agreement and pronominalization) that these
are semantically plural in spite of lacking any morphological marking (e.g. Bock et al. 2006).
Unlike forms like sheep or fish the collectives do not have a singular interpretation as well.
However, with collectives there is a potential mismatch between notional and grammatical
plural. The test case for this problem is subject–verb agreement, which shows whether the
subject head noun is lexically specified for the morphosyntactic feature of singular or plural.
Although there are a few collectives such as people that invariably trigger plural agreement
(with determiners, verbs, and pronouns), there is a lot of variation with regard to these lexi-
cal specifications across speakers and varieties. In general, BrE tends to show more plural
agreement than NAmE (e.g. Johansson 1979; Bock et al. 2006). Compare, for example, BrE
(3a) with NAmE (3b), from COCA. Note, however, that more recently, BrE seems to be
developing in the direction of NAmE (Bauer 1994).

(3) a. I don’t think the Royal Family are known for their intelligence. (Bock et al. 2006: 64)
b. CNN_LiveToday 2004: The royal family is not too happy about the publicity.

Other nouns may be marked for plurality, but do not necessarily designate plurals. Among
these are what Payne and Huddleston (2002: 340) call ‘bipartites’, that is, nouns that desig-
nate objects characteristically comprised of two parts. Among these are words denoting
articles of clothing worn on the lower part of the body, utensils with two inter-working parts,
and items associated with vision:
 7.2 Plural 123

(4) a. clothing: bermudas, bloomers, boxers, breeches, briefs, BVDs, capris, chinos,
corduroys, drawers, dungarees, flannels, galluses, jeans, khakis, knickers, leggings,
long johns, overalls, pajamas, pants, pantaloons, panties, pedal pushers, plus fours,
shorts, skivvies, slacks, suspenders, tights, tighty-whities, trousers, trunks, undies
utensils: bellows, callipers, clippers, cutters, forceps, pincers, pliers, scales, scissors,
b. 
secateurs, shears, snippers, tongs, tweezers
c. vision: bifocals, binoculars, glasses, goggles, spectacles

It appears that at least some of these can also be used without the final ‑s. We find this espe-
cially when the word functions as a premodifier to another noun, as in pant legs, or goggle lens,
but we also find the unmarked base in canonical head position of noun phrases:

Prevention 2009: Comb brows upward using a brow brush; trim stragglers straight across
(5) 
with a cuticle scissor. Then, with a slant-edged tweezer, pull strays from in between
eyebrows (the inner corner of your brow should line up with the inner corner of your
eye).
NBC_Today 2007: Don’t be afraid to layer it over a pretty blouse or a turtleneck, but
whatever you’re layering it over, make sure that’s sort of thin. And we love it over a
legging and a flat boot.

The question then arises how bipartites such as these are themselves pluralized. Payne and
Huddleston (2002: 341–2) remark that, ‘There is some uncertainty and variation among
speakers as to how readily they [viz. the bipartites] can also be applied to a plurality of such
objects’, but we find some regularity in the corpus data. Normally, to pluralize the bipartites,
it is necessary to use a partitive construction with a quantifier: two pairs of jeans, a set of
tweezers. We do not seem to find phrases like two leggings or three tweezers. However, we do
occasionally find these bipartite plurals used in lists of objects that suggest that a plural sense
is intended:

Harpers Bazaar 2010: The aesthetician used an array of products by Yon-Ka and Sonya
(6) 
Dakar—exfoliants and cleansers and masks—and all the regular tools: tweezers,
lancets, cotton pads.
Atlanta Journal Constitution 2010: Small sharp objects such as scissors, tweezers, nail
clippers and small tools may be carried onboard.

We turn now to other sets of nouns that might fall under the designation of pluralia tantum,
that is, nouns that have a plural form but no corresponding singular. As Acquaviva (2008:
ch. 2) points out this does not seem to be a single coherent designation. Looking first at for-
mal criteria, we can identify one group of pluralia tantum that end in the plural ‑s but lack a
corresponding singular base form without the ‑s (doldrums, dregs, shenanigans, hustings).
124 Noun inflection

Then, there is also a second group of nouns in ‑s which do have corresponding singular forms,
but where those forms have an entirely distinct meaning. These in turn can be divided into
those whose base is another noun (accommodations, damages, spirits) and those whose base
appears to be non-nominal (amends, belongings, eats, linguistics, regards, greens). Among the
forms on non-nominal bases, we frequently find adjectival forms in ‑ic (see (7a) below) and
‑able (e.g. adjustables, burnables, buyables, compostables, fashionables, floatables, gardenables,
honorables, immovables, see COCA for many more). In cases where the affixation of plural ‑s
appears to be category-changing, we might begin to wonder whether what we have is a sepa-
rate derivational affix, rather than the plural suffix.
We can also look at pluralia tantum from the point of view of semantics; that is, we find
several coherent semantic fields in which pluralia tantum are relatively common, including
names of academic fields (7a), diseases (7b), pseudo-diseases, or psychological states (7c),
and names for games (7d). There are, however, many miscellaneous items that don’t seem to
fit into neat semantic categories (7e). Some of the items in (7) behave like plurals morpho-
syntactically, in that they trigger plural subject–verb agreement, or are pronominalized by
plural pronouns. Examples are culled from Payne and Huddleston 2002, Quirk et al. 1985, and
our own files:

(7) a. Names for subjects: academics, acoustics, acrobatics, classics, cybernetics, economics,
ethics, genetics, gymnastics, informatics, linguistics, mathematics, phonetics, physics,
politics, pragmatics, semantics, statistics
Diseases: bends, chills, hemorrhoids, hiccups, hives, measles, mumps, piles, rabies,
b. 
rickets, shingles
Pseudo-diseases or psychological states: creeps, heebie-jeebies, jim-jams, jitters, runs,
c. 
shakes, shits, squirts, willies
d. Games: billiards, checkers, craps, darts, dominoes, draughts
Other pluralia tantum: accommodations, amends, annals, archives, arms, arrears, arts,
e. 
ashes, auspices, balls, bangs, banns, bowels, brains, civvies, clothes, confines,
congratulations, contents, credentials, customs, damages, directions, doldrums, dregs,
droppings, druthers, dues, eats, entrails, environs, fatigues, funds, goods, grassroots,
greens, grits, groceries, grounds, guts, hots, humanities, hustings, innards, intestines,
irons, letters, lodgings, looks, manners, minutes, news, nuptials, odds, outskirts, pains,
particulars, premises, quarters, regards, relations, remains, rights, roots, savings,
shenanigans, smarts, spirits, stairs, stays, surroundings, talks, teachings, thanks, trappings,
trimmings, troops, tropics, valuables, victuals, vittles, wages, wits, woods, writings

In some cases we have items that are just part of fixed idioms: to be in the doldrums, to have
one’s druthers, to have the hots for, etc.
A final topic we might raise here is the pluralization of non-count nouns. Non-count nouns
typically designate substances treated as undifferentiated (milk, oxygen), abstractions (truth),
 7.2 Plural 125

or aggregates of items, either uniform (rice) or variable (furniture), whose boundaries are not
conceptually salient. Just about any non-count noun can, however, be pluralized if it can be
reconceptualized as a bounded entity, as happens, for example, when we conceive of milk in
a container, or different types of rice; in cases such as these it is perfectly natural to think of
milks or rices.

New Yorker 2006: He tossed one of the milks in the air and tried to shoot it on the
(8) 
wing, . . .
Vegetarian Times 1998: Rice comes in the familiar brown and white varieties, but now
numerous aromatic rices are also sold in grocery stores.

There are very few non-count nouns that are not amenable to this sort of reconceptualiza-
tion, although Acquaviva (2008: 15) cites the word fun as one that is highly resistant.

7.2.2 Regular marking of the plural

In this section, we look at the regular marking of the plural from the point of view of orthog-
raphy (see also Chapter 4) and pronunciation (see also Chapter 9).

7.2.2.1 Spelling
The regular plural suffix is <‑s> added to the default form (9a). However, if the default
form ends in any one of the letters/letter sequences <ch, s, sh, x, z> then the suffix takes
the form <‑es> (9b).

(9) a. car ◆ cars

hat ◆ hats
translator ◆ translators
b. church ◆ churches
dish ◆ dishes
fox ◆ foxes
mass ◆ masses
waltz ◆ waltzes

If the suffix is <‑es> by this rule and the default form ends in a single <z>, then that <z> may
be doubled according to the normal rules of consonant doubling.

(10) fez ◆ fezzes

quiz ◆ quizzes
126 Noun inflection

With final <s> there is some variation, in that one can find forms that show doubling of the
<s>. In most cases the spellings showing doubling are clearly minority choices (e.g. <busses>,
<gasses> as against the much more common <buses>, <gases>). Most of the words with
final <s> have the final <s> in an unstressed syllable, where consonant doubling is less likely
anyway.

(11) alias ◆ aliases

atlas ◆ atlases
bus ◆ buses, busses
canvas ◆ canvases, canvasses
dais ◆ daises
f(o)etus ◆ f(o)etuses
gas ◆ gases, gasses
ibis ◆ ibises
iris ◆ irises
pancreas ◆ pancreases
plus ◆ pluses, plusses
rhinoceros ◆ rhinoceroses

Where the default form ends in <y> immediately preceded by a consonant letter, the plural
is normally made by replacing the <y> with <i> and then adding <es>.

(12) academy ◆ academies

buggy ◆ buggies
fancy ◆ fancies
hierarchy ◆ hierarchies
ideology ◆ ideologies
lullaby ◆ lullabies
mummy ◆ mummies
policy ◆ policies
sky ◆ skies
spy ◆ spies
sympathy ◆ sympathies
tabby ◆ tabbies

Where <y> is part of a vowel digraph, it is normally retained, and the plural with <s> is
formed:
 7.2 Plural 127

(13) buoy ◆ buoys

convoy ◆ convoys
day ◆ days
guy ◆ guys
railway ◆ railways
Thursday ◆Thursdays
X-ray ◆ X-rays

Where the <u> preceding the <y> is part of a <qu>, because the <y> is not part of a vowel
digraph, the plural with <ies> is still used:

(14) obloquy ◆ obloquies

soliloquy ◆ soliloquies

Proper nouns ending in <y> usually retain the <y> in plural forms.

(15) the four Marys

the two Germanys ~ the two Germanies
Say five Hail Marys.
How many Bettys do you know?
Of all the Februarys I remember, this has been the coldest.

There are no general rules for determining the plural form of words ending in single <o>
(words ending in <oo> like igloo, kangaroo, zoo, etc. make their plural regularly by adding
<s>). Some such words have an <s> plural (16a), others have an <es> plural (16b), and yet
others allow both (16c). Note that the (16a) words include those which have a vowel letter
before the <o>, abbreviations, and the names of peoples.

(16) a. albinos, avocados, bimbos, boleros, bonobos, boyos, cameos, casinos, cellos (also
celli), Chicanos, combos, curios, dos (‘parties’), dynamos, egos, embryos, Eskimos,
Filipinos, garbos (AusE), gigolos, gizmos, hairdos, impresarios, journos (AusE),
jumbos, kilos, memos, merinos, milkos (AusE), photos, placebos, pros, radios,
ratios, Romeos, silos, solos (also soli), sopranos (also soprani), stilettos, studios,
tangelos, tempos (also tempi), torsos, trios, turbos, vaqueros, videos, weirdos,
winos
b. 
buboes, dominoes, echoes, goes, heroes, negroes, noes, potatoes, tomatoes,
torpedoes, vetoes
c. archipelagos/archipelagoes, banjos/banjoes, buffalos/buffaloes (also buffalo, see
below), cargos/cargoes, desperados/desperadoes, flamingos/flamingoes, frescos/
128 Noun inflection

frescoes, grottos/grottoes, halos/haloes, hobos/hoboes, innuendo/innuendoes,

mangos/mangoes, mottos/mottoes, peccadillos/peccadilloes, salvos/salvoes,
volcanos/volcanoes, zeros/zeroes

The words in (16b) are all from the seventeenth century or earlier, while nineteenth-century
or later Italian and Spanish loans are in category (16a).
While there are many sub-cases of the words listed in (16a) which can be accounted for by
assigning them to particular categories, it seems that in the twenty-first century the default is
simply to use an <s> plural, and that learning spellers or learners of EFL need to learn the list
of exceptional items in (16b). The existence of so many <es> plurals here (or at least potential
cases) suggests that this has not always been the case, and that the <es> plural was once the
default on nouns ending in <o>.
There have been, and to a lesser extent continue to be, various fashions for using <’s>
rather than just <s> to mark the plural. Various subtypes can be recognized here. These vary
in how common they are. The ones listed in (17) and (18) can be found quite frequently in
our sources.

Things which are not thought of as being proper words (symbols, numbers, letters,
(17) 
abbreviations):
Mind your p’s and q’s.
Her academic record was full of A’s.
He has several β+’s; the 1960’s.
They all have M.A.’s.
I always use &’s in references.
My 1’s are just like my 7’s.

(18) Words which are being used as nouns because they are being mentioned:
If if ’s and but’s were apples and nuts.
There are too many and’s in this sentence.

Note that the linguist’s usage of italicizing instances of mention does not do away with the
possibility of using apostrophes in these cases.

(19) Things which are perceived as being marginal words:

I’ve been to too many of these do’s.

(20) Foreign words which end in a vowel letter other than <e>:
pizza’s, toga’s, wadi’s, piano’s, guru’s, ju-ju’s

(21) Occasionally with proper names ending in <s>:

keeping up with the Jones’s (or Joneses)
 7.2 Plural 129

In twenty-first-century usage all of these are rare in standard usage, though the apostrophe is
sometimes maintained to distinguish a’s from as, i’s from is. In non-standard usage, the type
illustrated in (20) persists, and is often extended to words which do not end in a vowel at all.
This is often termed the greengrocer’s apostrophe, since it is seen on market stalls and in
shops advertising wares such as pear’s, apple’s, and even, notoriously, golden deliciou’s. Such
apostrophes are simply considered wrong by prescriptivists.

7.2.2.2 Pronunciation
The regular sibilant suffix has three allomorphs: /s/, /z/, and /ɪz/ or /əz/ (depending on the
variety of English—we will use the transcription with /ɪ/ here). The allomorph /ɪz/ follows
base-final sibilant consonants (/s, z, ʃ, ʒ, ʧ, ʤ/), the /s/ follows any other base-final voiceless
consonants (/p, t, k, f, θ/), and the /z/ variant is the elsewhere variant, following all other
base-final voiced obstruents, all sonorant consonants and all vowels.

(22) base base final segment base base final segment base allomorphy
+ allomorphy + allomorphy
gas /s + ɪz/ cap /p + s/ home /m + z/
topaz /z + ɪz/ cat /t + s/ gun /n + z/
dish /ʃ + ɪz/ track /k + s/ song /ŋ + z/
rouge /ʒ + ɪz/ chief /f + s/ bottle /l + z/
church /ʧ + ɪz/ wreath /θ + s/ dog /ɡ + z/
judge /ʤ + ɪz/ spa /ɑː + z/
war /ɔː + z ‖ r + z/

This system is problematic for some nouns that show base allomorphy with the plural, as in
thieves, paths, and houses. This type is dealt with in the next section.

7.2.3 Regular suffix plus change in the base

Some nouns which end in voiceless fricatives make their plural form not only with the rele-
vant allomorph of the regular suffix, but with an additional change to a stem-final voiced
fricative. Thus the plural of thief with a final /f/ is thieves, with a /v/ and consequently a final
/z/. There are no instances of this happening with a final /ʃ/, so there are three sets here: final
/s/, final /f/, and final /θ/. They are all slightly different.
There is only one noun which shows a base alternation between /s/ and /z/ in the plural,
and that is house /haʊs/, houses /haʊzɪz/. Even this is not universal: in ScE the plural is regu-
lar: /haʊsɪz/.
Plurals for nouns ending in /f/ fall into three categories: (a) those that always change to
/v/, (b) those that variably change to /v/, and (c) those that never change to /v/. In most
cases, any phonetic change is reflected in an orthographic change.
130 Noun inflection

(23) Always /v/ Variably /v/ Never /v/

calves behalfs, behalves briefs
elves dwarfs, dwarves chiefs
knives halfs, halves fifes
lives hoofs, hooves oafs
loaves roofs, rooves proofs
sheaves scarfs, scarves reefs
shelves wharfs, wharves reliefs
thieves safes
wives serfs
wolves waifs

In addition to the examples in (23) note beef ◆ beeves (old-fashioned, poetic, or dialectal for
‘cattle’). Also note, staves, now considered the plural of stave, and thus a regular form, is ety-
mologically the plural of staff. Turves as the plural of turf is now extremely uncommon, since
turf is usually taken as an non-count noun. The form halfs is more likely to occur in the term
%
centre halfs or in %three halfs of bitter than in the expression a game of two halves. Some of
these forms are prescribed as part of the standard, for example roofs, though there is still a
great deal of variation.
The origin of this alternation is that the voiced fricatives were once allophones of the
voiceless fricatives which occurred intervocalically. When the voiced fricatives took on
phonemic status as a result of the influence of French on Middle English, the alternation
arose. However, since the voiced fricatives were only found intervocalically and in vocabu-
lary which predated the relevant period, words which end in <ff> were not affected, and
later loan words were not affected. Thus words like bluff, duff, gaff, huff, mastiff, midriff, puff,
riff, whiff all take regular plurals, as do aperitif, calif ‖ caliph, chef, clef, graph, leitmotif, serif,
shadoof.
Note that in general it is the common words that undergo fricative voicing. The plurals
in the first column of (23) have an average frequency of over 1300 in the BNC, while those
in the third column have an average frequency of under 200, with the middle column
showing intermediate values. That is, high frequency often has the effect of protecting
morphological irregularity. However, one word which does not fit into the general run of
the irregulars and is of low frequency in this range is sometimes heard with a /v/-plural: it
is the word giraffe, which may have higher frequency in the language of children than in the
language of adults.
Where /θ/ is voiced to /ð/ in the formation of the plural, there is no reflection of the dif-
ference in the spelling. Accordingly, there is little normative pressure on these words. Again
we find words which have the voiced plural, words which fluctuate between the voiced and
the regular plural, and words which show the regular /θ/ plural.
 7.2 Plural 131

(24) a. plurals in /ðz/: booth (for those who pronounce it with final /θ/), mouth, youth
(‘young man’)
b. both plurals found: lath, oath, path, sheath, truth, wreath
plurals in /θs/: berth, birth, breath, cloth, death, depth, earth, faith, growth, heath,
c. 
length, month, moth, smith, strength, tenth (and all similar fractions), youth (‘young
days’), ‑path, hyacinth, myth, shibboleth, zenith, and foreign words in -lith, -path etc.

The word bath has in BrE the plural /bɑːðz/ when it refers to a public swimming pool, but
/bɑːθs/ when it refers to domestic installations. Notice that clothes /kləʊðz, kləʊz ‖ kloʊðz,
kloʊz/ derives historically from cloth, though the relationship is now opaque.
The comment is often made that the regular plural is found when nouns such as leaf are the
heads in exocentric compounds such as the Toronto Maple Leafs (a sports team). Pinker
(1999) claims that this happens exclusively in names. Certainly this is not consistently true of
all exocentrics, as can be seen in (25).

(25) singular form final segments of pluralized form

broadleaf (plant) /fs/
cloverleaf (junction) /fs/, /vz/
cottonmouth (snake) /ðz/
frogmouth (owl) /θs/, /ðz/
loudmouth (person) /θs/, /ðz/
low-life (person) /fs/, /vz/
waterleaf (plant) /fs/

Many of the /θs/ and /ðz/ clusters are variably simplified in all but the most formal styles
with the omission of the dental fricatives (and perhaps some compensatory lengthening of
the alveolar fricative), so that clothes, for example, is regularly /kləʊz ‖ kloʊz/. The /fs/, /vz/
clusters are not subject to the same simplification, but word-final clusters /kst/ and /kt/, as
in text or effect, are: texts and effects are frequently pronounced [teks] and [ɪfeks],
respectively.

7.2.4 Plurals with vowel alternation

A small number of nouns have inherited from Germanic some plural forms that are created
by changing the vowel in the base. These cases are a remnant class of nouns which, at one
stage in their history, had a plural suffix containing [j] or [i], which had the effect of fronting
a back vowel in the base. The process is traditionally referred to as umlaut, and the plurals,
accordingly, as umlaut plurals, although all phonetic motivation for the alternations has long
since vanished.
132 Noun inflection

The relevant nouns in English are

(26) foot /fʊt/ ◆ feet /fiːt/

goose /ɡuːs/ ◆ geese /ɡiːs/
louse /laʊs/ ◆ lice /laɪs/
mouse /maʊs/ ◆ mice /maɪs/
tooth /tu:θ/, %/tʊθ/ ◆ teeth /tiːθ/
man /mæn/ ◆ men /men/
woman /wumən/ ◆ women /wɪmɪn/

Although there is some regularity in spelling here (with <oo> in the base form correspond-
ing to <ee> in the plural and <ou> in the base form corresponding to <i> in the plural), there
is far less of a pattern in the pronunciation. In any case, these are patterns that are not prone
to extension, except in facetious contexts.
Although there is no sign of loss of these irregular plurals in standard literal usage, there
are some places where there is variation in usage. Mouse denoting a piece of computing
equipment, for example, has either mice or mouses as its plural. Consider, too, the following
examples where Mickey Mouse is made plural. Here, the regular plural is more frequent than
the umlaut plural (e.g. with the regular plural having 214 as against 125 irregular plurals in the
155 billion word American English Google Books corpus).

(27) I have never seen so many Daffy Ducks and Mickey Mouses in one place. (McNab
2008: 477)

(28) One specialized in dreadful plastic kitsch; another toys, ten thousand Mickey Mice.
(Rathbone 2001: 26)
Goose in the literal sense usually has the plural geese (and where it does not, we would prob-
ably suspect a slip of the tongue), but note (29) where goose is not used literally.

(29) Todd didn’t need . . . wilder gooses to chase. (Anderson & Benson 1995: 381)
If it turns out that Qatar is innocent . . . our gooses could be cooked. (Sandford 2001: 300)

Similarly, where louse is not used literally (and for the purposes of humour and poetry), we
can find a regular plural:

(30) And that’s when those louses / Go back to their spouses / Diamonds are a girl’s best
friend ( Jule Styne, from Gentlemen Prefer Blondes, 1953)

The plural of sabre-tooth ‖ saber-tooth, which is an exocentric compound, is generally agreed

to be sabre-tooths ‖ saber-tooths rather than ?sabre-teeth ‖ ?saber-teeth, but there does not
 7.2 Plural 133

seem to be a regular pattern here, as can be seen from the other exocentrics in (31), where
the forms are as listed in the OED and Webster’s (W). All are so rare that they fail to occur
in the BNC.

(31) coltsfoot ◆ coltsfoots (W)

flat-foot (sl) ◆ flatfoots (also W and COCA), flatfeet (also COCA)
goosefoot ◆ goosefoots (‘plant’), goosefeet (‘hinge, junction etc.’)
pussyfoot ◆ pussyfoots (also W and COCA)
tenderfoot ◆ tenderfoots, tenderfeet (both in W and in COCA)

The other way round, spice is occasionally heard as the plural of spouse and hice as the plural
of house in joking contexts. Mongoose is sometimes erroneously thought of as containing
goose, and is pluralized as mongeese.
Note that where man is unstressed as the second member of a compound (or something
which was formed as a compound), the spelling of the plural remains <men>, but the pro-
nunciation of the plural is the same as the singular:

(32) postman /pəʊsmən ‖ poʊsmən/ ◆ postmen /pəʊsmən ‖ poʊsmən/

Scotsman /skɒtsmən ‖ skɑːtsmən/ ◆ Scotsmen /skɒtsmən ‖ skɑːtsmən/

7.2.5 Plurals in <n>

There are a few nouns which form their plurals in <n>, remnants of a much larger Middle
English group. The relevant nouns are ox, oxen; brother (‘monk’), brethren; child, children.
Brother (‘male sibling’) has a regular plural in <s>, and even brother ‘monk’ may have a regular
plural.
A very few <n> plurals beyond the standard ones are sometimes heard, usually as jokes.
Vax (the name of a brand of computer) is sometimes pluralized as Vaxen by initiates, a similar
case is boxen (for Unix hardware); sistren is sometimes used in parallel with brethren and
occasionally used in isolation. See the COCA examples in (33).

(33) Southern Review 1994: Oh, I was transported, sistren and brethren. What was I,
a. 
nineteen? And I had a splendid love, and I had a splendid city, and I had a splendid
life
Every dark desire 2007: The girl cut her eyes at Naomi and brushed past her. Or at
b. 
least tried to. ‘Good evening, sistren,’ Naomi greeted in a new purring voice. The
woman inhaled deeply as if smelling something particularly sweet; then she smiled.
People moved in, ebbing around them, some looking at Naomi with interest, others
with sneers.
134 Noun inflection

7.2.6 Unmarked plurals

Unmarked plurals are those where the plural form is the same as the base form. Some authori-
ties refer to these as ‘zero plurals’ (see e.g. Quirk et al. 1985), a term which we avoid since it
raises all the theoretical problems associated with affixes of zero form contrasting with a lack
of an affix. There are several distinguishable types of unmarked plural, which will be treated
separately below, though the classification is not a neat one.
In this treatment we ignore the fact that a noun like cattle is always treated as a plural,
which we assume is a matter of lexis, and we also ignore a structure like six foot three where
we assume that the syntactic construction takes a singular form rather than that foot is an
unmarked plural.
First there are very few nouns which always have a plural form indistinguishable from the
base form, but which do not fit easily into the major categories of unmarked plurals. These
are bob (obsolete BrE sl. ‘shilling’), craft (in the ‘vehicle’ sense), offspring, quid (BrE sl.
‘pound’), sheep, and their compound forms like aircraft. In modern usage, dice belongs in this
category, though Caesar’s Alea jacta est is still often quoted in the form the die is cast, and
speakers for whom this is a living singular–plural alternation have a totally irregular plural
here. Swine also belongs here when being used metaphorically to refer to a person or people:
He is such a swine/They are such swine. When referring to pigs, swine (like cattle) is always
plural.
Second, there are forms which show no change in pronunciation, but do show a change in
the written form. These include forms with reduced ‑man (like postman, Scotsman). In NZE
woman increasingly belongs to this class.
Next there are some semantic classes. The first of these includes the word fish (and its
compounds) and the names of specific fish such as cod, salmon, snapper, trout. There are sev-
eral factors that seem to contribute to whether or not the plural is marked with words for fish:
semantic, formal, lexical. There are, though, no hard and fast rules apparent in usage in this
area.
Semantic factors include the type of sea creature being referred to: words referring to
cetaceans, sharks, and shellfish (unless captured under some other generalization) tend to
have regular plurals (with, for example, lobster and shrimp showing variability), even when
they are being hunted or seen as a source of food: whales, dolphins, hammerheads, mussels,
clams, oysters, scallops etc. The word fish itself is usually used with an unmarked plural, a ten-
dency that becomes even stronger when it is compounded: blackfish, crayfish, dogfish, lemon‑
fish, etc. There is a large lexical component as to whether individual fish names take an
unmarked plural or a marked plural: cod, salmon, trout are clearly used most often with the
unmarked plural, while anchovies, bloaters, eels, herrings, sardines, soles are more likely to have
regular plurals. Even here, though, there is enormous variation. Fish that are named with
hypocoristic forms, whether these arise formally from suffixation or from clipping, always
take regular plurals: crays, kingies, spotties, etc. (Bauer 2009a).
 7.2 Plural 135

Where birds are concerned, there are a number of bird names which can have an unmarked
plural: capercaillie, duck, grouse, partridge, pheasant, pigeon, ptarmigan, quail. The regular plural
here seems to be more likely when several different species are referred to, as in (34), or when
individual birds are referred to.

(34) Determining the age of pinnated and sharp-tailed grouses (Amman 1944)

There is still a lexical element to the choice of plural here. In the expression a brace of ~, the
regular plural is preferred for partridges, pheasants, and pigeons, but the unmarked plural pre-
ferred for grouse and quail (American English Google Books corpus). Grouse when it means
‘complaint’ always takes the regular plural.
Where animals are concerned, the traditional description says that an unmarked plural is
used when wild animals are hunted for food (antelope) or for their hides/skins (mink), not
when they are hunted as vermin (see Sweet 1898: 44–5). If this was the rule, it is not clear to
what extent it still is, though unmarked plurals of many animal names seem more usual in
hunting contexts. This class might account for deer, which is rarely used in the regular plural.
Again, antelopes or elands often mean ‘types of X’, but there is a lot of variation from one lex-
eme to another. Bison, elk, and moose seem particularly rare in the regular plural form, but
take regular plural forms when these words are appropriated as names for human organiza-
tions or teams. Again, this could be the regularization of names.
There is a set of inhabitant words, most of which end in ‑ese, which are generally plural when
they are nouns but which can occasionally still be found used as singular forms as well. The
words include Chinese, Genovese, Japanese, Javanese, Maltese, Milanese, Portuguese, Swiss, Viennese,
Vietnamese, and some more. The most normal use of these words is in a phrase like the Vietnamese
to refer to the set of people who are Vietnamese, and these forms take plural concord:

(35) Her support for the Vietnamese continued all her life. (Dekker 2007)
The Viennese were regarded abroad as a good-natured, easygoing, and highly cultured
people. (Waugh 2008)

However, these words can occasionally also be used as singulars, making them part of the
unmarked plural class:

(36) A Vietnamese has stolen ‘Cambodian’ ID in this manner (http://www.mail-archive.

com/[email protected]/msg08303.html)
The inquiry came from a Japanese called Masami Tobari (http://www.memorial.krsk.
ru/eng/Dokument/Public/20030618.htm)

Such singulars are, though, extremely rare from native speakers of English at the current
period, and are often viewed as deprecatory in some way. The same is true to an even greater
136 Noun inflection

extent of the old, non-standard forms where Chinese, Portuguese, etc. was reinterpreted as
having a plural marker, so that the singular was Chinee, Portuguee, etc.
Other ethnic names can be found either with regular plurals or with unmarked plurals:
Apache(s), Bantu(s), Bedouin(s), Inuit(s) (Inuit is the plural form in the donor language),
Maori(s), Navajo(s), Shona(s). Sioux is spelled the same in the singular and the plural, and
pronounced /suː/ in the singular and /suː/ or /suːz/ in the plural. In Australia, tribal names
usually do not have plural marked, so that the singular and plural forms are homophonous,
for example Eora, Wiradjuri, Yolngu. Exceptions are the Koori(s) and Noongar(s), which have
both marked and unmarked plural forms, for reasons that are not quite clear.
In New Zealand there is an extra set of unmarked plurals, since words borrowed from the
Maori language are often left unmarked. For most nouns, plural is not marked on the noun in
the Maori language, and many Maori people feel strongly that, as a matter of respect for the
Maori language, these nouns should also be left unmarked when they are borrowed into
English. Thus nouns such as kaka, kea, tui (which denote birds) or kauri, pohutukawa, rimu
(which denote trees), and the word Maori itself, when denoting people, are often left
unmarked in the plural. A phrase such as the kauri can thus be ambiguous as to whether it is
singular or plural, even though it would not be ambiguous in Maori (where number would
be marked on the determiner). Although there is a strong prescriptive movement toward
unmarked plurals of Maori nouns in NZE (which may spread to nouns borrowed from other
non-European languages), the use of regular plurals is widespread in informal usage.

7.2.7 Foreign plurals

With a handful of languages, nouns borrowed from those languages come with their foreign
plural forms attached. However, not all nouns from the relevant languages necessarily retain
the plural from the foreign language concerned, and there is a certain amount of variation
about usage. While it is true that there are some nouns which almost always retain the foreign
plural, and others which virtually never use the foreign plural in English, this does not lead to
a useful classification for foreign learners (since the classes are random) or for native speakers
(since even the potentially relevant grouping cannot be discerned from the form of the sin-
gular). The vast majority of foreign nouns can be used with regular English plurals (even
nouns like datum which rarely occurs in the singular), and competing foreign plurals often
survive in specialized discourses. In the following we will discuss a number of donor lan-
guages and the plural forms that words from these languages take in English.

7.2.7.1 Greek
Where words from Greek are concerned, there are two major patterns of concern, illustrated
in (37). The first pattern comprises words ending in <on> that form their plural by replacing
<on> by <a>. The second of these patterns picks up some words which are actually Latin in
 7.2 Plural 137

origin, and some of the Greek words make their way into English via Latin, so ‘Greek’ here is
more of a mnemonic than a strict etymological comment. These words form their plural by
replacing the word-final rhyme by /iːz/.

(37) a. criterion ◆ criteria

 automaton, anacoluthon, criterion, and variably: ganglion, phenomenon,
prolegomenon
b. crisis ◆ crises
acropolis, amanuensis, amniocentesis, anacrusis, analysis, antithesis, apotheosis,
arsis, axis, basis, crisis, diærisis, dieresis, diagnosis, dialysis, emphasis, hypothesis,
metamorphosis, metathesis, nemesis, neurosis, paralysis, parenthesis, periphrasis,
psychosis, syllepsis, symbiosis, synopsis, synthesis, taxis, testis, thesis, thrombosis

A small number of Greek words belong to a slightly different pattern: apsis ◆ apsides, iris (in
the eye) ◆ irides (or a regular plural). Ibis, iris (a flower), mantis, metropolis, necropolis, trellis
have regular plurals. Pelvis, penis may have a word-based regular plural, or a stem-based ‑es
plural. Dais, loris are not Greek words, and have regular plurals.
The spellings <axes> and <bases> are ambiguous, either providing the regular plural of
axe and base or providing the Greek plural of axis and basis. The pronunciations are distinct,
however: /æksɪs/ versus /æksiːz/; /beɪsɪs/ versus /beɪsiːz/.

7.2.7.2 Latin
With Latin words there are four regular patterns, and some exceptional patterns. The regular
patterns are illustrated in (38) to (41). Words like (38) are Latin masculine nouns in ‑us,
which regularly show such plurals in English, especially in technical uses. The pronunciation
of the plural <i> varies between /iː/ and /aɪ/ depending on whether the traditional English
pronunciation of Latin is used or the classical pronunciation of Latin, and not all words vary
in the same way. Octopus is not a Latin noun of the appropriate form, so that octopi, while
extremely common as a plural form, albeit a non-standard one, is etymologically fallacious.
The Greek plural would be octopodes, but the regular English plural octopuses is standard.

(38) alumnus ◆ alumni

 bacillus, cactus, focus, fungus, hippopotamus, locus, nucleus, stimulus, stylus,
terminus

There are some Latin nouns, such as census, coitus, flatus, prospectus, quercus which belong to
a different declension class in Latin and thus do not take an ‑i plural. These normally take a
regular plural in English, if they are used in the plural at all.
Words like (39) are Latin neuter nouns in ‑um.
138 Noun inflection

(39) bacterium ◆ bacteria

addendum, curriculum, datum, desideratum, erratum, forum, labium, millennium,
ovum, referendum, spectrum, stratum, symposium

Words like (40) are Latin feminine nouns in ‑a. Words from this set regularly show the Latin
plural in English, especially in technical uses. There are four different pronunciations of the
suffix attested, and speakers seem to vary a great deal in which word can take each of the four
realizations: /ə ~ eɪ ~ iː ~ aɪ/. There are many foreign words borrowed into English which do
not belong to this class because they are not Latin: algebra is Arabic, drama is Greek, guerrilla
is Spanish, pizza is Italian, and so on. Genera and opera are Latin, but are already irregular
plurals (see (43) below).

(40) larva ◆ larvae

alga, alumna, amoeba, antenna, formula, vertebra

Words like (41) are Latin third declension nouns in ‑x. Of these, appendix and index frequently
also show the regular plural, often semantically distinguished from the Latin plural: appendixes
are parts of the body, not additions at the ends of books, and indexes are parts of books rather
than mathematical signs. These distinctions, however, are not rigorously maintained, and the
plural of appendix in particular is more variable than this simple statement suggests.

(41) matrix ◆ matrices

appendix, codex, cortex, index, matrix, vortex (and words ending in the suffix ‑trix, as in
animatrix, Madonnatrix)

Media, as in news media, is strictly speaking a plural of the kind in (39), the plural of
medium. It is often used as a singular in non-technical English, as shown in the BNC
example in (42).

New Internationalist n.d.: The media has added another dimension to racism by
(42) 
accusing Africans of importing AIDS.

Note the contrast with mediums, which is the plural of medium in the sense of ‘person who
claims to converse with spirits’. Although the artistic term is mixed media, in most other
senses, the two plurals are interchangeable.
There are also a number of Latin plurals which, as far as the grammar of English is con-
cerned, are irregular. Some common ones are listed in (43).

(43) a. corpus ◆ corpora

b. genus ◆ genera
 7.2 Plural 139

c. larynx ◆ larynges
d. opus ◆ opera
e. pharynx ◆ pharynges
f. species ◆ species

7.2.7.3 French
The majority of plurals in French are marked with an ‑s, and thus indistinguishable in writing
from English plurals. Those that require comment, therefore, are those that diverge from this
pattern in some way. There are two main types here, as illustrated in (44).

(44) a. plateau /plætəʊ ‖ plætoʊ/ ◆ plateaux ~ plateaus /plætəʊz ‖ plætoʊz/

b. rendezvous /rɒndeɪvuː ‖ rɑːndeɪvuː/ ◆ rendezvous /rɒndeɪvuːz ‖ rɑːndeɪvuːz/

Words which work like (44a) include adieu, bureau, chapeau, chateau (in French: château),
milieu, tableau, trousseau. Words which work like (44b) and show no change in the spelling
but a change in the pronunciation include chamois, chassis, corps, faux pas, patois, where the
final <s> is not pronounced in the singular.

7.2.7.4 Hebrew
A handful of nouns can take Hebrew plurals in ‑im. The main nouns for which this is found
in English are cherub, goy, kibbutz, seraph. Cherubim and seraphim are often restricted to use
where a biblical reference is intended, while cherubs would be the only possible plural of
cherub meaning ‘small, delightful child’.

7.2.7.5 Italian
There are some apparent Italian plurals in English, but their presence is to some extent mis-
leading. The first set of Italian plural forms are those belonging to musical terms, for example
alti, bassi, castrati, celli, concerti, contralti, libretti, mezzi, soprani, tempi, virtuosi. These are used
exclusively by musicians or people with a close connection to classical music, and the form
celli rarely appears except in musical scores. Such forms are probably better viewed as cases of
code-switching than as genuine English plurals. The regular English plural is always a possi-
bility here.
The next set are food terms such as macaroni, penne, ravioli, spaghetti, tagliatelli, tortellini.
While these forms are Italian plural forms, they act as singular non-count nouns in English,
like rice or porridge. That is, English has Here’s your spaghetti, eat it and not, as would be said
in Italian, eat them. Equally, English has How much spaghetti would you like?, and not *How
many spaghetti would you like?. Lasagna and lasagne (singular and plural respectively in Italian)
are used apparently randomly in English as the name of the dish. Confetti and graffiti, while
140 Noun inflection

not food terms, are used in much the same way as spaghetti etc., though graffito is available as
a technical term in art history or archaeology. Putti seems to be more used than putto, and one
of the putti seems more common than a putto. Again, this might be a case of code-switching.
The use of the term paparazzi is very confused in English: it may be used as a singular, as a
plural, and in plural use may add a regular plural. Panini and zucchini are the most surprising
since, though they are Italian plural forms, they are used in English as singular forms (with
regular plural marking added as required).

7.2.7.6 Others
The use of other foreign plurals (as / æsir from Norwegian, Festschrift / Festschriften from
German, hafiz / huffaz from Arabic) seems to depend on the bilingual competence of the
individual speaker/writer (and their willingness to show this competence). While unmarked
plurals are often possible in these cases, the regular English plural is clearly the norm. The use
of unmarked plurals in New Zealand usage of Maori words was discussed in Section 7.2.5.

7.2.8 Plurals of some complex lexical items

The general rule for complex lexical items is that they take the plural marker on the righthand
edge of the word: drop-outs, forget-me-nots, hardships, inputs, singer-songwriters, windmills. This
section is concerned with the exceptions, or potential exceptions (since there is some varia-
bility here), to this general rule. There are several sub-types.
First there are long-established phrases borrowed from French, where the French structure
has a noun followed by an adjective: attorney general, court martial, governor general, heir appar‑
ent, knight errant, letter patent, and so on. Normatively, these take the plural marking on the first
element, but there is a widespread tendency to put the plural marking on the righthand edge
in the regular manner. Many of these expressions are extremely rare or specialized legal terms,
which means they are treated as regular by those who are not familiar with them.
Next we have hyphenated expressions made up of a noun followed by a prepositional
phrase: coat-of-arms, commander-in-chief, daughter-in-law (and other parallel expressions with
‑in-law), grant-in-aid, maid-of-honour ‖ maid-of-honor, man-of-war. Parallel to these are some
expressions with noun-and-noun rather than a prepositional phrase: coach-and-six, gin-and-
tonic (and other drinks). Some of these take the kind of plural that the corresponding syntac-
tic phrase would take (e.g. commanders-in-chief, parallel to agreements in principle), and a few
can take only this kind of plural (coats-of-arms, men-of-war). The corpus evidence suggests
that the plural on the right edge is especially prevalent in spoken language. This is perhaps
particularly true of the ‑in-law series.
There are some compound titles which used to be inflected on both elements (Sweet 1898)
or on the first ( Jespersen 1914: 23). Many of these are little used today, and where they are
used, may be inflected regularly: knight templar, lord chancellor, lord justice, lord lieutenant, lord
mayor, major-general.
 7.3 Possessive/genitive 141

Agent nouns coined from phrasal verbs with the structure V‑er‑Prep have the plural mark
on the first element: hangers-on, passers-by, runners-up. However, even here, right edge plural
forms can be found: passer-bys, runner-ups. These are regarded as incorrect, and are still
minority forms, but they can be found even in edited text. Some people avoid the issue by
repeating the ‑er, and then marking the plural at the right edge: blower-uppers (cf. Bauer 1983:
71). See also example (45) from COCA.

San Francisco Chronicle, 1997: HUD owns homes, which are sold as-is without warranty.
(45) 
These fixer-uppers can be real bargains because the asking price will reflect the money
needed to be invested to improve the home.

Colloquially, the extra ‑er may even be doubled, as in (46).

(46) As you’ll recall he was a rule looker upperer in the first outing. (http://web.me.com/
nataliendpeter/Site/Orders_of_the_day/Entries/2010/1/24_Including_Black_Powder,_
part_2.html)
. . . allow only 5 seconds or something to eliminate the looker upperers amongst us
(http://www.wildaboutbritain.co.uk/forums/general-wildlife/49633-british-wildlife-quiz-
anybody-got-1-a.html#post446844)

A few appositional compounds with man or woman in the first element make both elements
plural: gentlemen-farmers, journeymen carpenters, menservants, women drivers, women doctors,
women teachers. This construction appears to be found only with man and woman (not, for
instance, with child), and with man, at least, is occasionally replaced by the regular right edge
marking.

7.3 Possessive/genitive
Historically, English had a genitive case, and its markers were part of the inflectional system.
What remains of that system is something written as <’s> or <’>, which we will refer to as
s-genitive. The genitive has among its functions to mark possession (hence the competing
label ‘possessive’) but this kind of marking does not behave entirely like typical inflectional
morphology since the marker also attaches to phrases. This phenomenon is known as the
‘group genitive’ (Quirk et al. 1985: 328) or ‘phrasal genitive’ (Payne and Huddleston 2002:
479). In contrast, the so-called ‘head genitive’ shows the marking on the head noun of the
NP. Consider the examples in (47).

(47) The kitten’s tail

The kitten I saw’s tail
142 Noun inflection

The kitten I was telling you about’s tail

Somebody important’s idea
Somebody else’s idea

Here we see that the <’s> is added not necessarily to a noun (which would be what we would
expect if it were inflectional) but to the entire phrase, and that its host may be of any word
class. Such cases are generally considered to be clitics rather than inflections, and many
scholars take this point of view (e.g. Carstairs-McCarthy 2002: 37). Payne (2011), however,
argues on the basis of the behaviour of coordinated single genitives in which a pronoun is the
second coordinate (e.g. *[Kim and you]’s children, as against the acceptable [Kim and her
husband]’s or [the woman next to you]’s children) that the interaction of head genitive marking
and phrasal genitive marking in such structures is impossible to accommodate under a clitic
analysis. A clitic analysis entails that the clitic be blind to the morphological makeup of the
second coordinate (e.g. Anderson 2005), which is, however, not the case. In a similar vein,
Payne and Huddleston (2002: 480–1) hold that the possessive in modern English is still
inflectional, on the grounds that it is so tightly enmeshed with plural marking (see below).
In view of this situation, one might want to avoid the term ‘genitive’ and use ‘possessive’
instead. The label ‘genitive’ runs counter to some expectations concerning inflectional end-
ings, and the term ‘possessive’ suggests a certain meaning of these constructions, which, as
shown in the next subsection, in many cases is not there. We will therefore use the grammati-
cally inspired term, that is ‘genitive’ and we will concentrate on those cases where it looks
most inflectional, that is, where the possessive marking is added directly to the noun.
We will also discuss the periphrastic alternative to the <’s>/<’> marker, namely the use of
the preposition of (of-genitive) since the two constructions are interdependent and can be
taken as two alternant realizations of the same underlying morphosyntactic category.
However, we will not discuss further intricacies of the syntax of genitives, such as the role of
definiteness, event nominalizations involving genitives, or oblique genitives (as in a friend of
Laurie’s). The reader is referred to general reference grammars for a treatment of these syn-
tactic aspects of genitive usage.

7.3.1 Semantics
As well as marking possession, the genitive is used to show a number of other semantic rela-
tionships between two nouns. In (48) we present one possible classification of meanings,
others can be found, for example, in Payne and Huddleston (2002: Sec 16.5.2), or Quirk et al.
(1985: 321–2).

(48) meaning s-genitive of-genitive

locative New Zealand’s scenery peace of mind
 7.3 Possessive/genitive 143

source mother’s love smell of bourbon

time/location winter’s day first day of Lent
causative the war’s roots the roots of the war
purpose wolf ’s bane day of rest
cause Zest 2011: an allegorical response to the devastation of war
the war’s devastation
partitive an hour’s delay lump of cheese
named after Hodgkin’s lymphoma bridge of sighs
essive Dublin’s fair city sign of the cross
topic season’s greetings book of birds

Part of the difficulty in dealing with the meanings of the genitive is that it is not clear that all
post-modifying of-phrases are genitive: That clown of a boy does not permit of an alternative
with <’s>, but uses markers which also mark possession. This introduces a certain random
quality into what is covered in this section.
There are also grammatical readings of the genitive, particularly when the noun modified
by the genitive is a deverbal nominalization. Then we find both subject-referencing genitives
(49) and object-referencing genitives (50):

(49) a. the train’s arrival, the arrival of the train

b. Newsweek 1995: Yigal Amir’s assassination of Prime Minister Yitzhak Rabin (‘Amir
assassinated Rabin’).
CNN_Politics 2000: The senator’s criticism of gratuitous sex and violence (‘The
c. 
senator criticized gratuitous sex and violence’).

(50) the prisoner’s release, the release of the prisoner (‘someone released the prisoner’),
John’s painting (‘John painted something’, or ‘someone painted John’), Lincoln’s
assassination (‘someone assassinated Lincoln’, or ‘Lincoln assassinated someone’)

The examples show that there is a certain amount of pragmatic interpretation of the genitives
in such cases, but the preference seems to be for s-genitives to be subject-referencing and for
of-genitives to be object-referencing where that interpretation is not blocked by the context.
Where both s‑ and of-genitives are present, the s-genitive must be subject-referencing and
the of-genitive object-referencing.
As is pointed out by Payne and Huddleston (2002: 475), it is possible to treat genitives
associated with nominalizations just like genitives associated with any other noun, so that,
for instance, if each student in the class had been assigned an assassination to write about,
Lincoln’s assassination could be the assassination studied by the student called Lincoln, even
if it involved the assassination of Martin Luther King.
144 Noun inflection

7.3.2 Spelling
The major pattern for singular possessives is to add <’s> to the base form of the noun.

(51) car ◆ car’s hero ◆ hero’s

cat ◆ cat’s house ◆ house’s
child ◆ child’s mummy ◆ mummy’s
cox ◆ cox’s waltz ◆ waltz’s
dog ◆ dog’s wife ◆ wife’s
fish ◆ fish’s wolf ◆ wolf ’s

Note that the rule applies independent of the last letter of the base to which the <’s> is added,
and that there are no changes to the form of the bases as there are with the plurals (see Section
7.2): wolf’s must be possessive and wolves must be plural. There are very few exceptions to this
general principle, though for goodness’ sake is one of them.
Where the base is a proper noun and ends in <s> there is generally a choice. Either the
main rule can be followed (as in (51)), or a simple <’> can be added to the <s>. Note that
irrespective of which spelling is chosen (e.g. <Dickens’> or <Dickens’s>), speakers may read
out the form employing the [ɪz] allomorph.

(52) Aristoteles ◆ Aristoteles’ ◆ Aristoteles’s

Burns ◆ Burns’ ◆ Burns’s
Dickens ◆ Dickens’ ◆ Dickens’s
Euripides ◆ Euripides’ ◆ Euripides’s
James ◆ James’ ◆ James’s
Jesus ◆ Jesus’ ◆ Jesus’s
Matthews ◆ Matthews’ ◆ Matthews’s

Which form is preferred is variable. Greek names are perhaps most likely to be marked sim-
ply with <’>, while English names, especially given names, are most likely to have <’s>. With
hypocoristic names like Bas, Bess, Les, the <’s> version (and the corresponding /ɪz/ pronun-
ciation) is virtually obligatory. Names like Clausewitz and Fritz, which end with the letter <z>
but the sound /s/ take the <’s> variant.
Where a plural possessive is required, if the plural form does not end in <s>, then <’s> is
added.

(53) children ◆ children’s

men ◆ men’s
mice ◆ mice’s
 7.3 Possessive/genitive 145

oxen ◆ oxen’s
teeth ◆ teeth’s
women ◆ women’s

Where the plural form ends in <s>, then a simple <’> is added.

(54) aviatrices ◆ aviatrices’

cats ◆ cats’
dogs ◆ dogs’
indices ◆ indices’
mummies ◆ mummies’
wives ◆ wives’
wolves ◆ wolves’

7.3.3 Pronunciation
The pronunciation of genitive <’s> is generally the same as the pronunciation of the regular
plural, as shown in (55a) (see also Section 7.2.2.2), with two groups of systematic exceptions
concerning bases ending in an alveolar fricative. In both of these groups we observe a haplology
effect (see Chapter 9 for a discussion of the notion of haplology). In the case of the genitive
<’s>, this effect is that only one alveolar fricative surfaces. One group of haplological genitive
forms are words ending in morphemic ‑s, that is regular plural nouns, as in (55b), or other
kinds of morphemic ‑s, as in (55c) and (55d), both taken from Stemberger (1981: 793).
Another group is proper nouns and abbreviations, at least in some dialects, as shown in
(55d–f), examples from COCA, where we find variable pronunciations with final /ɪz/ or
simply final /z/ or /s/.

(55) cat’s ◆ cats ◆ cats’ /kæts/ a.

dog’s ◆ dogs ◆ dogs’ /dɒɡz ‖ dɔːɡz/
horse’s ◆ horses ◆ horses’ /hɔːsɪz ‖ hɔːrsɪz/
church’s ◆ churches ◆ churches’ /ʧɜːʧɪz ‖ ʧɜːrʧɪz/
b. for politics’ sake
c. linguistics’ contribution to the study of language
d. The song reader (2003): a novel: We’d moved into our apartment on the second floor
of Agnes’ house when I was just a baby.
e. CNN_Event 1999: Yes, the diary was found in Harris’ house.
146 Noun inflection

Agricultural Research 1999: Sharpley was recently appointed to coordinate ARS’

f. 
contribution to a new national program with NRCS

7.3.4 s-genitive versus of-genitive

7.3.4.1 Variability and non-variability
The use of s-genitives is often in variation with the use of periphrastic genitives with the
preposition of, so that we can find both of the alternatives in (56), both from the BNC.

Mitcham: Age Concern England, 1989: As far as possible, the daily lives of people
(56) 
should be determined by their own decisions
British Journal of Social Work, 1991: a final section considers the quality of people’s lives
as perceived by their relatives and others

This raises the obvious question on which basis speakers decide which variant to take. First of
all there seem to be constructions where no variation can be observed (when trying to hold the
meaning constant). This holds for partitive genitives, that is those in which the possessum nar-
rows down the quantity or number of the referent(s) of the non-head, as in (57a) and (57b).

(57) *a cake’s piece a piece of cake a.

*a bread’s slice a slice of bread
*a beer’s barrel a barrel of beer
*a book’s copy a copy of a/the book
*a chicken’s half a half (of a) chicken
*(an) advice’s word a word of advice
*(a) garlic’s clove a clove of garlic
*(a) jam’s jar a jar of jam
*my students’ some some of my students b.
* my childrens’ one one of my children
*his players’ three three of his players

Equally, no variation is observable with personal pronouns as possessors. Thus *the house of
me is impossible, and where a pronoun combines with of, the construction is not necessarily
synonymous with a genitive involving the possessive determiner, as in examples like the
thought of me versus my thought, or the picture of me versus my picture.
Other non-variable genitives are the s-genitives called ‘post-genitive’ (58a), ‘local genitive’
(58b), and ‘independent genitive’ (58c) by Kreyer (2003), and the of-phrases that modify a
pronominal head (58d) or complement a non-nominal non-head (58e).
 7.3 Possessive/genitive 147

(58) a. a friend of mine

a student of hers
b. We meet at Laurie’s.
She is staying at Frank’s.
c. Your house is bigger than Fred’s.
My car is smaller than the doctor’s.
e. My house is as big as that of the mayor.
*My house is as big as the mayor’s that.
f. the idea of providing the food
*providing the food’s idea

Turning to those constructions in which both variants are possible and synonymous, the
choice of the synthetic versus the analytic genitive is probabilistic. There are many different
factors which tend to influence the preference of one form over another, but no single factor
which can guarantee a particular form. Among the most important determining factors we
find animacy of possessor, meaning relationship, semantic category of the possessor, length
of possessor, length of possessum, sentence length, final segment of the possessor, and previ-
ous mention. In the following we will briefly review these factors, as they have emerged in
studies such as Kreyer (2003), Hinrichs and Szmrecsanyi (2007), Rosenbach (2005), or
Szmrecsanyi (2006).

7.3.4.2 Animacy
The higher up the animacy hierarchy the possessor is, the more likely it is that the synthetic
possessive will be used (e.g. Rosenbach 2005). This tendency is illustrated with the examples
in (59), with number of occurrences in the BNC given for comparison. The data are ordered
according to the proportion of s-genitives.

(59) s-genitive N of-genitive N proportion

of s-genitive
the house’s 56 of the house 3299 0.02
the idea’s 15 of the idea 296 0.05
the committee’s 314 of the committee 629 0.5
the nation’s 630 of the nation 681 0.93
the ant’s 8 of the ant 6 1.33
the frog’s 14 of the frog 10 1.4
the teacher’s 352 of the teacher 195 1.81
the dog’s 210 of the dog 113 1.86
148 Noun inflection

7.3.4.3 The meaning relationship

As already mentioned, object-referencing relationships tend more to be encoded by the of-
genitive, while the subject-referencing meaning has a greater proportion of the synthetic
construction. Where both subject-referencing and object-referencing genitives are found in a
single NP, the subject-referencing is marked with the synthetic possessive, and the object-
referencing is marked with the of-genitive, as in The newspaper’s publication of the story.

7.3.4.4 Semantic category of the modifier

The synthetic genitive is particularly frequent where the possessively-marked noun denotes
time or place: the country’s population, summer’s end, tomorrow’s dinner, the university’s science
building, a week’s pay (see e.g. Sweet 1898: 52; Quirk et al. 1985: 324). For example, we find 147
attestations in COCA of the country’s population, but only 30 of the corresponding of-genitive
construction.

7.3.4.5 Length effects

There are two kinds of length effects, one observable in the NP itself, and one emerging from
the discourse context. The context effect is straightforward. The longer the sentence in which
we find the pertinent NP, the more speakers or writers prefer the of-genitive. This suggests a
processing advantage for the of-genitive. Processing constraints are also most likely responsi-
ble for the fact that longer possessor phrases have a tendency towards the of-genitive whereas
length in other parts of the noun phrase (prenominal modifiers, nominal complements)
makes the s-genitive more likely. In both cases, the longer constituent is placed towards the
end of the NP, which follows the principle of end-weight and facilitates processing (e.g.
Hawkins 1994). Consider the data in (60) for illustration.

(60) a. Long possessor

the analysis of [the full data set]
the phone number of [the guy who lives next door]
b. Length elsewhere
Germany’s [first two very successful attempts]
the committee’s [decision to withdraw the proposal]

7.3.4.6 Final segment

If the final segment of the possessor is a sibilant, the chances rise that the of-genitive is used.
This might appear to be functionally motivated by the avoidance of ambiguity in examples
such as /ðə skuːlz dɔː(r)z/, which could be the school’s doors or the schools’ doors. To choose
the doors of the schools makes clear what the message is. However, only a minority of these
cases involve regular plural nouns, and the effect also holds for other sibiliants, that is for /ʃ, ʒ,
 7.3 Possessive/genitive 149

tʃ, dʒ/. This suggests that we may be dealing not with ambiguity avoidance but with a phono-
logical-articulatory planning difficulty.

7.3.4.7 Previous mention

Following up on a claim by Altenberg (1982), Szmrecsanyi (2006) has shown that previous
mention of a particular construction favours the use of the same construction in the next
genitive choice context. This can be explained as an on-line priming effect.

7.3.4.8 Other factors

Genitive choice is also susceptible to stylistic or textual variation. Thus, texts with a higher
lexical density (as measured by type–token ratio) have more s-genitives than texts with lower
lexical density (Szmrecsanyi 2006: 97).
Another factor is nestedness. Sequences of synthetic possessives, such as His wife’s cousin’s
son’s award are rare, whereas strings of of-modifiers are not. Where possible, therefore, multi-
ple modifiers in nested genitives are usually distributed so that there will be at most one
synthetic possessive in the set.
There are also preferences linked to specific lexical items. Some phrases with genitives
are more or less fixed lexical items and show preferences which cannot be explained by the
determinants just discussed. Some examples are given in (61), with figures showing occur-
rence in COCA.

(61) decency’s sake 5 the sake of decency 0

France’s republic 0 the republic of France 5
kings’ king 0 king of kings 71
life’s love 2 love of [poss. pronoun] life 451
sun’s rays 242 rays of the sun 127

Finally, there are differences between varieties of English. Szmrecsanyi (2006) finds that
within the British Isles speakers from the Hebrides, the Midlands, or Wales have a preference
for of-genitives, in contrast to speakers from the southeast of Britain. Differences between
NAmE and BrE have also been noted, to the effect that the synthetic variant is more frequent
in NAmE, but the strength of the effect varies in interaction with other variables. There is
also ongoing change, such that the s-genitive is being more widely used (Ainsworth 1992;
Rosenbach 2002).
 chapter 8

Function words: pronouns,

determiners, wh‑forms, deictics

8.1 Prospectus
In this chapter we will consider all other English word classes that might conceivably be
subjected to morphological analysis, among them pronouns, determiners of various sorts,
wh‑words, and deictics. We will not be concerned with the way in which the particular forms
are used except where this impacts on our analysis: for example, the fact that we can be used
as a royal or authorial plural as in (1) and (2) is not relevant to the general pattern of pro‑
nouns in English, and is not part of our remit. Similarly, the use of they as a singular, as in (3),
while of great prescriptive and descriptive interest, does not impact on the way in which the
system is structured. Nor does the use of feminine pronouns to refer to ships or machines, or
the variable use of she or he as an impersonal pronoun in place of it in some varieties of
English (for the latter see Pawley 2002, 2008; Siemund 2007).

(1) We are not amused.

(2) In this paper we will illustrate . . . [in a paper with a single author]
(3) Everybodyi is hoping that theyi will win.

We begin with pronouns and then look at determiners, wh‑words, and deictics.

8.2 Pronouns
8.2.1 Personal pronouns
The first set of personal pronouns is laid out in (4). The major difficulty here is a matter of
nomenclature. Traditionally these are referred to as nominative and accusative pronouns
(Payne and Huddleston 2002: 426) or subjective and objective pronouns (Quirk et al. 1985:
346). But if, following Hudson (1995), we query the whole notion of case for English, such
 8.2 Pronouns 151

nomenclature seems misleading (though see Quinn 2005 for a more nuanced view). We here
propose the labels sole subject form (for pronouns like I) and the default form, on the
grounds that for many speakers the only place where pronouns like I occur regularly is when
they are uncoordinated subjects of finite verbs, and the pronouns like me occur unless there
is some reason to have a pronoun from the other series.
In (4) we present both standard and alternative forms of the personal pronouns. The
forms occurring in the column marked ‘alternatives’ are dialectal (archaic, sometimes) or
register-specific forms of differing degrees of generalization. See, for example, Wright (1905),
Wentworth (1944), Orton et al. (1978) and the OED, and for a wider look which includes
creoles, see Wales (1996: 90). IrE is notable for allowing reflexive forms (see Section 8.2.3) to
be used without antecedent in place of the standard forms (Filppula 2004: 93).

(4) Sole subject Default

Standard Alternatives Standard Alternatives
1st person singular I ch, ich, he, her me us, I
2nd person singular you thou you thee
3rd person singular masculine he him (ə)n
feminine she shoo, hoo her
neuter it it (ə)n
1st person plural we us /ʌs/ ~ /ʌz/
2nd person plural you ye, yous, yes you ye, yous, yes
/jiːz/, y’all, /jiːz/, y’all,
you-all(s), you you-all(s), you
guys, yous guys, guys, yous
you ’uns guys, you ’uns
3rd person plural they them

Although various forms are recorded in (4), there can be a lot of phonetic variation, for
example in the form of the pronoun I, and many of the pronouns have a distinct form in
unstressed position, where pronouns are often found (see, for example, Wales 1996: 14).
In addition to the pronouns in (4), we should add the paradoxically named impersonal
personal pronoun, one, usually denoting some unspecified person, but in a socially very nar‑
row dialect of BrE indicating the first person singular.

8.2.2 Possessive pronouns

The possessive pronouns are set out in (5). Again nomenclature is a problem. Payne and
Huddleston (2002: 426) talk about dependent and independent possessives (for my and
mine respectively), but many syntactic theories see the possessive as the head of the
construction, so that the label ‘dependent’ seems contradictory. Quirk et al.’s (1985)
­
152 Function words: pronouns, determiners, wh‑forms, deictics

determinative function and independent function is better. Wright (1905) uses conjunctive
and disjunctive (respectively), Kruisinga (1932) uses attributive and independent, Poutsma
(1916) uses conjoint and absolute. An alternative terminology would see these as possessive
adjectives and possessive pronouns respectively, or as prosodically weak and strong. We will
use the terms determinative for those possessives like my that behave syntactically as
determiners and absolute for those like mine that can occur in predicative contexts. The
possessive pronouns are presented in (5):

(5) Determinative Absolute

Standard Alternatives Standard Alternatives
1st person singular my me mine mines, my
2nd person singular your thy, thee yours thine, yourn
3rd person singular masculine his he his hisn
feminine her hers hern, hersn
neuter its it, his its
1st person plural our wə(r), us, we ours ourn
2nd person plural your ye, you yours yourn, you-all’s,
y’all’s, your-all’s,
yous guys’, your
guys’
3rd person plural their theirs theirn

The forms that are available allow for greater regularity than is present in standard varieties,
and some dialects actually exploit this. In particular, some dialects have consistent ‑n forms
in the absolute forms, others have consistent ‑s forms, while standard Englishes have a mix‑
ture. Note also that absolute forms that contain apparently determinative forms (e.g. you
all’s) are sometimes disfavoured; a double possessive (your-all’s) is a relatively frequent
outcome.

8.2.3 Reflexive and emphatic pronouns

The reflexive pronouns are given in (6).

(6) Standard Alternative

myself
yourself thyself
himself hisself
herself
itself
 8.3 Determiners 153

ourselves usselves
yourselves
themselves theirselves

Apart from the variation illustrated above, there is variation in the phonetic form of the ‑self
element: often the /f/ or /v/ is elided or the /l/ is elided. Sometimes the /l/ is replaced by
/n/ when there is no /f/ or /v/. In some contexts there may not be a separate set of reflexive
pronouns, the default pronoun illustrated in (4) being used instead.

8.3 Determiners
The demonstrative determiners (or ‘determinatives’, using the terminology of Payne and
Huddleston 2002) are laid out in (7). Yon is old-fashioned or dialectal; it is no longer part of
the standard language. It appears to be used to point out objects which are within sight but
further away than those covered by that.

(7) proximal distal super-distal

singular this that yon
plural these those yon

Yon may take on the form yond or yonder (though yonder is also an adjective and an adverb).
Some dialects use this here and that there in place of simple this and that. This may be replaced
in some dialects by thick(y) or some derived form, that may be replaced by thack(y), thon. In
ScE those is replaced by they.

8.4 Wh‑words
Wh‑forms are used in English both as the initial element in relative clauses and in wh‑­
questions; traditionally this set of forms is referred to in the former case as ‘relative pronouns’
and in the latter as ‘interrogative pronouns’. Also included in the traditional designation of
relative pronouns is the form that, which is better analysed as a conjunction (e.g. van der
Auwera 1985; Huddleston et al. 2002: 1056–7) or complementizer (Carnie 2006). As this
form is generally morphologically invariant in relative clauses, we do not discuss it here,
although note that in ScE that’s is possible in possessive contexts.
In both relative and interrogative contexts there is a great deal of variation in usage in the
contemporary language, as well as regional and diachronic variation. In (8) and (9) we
present a rather full, conservative, and standard set of wh‑words, on which we add comments
154 Function words: pronouns, determiners, wh‑forms, deictics

below. The data in (8) are divided up according to whether the wh‑form occurs in a restric‑
tive (defining) or non-restrictive (parenthetical) relative clause.

(8) Relative clauses

Restrictive Non-restrictive
rational beings, other referents rational beings, other referents
higher animals higher animals
Subject who which, when, who which, when,
Non-subject whom, who where, why whom, who where, why
Possessive whose

(9) Interrogatives
rational beings, higher animals other referents
Subject who which
which what
Non-subject who which
whom what, when, where, why, how
which
Possessive whose

Whom as direct object is infrequent, perhaps more so in interrogative than in relative con‑
texts, especially in spoken English. It is often replaced by who when referring to a direct
object. Whom is used more consistently in both relative clauses and interrogatives as a non-
preposed object of a preposition (e.g. to whom), although in some varieties (e.g. NZE) which
can be used following a preposition even for an animate referent.

(10) 60,000 refugees, of which half could die (Radio New Zealand 1994)

Nonetheless, whom remains part of the system in non-subject contexts, especially in very
conservative or formal written style.

8.5 Compound determinatives

Here we consider a set of non-canonical compound forms (see Chapter 19.2.3) which
denote indefinites. It is again difficult to decide what to call these items. They consist
of a quantifier as first element and a semantically light noun (thing, body) or pronoun
(one) as second element. For lack of a better term, we adopt the terminology of
Payne and Huddleston (2002: 423) and call them ‘compound determinatives’. They are
set out in (11).
 8.6 Deictic pro-forms 155

(11) Rational beings Non-personals

nobody no one, no-one nothing /nʌθɪŋ/
somebody someone something
everybody everyone everything
anybody anyone anything

8.6 Deictic pro-forms

We include here the set of forms which might be viewed as pro-adverbs referring to a particu‑
lar temporal or locative space. The relevant simple forms are set out in (12).

(12) Spatial location Temporal location

proximal distal interrogative proximal distal interrogative
location here there where now then when
source hence thence whence
goal hither thither whither

Note that hence is frequently used with a meaning equivalent to ‘therefore’ rather than a spe‑
cifically spatial meaning. Thence and whence have similar functions.
Here and there also enter into several non-canonical compound forms, some of which are
illustrated in (13).

(13) elsewhere, henceforth, hereabout(s), hereafter, hereby, hereto, hereupon, herewith,

hitherto, thenceforth, thereabout(s), thereafter, thereby, therefore, thereupon

Note that many of these compounds which are built on the spatial locatives set out in (12)
nevertheless make reference to temporal location. Therefore is used as a conjunction rather
than as an adverb; hereafter and thereafter have uses as nouns.

8.7 Analysis of the forms

However we try to analyse the various forms collected together here, we have to recognize
that there is little in English function words to signal their morphological relatedness. My and
his have little in common to demonstrate their possessiveness, for instance.
Furthermore, although we can see various tendencies in the formal and semantic analysis
of these forms (and generative syntactic theory sometimes tries to exploit these tendencies;
see for example Di Sciullo 2005; Klinge 2005b), it seems unlikely that most of them are
156 Function words: pronouns, determiners, wh‑forms, deictics

actively parsed by contemporary speakers. The exceptions are the compound forms, and
even there, there is often little semantic connection between the second element when it
occurs in the compound and when it occurs as an independent form. For example, therefore
does not have any obvious semantic link with for/fore. With a form like hence there is a certain
amount of evidence from usage that the force of the word is not appreciated by speakers,
since many of them say from hence, and from whence, even though hence and whence include a
‘from’ meaning.

(14) BNC. Pamela, or the Reform Rake 1987: Pamela shall not stir from hence
COCA. Analog Science Fiction & Fact 2006: They will be taken up to the stars from
whence we came
BNC. The Dyke & the Dybbuk 1993: it sends me back, in some discomfort, to the place
from whence I came

We also need to consider the frequency of the elements dealt with in this chapter. The basic
forms are all so frequent that they no longer require specific parsing for appropriate use or for
understanding.

8.7.1 Initial <th>, <h>, and <wh>

A number of the forms have an initial <th> (phonologically /ð/), something they share with
the definite article the. Indeed, we could see the as the minimal form here, with only (in most
instances) the minimal vowel /ə/ supporting the /ð/ while words such as thou, thee, thy,
thine, they, them, their, theirs, this, that, these, those, there, thence, thither, then all have some
supporting material. Historically, the analysis of this initial /ð/ as being an indication of
deixis seems valid. It is also notable that the only words of English with initial /ð/ are in this
set (if though can be added). However, thither, now a rather rare member of the set and often
occurring in the phrase hither and thither (more than a third of the uses attested in COCA),
is pronounced with an initial /θ/ (and thus removed from the set) in some varieties of ScE
and NAmE.
Initial /h/ in the deictic set of here, hither, hence seems to indicate proximity, especially
where it contrasts with initial /ð/ and <wh>. However, the initial /h/ in him and her, and in
who cannot belong to the same grouping, since these words are not used for proximal
objects.
The initial <wh> (more usually these days pronounced as simply /w/, though the <wh>
spelling persists) is a regular concomitant of interrogativity in why, when, what, which, whither,
whence, where. However, some of these forms are also used to introduce relative clauses where
there is no longer any trace of interrogativity. At some level this is explained historically by
the fact that the relative use developed from the interrogative one in the late Middle English
 8.7 Analysis of the forms 157

period, possibly with some influence from French (Fischer 1992: 301; Fischer and Van der
Wurff 2006: 128). In who, whom, whose the <wh> spelling persists, but the pronunciation as
/h/ masks any relationship.
Although there is thus little empirical synchronic evidence for morphological complexity,
both Di Sciullo (2005) and Klinge (2005a) have tried to make an argument for analysing the
wh‑forms and deictics as complex; the former, for example, claims that wh‑ in wh‑words is a
quantifier and the th‑ in words like this and that a determiner. What remains (e.g. ‑o in who,
‑at in what or that) is analysed in terms of features like [+/–human] or [+/–proximate].
Whatever the benefits of such analyses, it is nonetheless clear that these elements are not
morphemes in the classic sense of the term, and it is uncertain what is to be gained by analys‑
ing them as complex.

8.7.2 Oblique <m>

The final <m> in him, them, whom is historically a dative inflection which by the end of the
Middle English period indicated any oblique case, rather than strictly a dative. Given that we
have avoided an analysis in terms of case, the precise meaning of this /m/ becomes difficult
to specify, and must be obscured by the loss of whom in most environments.

8.7.3 Final <n>, <r>, <s>, and <t>

A final /n/ indicates possession in mine, thine, and is generalized to the absolute forms in
some dialectal forms. However, the final /n/ of then, when, yon has no such correlation.
A final historical /r/ (variously treated in non-rhotic varieties of English) indicates posses‑
sion in her, their, your. Again, this is historically correct, but there are too few forms for a
pattern to emerge in contemporary English.
A final <s>, usually pronounced /z/, but sometimes /s/ depending on the phonotactic
environment in which it is found, indicates possession in his, its and in the absolute posses‑
sive forms (e.g. yours, hers, etc; see (5)). This is supported by the possessive marker on nouns,
and is probably robust and recognizable.
Historically, the final <t> in it and what indicates neuter gender, but that pattern is not
recuperable in the modern forms.

8.7.4 Thou and you

The spelling <ou> in thou and you might appear to show a relationship (despite the differ‑
ent pronunciation in modern English), but since, historically speaking, thou is nominative
while you is accusative, thee is accusative while ye is nominative, there is no usable pattern
here.
158 Function words: pronouns, determiners, wh‑forms, deictics

8.7.5 Analysability and morphemic status

In principle, therefore, forms like h‑ence, th‑i‑ne, th‑ither, wh‑o‑m, you‑r‑s are analysable as
shown into partly helpful elements. It would, though, be a mistake to assume that these
elements are morphemes, available to the introspection of the modern speaker. Rather,
what we have here are sub-morphemic elements which may set up what Hockett (1987)
calls resonances. While there may be some value in drawing some of the generalizations to
the attention of sophisticated second language learners at a suitable level of expertise, fun‑
damentally, all the non-compound forms are best treated as morphologically unrelated
members of their sets.
 Pa rt III

Derivation
 chapter 9

Derivation: phonological
considerations

9.1 Prospectus
This chapter gives an overview of phenomena in derivation where morphology and phonol-
ogy interact. While such phenomena will also be dealt with in some detail in the chapters to
follow, the present chapter provides a general introduction to the role of phonology in English
derivation and the principles that govern the morpho-phonological behaviour of many of the
morphological categories to be described in the upcoming chapters.
There are some rather isolated cases of derivation in which the relevant meaning is solely
expressed through a manipulation of the base-internal phonological structure. This is, for
example, the case with some umlauted causative verbs (fall ◆ fell, lie ◆ lay, rise ◆ raise, sit ◆ set)
or with certain very rare nominalizations (e.g. abide ◆ abode, sing ◆ song). We do not consider
these cases of unproductive morphology in this chapter.
After a general discussion of the role of phonology in English derivation in the next sec-
tion, we will deal with the problem of allomorphy in Section 9.3, and with particular kinds of
phonology–morphology interactions in Section 9.4 (on the role of prosody), Section 9.5 (on
haplology), Section 9.6 (on prosodic morphology) and Section 9.7 (on phonological selec-
tional restrictions).

9.2 Phonology in derivational morphology

English derived words are related to their bases in at least two respects, semantically and
phonologically. In other words, a derived word normally shares both a meaning compo-
nent and parts (or all) of its form with its base. A look at a sample of derived words
reveals, however, that the formal relation, that is the phonetic or phonological relation
between base and derivative, is not necessarily straightforward. The issue is how faithful
the derivative is to the phonological structure of the base. The words in (1) illustrate dif-
ferent cases:
162 Derivation: phonological considerations

(1) a. help ◆ helpless, blue ◆ blueness, test ◆ test-wise, assess ◆ assessment

b. govern ◆ government, rest ◆ restless, exact ◆ exactly
c. cónjugàte ◆ cònjugátion, cólony ◆ colónial
d. feminine ◆ feminize, mercury ◆ mercurate
e. cylinder ◆ cylindric, enter ◆ entrance
f. serene ◆ serenity, meter ◆ metrical
g. endocentric ◆ endocentricity, erase ◆ erasure
h. Jacob ◆ Jake, Alfred ◆ Freddy, celebrity ◆ celeb, professor ◆ prof, beer + nirvana ◆
beervana, Japanese + English ◆ Japlish

The words in (1a) give examples of affixation in which the base is phonologically fully
recoverable from the derivative. The suffixes in the derivatives form syllables of their own,
and the suffix is phonologically not integrated into the base, which means that the mor-
phological boundary and the phonological boundary (i.e. the syllable boundary) coincide.
This is an interesting fact, as in many cases, the first consonant of the consonant-initial
suffix could in principle be syllabified as the second consonant of an onset that starts with
the base-final consonant, as in *hel.pless or *tes.twise. But this does not happen; the suffix
does not form a phonological unit with its host. Affixes that display this kind of inert
behaviour have been called ‘non-cohering’ (e.g. Dixon 1977: 93; Kaisse 2005: 28) and are
generally consonant-initial.
The cases in (1b) present base-derivative pairs where, phonetically, the bases are not
straightforwardly recognizable in the derived word, that is, the derivatives are phonologi-
cally not fully transparent. For example, government is mostly pronounced /gʌvmənt/ or
/gʌvəmənt ǁ gʌvərmənt/, and restless and exactly are often pronounced without a /t/.
Quite frequently such phonological opacity goes together with semantic opacity, as, for
example, in the case of government, which does not denote ‘action of verbing’ (as is
standardly the case with ‑ment derivatives), but rather denotes the people who govern,
or, more generally, ‘political authorities’. Notably, these particular cases of phonological
opacity are idiosyncratic, that is, they reflect the history of individual words, or they
concern phonetic simplifications that are highly dependent on the frequency of individ-
ual words. For example, while exactly is rather frequent and easily loses its /t/, conjunctly
is of low frequency and is unlikely to occur without its /t/. In what follows we will not
dwell on such idiosyncrasies but focus on the more systematic patterns of phonological
opacity.
We should note that phonological variation as a result of morphological contiguity varies
from the categorical to the extremely variable. In the literature, such matters are sometimes
dealt with as a distinction between lexical and post-lexical phonology (see e.g. Mohanan
1986). Unfortunately, a distinction is very often difficult to draw, especially when the same
 9.2 Phonology in derivational morphology 163

process (say, for example, the assimilation of a nasal to the place of articulation of a following
consonant) may be involved in both lexical and post-lexical phonology. The deletion of
/t/ mentioned above may not be strictly morphological in the sense that it might occur
in the syntactic phrase the exact lenition as well, and be subject to factors such as the speed
of speech, the formality of the occasion, the audience for the speech event and so on, but
it does have the effect of masking the transparency of morphological processes. We will
not discuss such matters further here, but our focus is on those processes which affect
phonemic structure rather than, for instance, degrees of nasality or aspiration. This is in
line with the notion of morpho-phonemic variation being involved. We take this slightly
further than normal in that we do consider some instances which are not categorical but
variable.
Examples (1c) to (1h) show morpho-phonological alternations that affect words of the
same morphological category and that are systematic in nature; (1c) gives examples of stress
shift, (1d) shows the truncation of the base under suffixation, (1e) illustrates some extreme
cases of resyllabification, (1f) contains instances of vowel alternations, and (1g) of consonant
alternations. In these instances the phonology takes the morphological strings and treats
them as single units; affixes which are or can be involved in such strings have been termed
‘cohering affixes’. Which affixes are classified as cohering and which ones as non-cohering is
not always clear, however. Some authors (e.g. Kaisse 2005) use the possibility of stress shifts
as a criterion, but if syllabification is used as a criterion, all vowel-initial suffixes are cohering.
And even consonant-initial suffixes are capable of causing phonological alternations (deep ◆
depth, marked ◆ mark[ə]dness) or of showing syllabic integration with the base. For example,
adverbial ‑ly does not geminate in frailly but coalesces with base-final syllabic /l/ to form an
onset, something which also occurs in gently, and the suffix could therefore be classified as
cohering.
Returning to our different kinds of phonology–morphology interaction in (1), we see
an extreme case of phonology–morphology interaction in (1h), often referred to as ‘pro-
sodic morphology’. Here, a morphological category is primarily or exclusively character-
ized by a manipulation of the prosodic structure (instead of by affixation or conversion),
for example by deleting phonological material to create derived words of a certain pro-
sodic structure.
To summarize, the examples in (1b–h) illustrate patterns in which a given base stands in a
phonologically non-trivial relationship to its derivative. As we will see, individual morpho-
logical categories often display a kind of relationship that is peculiar to this category. These
relationships will be further investigated in Sections 9.3 to 9.6.
Phonology may also play another role in word-formation patterns. For example,
affixes may have different phonetic realizations on a systematic basis, that is allomorphs,
and the distribution of these allomorphs may be phonologically governed, as shown for
‑al and in‑ in (2). Such cases of affix allomorphy will be discussed in more detail in
Section 9.3.2.
164 Derivation: phonological considerations

(2) a. ‑al: polar / *polal versus *labiar / labial,

b. in‑: inaccessible ◆ illegal ◆ improbable

Finally, phonology may have a say in imposing selectional restrictions on combinations of bases
and affixes. For example, the nominal suffix ‑al (as in refusal) can be found only on bases that are
stressed on the final syllable (consider arríval, bestówal, renéwal). Section 9.7 will take a closer
look at the kinds of restrictions that can be observed with different derivational categories.
Before turning to the discussion of the individual morpho-phonological patterns, a word
is in order as to the kind of approach we are taking here. Phonological alternations in deriva-
tional morphology, such as stress shifts (e.g. prodúctive ◆ productívity) or base truncations
(e.g. feminine ◆ feminize), have attracted a lot of attention in the theoretical literature, mostly
under the label of ‘Lexical Phonology’. In much of this literature, a serial derivational approach
is predominant, which is also reflected in the terminology that is widely used to talk about
the pertinent phenomena. For example, the term ‘stress shift’ implies that a given stress is
shifted from the position it occupies in the base to another position in the derivative when an
affix is added to that base. More recent work in morpho-phonology has shown, however, that
such a serial-derivational approach is by no means the only possible one and that, for exam-
ple, output-oriented approaches, or approaches taking into account other related words apart
from the base, may often be able to provide a more insightful account of the empirical facts.
In this chapter, we will continue using the established terminology, but without a commit-
ment to the derivational approaches that underlie this terminology. We will also approach
the phonological facts from a morphological perspective. Thus, we will conceptualize
­morpho-phonological alternations as kinds of base allomorphies and affix allomorphies.
Lexical Phonology has also capitalized on the idea of lexical strata with two (or sometimes
three) levels of affixation, with the affixes of each level assumed to show the same kind of
behaviour concerning, for example, stress shift, syllabification, trisyllabic laxing, velar soften-
ing, etc. More recent work (e.g. Plag 1999; Raffelsiefen 1999, 2004, see also the chapters to
follow in this volume) has shown, however, that each derivational category comes with its
own set of phonological restrictions and properties, and that the stratal division of the lexi-
con is at best a gradient in psycholinguistic or statistical terms (Hay 2002; Plag and Baayen
2009) and at worst non-existent in structural terms (see Chapter 27). We will therefore not
frame our discussion in stratal terms.

9.3 Allomorphy
As defined in Chapter 2, allomorphs are phonologically distinct variants of a morpheme that
occur in complementary phonological environments. The cases in (1b) show, however, that
under a wider definition, allomorphs may also occur in free variation. Furthermore, we will
see that different allomorphs of the same affix may also be morphologically conditioned.
 9.3 Allomorphy 165

The examples in (1c) to (1g) illustrate that allomorphy in English derivation is often
observable with bases, but affixal allomorphy may also occur, as shown in (2). In the fol-
lowing we will first deal with base allomorphy, then turn to affix allomorphy. As mentioned
in Chapter 2, there are also elements, what we have called extenders, whose status as part
of the affix, or as part of the base, or as part of neither, is not entirely clear. In the former
case, we would treat these cases as base allomorphy, in the second as affix allomorphy, and
in the third we would need a new kind of category, probably similar to what has been called
a ‘linking element’ with German and Dutch compounds. We will discuss extenders in
Section 9.3.3.

9.3.1 Base allomorphy

Different kinds of more or less systematic base allomorphy can be distinguished. We will
briefly introduce them here and discuss the more intricate ones in greater detail in the sec-
tions to follow.

9.3.1.1 Prosodic restrictions and adjustments

Prosody becomes relevant when morphological categories impose prosodic restrictions on
their derivatives. These restrictions can have two kinds of consequences. First, the restric-
tions may be satisfied by choosing only those bases which meet the relevant criteria. For
example, for nominalizations in ‑al only those verbs that end in a stressed syllable can be
bases, such that ‑al derivatives end in a trochee (arríval, refúsal). Simplifying somewhat, the
suffix ‑ize allows only bases that end in a trochee, or, under specific circumstances, end in a
dactyl (rándomize, hóspitalize). This kind of selectional restriction will be dealt with in
Section 9.7.
The second possible consequence of prosodic restrictions may be that in order to satisfy
the restrictions, certain adjustments are made, so that the derivative fits the restrictions, but
does so at the cost of a loss of phonological transparency, that is by sacrificing some of the
phonological identity of the base. The most widely discussed adjustment comes under the
label of stress shift and concerns patterns in which base and derivative differ concerning
which syllable(s) carry stress. Stress shift is thus basically a consequence of some stress-
related prosodic restriction imposed on the derivatives of a particular morphological category
and generally happens only in those cases where the restriction is not trivially satisfied any-
way. For example, ‑ity is said to be stress-shifting, but in fact shifts stress only in those cases
where the stress does not end up on the antepenult syllable of the derivative anyway
(cf. prodúctive ◆ productívity versus inténse ◆ inténsity).
Another way of satisfying prosodic constraints is the insertion of phonological material,
or the deletion of phonological material. Both types of solution can be found in English
derivational morphology. For example, we find prosodically conditioned epenthesis with
adverbial ‑ly (amaz[ə]dly, suppos[ə]dly), and prosodically conditioned truncation with ‑ize
166 Derivation: phonological considerations

(patina ◆ patinize, summary ◆ summarize). Truncation and epenthesis will be discussed in

more detail in the sections on truncation (Section 9.3.1.2) and on vowel alternations
(Section 9.3.1.4).
The problems involved with morphologically-induced stress shift are quite intricate and
we will discuss them in greater detail in Section 9.4. In this section we only lay out the most
basic properties of this type of alternation. The examples in (3) illustrate stress shift for a
number of suffixes. In general, stress shift is restricted to suffixation and is found only with
non-native suffixes (but not all non-native suffixes show it). There are instances of stress shift
with prefixes, too, such as ímpious, ímpotent, préface, préfix, prémature (BrE), réplicate, but
these are individually lexicalized, and in many cases the sense of the prefix is lost, or the status
of the forms as morphologically complex is questionable anyway, as in ímpetus, préfect, rém‑
edy. Because these are so unpredictable, we do not deal with them in any detail in this chapter.
As we can see in (3), there are suffixes that attract the main stress of the derivative to them-
selves (such as ‑ee or ‑ese), and suffixes that require a certain syllable of the derivative to carry
the main stress.

(3) a. ‑ation: persónifỳ ◆ persònificátion, rándomìze ◆ ràndomìzátion

b. ‑alA: président ◆ presidéntial, súicide ◆ sùicídal
c. ‑ese: jóurnal ◆ jòurnalése, Japán ◆ Jàpanése
d. ‑ity: mónstrous ◆ mònstrósity, prodúctive ◆ pròductívity

Of necessity, imposing a particular main stress position on a different syllable involves the
creation of a different prominence pattern as a whole, which raises the question of how the
stress distribution in the base relates to the distribution of stresses in the derivative. This is
the problem known as ‘stress preservation’. In (3c), for example, journalese preserves the main
stress of the base as a secondary stress in the derivative, while Japanese completely destresses
the syllable that carries the main stress in the base.
Quite often, the stress pattern imposed on the derived word leads to vowel qualities differ-
ent from those of the corresponding vowels of the base. For ease of discussion, we will talk
about vowel qualities in terms of ‘tense’ and ‘lax’, following the terminology of Jakobson et al.
(1951). The terminology is phonetically obscure (Lass 1976), but no more awkward to use
than the alternatives (long/short, checked/unchecked, [±ATR]) and it allows us to keep
well-known terms such as ‘trisyllabic laxing’. Using this terminology, the schwa in the last
syllable of president corresponds to a full vowel under ‑al suffixation (presidential), and the
tense vowel of explain corresponds to a schwa in explanation, and to the lax vowel /æ/ in
explanatory. Note also that it is not always possible to assign a given affix unambiguously to
the class of stress-shifting suffixes, as the pertinent derivatives may show variable behaviour,
as is shown later in this chapter. Table 9.1 summarizes the basic patterns involved in prosodic
adjustments.
 9.3 Allomorphy 167

Table 9.1 Prosodic adjustments: overview

Type of adjustment Selected affixes concerned Example
none ‑ful, ‑less, ‑wise resourceful, hopeless,
cornerwise
selectional restrictions ‑aln, ‑er, ‑ize arrival, electioneer, randomize
truncation ‑able, ‑al, ‑ize manipulable, liberal, feminize
epenthesis ‑ly, ‑ness supposedly, markedness
stress shift ‑ala, ‑ation, ‑ese, ‑ity parental, computation,
Japanese, partiality

9.3.1.2 Base truncation

Base truncation is the name of a phenomenon in which final segments of bases are no longer
present in the derivative. Consider the examples in (4).

(4) a. demonstrate ◆ demonstrable, navigate ◆ navigable

b. philosophy ◆ philosopher, astronomy ◆ astronomer
c. emphasis ◆ emphasize, summary ◆ summarize
d. chlorine ◆ chloral, ventriloquy ◆ ventriloqual
e. Malta ◆ Maltese, China ◆ Chinese

It has been claimed (e.g. by Aronoff 1976) that base truncation concerns the deletion of mor-
phemes, as seems to be the case in (4a), where the suffixation of ‑able seems to trigger the
deletion of the verbal suffix ‑ate. Similarly, data such as (4b) suggest the deletion of a suffix ‑y.
The facts can, however, also be explained in a generative type of approach by considering
both ‑ate and ‑able derivatives, as well as the ‑er formations in (4b), as being derived from
bound bases (e.g. Raffelsiefen 2004: 121). Such bound bases could then also be used to
account for other kinds of formations (e.g. toler-ate, toler-able, toler-ant, etc.).
A similar analysis applies to cases where inflectional endings from the classical languages
are deleted before the addition of derivational endings. Etymologically speaking, this is prob-
ably the result of patterns in classical morphology, but it leaves apparently unmotivated
truncation in English where the deleted material is neither inflectional nor affixal. Examples
are cerebellum‑ar, genera‑ic, nebula‑ous, nucleus‑ar, phenomenon‑al, scrotum‑al.
The data in (4c) to (4e) demonstrate that the rhymes of final unstressed syllables
can undergo deletion with the suffixes ‑ize, ‑al, and ‑ese. Deletion in (4c) may be analysed
as a prosodic optimization strategy, avoiding a stress lapse, that is a sequence of two
unstressed syllables, before ‑ize. The exact conditions under which the truncations illus-
trated in (4c) to (4e) occur are laid out in the chapters where these suffixes are described
in detail.
168 Derivation: phonological considerations

The examples in (4c) to (4e) also nicely show that stress shift is independent of other
kinds of morphology–phonology interactions, as all three suffixes can induce base trunca-
tion, but they differ in their stress-shifting capacity: ‑ize is generally not stress-shifting, while
adjectival ‑al often causes stress to move to the antepenult syllable of the derivative, and ‑ese
is auto-stressed.

9.3.1.3 Syllabification, gemination, and degemination

As already mentioned in Section 9.2, cohering suffixes form a phonological unit with their
host, which entails that syllabification can work across the morpheme boundary, especially if
the suffix is vowel-initial. The whole question of syllabification is particularly controversial in
English, with disputes in the literature as to where syllable boundaries go and what the proper
techniques for discovering them are. Moreover, there is a certain amount of evidence that
syllabification and resyllabification may be variable in some words, either from dialect to
dialect or from speech-style to speech-style. Claims about resyllabification therefore have to
be made with appropriate care and hedging. Nevertheless, there seems to be some degree of
agreement, which we will try to exploit in what we say. One such point of agreement is that
resyllabification in English derivation is restricted to vowel-initial suffixes. Consonant-initial
suffixes generally do not induce resyllabification (but note what was said concerning the
behaviour of ‑ly just above), and prefixes generally do not cause resyllabification either, as
can be seen with dis.please or mis.teach that have base-initial aspiration of the plosive, an indi-
cation of a preceding syllable boundary. However, some individual prefixed forms, such as
mi.stake, su.burbia do show resyllabification, but such phonological opacity is often accompa-
nied by semantic opacity. In those cases where we do find resyllabification with a prefix and
semantic opacity is not involved, resyllabification seems to be favoured in foot-medial envi-
ronments, as in (dì.sa)footllów, (ì.nex) footpérienced (Giegerich 1999: 279–80).
In spite of all controversies, it is relatively clear that there are some suffixes where many
speakers syllabify a derived word in such a way that a base-final segment then forms the onset
of the final syllable of the derived word, as shown in (5).

(5) irri.t‑ate, adver.t‑ize, coura.ge‑ous, esca.p‑ee, ea.t‑er, mea.t‑y

Where there is a final <er> pronounced /ə ‖ ər/, the /r/ always surfaces as the onset to the
next syllable when a vowel-initial suffix follows, whether the support vowel for the /r/ is
written and pronounced (as in (6a)) or not (as in (6b)).

(6) a. cylinderish, disasterish, entering, metering

b. cylindric, disastrous, entrance, metrical

A related matter is whether base-final sonorants are syllabic in the derivative. As a general
tendency, it is the native suffixes that favour the syllabic treatment. What complicates matters
 9.3 Allomorphy 169

further, however, is the fact that even within a given morphological category, syllabification
of base-final sonorants is often not uniform. For example, with ‑ess, we find non-syllabic
tigress, but also syllabic pantheress, we find non-syllabic huntress, paintress, waitress, but also
syllabic farmeress, speakeress, teacheress. Formations in ‑er also seem variable, with non-­
syllabic fiddler and smuggler, and variably syllabic bottler and meddler. For ‑ous, we find disas‑
trous and monstrous alongside syllabic murderous and feverous. Adjectival derivatives in ‑y
also show variability, sometimes even with the same word (angry, buttery, leathery,
wintery/wintry, wriggly). In sum, it is hard to establish larger generalizations across and
within morphological categories beyond those stated above. Most of the alternations are
simply lexically governed, with similar words showing a tendency to undergo the same
kinds of alternation.
With final nasals, there are other problems. Looking first at examples with a sequence of
two nasals, the sequence may be present only in a subset of derivatives with vowel-initial
suffixes and not in the base form. Examples are given in (7).

with/n/: autumnal, columnar, % columnist, condemnable, condemnatory, condemnation,

(7) 
damnable, damnation, hymnal, % limner, limnology, solemnity, % solemnize
without/n/: autumn, column, condemn, condemning, damn, damning, hymn, limn, limning,

solemn

Final syllabic /m/ is usually resyllabified into the next syllable when the affix begins with a
vowel, as in baptismal, chasmic, orgasmic, phantasmagorical, rhythmic. According to the OED
this happens with native adjectival ‑y as in chasmy and phlegmy. This does not happen, how-
ever, before ‑ing (orgasming).
Another problem concerning syllabification is gemination and degemination. This prob-
lem emerges, for instance, with /l/‑final bases that take /l/‑initial suffixes. Here we can
observe non-uniform behaviour of different suffixes and even of the same suffix with differ-
ent speakers or at different levels of formality, speech tempo, and so on. In the style repre-
sented in pronouncing dictionaries, adverbial ‑ly, for example, geminates with stalely and
vilely, but not with fully and really, nor with all suffixed bases ending in ‑al ( federally, globally,
spiritually), and variably with, for example dully and wholly (see Bauer 2001: 82 for more
data). If the base ends in a syllabic /l/, no gemination occurs, as shown in gently, idly, singly.
The suffix ‑less, on the other hand, hardly ever degeminates in this style, substantiating its
status as non-cohering (petalless, pupilless, sealless, soulless).

9.3.1.4 Vowel alternations

There are many derivatives in which one vowel has a quality different from that of the corre-
sponding vowel in the base. We can distinguish different types of phenomena, with the tradi-
tional terminology labelling the respective phenomenon being based on what happens to the
170 Derivation: phonological considerations

base vowel when the suffix is added: vowel strengthening, vowel laxing, trisyllabic laxing,
trochaic laxing, and vowel epenthesis. We will discuss each in turn.
Example (8) illustrates vowel strengthening, which is a term that refers to a pattern where
a reduced vowel of the base corresponds to a full vowel in the derivative. This kind of alterna-
tion is usually the consequence of a stress shift to the vowel in question.

(8) at[ə]m ◆ at[ɒ]mic ‖ at[ɑː]mic, tot[ə]m ◆ tot[e]mic, Milt[ə]n ◆ Milt[ɒ]nic ‖ Milt[ɑː]nic

◆ Milt[əʊ]nian ‖ Milt[oʊ]nian, aut[ə]mn ◆ aut[ʌ]mnal, manag[ə]r ◆ manag[ɪə]rial ‖
manag[ɪ]rial, r[ ̍ɪə]l ‖ r[i:ə]l ◆ r[i æ]lity
̍

An obvious problem in a derivational approach is to understand which full vowel is cho-

sen, given that the schwa in the base is a vowel that does not contain any information as to
which vowel it might turn into under suffixation. The generative literature has proposed
many different solutions to the problem, none of them being satisfactory (see Giegerich
1999: ch. 5 for a summary and discussion). The most adequate account seems to be one
that makes crucial reference to the orthographic form (see Raffelsiefen 1993: 73–4,
Giegerich 1999: ch. 5). In such an approach, the full vowel is realized as suggested by the
normal spelling-to-pronunciation mapping of the respective letter, given the prosodic
structure of the derivative. This analysis also entails the correct prediction that illiterate
speakers have no way of predicting the right quality of the stressed vowel of atomic if they
only know atom. Furthermore, it makes the correct prediction that there will be variation
as to the pronunciation of some of the vowels given in (8). Giegerich (1999: 163) reports
the results of an experiment in which speakers indeed produced variant forms that satis-
fied both the constraints of the spelling–pronunciation relation and of the morphological
category in question. For example, Handelian came out with [iː], [e] or [eɪ] as its stressed
vowel.
Let us turn to vowel reduction, which can be viewed as something like the reverse of the
vowel strengthening pattern in that the tense vowel or diphthong of the base corresponds to
a lax vowel in the derivative. However, the vowel in question is stressed in both base and
derivative. This is illustrated in (9):

(9) a. convene ◆ convention, deceive ◆ deception, describe ◆ description, produce ◆ production,

retain ◆ retention
b. destroy ◆ destructive, receive ◆ recipient, reside ◆ residual
c. deep ◆ depth, five ◆ fifty, thief ◆ theft, wide ◆ width

In (9a) and (9b) we find non-native bases, in (9c) native bases. Although the historical
roots of the two patterns are completely different, the analysis for both sets of words can
be the same. The phonological alternations are not productive with the kinds of words
 9.3 Allomorphy 171

under discussion, and vowel laxing is therefore best viewed as morphologically and lexically
conditioned base allomorphy. Note also that the quality of the lax vowel cannot be straight-
forwardly derived from the quality of the tense vowel. Morphological or lexical conditioning
is also supported by the fact that the laxing of the vowel is often accompanied by unpredict-
able behaviour of some of the consonants, as in width (where the <d> is variably voiced), and
in deception and production (with an etymologically, but not phonologically, motivated base-
final unvoiced plosive in the derivative).
Trisyllabic laxing is the name of a generalization according to which certain classes of
derived words require that vowels that are three syllables or more from the end of the word
be lax, provided the syllable following the pertinent vowel is unstressed.

(10) a. ser[iː]ne ◆ ser[e]nity, vain ◆ v[æ]nity

b. expl[eɪ]n ◆ expl[æ]natory, comp[iː]te ◆ comp[e]titive

Although trisyllabic laxing has been ascribed the status of a rule in Lexical Phonology and in
Chomsky and Halle (1968), its status is somewhat obscure. Thus, it is not clear when it actu-
ally applies. Part of the problem is that even within morphological categories that frequently
show it, we find variability. This is illustrated in (11) for the suffixes ‑ity, ‑cy, and ‑al.

(11) a. ‑ity: serene ◆ ser[e]nity, but obese ◆ ob[iː]sity, obscene ◆ obsc[e]nity ~ obsc[iː]nity
(Wells 2008)
b. ‑cy: private ◆ pr[ɪ]vacy ǁ pr[aɪ]vacy, but pirate ◆ p[aɪ]racy, diplomat ◆ dipl[əʊ]macy ǁ
dipl[oʊ]macy
c. ‑al: nation ◆ n[æ]tional, but ceremony ◆ cerem[əʊ]nial ǁ cerem[oʊ]nial

A related phenomenon is trochaic laxing. The observable vowel quality alternations are the
same as with trisyllabic laxing, but the environment is different. It concerns the formation of
trochaic disyllables, as in cone /kəʊn ‖ koʊn/ ◆ c[ɒ]nic ‖ c[ɑː]nic or cycle /saɪkl/ ◆ c[ɪ]clic.
Again we find that this alternation is lexically specified and highly variable even for single lexi-
cal items (see, for example, Collie (2008) on c[aɪ]clic ~ c[ɪ]clic).
We turn finally to a type of vowel alternation in which zero alternates with schwa, a phe-
nomenon that can be conceptualized as either vowel epenthesis or as allomorph selection.
This phenomenon can be observed with ‑ness and with ‑ly (see Miner 1975; Wiese 1996b for
details and more data).

(12) a. 
absorbed ◆ absorb[ə]dness, marked ◆ mark[ə]dness, well-formed ◆
well-form[ə]dness
b. amazed ◆ amaz[ə]dly, marked ◆ mark[ə]dly, supposed ◆ suppos[ə]dly
172 Derivation: phonological considerations

Table 9.2 Vowel alternations: overview

Type of adjustment Selected affixes concerned Examples
vowel strengthening ‑an, ‑ic, ‑ity Devonian, satanic, majority
vowel laxing ‑ation, ‑ent, ‑th observation, prevalent, depth
trisyllabic laxing ‑ala , ‑ory, ‑ity criminal, explanatory, profanity
trochaic laxing ‑ic conic
vowel epenthesis ‑ly, ‑ness markedly, markedness

The pertinent mechanism that triggers epenthesis, or the selection of a different allomorph
of the participle ending, makes reference to the foot structure of the base. Derivatives ending
in ‑ness and ‑ly are apparently subject to a restriction for bases of more than one syllable to
have a trochee before the suffix. This restriction is trivially satisfied in words such as develope‑
dness, polishedness, astonishedly, or hurriedly, and is satisfied in words such as markedly through
the selection of a particular allomorph of the verbal participle that would not be licensed by
the phonology of the verbal base (mark[ə]d‑).
Table 9.2 summarizes the vowel alternations.
Table 9.3 presents a different view of vowel alternations, concentrating on the particular
vowels which alternate rather than on the types of morphological process involved. The vow-
els /ɪə/, /eə/ and /ʊə/ are not included in Table 9.3 because they can all be seen as variants
on sequences of vowel and /r/, from which they derive historically. Similarly /ɜ:/ is not
listed.

9.3.1.5 Consonant alternations

There are many kinds of consonant alternation, with some of them being given their own
labels, such as velar softening or spirantization. The examples in (13) illustrate velar softening
with the suffixes ‑ity, ‑ify, ‑ize, and ‑ism.

(13) a. atomic ◆ atomicity, iambic ◆ iambicity, opaque ◆ opacity

b. opaque ◆ opacify, silica ◆ silicify
c. classic ◆ classicize, erotic ◆ eroticize, historic ◆ historicize
d. classic ◆ classicism, historic ◆ historicism, romantic ◆ romanticism

The alternation has been described as a rule according to which a base-final /k/ is realized as
/s/ before certain suffixes. It may appear that this rule is basically an assimilation rule, as the
suffixes all seem to start with a front vowel. This does, however, not hold for ‑ize. The idea of
a neat phonological rule is also called into question by the fact that not all words with base-
final /k/ undergo velar softening. Some that do not are listed in (14).
ernations between vowels
e æ ʌ ɒ ‖ ɑː ʊ iː ɑː ‖ æ ɔː uː eɪ aɪ ɔɪ aʊ

compel
compulsion
long coffee
length caffeine

succeed peace bleed heat

t success pacifist blood hot
France class
s French classify
broad Gaul war psalm
breadth Gallic warrior psalter
root moon school tutorial
radical month scholar tutor
retain table glass slay two
le retention tabular glazier slaughter twain
see
sight
destroy law choose
destruction lawyer choice
flower doubt
floral dubious
holy no globe Aberdeen ghastly vocal vocal
halloween none globular Aberdonian ghost voice vowel
telegraph acid author Elizabeth particle author statute able shire mouth
telegraphy acidic authority Elizabethan particular authority statutory ability Cheshire Portsm
174 Derivation: phonological considerations

(14) zinc ◆ zin[k]ify, anarch(y) ◆ anarchism, monarch ◆ monarchism, Masoch ◆

masochism

Plag (2003: 210–12) discusses possible analyses and demonstrates that no principled phono-
logical or orthography-based account of these facts is possible. This is because the whole set
is etymologically motivated, and the rule does not apply in etymologically irrelevant envi-
ronments. As was the case with other alternations discussed above, which words participate
is ultimately lexically specified, but one regularity persists into English, namely that bases
ending in ‑ic, consistently show the alternation (pace Ohala (1974), who claims that in experi-
mental conditions, native speakers do not consistently use velar softening in relevant items
that are unfamiliar).
Another consonantal alternation that is quite frequent in English derivation is spirantiza-
tion. It is a pattern where base-final /t/ and /d/ in the free base correspond to /s/, /ʃ/ or /ʒ/
in the derivative, and base-final (underlying) /r/ corresponds to /ʒ/. Example (15) illustrates
this for some of the morphological categories where the alternation can be found.

(15) a. /t/ ~ /s/: diplomat ◆ diplomacy, pirate ◆ piracy, president ◆ presidency

b. /t/ ~ /ʃ/: equate ◆ %equation, erupt ◆ eruption, president ◆ presidential, protect ◆
protection, radiate ◆ radiation
c. /t/ ~ /ʒ/: equate ◆ %equation
d. /d/ ~ /s/: conclude ◆ %conclusive, corrode ◆ %corrosive, persuade ◆ %persuasive
e. /d/ ~ /ʒ/: conclude ◆ conclusion, erode ◆ erosion, persuade ◆ persuasion
f. /d/ ~ /z/: corrode ◆ %corrosive, divide ◆ %divisive, divisible, persuade ◆ %persuasive
g. /r/ ~ /ʒ/: adhere ◆ %adhesion, cohere ◆ %cohesion
h. /r/ ~ /s/: adhere ◆ %adhesive, cohere ◆ %cohesive
i. /r/ ~ /z/: adhere ◆ %adhesion, cohere ◆ %cohesion
j. /s/ ~ /z/: spouse ◆ %spousal

Note that with /t/, the voicing specification of the segment in question is normally the same
in both forms, but not always, as shown by the example in (15c). Equation has a voiced frica-
tive in some varieties instead of the expected unvoiced one. The reverse voicing patterns are
also attested. Voiced /d/ regularly alternates with unvoiced /s/ in ‑ive derivatives, but persua‑
sive is also attested with a voiced base-final fricative, as shown in (15f). As far as we know
there is no good description of the variation which clearly exists here, let alone any insight
into whether there is any patterning to the variation.
Another alternation concerns suffixes, or allomorphs of suffixes, that have an initial front
vowel or glide, such as ‑ial, ‑ion, or ‑ure. They trigger palatalization of the base-final segment,
sometimes accompanied by voicing, as shown in (16).
 9.3 Allomorphy 175

Table 9.4 Velar softening, spirantization, and palatalization: overview

Alternation (Selected) affixes concerned Examples

velar softening
/k/ ~ /s/ ‑ify, ‑ism, ‑ity, ‑ize silicify, classicism, elasticity, eroticize
spirantization
/t/ ~ /s/ ‑y piracy
/t/ ~ /ʃ/ ‑ation eruption
/t/ ~ /ʒ/ ‑ation %
equation
/d/ ~ /s/ ‑ive %
conclusive
/d/ ~ /ʒ/ ‑ation conclusion
/d/ ~ /z/ ‑ive %
persuasive
/r/ ~ /ʒ/ ‑ation %
adhesion
/r/ ~ /s/ ‑ive %
adhesive
/r/ ~ /z/ ‑ive %
adhesive
/s/ ~ /z/ ‑al %
spousal
palatalization
/s/ ~ /ʃ/ ‑al, ‑ion circumstantial, convulsion
/s/ ~ /ʒ/ ‑ation precision
/z/ ~ /ʒ/ ‑ation, ‑ure confusion, composure
/t/ ~ /tʃ/ ‑ure departure, implicature
/ɡ/ ~ /dʒ/ ‑y, ‑ize analogy, apologize

(16) a. /s/ ~ /ʃ/: circumstance ◆ circumstantial, %erase ~ erasure

b. /s/ ~ /ʒ/: concise ◆ concision, precise ◆ precision
c. /z/ ~ /ʒ/: confuse ◆ confusion, excise ◆ excision, %erase ◆ erasure, expose ◆ exposure
d. /t/ ~ /tʃ/: depart ◆ departure, implicate ◆ implicature, moist ◆ moisture
e. /ɡ/ ~ /dʒ/: analogue ◆ analogy, monologue ◆ monologic

Note that the set of pairs that show a certain alternation can be extremely small and even
within the small set there may be variation. For example, monologize is attested with both of
the alternating sounds. A summary of the alternations discussed so far is given in Table 9.4.
The inclusion of a suffix in the table does not mean that all potentially pertinent forms show
the respective alternation, but that at least one or two do so.
We should also note the rather unusual alternation between <w> and <v>, which extends
to <f(f)> and <v> in proper names: Shaw ◆ Shavian, Aronoff ◆ Aronovian.
176 Derivation: phonological considerations

A quite peculiar set of consonantal phenomena concern certain combinations of nasals

with one other consonant. These combinations appear as combinations in certain derived
words, but that other consonant is missing from the free base, or from other kinds of deriva-
tives. The phenomenon is restricted to very few words, and is illustrated in (17) through (19).
Item (17) concerns the clusters <gn> and <gm>, (18) shows <mn>, and (19) <mb>.

(17) /n/or/m/ /n/or/m/ /ɡn/or/gm/

assign assignment assignation
design designer designate
paradigm paradigmer paradigmatic
sign signing signal

(18) /m/ /m/ /mn/

autumn autumnish autumnal
hymn hymnless hymnic
solemn solemnly solemnize

(19) /m/ /m/ /mb/

bomb bomber bombastic, bombard
crumb crumby crumbly

A final set of consonantal alternations concerns cases where we find singleton consonants
in the derivative that do not surface in the base, as in Plato ◆ Platonic, aroma ◆ aromatic.
Such extenders raise the question of whether they belong morphologically to the base or
to the affix, and they will therefore be dealt with, together with similar pertinent cases in
Section 9.3.3.

9.3.1.6 Classically motivated allomorphy

There is a set of irregular instances of stem allomorphy inherited from Latin, but apparently
used morphologically in English. From the English point of view, these variants are only
partly predictable in terms of patterns in the lexicon, some of which have been discussed
above. Some examples of such alternations are provided in (20).

(20) assume assumption

conceive conception
conjoin conjunction
deduce deduction
evolve evolution
move motion
 9.3 Allomorphy 177

To the extent that they involve vowel alternations, these can be seen as special cases of what
we might term ‘classical ablaut’ in English. Here we have instances of vowel change (not
infrequently accompanied by consonant change) which reflect the apophony of inflectional
systems in the classical languages (mainly Latin), and sometimes the changes between Latin
and French. The variation is not motivated in English, and it is probable that these patterns
should be seen as part of English etymology rather than as part of English morphology.
What we find is that the same historical element arises in several different forms, either
because of the phonological changes which have affected it in the transition between the
various donor languages, or because of the patterns of apophony in the individual languages.
Some examples are provided in (21), with the common etymological elements marked in
bold font.

(21) Greek ‘bear, carry’: semaphore, periphery

Greek ‘work’: ergonomics, organ, metallurgy
Latin ‘sit’: session, preside, sedentary, supersede
Latin ‘act’: agent, action, intransigent
Latin ‘fall’: cascade, accident, decadence
Latin ‘law’: legal, loyal, privilege
Latin ‘rule’: regular, rigorous, direct, rule
Latin ‘take’: capture, conceive, conception, recuperate
Latin ‘throw’: project, ejaculate
Latin ‘tip’: apex, apical
Latin ‘touch’: tactile, contiguous, contaminate, tangible
Latin ‘vessel’: vase, vessel
Latin ‘year’: annual, biennial

9.3.1.7 Base allomorphy: a summary

In this overview of base allomorphy in English derivation we have seen that there are recur-
rent patterns of morpho-phonological alternations. Each of these patterns is regular in the
sense that it can be described in terms of a systematic correspondence between basic and
derived phonological representations. However, it has also become clear that the domains in
which the respective alternations can be observed are not easily predictable. Certain groups
of affixes may show a particular type of alternation, but the next type of alternation one looks
at is unlikely to be found with exactly the same set of affixes. Furthermore, even within a
given morphological category, the alternation does not consistently apply to all words that it
could apply to, and even individual words may vary across or within speakers. The following
chapters will provide much more empirical evidence for these conclusions, when each deri-
vational morphological category is described in detail.
178 Derivation: phonological considerations

9.3.2 Affix allomorphy

When talking about affix allomorphy in derivational morphology, one faces the general prob-
lem of distinguishing allomorphs from morphemes. In other words, a primary consideration
is whether two formatives express the same meaning. If that is the case, the two formatives
can be considered as two potential allomorphs. Unfortunately, the semantics of many affixes
is quite intricate and it is often not straightforward to tell whether one is dealing with two
allormorphs of one affix, or with two different affixes. Our decisions are based on large-scale
empirical evidence and are as theory-independent as possible.

9.3.2.1 Suffix allomorphy

There are many derived words in which the derivative seems to feature a variant of the suffix
in question, or where, under an alternative analysis, the base shows a variant form. In many
cases, the decision about which analysis is preferable is unclear or even impossible to make.
In this section we discuss only the clear cases, that is those suffixes where we can make a case
for suffix allomorphy.
Some suffixes have only distinct orthographic variants that do not show any phonetic–
phonological distinctness: <able> ~ <ible>, <er> ~ <or>, <ant> ~ <ent>. Their distribution
will be discussed solely in the respective chapters. With regard to the real allomorphies, we
will see in this general overview that phonological and morphological restrictions are mostly
responsible for the distribution of the allomorphs. For full discussion of the details involved
in the respective allomorphy, the reader is referred to the individual chapters where these
suffixes are treated at length. We first discuss some examples of suffixes whose realization is
subject to phonological constraints. These constraints may involve restrictions on the seg-
mental or on the prosodic level.
The suffix ‑al surfaces variably as <ar> or <al>, depending on the phonology of the
base. In general, bases that end in an /l/ (or /l/ plus schwa, the latter being truncated)
prefer the rhotic variant, as shown in (22). The restriction may also be expressed in
prosodic terms as one against identical syllabic constituents filled by a liquid not in a
cluster in the derived word. This would also account for the allomorph selection shown
in (22b), but this restriction is only variably satisfied (see Raffelsiefen 1999 and Chapter
14 for details).

(22) a. arteriole ◆ arteriolar, corolla ◆ corollar, enamel ◆ enamelar, fibrilla ◆ fibrillar, nodule ◆
nodular, protocol ◆ protocolar
b. alga ◆ algal versus column ◆ columnal ~ columnar, lamin ◆ laminal ~ laminar

The nominal suffix ‑ery has two variants, <ery> and <ry>. The distribution is prosodically
governed, with ‑ry attaching to disyllabic feet, and ‑ery to monosyllabic feet, as shown in
(23a) and (23b), respectively. Counterexamples exist, however, as in (23c).
 9.3 Allomorphy 179

(23) a. chíckenry, phéasantry, rábbitry

b. hénnery, píggery, róokery
c. mónastery, láundry

The nominal suffix ‑ation has the allomorphs ‑ation (industrialization), ‑cation (personifica‑
tion), ‑ion (hyphenation), ‑sion (recursion), and ‑tion (absorption). The allomorphs ‑ation,
‑cation, and ‑ion are productive and they exemplify morphologically conditioned allomor-
phy: ‑ation attaches to ‑ize, ‑cation to ‑ify (with concomitant vowel laxing), ‑ion to ‑ate (with
concomitant palatalization). This is illustrated in (24).

(24) a. randomize ◆ randomization, verbalize ◆ verbalization

b. personify ◆ personification, magnify ◆ magnification
c. carbonate ◆ carbonation, concentrate ◆ concentration

There are also cases in which suffix allomorphy is induced by the suffixation of already suf-
fixed words. For example, ‑ic has allomorphs that only emerge if the ‑ic derivative undergoes
suffixation by ‑ity or ‑an (in the allomorph ‑ian), following the rules of velar softening and
palatalization, respectively, as in electricity and electrician. The suffix ‑ify has an allomorph
/ɪfɪ/ if combined with the pertinent allomorph of ‑ation, as in magnification, personification.

9.3.2.2 Prefix allomorphy

There are two fundamental types of prefixal allomorphy in English, a type which is based on
allomorphy in the classical languages and a type which is based on assimilation which func-
tions as a part of English word-formation. This second type has two sub-types (although the
phonetic facts are not well described): they are commonly termed obligatory and optional
assimilation (e.g. Weber 2002).
The classical type works with a very few non-native prefixes whose allomorphs are origi-
nally borrowed from the donor languages. Typically, these involve prefixes which vary
between a final vowel and a final consonant, with the consonant-final prefix occurring prevo-
calically (sometimes also before /h/). Some examples are given in (25). (See Chapter 19 for
discussion of the instances with ‑o‑ in neoclassical formations.)

(25) a-sexual an-hydrous an-aerobic

a-historical
anti-body ant-helion ant-acid
anti-hero
bi-labial bin-ocular
dia-chronic di-optre ‖ di-opter
180 Derivation: phonological considerations

endo-cardium end-arteritis
exo-cellular ex-arteritis
exo-arteritis
hypo-dermic hyp-aesthesia ‖ hyp-esthesia
hypo-allergenic

With the prefix in‑ we have a case of so-called ‘obligatory’ assimilation. The negative
prefix in‑ displays assimilation with a following labial consonant (impossible, immature,
imbalance). With a following /r/ or /l/, the writing system sees an assimilation of the
final nasal of the prefix to the following sonorant, but phonetically the nasal is deleted:
irrelevant /ɪreləvənt/, illegal /ɪliːɡəl/. Before a vowel, /h/, or any other alveolar conso-
nant, the form of the prefix is in‑ /ɪn/. This is usually seen to be the limit of the obliga-
tory assimilation, but this takes no account of the fact that the prefix can also occur
before a labio-dental (infrequent), before a post-alveolar (injudicious), before a velar
(incautious). The in‑ prefix appears not to occur before a dental ([θ, ð]). In these cases,
the facts of variation are unfortunately unclear, except that it is clear that there is varia-
tion. Either the nasal can stay unassimilated as [n], or it can become assimilated to the
point of articulation of the following consonant ([ɱ], [ ɲ], [ŋ]), or there can be a double
articulation.
The instances of so-called obligatory assimilation are ‘obligatory’ largely in as much as the
assimilation is expected to be marked in the orthography. However, in COCA we find exam-
ples like inbelief (see discussion in Chapter 17) which suggest that the orthographic obligato-
riness may not be automatic.
The kind of assimilation before labiodentals, post-alveolars, and velars is like the
optional assimilation we find with un‑. There is no orthographically sanctioned assimila-
tion of the final <n> of un‑, but the /n/ may be assimilated to the place of articulation of
any following consonant, as in unbelievable, unfavourable, unthinking, unjust, unusual,
unconscious, unwanted. Whether the rates of assimilation and the factors influencing
assimilation (formality, speech tempo, regional dialect, etc.) are the same for incautious
and unconscious is not clear.
The prefix en‑ and the neoclassical element syn‑ both show orthographically-licensed
assimilation before bilabials (emplane, symbiosis), but en‑ does not before /r/ or /l/ (enlist,
enroll), and syn‑ does not seem to be found in relevant environments. The historical Latin
prefix con‑ in words of Latin origin behaves like negative in‑ (condemn, compel, collate, correct),
but is no longer recognizable as an English prefix. The Latin prefix ad‑, no longer a prefix in
English, has an even wider range of forms.
There are a few other cases of assimilation involving prefixes, but they are not systematic,
and are variable, and can be found in places where there is no morphological boundary or a
word boundary. They include %mi[z]anthropy, %mi[ ʃ ]chance, %mi[ ʃ ]shapen.
 9.3 Allomorphy 181

9.3.3 Extenders
Many derived words are characterized by seemingly meaningless formatives that occur
between bases and particular suffix allomorphs. We have introduced the term extender for
these elements (see Chapter 2.3.2). Some examples are collected in (26). The examples in
(26a) show extenders directly on a base, those in (26b) extenders on what might be analysed
as other extenders. This is done merely for the sake of clarity, rather than to make a strong
morphological claim. The sequences of extenders in (26b) could just as easily have been
analysed as single, longer extenders.

(26) a. extender suffixes affected examples

‑at‑ ‑ic, ‑ism, ‑ure phlegmatic, lymphatism, signature
‑c‑ ‑able, ‑ation, ‑ism, ‑y applicable, justification, witticism,
chieftaincy
‑e‑ ‑an, ‑ous Caesarean, gaseous
‑i‑ ‑al, ‑an, ‑ary, ‑ation, ‑er tutorial, grammarian, presidentiary,
distanciation, hosier
‑in‑ ‑al, ‑ation attitudinal, pagination
‑it‑ ‑ation, ‑ive, ‑ize, ‑or, ‑ure competition, sensitive, sensitize,
competitor, divestiture
‑l‑ ‑ese Congolese
‑n‑ ‑ese, ‑ic, ‑ism, ‑ist, ‑ize Balinese, Platonic, Platonism,
tobacconist, Platonize
‑od‑ ‑ic spasmodic
‑s‑ ‑ation recursion
‑t‑ ‑eer, ‑ic, ‑ism, ‑ist, ‑ize, ‑ure muleteer, dramatic, egotism,
dramatist, dramatize, mixture
‑u‑ ‑al, ‑ous processual, sensuous
‑ul‑ ‑ar cellular

b. extender extender examples

‑c‑ ‑at‑ sensificatory
‑it‑ ‑i‑ dietitian
‑n‑ ‑i‑ Torontonian
‑v‑ ‑i‑ Peruvian
182 Derivation: phonological considerations

Two main questions emerge from such data: where does the extender belong, and what kinds
of generalizations are possible as to when which extender occurs? We will deal with each
question in turn.
Especially in the early generative literature (e.g. Chomsky and Halle 1968; Aronoff 1976),
there are attempts to solve the segmentation problem by treating the extender as part of the
base and devising a number of deletion or truncation rules to derive the attested shape of the
derivatives. As pointed out, for example, by Szymanek (1985), such attempts fail to provide
an accepted methodology to arrive at a convincing theory-independent account of the allo-
morphy. For example, the postulation of a stem presidenti‑ may nicely give us presidential
when ‑al is attached, but this solution necessitates the deletion of the extender in other for-
mations (e.g. presidentship). Even worse, the base sense takes an extender ‑it‑ in sensitive but
‑u‑ in sensual, so the bases are not necessarily consistent in which extender they take. Similar
arguments hold for other extenders in other morphological categories. In the chapters that
follow we discuss for each morphological category the status of potential extenders. In most
cases, it is impossible to come up with a clear answer to the segmentation question, but
sometimes distributional patterns can be used to justify a particular analysis independently.
Let us turn to the second question, the predictability of occurrence. Very often, extenders
are found with forms borrowed from Latin or Greek, and as such, English has inherited only
a limited number of extenders, often recurrent across morphological categories. Also due to
these historical roots in loan words, the appearance of extenders is largely lexically governed
and not predictable. For example, it is not evident from a present-day English perspective
why aroma takes an extender when suffixed by ‑ize (aromatize), while patina does not (pati‑
nize). There are, however, certain pockets of regularity that can be either phonologically or
morphologically characterized. Thus, with the exception of doctor and pastor, agent nouns in
‑or generally take the extender ‑i‑ when suffixed with ‑al (ambassadorial, appressorial, directo‑
rial, editorial, fornicatorial, narratorial, professorial, procuratorial, proprietorial, senatorial).
Such patterns, where existent, will be identified in the descriptions of individual morphologi-
cal categories in the chapters to follow.

9.4 Prosody
As already mentioned, many suffixes impose a particular prosodic structure on their deriva-
tives. In other words, many morphological categories are characterized not only by their
members sharing a specific affix, but also by sharing a specific prosodic structure. In order to
satisfy the restrictions imposed by such structural templates, derivatives often show phono-
logical characteristics that differ from that of their bases. In the previous sections we have
already seen that these characteristics may concern stress shifts, truncation, vowel alterna-
tions, or consonant alternations. In this section we will have a closer look at allomorphies
that have their roots in prosodic adjustments. As mentioned above, prosodic adjustments
 9.4 Prosody 183

can be made through stress shift, or through truncation or epenthesis. We have already dis-
cussed the latter two by way of examples from ‑ize, ‑ness, and adverbial ‑ly, and will concen-
trate on stress adjustments in this section.

9.4.1 Auto-stressed affixes

Auto-stressed affixes are characterized by the fact that each of them carries a specific stress:
either primary or secondary. This can happen with prefixes and suffixes. As follows from
general English stress rules, tense vowels and diphthongs carry stress by default (at least
secondary stress). Affixes with such vowels predictably carry secondary stress, as is the case
with, for example, the suffixes ‑ate, ‑ite, ‑oid, and ‑ize, or the prefixes de‑, post‑, and pre‑. In
other words, there is nothing morphological about this stress-behaviour. However, there are
also affixes that may carry secondary or primary stress without there being a purely phono-
logical reason. We will first discuss the pertinent prefixes, then the suffixes.

9.4.1.1 Prefixes
A closer look at prefixes such as dis‑, in‑, or mis‑ reveals that their stress behaviour is variable
and seems to be influenced by the base to which they attach. Thus, for all three prefixes we
find unstressed, secondarily stressed, and main stressed attestations. This is shown in (27):

(27) a. unstressed: disown, ingenious, immobile, mistake, misanthropist

secondary stress: disallow, disendow, illegitimate, inconsistent, mispronounce,
b. 
missell, misname
c. main stress: disparate, indolent, infamous, misfit

All three prefixes have secondary stress as their default, but two patterns emerge with regard
to the unstressed and secondarily stressed variants. The prefixes dis‑ and in‑ are unstressed
when adjacent to a main stress, which can be interpreted as a stress clash effect. Words with
main stress on these two prefixes are very rare and involve words with bound bases. The
prefix mis‑, in contrast, rarely occurs unstressed or main stressed, and if so, only in words with
bound bases. Thus, mis‑ is not subject to stress clash repair.
The polysyllabic prefixes, such as anti‑, hyper‑, mini‑, or semi‑ seem to variably occur either
with a primary stress (ánticline, hýpertext, mínivan, sémigloss) or with a secondary stress (ànti-
clóckwise, hỳperáctive, mìni-róundabout, sèmifínal). If they occur with primary stress, the
stressed positions of the base are preserved (as secondary stresses).

9.4.1.2 Suffixes
Apart from suffixes that are secondarily stressed due to their segmental phonological makeup,
there are some suffixes that are main-stressed. The list of pertinent suffixes is rather short:
184 Derivation: phonological considerations

‑ana, ‑ee, ‑eer, ‑ese, ‑esque, ‑ette, ‑teen. In Caribbean and IrE, the suffix ‑ize is also main-stressed.
The suffix ‑ess may be stressed in some words, for example (variably) in stewardess, but is
usually not stressed (goddess, tigress). Another candidate for the status of main-stressed suffix
is the suffix ‑ation, particularly with its allomorphs ‑ation and ‑cation, but if we take all its
allomorphs into consideration, that is, including ‑ion, which attaches, for example, produc-
tively to verbs in ‑ate, the best analysis seems to be the one that says that ‑ation derivatives
have their stress on the penultimate syllable (hyphenátion, convérsion). The suffix is therefore
better classified as stress-shifting, and will be treated in Section 9.4.2.
The obvious question with the auto-stressed suffixes, phrased in procedural terms, is what
happens to the stresses of the bases when the stress shifts to the suffix. In general, we can
observe stress preservation, that is, what is a stressed syllable in the base is also a stressed
syllable in the derivative, with the complication that primary stress is demoted to secondary.
This is illustrated in (28a). However, we also observe (sometimes variably) non-preservation
of stress with at least some of the suffixes, as shown in (28b) for ‑ana, ‑ee, and ‑ese.

(28) a. 
Hóllingworth ◆ Hòllingworthiána, cóntact ◆ còntactée, áuction ◆ àuctionéer,
Américan ◆ Amèricanése, Chómsky ◆ Chòmskyésque, lémon ◆ lèmonétte
b. Níxon ◆ Nixòniána, Thómpson ◆ Thompsòniána, invést ◆ ìnvestée, emplóy ◆ employée
~ èmployée, Japán ◆ Jàpanése, Tàiwán ◆ Tàiwanése

In Section 9.4.3 we will have a closer look at stress-preservation and the principles that
govern it.

9.4.2 Stress-shifting affixes

There are quite a number of morphological categories that produce a relatively fixed prosodic
output in their derivatives. Affixation with these categories then necessarily involves a com-
plex relationship between the base and the derivative in all those cases where the base, as part
of the derivative, does not satisfy the prosodic restrictions imposed by the affix. From a pro-
cedural perspective, the affixes in question are capable of shifting the stress as found in the
base to a position different from where the stress is in the base. It is only the suffixes that
productively induce such prosodic adjustments in English.
We can distinguish between different kinds of stress-shifting suffixes, sometimes referred
to as strong, weak, and long retractors (Liberman and Prince 1977; Hayes 1982). In this termi-
nology, ‘retraction’ refers to the movement of stress away from the end of the word. The term
is not restricted to morphologically complex words, but, as we will see, it is evident that
retraction is often dependent on the morphological structure. We restrict our discussion to
morphologically complex words, where strong retractors are suffixes that force the stress of
the derivative always to end up on the same syllable counting from the word end, that is
either on the antepenultimate, or on the penultimate syllable, irrespective of the syllabic
 9.4 Prosody 185

structure of the syllables involved. For example, ‑ation derivatives all have stress on the penul-
timate syllable, irrespective of the syllabic structure. Examples of strong retraction, including
‑ation derivatives, are given in (29).

(29) a. ‑ation: cohésion, conventionalizátion, expectátion, personificátion, starvátion

b. ‑ity: absúrdity, acceptabílity, obésity, productívity

Weak retraction is a pattern that is characterized by the stress position being co-determined
by the weight of the penultimate syllable. In weak retraction, if the penult is heavy, stress falls
on the penult, and if the penult is light, stress falls on the antepenult. In (30) we provide
examples with suffixes that are supposedly weak retractors, among which we often find bound
bases whose stress pattern is hard to discern independently anyway. In such cases the notion
of stress shift is somewhat inappropriate. Furthermore, counterexamples to the proposed
patterns can be found, as demonstrated for ‑oid in (30d). Other productive suffixes that show
weak retraction (often variably, though) are the adjectival ones, ‑ary, ‑ate, ‑ative, ‑atory, ‑ile,
‑ory, and ‑ose (see further Chapter 14), and nominal ‑ary, ‑ory, ‑ose. Note that with polysyl-
labic suffixes the picture becomes more complicated, as the syllable whose weight is assumed
to be decisive for stress assignment may no longer be the penultimate, but the antepenulti-
mate, see (30e).

(30) a. ‑ide: cýanide peróxide

b. ‑ine: elephántine, ívorine, labyrínthine, víctorine
c. ‑ite: ánnamite, archimándrite, molýbdenite, sélenite
d. ‑oid: archándroid, méteoroid, but bénz[iː]oid, Cáledonoid
e. ‑ary: eleméntary, probátionary, státionary

Long retraction refers to words whose stress ends up on the third syllable before the suf-
fix. This is illustrated in (31) for the adjectival suffix ‑ory. It seems, however, that such
cases are best analysed as simple cases of stress preservation (see also Liberman and
Prince 1977: 277).

(31) antícipatory, artículatory, hallúcinatory, manípulatory, stípulatory

Also with long retraction, the same suffix is able to show different kinds of behaviour, some-
times even with the same word, for example compénsatory (weak retraction) ~ cómpensatory
(long retraction), respíratory (weak retraction) ~ réspiratory (long retraction), eleméntary
(weak retraction, see (30e)) ~ dísciplinary (long retraction). Such data call into question the
neat classification of suffixes into the various classes.
186 Derivation: phonological considerations

A particularly intricate case of stress-shift behaviour is the suffix ‑able, which nicely show-
cases the complexities of the matter, as illustrated in (32) and (33). In (32a) and (32b) we find
cases of the most prevalent pattern, which is for there to be no stress shift, no matter whether
we have iamb-final bases (32a) or trochee-final bases (32b). In (32c) we illustrate cases of lexi-
calized stress shifts with bases that normally would not undergo stress shift.

(32) a. abrídge ◆ abrídgeable, abúse ◆ abúsable, besmírch ◆ besmírchable, desíre ◆ desírable,

decláre ◆ declárable
b. álter ◆ álterable, ánswer ◆ ánswerable, búdget ◆ búdgetable, bálance ◆ bálanceable,
chállenge ◆ chállengeable
c. %admíre ◆ ádmirable, %compáre ◆ cómparable, %prefér ◆ préferable

Only with bases of three or more syllables that do not end in an iamb or a trochee do we
find more regular stress shifts with ‑able. There are three different patterns, given in (33).
The first set of words has bases that end in a dactyl. These words do not show stress
shift. If, in contrast, the antepenultimate syllable of the derivative is heavy, we find two
different patterns. One subset of words seems to stress shift categorically (33b), the other
variably (33c).

(33) a. jét.ti.so.na.ble, mó.ni.to.ra.ble, ré.gis.te.ra.ble

b. al.lo.cá.ta.ble, a.ro.ma.tí.za.ble, ar.ti.cu.lá.ta.ble, ca.te.go.rí.za.ble, com.mer.cia.lí.za.ble,
cul.ti.vá.ta.ble, do.cu.mén.ta.ble, ex.tra.dí.table, ma.ni.pu.lá.ta.ble, pre.di.cá.ta.ble
c. á.na.ly.za.ble ~ a.na.lý.za.ble, cér.ti.fy.a.ble ~ cer.ti.fý.a.ble, drá.ma.ti.za.ble ~ dra.ma.tí.
za.ble, i.dén.ti.fi.a.ble ~ i.den.ti.fí.a.ble, í.te.mi.za.ble ~ i.te.mí.za.ble, ló.ca.li.za.ble ~
lo.ca.lí.za.ble, per.só.ni.fy.able ~ per.so.ni.fý.able, ré.cog.ni.za.ble ~ re.cog.ní.za.ble

Very often, the stress-shifted derivatives in productive morphological categories are lexi-
calized exceptions, as we saw with ‑able. Another example of this is ‑ate, which has been
claimed to be a strong retractor in spite of the many exceptions (e.g. Liberman and Prince
1977: 275), but the evidence from neologisms strongly suggests that stress shift is generally
not productive (Plag 1999). The same holds for ‑ify, and ‑ize (Plag 1999), which have
sometimes been treated as stress-shifting (e.g. Zamma 2003). Kettemann’s (1988) results
demonstrate, however, that even those suffixes that do not productively shift the stress
may induce stress shifts for very narrowly defined subsets of similar words, as shown by an
experiment in which subjects had to attach ‑ify to bases ending in /ɪd/, and frequently
created stress-shifted forms in obvious analogy to the stress-shifted lexicalized forms
humídify or solídify.
As the above examples have shown, in spite of the many claims to the contrary, it is often
not clear which suffixes belong to which of the four classes, non-stress-shifting, weak
 9.4 Prosody 187

retractor, strong retractor, long retractor. Many suffixes show stress shift for a subset of words,
and no stress shift, or a different kind of stress shift, for other words even though the differ-
ently treated bases have the same prosodic structure. Thus, Liberman and Prince’s (1977:
276) statement still holds that ‘the emerging picture is that retraction in complex words is
largely controlled by suffix type, but admits of considerable lexical variation, particularly
among the less productive morphological categories’.
This leads us to another source of apparent irregularities in the prosodic behaviour of
derivatives, the relationship of derivatives to other derivatives. Traditional analyses of pro-
sodic alternations have concentrated on the relationship between the less complex form, that
is, the base, and the more complex form, the derivative. However, if we also take other com-
plex forms into consideration as potential bases, many apparent exceptions turn out to be in
regular correspondence with patterns in those other potential bases. Thus, stress is not shifted
in humídify if, instead of the base húmid, we take the corresponding ‑ity derivative to be the
relevant related form: humídify ◆ humídity (cf. solídify ◆ solídity). Similarly, the variable stress
of démonstrable ~ demónstrable may be a consequence of the competing related forms démon‑
strate and demónstrative. This kind of approach, or the phenomenon that it tries to capture,
has come under various names in the literature (with variant underlying theoretical assump-
tions), namely ‘stem selection’ (Raffelsiefen 2004: 95, with many more such examples and
explicit analyses), ‘paradigm uniformity’ (e.g. Steriade 2000), ‘multiple correspondence’ (e.g.
Burzio 1998), or the ‘split-base’ effect (Steriade 1999).

9.4.3 Stress preservation

As mentioned earlier, prosodic alternations not only pose the question of whether the derived
word is stressed on the same syllable as the base, but also raise the problem of where the
derivative is main stressed. There is also the problem of what happens to the main stressed
syllable of the base when main stress is shifted to another syllable under suffixation. This
problem has been given the name of stress preservation, but we will see that this derivational
terminology and perspective is not always helpful, as it narrowly focuses on the relationship
between the less complex base and its more complex derivative. In contrast, one can approach
the problem from a more general perspective of lexical relatedness, according to which there
are constraints on how the prosody of derived words must correspond to the prosodies of
different kinds of related word (see also Chapter 23.2).
In cases of stress shift, we frequently find a pattern according to which the primary stress
of the base corresponds to a secondary stress in the derivative. Such cases are illustrated in
(34) for ‑ation and ‑ity.

(34) a. ‑ation: ábdicate ◆ àbdicátion, mágnify ◆ màgnificátion, rándomìze ◆ ràndomizátion

b. ‑ity: degrádable ◆ degràdability, sensátional ◆ sensàtionálity, táciturn ◆ tàcitúrnity
188 Derivation: phonological considerations

However, we also find patterns in which stress preservation is systematically absent, as in

(35). Failure to preserve stress can be generally observed if the syllable in question is imme-
diately adjacent to the derivative’s main stress. The examples in (35) show both preservation
and lack of preservation.

(35) a. ‑ese: Cantón ‖ Cánton ◆ Càntonése versus Japán ◆ Jàp[ə]nése

b. ‑ity: degrádable ◆ degràdabílity versus canónic ◆ càn[ə]nícity, cómplex ◆
c[ə]mpléxity
c. ‑an: Ámazon ◆ Àmazónian versus phonétic ◆ phòn[ə]tícian, Apóllo ◆ Àpp[ə]lónian

As noted by many authors, there is also a third pattern, comprising forms that do not
conform to either of the above two patterns. Such forms preserve stresses (at least varia-
bly) in spite of the fact that these stresses are adjacent to a main stress in the derivative.
This is shown in (36). In accordance with perhaps most of the literature, we see reduction
to schwa as evidence for unstressedness, a full vowel as evidence for at least some degree
of stress. Those analysts who would not assign stress to the vowels in question have to
deal with the question of variable vowel qualities instead. The problem remains essentially
the same.

(36) auth[e]ntícity, cre[eɪ]tívity, exp[e]ctátion, exp[ɔː]rtátion, f[æ]scístic,

impl[ɑː ǁ æ]ntation

Notably there is no aberrant behaviour in the opposite direction, that is, there is no complete
loss of stress when not adjacent to a main stress. This fact, among many others, shows that
prosodic faithfulness is an important principle in English derivational morphology that is
overwritten only by specific requirements of morphological categories, or by strong interven-
ing phonological principles, such as the avoidance of stress clashes (see, e.g. Pater 2000).
The many lexical exceptions to stress preservation can be understood better if we take
frequency into account. Collie (2008) investigates a corpus of 198 five- and six-syllable
complex words with three pre-tonic syllables and various suffixes (taken from Jones 2003),
in which regular stress preservation would have been expected to occur on the second
syllable counting from the left, as in anticipation or sensationality. The majority of the forms
indeed show the predicted behaviour, but about one-quarter have variable stress preserva-
tion (authórity ◆ àuthoritárian ~ authòritárian), and an additional 4 per cent are listed with
stress on the first syllable (repátriate ◆ rèpatriátion). An investigation of the frequencies of
the bases and derivatives involved reveals that a higher frequency of the base and a lower
frequency of the derivative favour stress preservation. In other words, stress preservation
is the default for productive derivation, and failure to preserve stress correlates with
lexicalization.
 9.5 Haplology 189

9.5 Haplology
In this section we discuss a family of rather widespread suffixal phenomena that are all char-
acterizable by haplology, that is by the avoidance of identical phonological structure in mor-
phologically complex words, usually coming about by the addition of a suffix. Where we find
haplology it typically arises in a particular morphological environment, not just a phonologi-
cal environment: that is, the particular affixes involved are important for the application (or
non-application) of haplology. We have already discussed one pertinent case, the allomorphy
of the suffix ‑al, but the phenomenon is much more common, as the data in (37) to (39)
show. There are three different sorts of haplological effect, the first of which is the choice of a
particular allomorph as in (37). The only suffix that has this possibility in store to handle a
haplology effect seems to be adjectival ‑al, which we have already discussed in Section 9.3.2.1.
We repeat the pertinent data in (37) for convenience.

(37) arteriole ◆ arteriolar, corolla ◆ corollar, enamel ◆ enamelar, fibrilla ◆ fibrillar, nodule ◆
nodular, protocol ◆ protocolar

The second type of effect is the phonological adjustment of the base, which is much more
common, as shown in (38a). The constraint against identical onsets leads to truncation with
‑ess and ‑ize. This constraint may interact with other prosodic constraints, for example with
those on foot structure. A comparison of the data in (38a) with those in (38b) shows, for
example, that identical onsets in ‑ize‑formations are not repaired with trochaic bases, but
only with dactylic ones (see Plag 1998, 1999 for a full analysis).

(38) a. adventure ◆ adventuress (*adventu.re.ress), murderer ◆ murderess (*murde.re.ress),

sorcerer ◆ sorceress (*sorce.re.ress), feminine ◆ feminize (*femi.ni.nize), metathesis ◆
metathesize (*metathe.si.size)
b. classicize, canonize, candidize, mirrorize, potentize, strychninize

Finally, there is the simple impossibility of attaching the respective suffix if the morphologi-
cal category does not allow the kinds of adjustments that would be necessary to avoid the
offending sound sequence. This is illustrated in (39) for ‑age, nominal ‑al, ‑ee, ‑eer, ‑ish, and
‑ity. Of these, the suffixes ‑age, nominal ‑al, ‑eer, ‑ette, ‑ish do not allow their derivative to end
in a syllable with the same onset and coda, which leads to lexical gaps (39a). Derivatives in
‑ity do not allow the same onsets in the antepenult and penult (39b), and derivatives in ‑ee
may not have a high front vowel immediately preceding the suffix (39c).

(39) a. *lar.geage, *we.dgage, *con.cea.lal, *in.ha.lal, *in.stil.lal, *rebe.lal, *re.vea.lal, *re.vol.ve.reer,

*mor.ta.reer, *close.tette, *carpe.tette, *fi.shish, *mu.shish, *rub.bi.shish
190 Derivation: phonological considerations

b. *can.di.di.ty, *sor.di.di.ty, *splen.di.di.ty versus flo.ri.di.ty, hu.mi.di.ty, ra.pi.di.ty

c. *carryee, *freeee, *pityee, *seeee

Obviously, the force of an argument based on apparently non-existent forms depends largely
on the assumed productivity of the respective suffix and the size of the sample, but it seems
that for the categories mentioned a convincing case for lexical gaps can be made. At the other
end of the scale, many of the attested forms showing haplology effects may be lexicalized, but
the evidence from low frequency forms and neologisms shows that haplology is a pervasive
and complex phenomenon, since it undoubtedly occurs productively in a number of mor-
phological categories.

9.6 Prosodic morphology

The term ‘prosodic morphology’ is used for morphological categories that are expressed
predominantly or exclusively through prosody, that is by manipulations of, or restrictions on,
the metrical or syllabic structure of the potential derivatives. This is, for example, the case
with clippings, with certain types of hypocoristics, with blends, and with expletive insertion.
We will briefly discuss each of these; for more detailed analyses the reader is referred to
Chapters 18, 19, and 20.

9.6.1 Clippings
With clippings the relationship between a derived word and its base is expressed not by add-
ing material (for example in the form of a suffix or prefix), but by the conspicuous and signifi-
cant absence of phonetic material in the derived word. In (40a) this is illustrated for first
names, in (40b) for common nouns.

(40) a. Pat ◆ Patricia, Pete ◆ Peter, Lon ◆ Alonzo, Trish ◆ Patricia

b. fab ◆ fabulous, fax ◆ facsimile, exec ◆ executive, lab ◆ laboratory, pro ◆ professional,
ump ◆ umpire

Clippings have often been claimed to be irregular and highly idiosyncratic (for example,
Dressler and Merlini Barbaresi 1994; Dressler 2000), but more recent work, for example
Lappe (2007), has shown that such claims are ill-founded. Lappe’s analysis of large sets of
data reveals that there are severe restrictions on both the prosodic shape of clippings and the
phonological relationship between the clipping and its base.
Thus, most clippings are exactly one syllable long, no matter how long the base word is.
Furthermore, there is a strong tendency to have both onset and coda filled, and an obligatory
 9.6 Prosodic morphology 191

requirement to have a heavy rhyme, with the effect that forms with only a consonant and
lax vowel, or with only a long vowel or diphthong are ungrammatical (for example (*M[e] ◆
Melanie, or *I ◆ Ida).
The specific restrictions concerning the faithfulness of the clipping to its base allow for
some variation, a fact that has undoubtedly contributed to the impression that clippings are
idiosyncratic in their behaviour. For instance, the phonological material for name clippings
may come from, or ‘anchor to’, either the first syllable or from a stressed syllable (cf. Pat ◆
Trish ◆ Patrícia, Gail ◆ Ábigàil). The surviving material in common noun clippings, in contrast,
is overwhelmingly taken from the first syllable. If a common noun base starts with an
unstressed syllable followed by a stressed syllable, two syllables (instead of only one) can
make it into the clipping, with no stress shift (celéb ◆ celébrity, exéc ◆ exécutive).
Exceptions to these prosodic patterns are extremely rare (e.g. Ron ◆ Aaron), but these cases
often involve clippings that are homophonous to regular clippings of other bases (cf. Ron ◆
Ronald). The surviving material consists of contiguous segments starting from the left edge
of the relevant syllable. The heavy rhyme constraint explains why, especially with lax vowels,
the first consonant of the following syllable also survives (as in Pat ◆ Pa.tri.cia): it makes the
syllable of the derivative heavy. The segmental contiguity requirement is sometimes violated,
as shown in forms such as in Ike ◆ Isaac or Floss ◆ Florence, where segments are skipped.
Notably, skipping is systematic with derivatives that would otherwise end in /r/ (*Flor).

9.6.2 Hypocoristics in ‑ie

This suffix is variably spelled <ie> or <y>, sometimes even <ey> and <ee>, and we use ‑ie as
shorthand for all orthographic variants. Derivatives with this suffix are mostly disyllabic with
stress on the first syllable, followed by an unstressed syllable (Ándie, Jénny, Láurie). This
means that there must be serious prosodic and segmental adjustments with very many
potential base words. As was the case with clippings of proper names, in ‑ie derivatives, mate-
rial from the first syllable (Australian ◆ Aussie) or from a stressed syllable can survive (umbrella
◆ brollie, tobacco ◆ baccie). We also find similar segmental alternations as with clipped names
(e.g. Nathaniel ◆ Natty), which, together with the anchoring facts, suggests that clippings may
serve, and frequently do serve, as bases to hypocoristics in ‑ie, instantiating the same con-
straints and inheriting the same structures, modulo the additional complications that come
with the suffixation of ‑ie. One such complication is that there is never a complex onset pre-
ceding ‑ie, even if the base has a complex onset in its second syllable (Andrew ◆ Andy, not
*Andry).

9.6.3 Expletive insertion

The final case of morphology where prosodic units and prosodic restrictions are chiefly
responsible for the shape of the derived word is expletive insertion. Its status as a
192 Derivation: phonological considerations

morphological (as against a potentially syntactic) process is somewhat controversial and will
be discussed together with its semantics in more detail in Chapter 18.2.7. Here we concentrate
on the prosody of the process (see also Hammond 1999: 161–4; Plag 2003: ch. 4 sec. 6). In
this type of formation, an expletive is inserted into the base, as shown in (41).

(41) èco-fucking-nómics
àtmos-fucking-phére
emàscu-bloody-láted
pròpa-fucking-gánda
hàndy-bloody-cáp
trànsconti-bloody-néntal
V̀-bloody-IṔ
àny-fucking-thíng
dìm-fucking-wít
kùng-fucking-fú
mìd-fucking-wést
ùr-fucking-báne

The point of insertion of the expletive is governed by the prosodic structure of the base, such
that the expletive must be inserted between two feet. This restriction excludes, for example,
trochaic disyllables or dactylic trisyllables as potential bases (cf. Áspen ◆ *Às-fucking-pen,
énemy ◆ *éne-fucking-my ◆ *é-fucking-nemy), as the expletive would not have two feet surround-
ing it. Similarly, the expletive may not appear between an unstressed syllable and a following
stressed syllable, as benígn ◆ *be-fucking-nígn or agénda ◆ *a-fucking-génda.

9.7 Phonological selectional restrictions

There are many suffixes in English that are subject to constraints concerning the kinds of
phonological structures possible bases may exhibit. By way of illustration, we look at nomi-
nal ‑al, ‑eer, ‑en, and ‑ify. The examples in (42) illustrate a selection of forms in ‑al that are
attested in COCA (those that start in one of the first four letters of the alphabet). Apart
from burial (which, etymologically speaking, contains a different affix), all bases obey the
constraint that the suffix only attaches to bases that have their main stress on the final
syllable.

(42) accrual, acquittal, appraisal, approval, arousal, arrival, avowal, bequeathal, bestowal,
betrayal, betrothal, burial, carousal, committal, conferral, construal, deferral, demurral,
denial, deposal, disbursal
 9.8 Summary 193

Derivatives in ‑eer that are based on free bases (instead of bound roots) are all derived from
bases that end in a trochaic foot, as shown in (43).

(43) allotmenteer, budgeteer, cameleer, canyoneer, concessioneer, conventioneer,

envisioneer, imagineer, leafleteer, planeteer, summiteer

The suffix ‑en shows a segmental as well as a prosodic restriction. It only attaches to bases
that are monosyllabic (or else end in an iamb) most of them also ending in an obstruent.
Consider (44).

(44) awaken, blacken, brisken, broaden, crispen, deepen, lengthen, outen, quicken, ripen,
strengthen, swiften, toughen

Finally, the verbal suffix ‑ify is sensitive to foot structure. Derivatives need to end in a dactyl,
and given the inability to shift stress or to truncate, this restricts the suffix to three types of
bases: those that are monosyllabic (45a), those that end in an iambic foot (45b), and those
that end in a trochee that itself ends in /i/ so that the final vowel and the suffix-initial vowel
can coalesce (45c).

(45) a. artify, massify, jazzify, sinify, trustify, tubify, youthify

b. bourgeoisify, opacify
c. gentrify, nazify, yuppify

One general question that such phonological restrictions raise is whether they should be
formulated in an input-oriented or output-oriented fashion. This question has been discussed
in a number of publications (e.g. Plag 1999; Raffelsiefen 1999; Pater 2004), with the general
conclusion that there are at least some phenomena where output restrictions are better able
to capture the empirical facts, while it seems that all input restrictions can be satisfactorily
reformulated as output restrictions.

9.8 Summary
In this chapter we have taken a bird’s eye view of a plethora of instances of morphology–
phonology interaction and tried to illustrate some general principles that are at work in these
interactions. As became clear, semi-regularity or lexical government prevails in this area and
variation is all-pervasive. This calls into question the theoretical and psycholinguistic status
of many of the generalizations that can be found in the literature and raises the question of
how the variability can be accounted for.
194 Derivation: phonological considerations

Overall, it seems that, very generally speaking, there are conflicting demands that have to
be met by derived words in the mental lexicon. On the one hand (and abstracting away from
conversion), derivatives, being new or different words, must be different in shape and in
meaning from their base, but not too different, so that the relationship between base and
derivative is still recoverable. Affixation seems to be a way to achieve this compromise. On
the other hand, the words of a morphological category must be sufficiently similar to each
other in shape and meaning in order to count as members of the same morphological cate-
gory; this is already achieved by simply adding an affix, but quite frequently it also involves
more intricate relationships between the words, or subsets of the words, of a given morpho-
logical category.
The similarity space between bases and their derivatives, and between derivatives of the
same category is potentially vast, but not unconstrained, as increasing phonological dissimi-
larity leads to decreasing cohesion within a category, and to a loss of the consistent pairing of
form and meaning that is at the heart of any morphology. Within this similarity space, derived
words enter into similarity relationships not only with their respective bases and other mem-
bers of their own category, but also with members of other categories. This is a crucial mecha-
nism by which much of the variation and semi-regularity emerges that we find in the
morpho-phonology of derived words in English.
Other sources of variation are the many lexical idiosyncrasies that come with large-
scale borrowing of complex words from other languages, and the diachronic phonological
and lexical developments that may affect the relation between form and meaning in
­complex words.
 chapter 10

Derived nouns: event,

state, result

10.1 Prospectus
In this chapter we discuss derived nouns denoting events, states, results, as well as related
semantic categories such as products and means, specifically nouns formed with the suffixes
‑ing, ‑ation (and its variants ‑cation, ‑ition, ‑ion, ‑sion, ‑ution, ‑tion), ‑ment, ‑al, ‑ure, ‑ance (and its
variants ‑ence, ‑ancy, ‑ency) and nouns formed by conversion. For ease of exposition in what
follows, we will refer to these as event/state/result nouns, although it should be kept in mind
that such nouns do sometimes have readings that go beyond events, states, and results. Nouns
whose primary function lies in denoting participants in events (e.g. agent, patient) or quali-
ties, collectives, and other abstracts will be treated in subsequent chapters. We begin with the
formal properties of the processes in question, including the kinds of bases that undergo
affixation or conversion, their syntactic category, their status as native or non-native, free or
bound, as well as phonological or prosodic restrictions on bases. We then move on to the
range of semantic interpretations available for the different kinds of formation. All examples
in this chapter are taken from COCA, unless otherwise specified.
There is, of course, a vast amount of literature on the syntax of English nominalizations,
much of it treating deverbal nouns in ‑ing (Lees 1960; Chomsky 1970; Pullum 1991; Yoon
1996), but also other event, state, and result nouns (Grimshaw 1990; Alexiadou 2001; Baker
2003; Heyvaert 2003, among many others). In this volume we are primarily concerned with
the phonological and morphological properties of English nominalizations as derived words.
We treat their syntactic properties only insofar as we are interested in their potential to refer-
ence one or another argument of their base when it is a verb. This largely syntactic literature
will therefore not be the subject of further discussion.

10.2 Formal considerations

Derived event/state/result nouns are formed both by affixation and conversion. The vast
majority are formed on non-auxiliary verbal bases, although we find occasional formations
196 Derived nouns: event, state, result

on adjectives, and even nouns. Non-native verbs are by and large nominalized through
­affixation processes of varying levels of productivity, while native verbs tend to favour nomi-
nalization by conversion. There is, however, some overlap in base type between the two
­processes. All non-auxiliary verbs, whether of native or non-native origin, have at least a
nominalization in ‑ing, and most verbs have one or more other nominalizations as well.

10.2.1 Non-native affixation

The principal non-native affixes for deriving event/state/result nouns are ‑al, ‑ance, ‑ment,
‑ation, and ‑ure. Of these only ‑ation displays a substantial degree of productivity in contem-
porary English, and this by virtue of prior affixation. We discuss nominalization by ‑al, ‑ance,
‑ure, and ‑ment first, and then look more closely at the more productive ‑ation.

10.2.1.1 Minimally productive suffixes: ‑al, ‑ance, ‑ment, and ‑ure

The suffixes ‑al, ‑ance, ‑ure, and ‑ment are largely exemplified by established words in contem-
porary English and show little productivity. We first discuss the combinatorial properties of
these suffixes and then their phonology.
The suffix ‑al attaches to verbs only, a vast majority of them being non-native, with only
occasional native bases being attested (e.g. bequeathal, betrothal). Examples like the following
are found in COCA.

(1) accrual, acquittal, appraisal, approval, betrothal, burial, carousal, committal, construal,
deferral, demurral, denial, deposal, disbursal, espousal, observal, portrayal, proposal,
rebuttal, recital, recusal, redressal, referral, refusal, rehearsal, removal, renewal, revisal,
submittal, survival, transferral, transmittal, traversal, trial, upheaval, withdrawal

The suffix is found in the occasional novel form (e.g. overthrowal, COCA).
What we cite as the suffix ‑ance covers six different variants, ‑ance, ‑ence, ‑ancy, ‑ency, ‑ce
and ‑cy. The distribution is either heavily dependent on the morphology of the base, or else
not entirely predictable. The first problem with these variants is the spelling with either <a>
or <e>, as in dependence versus resonance. We consider this largely dependent on etymology
and as rather random from the perspective of present-day English, and therefore we use the
more frequent <a> in the citation form for this suffix.
The second problem is the variation between the presence or absence of a final ‑y, as in
consistence versus consistency. In general, there seems to be little or no difference between ‑y-full
and ‑y-less variants (cf. the synonymous pairs abundance ◆ abundancy, ascendence ◆ ascendency,
consistence ◆ consistency, dependence ◆ dependency, relevance ◆ relevancy), but some lexicalized
pairs do exist that have different meanings (emergence ◆ emergency, residence ◆ residency).
The final problem is the presence or absence of ‑an‑ in the suffix, or, in other words,
whether the suffix is realized as ‑ce/‑cy or rather ‑ance/‑ancy/‑ence/‑ency. In general, adjec-
tives ending in ‑ant or ‑ent take ‑ce/‑cy, while other bases take one (or more) of the other set.
 10.2 Formal considerations 197

This situation can be analysed in different ways. Apart from an analysis in terms of suffix
allomorphy (e.g. ‑ance versus ‑ce), one could also posit bound roots as bases (e.g. redund-
ancy). Alternatively we could analyse the alternation as a case of haplology (see Chapter 9.5),
such that the non-appearance of ‑ant-ance is accounted for by the avoidance of the repetition
of the same phonological string at the end of the nominalized form. Finally, in a paradigmatic
approach (see Chapter 23.4) no reference is necessary to where potential morpheme bounda-
ries are. Instead there is the regular correspondence relationship between the morphological
categories of ‑ant adjectives with ‑ate verbs and with ‑ance nominalizations.
What confuses matters further is the fact that, perhaps in analogy to the deadjectival forma-
tions, the suffix ‑y, if attached to certain words ending in <t>, leads to a final string <cy>. These
formations are dealt with in Section 10.2.5 below (see also Chapter 12.2.5 for discussion).
The examples in (2) and (3) illustrate the use of the suffix ‑ance, with the forms in (2)
showing y-less formations, the ones in (4) the y-full formations.

(2) ‑ance/‑ence: abettance, abhorrence, abidance, ascendance, clearance, compellance,

compliance, condolence, disobeyance, divergence, effervescence, endurance, entrance,
exfoliance, guidance, hesitance, hindrance, illuminance, impedence, nurturance,
observance, occupance, occurrence, performance, quittance, reactance, reminiscence,
securance, severance, tolerance, transmittance, utterance, variance, vengeance

(3) ‑ancy/‑ency: absorbency, ascendency, claimancy, coherency, compliancy, concordancy,

concurrency, dependency, deviancy, divergency, dominancy, expectancy, exultancy,
hesitancy, inhabitancy, inherency, insistency, irritancy, luxuriancy, occupancy, pendency,
persistency, precedency, radiancy, recurrency, repellency, residency, resistancy,
retardancy, stagnancy, sufficiency, tendency, transcendency, vacancy

The suffix attaches to adjectives and verbs and, due to the regular correspondence of the
pertinent derivatives just discussed, it is often unclear what the base for a given form actually
is. Thus, sometimes, a triad of forms in ‑ate, ‑ant, and ‑ance/‑ancy is attested (hesitate, hesitant,
hesitancy; alternate, alternant, alternance; participate, participant, participance), which makes it
impossible to decide whether the base of the ‑ance nominal is the verb in ‑ate or the corre-
sponding adjective in ‑ant (or, in fact, a bound root). While most bases are non-native, ‑ance
shows a small number of formations on native bases, mostly established forms (hindrance,
riddance) with only the occasional apparent novel form (believance, coming outtance).
The suffix ‑ure attaches to verbs (see (4a)), but is found frequently on bound roots (4b),
as well as occasionally on nominal bases (4c). It is occasionally attested with extenders, as in
‑at‑ure (signature, curvature, entablature), ‑it‑ure (divestiture, penditure), and ‑t‑ure (mixture).

(4) a. ‑ure on verbs: accenture, annexture, closure, composure, contracture, creature,

curvature, departure, disclosure, erasure, expenditure, exposure, failure, fixture,
flexure, gravure, implicature, jointure, licensure, mixture, orature, pleasure, pressure,
procedure, sculpture, seizure, signature, vesture
198 Derived nouns: event, state, result

b. ‑ure on bound roots: comfiture, culture, embrasure, fissure, furniture, garniture,

gesture, investiture, lecture, puncture, rupture, structure
c. ‑ure on nouns: candidature, architecture, magistrature, portraiture

The suffix ‑ment is primarily deverbal (5a), but appears occasionally on bound bases (5b).
Only a few forms are found based on adjectives (5c), or nouns (5d). There is a large majority
of non-native bases, but native bases are by no means unusual (e.g. hangment, hatchment,
oddment, settlement, shipment, startlement, unfoldment, upliftment, upsetment, wonderment, wor‑
riment, worsenment).

(5) a. ‑ment on verbs: abandonment, abasement, abatement, assemblement, assess­ment,

changement, chastisement, citement, commandment, development, dis­
advantagement, embitterment, emboxment, embracement, englobement,
engulfment, evolvement, excitement, extinguishment, fakement, feignment,
lodgement, lopment, malignment, management, statement, tracement, trapment,
treatment, unfoldment
b. ‑ment on bound roots: compartment, ornament, rudiment
c. ‑ment on adjectives: embeddedment, hardyment, insensiblement, oddment,
scarcement, surement
d. ‑ment on nouns: illusionment

As the examples in (1)–(5) illustrate, these non-native nominalizing suffixes appear to have a
preference for disyllabic bases, and among the disyllabic bases, for those that consist of an
iambic foot. Whether this phonological pattern is actually a restriction on these suffixes, or
merely an artefact of the tendency of non-native verbs to display this iambic pattern is diffi-
cult to determine.
The suffix ‑ment can perhaps be taken as illustrative in this regard. The vast majority of its bases
consist of an iambic foot: abatement, abutment, allotment, appeasement, bombardment, enticement,
equipment, investment, procurement, regalement, resentment. Trochaic bases are attested, but far less
frequently: banishment, bafflement, languishment, punishment, ravagement, settlement. Where it
attaches to bases that are longer than two syllables, the first syllable has a tendency to belong to a
prefix like en‑ or dis‑: entanglement, encouragement, disappointment. In addition, ‑ment attaches to
a number of monosyllabic bases: movement, pavement, payment, placement, statement. Interestingly,
where it attaches to native bases (see Section 10.2.1.2), they display a variety of prosodic forms,
including iambs (uplíft, upsét), trochees (wónder, wórry), and monosyllabic bases (hang, ship),
thus suggesting that the overwhelming preponderance of iambic disyllables as bases might be
seen as an artefact of the typical stress pattern of non-native verbs.
Similarly, the suffix ‑al is frequently reported as selecting bases with final stress (e.g. Plag 2003);
the sole counterexample to the generalization is the word burial, which etymologically speaking
 10.2 Formal considerations 199

does not have the same suffix as, say, arrival, refusal. Synchronically, however, this difference is
invisible, and burial is an exceptional word in the series.
Although the vast majority of forms with these suffixes are established, there are neverthe-
less some low frequency and novel forms that appear in corpus data. Low frequency forma-
tions that are attested with the suffixes ‑al, ‑ance, and ‑ure, either duplicate the meaning of an
existing form (6), or display a nuance of meaning that differs from an established nominaliza-
tion (7).

(6) a. Jennifer Eight, 1992: I don’t know nothing about this “Jenn-ifer” girl, cept what some
of the guys told me—but principal feature of the case was a gruesome displayal of
the body.
b. Journal of Church and State, 1991: Even if applied to civil courts, it has been said
that such rules extend only to “real material jurisdiction in its strict meaning.”
( . . . ) not extending to disregardance of the law ( . . . ) nor even to denial of
natural justice ( . . . ).
Twentieth century British history. Oxford: OUP, 1991: “The internal dissension […]
c. 
would be ludicrous if it were not a sinister symptom of the decadent disrupture of
British public opinion, owing to lack of faith either in the present order or in any
consistent and comprehensive principles of reconstruction.”

(7) a. Ear, Nose, and Throat Journal, 2005: A diagnosis of bilateral peritonsillar abscess was
suspected when fluctuance was elicited on finger palpation.
Denver Post, 2009: Get with the lingo. There’s a new word in the theater glossary—
b. 
“revisal.” That‘what they’re calling the revised “Molly Brown” musical. The book has
been overhauled.

The OED cites both disrupture and revisal, with last dates of occurrence in the late nineteenth
century, but it appears that they were never frequent or established forms. Thus their occur-
rence in contemporary corpora might be taken as a sign of marginal productivity. Although
some of these forms have older citations in the OED, it is nevertheless unlikely that speakers
have been exposed to them and more likely that they occur as spontaneous coinages in the
texts captured in COCA.
Unlike ‑al, ‑ance, and ‑ure, where the number of new forms is negligible, we find more
novel or low-frequency forms for ‑ment in COCA and BNC:

(8)  OCA: accruement, addressment, alertment, ambushment, approvement, assemble­

C
ment, assurement, attirement, ceasement, completement, conjoinment, contestment,
dischargement, divorcement, fakement, financement, jugglement, lancement,
omitment, razement, renouncement, revealment, unfoldment, worriment
BNC: humblement, intendment, regardment
200 Derived nouns: event, state, result

Although quite a few of these forms are attested in the OED, none were ever in wide use, and
we might again consider them as sporadic and spontaneous creations where they occur in the
corpora.
We find that these forms rarely differ in meaning from the more established nominaliza-
tion for a given verb, as the examples below suggest:

(9) a. A BC Good Morning America 1999: In light of recent race-related crimes across
the United States and movements to eliminate affirmative action, what is your
stand on that as far as the continuation or the ceasement, I guess, of affirmative
action?
ABC Nightline 1990: And I think an indication of that is the fact that Minister de
b. 
Klerk, in his historic speech, made the omitment of not mentioning anything about
black education.
CBS Face the Nation 1993: I want 10 years in prison without parole for possessing a
c. 
firearm during the commission of a violent crime or a drug felony. I want 20 years
for dischargement with the intent to do bodily harm.
d. NPR 2000: It’s because we really start from scratch with no capital. It’s hard to find
financement.

There are rare cases, however, in which the nominalization in ‑ment is clearly intended to be
distinguished semantically from the more commonly occurring nominalization of a particu-
lar verb.

(10) a. Dateline NBC 2008: Doctors wanted to give the babies a chance to grow and get
stronger and give surgeons a chance to run tests to see if the rare separation surgery
would be possible. This type of conjoinment is one of the most complicated to
separate.
Symposium, 1994: Revealment in relation to the detective story is the moment of the
b. 
unravelling of the mystery plot: the little old lady is shown to be a vile poisoner, the
composed butler is a fiendish gambler with a propensity for murder.

In these examples, the nominalization conjoinment is clearly intended as a technical term for
a medical condition, and revealment is coined as a literary term in distinction to the more
usual nominal revelation.
To the very small extent that new forms can be created with the prefix en‑ (see Chapter 13),
nominalization in ‑ment is likely, unless the derivation also ends in the suffixes ‑ate, ‑ize, or
‑ify. We find in COCA forms like envotement and emplotment which may be novel forms;
neither the form in ‑ment nor the corresponding verb is attested in the OED. COCA also
attests several verbs in en‑ that are not recorded in the OED and might be likely candidates for
 10.2 Formal considerations 201

nominalization in ‑ment, although the forms are not attested: enchamber, enchange, enclod,
encloth, enflood, enfog, enmagick, enpool, enspirit, entragick, envoice, enweb.

10.2.1.2 Productive affixal nominalization: ‑ation

The suffix ‑ation has a number of different variants (‑ation, ‑cation, ‑ion, ‑ition, ‑iation, ‑sion,
‑ution, ‑tion), and we select the form ‑ation as the citation form as this string is common to all
productive variants.
The suffix is the only one of the non-native nominalizing affixes that displays clear produc-
tivity in modern English. It is mostly found on verbs (see (16) and (17) below), although
occasional forms do appear on bound bases (11):

‑ation on bound bases: ambition, aspersion, benediction, benefaction, companion,

(11) 
compassion, constellation, contagion, contraption, contusion, diction, dimension,
dominion, duration, ebullition, excursion, jurisdiction, monition, munition, nutrition,
perdition, sedition, vocation, volition

There are also occasional forms on adjectives (12) and nouns (13). Some of these are loans
rather than English formations, but have to be analysed as having a nominal base in syn-
chronic structure; others are genuine English formations on a noun.

(12) ‑ation on adjectives: adequation, contrition, discretion, distinction, erudition, explicitation,

extinction, festivation, humanation, inanition, inchoation, palmation, precision

‑ation on nouns: artefaction, concretion, intellection, metalation, ozonation, placentation

(13) 

The vast majority of words in ‑ation are formed on non-native bases, but occasional forms on
native bases can be found, as illustrated in (14):

‑ation on native bases: backwardation, blackalation, bolsteration, botheration, flirtation,

(14) 
flotation, gatheration, harvestation, hateration, holleration, starvation

Occasionally we see extenders in these forms, as in blackalation or hateration. Some of these

are clearly meant to be humorous, clever, or jocular formations.

Rolling Stone 2002: The queen of hip-hop soul hits Number One with “Family Affair,”
(15) 
denouncing holleration and hateration.

The derivatives of this morphological category all have their main stress on the penultimate
syllable, the suffix thus belongs to the set of stress-shifting suffixes (see Chapter 9 for more
discussion).
202 Derived nouns: event, state, result

The main area of productivity for this affix lies in its status as the default nominalizer of verbs
formed with the suffixes ‑ize and ‑ify. As these are the predominant affixal means of forming
new verbs in contemporary English, they provide a source of new bases for ‑ation. Any verb that
is created with the suffix ‑ize will be nominalized using the variant ‑ation, while any verb created
with the suffix ‑ify will be nominalized with ‑cation (and vowel change in the verbal suffix). The
forms in (16) and (17) are only a few of the forms from COCA that we could cite.

Nominalizations on ‑ize: aerosolization, aristocratization, bilingualization,

(16) 
christmasization, corporativization, cyclomidization, dollarization, emblem­atization,
figuralization, geometrization, hausaization, laicization, micral­ization, numinization,
povertization, protestantization, reflectorization, skeletonization, solubilization, teddy-
bearization, vascularization

Nominalizations on ‑ify: adultification, aridification, celebrification, crechification,

(17) 
dignification, disneyfication, etherification, floridification, glassification, humidi­fication,
jazzification, mall-ification, mississippification, parentification, rarification, sandi­
fication, sportification, spaghettification, townification, uglification, zombification

COCA does, however, also attest a handful of apparently novel forms on other sorts of bases
including bolsteration, complexation, distanciation, effemination, endorsation, expection, expendi‑
tion, festivation, harvestation, hateration, judgitation, and ozonation. There are a mere handful
of forms in ‑izement (advertizement, aggrandizement, enfranchisement) or ‑izance (cognizance
and its various derivatives), all of them item-familiar and of relatively high frequency. Of
these, only enfranchisement is an English formation. The contemporary data culled from
COCA thus corroborate the results of Bauer (2001) which assessed the productivity of nomi-
nalizing affixes in English on the basis of data from the OED (2nd edn) and the Wellington
Corpus of New Zealand English and found ‑ation to be the only productive suffix. There are
no examples with the suffix ‑ify plus other nominalizing affixes.
There are some nominalizations in ‑age which also belong semantically to this chapter (for
example, dosage, drainage, slippage, spillage, stoppage), but which are treated in Chapter 12
with other words taking the same suffix.

10.2.2 Nominalizations with ‑ing

All non-auxiliary verbs in English, regardless of their origin or other means of nominaliza-
tion, have nominal forms in ‑ing. We are chiefly concerned here with the ‑ing form that can
occur with a determiner, and that takes its complement in the form of a prepositional phrase,
typically with the preposition of, for example the closing of the Suez Canal or the teaching of the
arts. What we call the nominal ‑ing form therefore has the external syntax of the other forms
of nominalization that we discuss here, albeit a more active and typically less lexicalized
semantics, a subject we will take up in Section 10.3.
 10.2 Formal considerations 203

What is perhaps more interesting is that while all verbs have a nominalization in ‑ing, a
significant number of verbs have only a nominalization in ‑ing, having neither a nominaliza-
tion formed with one of the affixes discussed above, nor a nominalization formed by conver-
sion (at least according to standard dictionaries). Among these are the majority of verbs that
are themselves derived from nouns, either by conversion (18) or by affixation of de‑, un‑,
inter‑, or over‑ (19).

denominal: antique, apprentice, bandage, bargain, bicycle, boss, bridge, bully, calve,
(18) 
can, chauffeur, coach, dust, escort, eye, ferret, garrotte, glue, group, gut, hammer, hay,
host, joke, journey, ladle, lash, log, manicure, mark, network, package, paint,
parachute, peel, picnic, pilot, pocket, quiz, referee, rhyme, rivet, robe, shepherd,
shingle, smudge, snack, splinter, star, tape, thatch, tree, vacation, wallpaper, widow,
witness, yoke

(19) a. prefix de‑: debone, debug, defog, defraud, defrost, defuzz, degas, degerm, deglaze,
degrease, degrit, degum, dehair, dehead, dehorn, dehull, dehusk, deice, deink, demast,
denude, desex, desprout, destarch, destress, detassle, detusk, devenom, dewater,
dewax, deworm
prefix un‑: unbolt, unbosom, unbuckle, unbutton, unchain, unclamp, unclasp, unclip,
b. 
unfasten, unfold, unglue, unhinge, unhitch, unhook, unlace, unlatch, unleash, unlock,
unnerve, unpeg, unpin, unscrew, unshackle, unstaple, unstitch, untie, unzip
prefix inter‑: interlard, interleave, interlink, interlock, intermingle, intertwine,
c. 
interweave

In addition, there are a significant number of both native and non-native underived verbs
(20), (21) that seem not to have a nominalization other than the ‑ing form:

native underived: ask, bake, clean, draw, drench, dwell, earn, flit, follow, forbid, gather,
(20) 
hang, hear, learn, mean, mourn, open, spell, teach, wed, write

non-native underived: blanch, board, broil, cavort, char, covet, crochet, curry, daunt,
(21) 
devour, extoll, fester, gloat, hurtle, lurk, perish, pierce, quit, quiver, rankle, render,
squash, stuff, suffer, thrive, tremble, waver

10.2.3 Conversion
In addition to nominalization with the affixes discussed above, English frequently uses con-
version to derive nouns from verbs. Indeed, the majority of native verbs have a conversion
nominalization. A selection of verbs which form nominalizations using the process of
­conversion is given in (22).
204 Derived nouns: event, state, result

(22) answer, bark, beat, bend, bet, bid, boast, break, breed, build, burn, bust, cough, creep,
cut, dare, deal, dig, do, dream, drink, drive, drop, end, fall, fart, feed, fling, flight, fold,
forecast, freeze, grind, grip, heap, help, hit, hold, hope, kick, kill, knock, laugh, lead,
leap, leave, like, life, need, neigh, nip, play, proof, pull, raise, reach, rest, ride, ring, rise,
run, sail, say, sale, set, shake, shave, span, spare, speed, spill, spin, spit, split, spread,
spring, sprout(s), stand, stick, sting, stink, sweep, swell, swim, swing, take, talk, tear,
thank(s), throw, thrust, tread, trust, turn, wait, walk, want, wash, watch, weave, win,
wish, wonder, work, worry, yawn

A non-negligible number of non-native verbs nominalize by conversion as well, as the exam-

ples in (23) illustrate. For verbs among the non-native portion of the English lexicon, nomi-
nalization by conversion is not nearly as frequent as by suffixation of ‑ation, but is somewhat
more frequent than affixation with ‑ment, ‑al, ‑ance, ‑ure, and the other affixes mentioned in
Section 10.2.1.1. Many of the words in (23) have parallel nominalizations with affixes, some-
times with the same meaning, sometimes with a different meaning.

(23) acclaim, accord, act, advance, affront, blame, canter, capture, caress, cascade, collapse,
comfort, cruise, decline, delay, design, desire, effect, elapse, embrace, escape, esteem,
exchange, finish, haunt, lament, laud, marvel, measure, mount, move, notice, offer,
order, preserve, press, pretend, process, profit, regard, regress, regret, reign, relapse,
remark, reserve, respect, rest, result, return, reverse, review, revolt, reward, rhyme,
sense, sojourn, study, supply, support, surge, surmise, surprise, surrender, touch, trace,
travel, visit, warrant

Conversion has clearly been productive in both the native and the non-native sectors of the
lexicon in earlier stages of English, but its status in contemporary English is harder to estab-
lish. Whether it continues to be productive depends largely on the creation of new verbs that
are neither formed with the suffixes ‑ize and ‑ify, nor are themselves formed by conversion
from adjectives or nouns. The creation of nouns by conversion is far less productive in cur-
rent English than the creation of verbs. Nevertheless examples such as a commute, a hijack, a
work-to-rule (BrE) (all twentieth-century examples) indicate that verb to noun conversion is
available.

10.2.4 Nominalization by stress shift

There are quite a few disyllabic verbs, almost all non-native, that form their nominaliza-
tions by stress shift, changing from an iambic pattern to a trochaic one. Examples are given
in (24).

(24) accent, addict, address (NAmE), affect, affix, alloy, ally, annex, augment, combat,
combine, commune, compress, console, consort, construct, contest, contract, contrast,
converse, convert, convict, decrease, defect, discharge, discount, discourse, escort,
 10.2 Formal considerations 205

essay, excerpt, excise, exploit, export, extract, ferment, finance, indent, insert, insult,
intern, intrigue, invite, mandate, object, permit, pervert, proceed, process, produce,
progress, project, prolapse, prospect, recess, recoil, record, redress, refill, refund, refuse,
research, retard, rewrite, segment, subject, surmise, survey, suspect, torment, transfer,
transport, update, upgrade, uplift, upset

It is assumed in these cases that the noun is derived from the verb, rather than the opposite.
As Kiparsky (1982b: 13) points out, for nominalizations that have lexicalized meanings, a
semantically distinct verb can be derived from the derived noun, which then maintains the
stress pattern of the noun: so protéstV ◆ prótestN ◆ prótestV.
Several trisyllabic or longer verbs exhibit nominalization with stress shift as well, as
in (25).

(25) álternàte (BrE), attríbute, envélope, réprimand

Note that in BrE, alternate has first syllable stress for the verb and second syllable stress for
the noun, which is not what we might normally expect. The other examples do not show a
consistent position of stress for the verb.
Sometimes the main stress in the noun and the verb do correspond, but the secondary
stress changes. This happens not only with noun–verb pairs, but also with adjective–verb
pairs, which are added here to give a fuller picture of the phenomenon. The last syllable in
these words takes a secondary stress when the word is a verb, but is unstressed when it is a
noun or an adjective. Examples are given in (26).

(26) advocate (N, V), affricate (N, V), aggregate (A, N, V), appropriate (A, V), aspirate (N,
V), associate (N, V), certificate (N, V), complement (N, V), compliment (N, V),
confederate (A, N, V), consummate (A, V), coordinate (A, N, V), deliberate (A, V),
desolate (A, V), document (N, V), duplicate (A, N, V), elaborate (A, V), estimate (N, V),
experiment (N, V), geminate (A, N, V), graduate (N, V), implement (N, V), increment
(N, V), intercept (N, V), moderate (A, V), numerate (A, V), ornament (N, V),
prophesy (N, V), regiment (N, V), supplement (N, V), triplicate (A, V) (and others
similar)

The phenomena illustrated in (24)–(26) are variable in contemporary English. Not all speak-
ers show the stress differences on the same set of words, and there is also a diachronic shift
with words both being added to and deleted from the sets. Some of the deletions occur when
the nominal or verbal use of the word vanishes; in other cases both the nominal and the
verbal forms come to take the same stress pattern.
Another phenomenon that might be discussed under the topic of stress change is the
group of nominalizations formed from particle verbs like put down, blow up, burn out, fuck up,
and so on. Whereas particle verbs can display various stress patterns in sentential contexts
depending on whether the particle appears adjacent or non-adjacent to the verb, the
206 Derived nouns: event, state, result

nominalizations always display a trochaic stress pattern: pút down, blów up, etc. A similar
stress pattern occurs in the examples in (27), which have particles or prepositions in initial
position (and whose derivational history is not always clear).

(27) bypass, download, income, inlay, input, outlay, output, overcharge, overlap, overlay,
overload, overlook, override, overrun, overturn, throughput, underlay, underline,
underscore, upgrade

10.2.5 Miscellaneous
A few nominalizations, predominantly from non-native verbs, exhibit devoicing of a final
consonant: abuse, excuse, extent, descent, belief. For fuller treatment of this pattern see Chapter 25.
Several nominal forms corresponding to Latinate verbs are based on different stem forms of
the original Latin verbs: defend ◆ defense, expend ◆ expense, respond ◆ response, suspend ◆ sus‑
pense, transcribe ◆ transcript, exceed ◆ excess, perceive ◆ percept. Some of these also exhibit stress
shift (e.g. percéive ◆ pércept, transcríbe ◆ tránscript).
The suffix ‑y derives a limited number of pertinent nominalizations (e.g. advocacy, assem‑
bly, burglary, beggary, conspiracy, delivery, determinacy, entry, felony, gluttony, inquiry), but also
abstract nouns of various types, treated in Chapter 19.2.5. That chapter also contains a discus-
sion of the allomorphy patterns involved with this suffix and of its relation to other suffixes
such as -ery, and ‑ity.
A handful of event nominalizations in ‑th still exist, among them: birth, growth, health,
stealth, tilth, troth, wealth (for deadjectival formations with this suffix, see Chapter 12). This
suffix is clearly unproductive, and has been so for several centuries (Marchand 1969).
Finally, a number of verbs also display nominalizations with lexicalized plurals, for exam-
ple, forms like amends, arms, congratulations, creeps, digs, droppings, earnings, eats, lodgings,
looks, regards, savings, staggers, surroundings, works, writings (see Chapter 8.2.1). None of
these processes, probably except for those on bases in ‑ing, exhibit productivity in contem-
porary English.

10.3 Semantic considerations

In what follows, we will assume that nominalizations typically display a range of meanings,
with particular nuances determined by the semantic characteristics of the base combined
with the syntactic context in which the nominalization is used. We will therefore typically
discuss what we will refer to as ‘readings’ or ‘interpretations’, rather than ‘meanings’ or ‘deno-
tations’, and give, wherever possible, specific citations in which the reading in question is
displayed.
 10.3 Semantic considerations 207

10.3.1 Range of readings

The basic division that is normally made is between eventive readings and result readings
(e.g. Grimshaw 1990). All ‑ing nominals have an eventive reading, as do many, if not most,
affixal nominalizations. As has been pointed out frequently, the eventive reading of non-ing
nominalizations is most easily available when the nominalization displays the full argument
structure of the corresponding verb,: as in the professor’s demonstration of the technique.
Some theorists (e.g. Grimshaw 1990) distinguish between what they call ‘complex event’
and ‘simple event’ readings, the former being a syntactic context in which all verbal argu-
ments are maintained in the phrase headed by the nominalization (e.g. the Romans’ destruc‑
tion of the city) and the latter a syntactic context in which one or more of the expected verbal
arguments are absent (e.g. the quick destruction of the city). It is unclear whether there is any
real distinction here, there being no independent way to identify the complex event reading
from the simple except by determining how many arguments are present. So we will simply
distinguish eventive from non-eventive interpretations in what follows.
Syntactic context plays the key role in the precise interpretation of any nominalization.
For example a nominalization like demonstration clearly denotes an action in a sentence like
(28a). But (28b) has what is usually termed a ‘factive’ reading (that is, ‘the fact that the profes-
sor demonstrated the technique …’), and in (28c), what is in question is the manner in which
the action is performed.

(28) a. The professor’s demonstration of the technique proceeded smoothly.

b. The professor’s demonstration of the technique was a scandal.
c. The professor’s demonstration of the technique was deft.

All non-modal verbs exhibit ‑ing nominalizations. Nominalizations in ‑ing characteristically

display eventive readings and are less prone to lexicalization or semantic drift, although there
are a few that exhibit result, product, or means interpretations, as examples given below will
suggest. Where there is lexicalization, it often involves narrowing of the verbal denotation to
a specific context. So, for example, mouldings are specific sorts of things that are moulded,
namely decorations on walls, kindling is small pieces of wood that help to kindle a fire, a
finding is something that is found as an outcome of scientific research or a legal proceeding.
For native verbs whose only other nominalization is by conversion, the ‑ing nominaliza-
tion is usually the only eventive nominalization, the conversion nominalization being more
likely to display aspectual or result readings (see below), but not an eventive interpretation.
Systematic exceptions here are verbs denoting emotions (love, hate, fear, hope, trust, dislike).
These are stative verbs, and stative verbs are usually said to be awkward with eventive read-
ings. However, when an eventive reading is needed, there is no problem with using the ‑ing
nominalization:
208 Derived nouns: event, state, result

(29) COCA. CNN_King 2004: We have a lot more loving of our neighbor to do
BNC. Cole, the Ladykiller, 1993: I make you feel suffocated, do I, with my loving?

Otherwise, the conversion nominal is more usual:

(30) Her love of music; my fear of snakes

The general principle with conversion nominalizations is that they may exhibit any of the
readings generally associated with nominalizations (see below). Nevertheless, there are some
verbs whose conversion nominalizations appear to resist an eventive interpretation: e.g. *her
drive of the car, *the sailor’s heave of the rope. The examples in (31) are interpreted as eventive
because they display full argument structure, and those in (32) because of the presence of the
temporal adjectives constant or frequent.

Denver Post 2004: Last year, just before the 50th anniversary of Sir Edmund Hillary’s
(31) 
climb of Mount Everest, the Sherpas wanted to put up an Internet cafe at 18,500 feet,
base camp on Everest.
PBS Newshour 1995: Technically, in Pennsylvania, you only have to have one license

plate on your car, so the officer was improper on his stop of this car, but while he
stopped the car, he observed in the back seat over 200 pounds of cocaine.
Ind_Geraldo 1992: But, of course, the fact that he was involved or may have been

involved in that robbery two weeks before the crime doesn’t really support his murder
of your mom two weeks later.
Environment 1990: Bolivia’s swap of $650,000 is tiny in comparison to the overall

commercial debt of $500 million.

American Heritage 1998: We need to learn to understand our technological behavior as

(32) 
a constant blend of these very different modes of consciousness.
Consumer Reports 2007: [ . . . ] Bouts of coughing unrelated to colds or the frequent

spread of any symptoms to the chest when you do have a cold.
San Francisco Chronicle 1991: “[ . . . ] I can smell a skunk that was killed, and I know how

many hours before it was killed,” he said, in a conversation punctuated by the frequent
crash of beer empties in a trash can behind the bar.

Interestingly, it seems easier to find non-native conversion forms in eventive contexts. Forms
such as report, surrender, review, catch, chase, appeal and others occur easily with an eventive
reading, as the examples in (33) suggest.

American Spectator 2008: When it was adopted, his report of events to the governor of
(33) 
Virginia placed it after fairly trivial matters, such as the authorization of some
lighthouses.
 10.3 Semantic considerations 209

ABC Primetime 1993: Tim’s role in all of this didn’t end with his surrender of those

videotapes.
Driftless 2008: The dark foliage above him seemed to draw nearer and a spirit of fatigue

invaded his senses, disrupting his review of recent events.
Sports Illustrated 2008: Even then, it took the replay official to determine that the ball

had crossed the plane of the goal line on wideout Santonio Holmes’s catch of Ben
Roethlisberger’s four-yard pass.
Entertainment Weekly 2006: Clay gave up his chase of the Mister Softee truck and stood

with one foot on the sidewalk and the other planted in the gutter, watching as it
swerved into the center lane of Boylston Street, still tinkling.
Washington Post 2008: Fu […] said Bush’s appeal to Chinese officials on religious

freedom and human rights was mostly for the benefit of his domestic critics.

Non-eventive interpretations can take a number of forms, among them results (the outcome
of verb-ing), products (the thing or stuff that is created or comes into being by verb-ing),
instrument (the thing that verb-s, a way of verb-ing), locations (the place of verb-ing),
agents (people or person who verb‑s), measure terms (how much is verb-ed), paths (the
direction of verb-ing), patients (thing verb-ed, thing affected or moved but not created by
verb-ing), states (the state of verb-ing or being verb-ed), and what might be called
‘instances’ (an instance of verb-ing). We provide a few examples of each of these in (34)–
(43). The (a) examples are lexical items with the potential to display the reading in question,
given the right context. The (b) examples are attested citations from COCA illustrating that
reading. It should be kept in mind that these same nominalizations can exhibit other read-
ings, given other syntactic contexts.

(34) result: the outcome of verb-ing

a. 
acceptance, acquittal, addition, adherence, adjunction, aggravation, agitation,
alteration, amelioration, americanization, arousal, arrival, articulation, assassination,
assessment, canonization, captivation, castigation, certification, collision,
communication, completion, compression, computation, cooperation, correlation,
degradation, evaluation, exertion, hesitation, impeachment, improvement, inflation,
integration, movement, omission, operation, ordainment, ossification, procurement,
reaction, reclamation, recognition, recurrence, repulsion, retrieval, revilement,
revolution, rotation, satisfaction, stabilization, stimulation, stratification,
submergence, surrender, transformation, understanding, unification, unionization,
upset, veneration, vindication
USA Today 2009: Still, most population specialists contend that the most likely long-
b. 
term result will be the “browning of America,” with the different races meeting and
mingling joyfully as such groups always have done—but perhaps not without some
lingering acrimony before complete acceptance is achieved.
210 Derived nouns: event, state, result

New Yorker 2006: His aunt had lived to be ninety, and he was sixty-six. His son was
twenty-four, which, he quickly computed, was the difference in age between his aunt
and him. The computation meant nothing.
Time 2009: By that time, she was also married to a high school sweetheart, Kevin Edward
Noonan—a marriage that ended in divorce in 1983.
African Arts 2009: For the Oromo, the lower body is connected to the past through its
link to the ground, to birthing, and to containment.
Harpers Bazaar 2009: The result was the biggest seizure ever in New York City: 19
tractor trailers filled with counterfeit clothing and footwear with an approximate street
value of $10 million.

(35) product: the thing or stuff that is created or comes into being by verb-ing
a. allotment, allowance, appendage, blot, brew, bruise, building, creation, description,
drawing, embroidery, etching, gouge, growth, jumble, knowledge, pavement, puke,
saute, scrape, slit, smear, spatter, spill, splatter, supply, tear, thought, wound
Southwest Review 2009: Running my hand over that cold carving was like touching
b. 
the new contours of my sister’s face.
USA Today 2009: The sticky-sweet concoction combines gin, cherry brandy,
pineapple juice, Cointreau, grenadine and other ingredients.
Southwest Review 2005: Back on the couch he fingered the stuffing the cat had clawed
out, making the tear measurably worse.

(36) instrument: the thing that verbs, a way of verb-ing

a. accompaniment, adornment, advertisement, clog, conveyance, cure, decoration,
distraction, entertainment, feed, flattery, greeting, imagination, kindling, leavening,
refreshment, tie, wrap
Men’s Health 1994: He sends you to a cardiologist, who confirms your worst fears:
b. 
An artery that feeds the heart is blocked; you may be in danger of a heart attack. #
It’s obviously a scary diagnosis, but the doctor recommends angioplasty, a relatively
simple and painless procedure to pinpoint and clear the clog.
Happiness Key 2009: For decoration, three turquoise seahorses descended the wall
at a forty-five-degree angle.
Fantasy & Science Fiction 2009: “There’s not much out here in the way of
refreshment, Mister Lawyer. Your drivers could each use a plate of cookies.”

(37) location: the place of verb-ing

a. camp, dump, dwelling, exhibition, haunt, mooring, perch, residence, rise, roost,
seat, wallow
New Yorker 2006: But she did open the door at last, and beckoned him inside. They
b. 
stood in the entry, quite close.
 10.3 Semantic considerations 211

American Spectator 2009: It is in the same building as the dwelling of Irving Kristol and
Gertrude Himmelfarb.
Life is Short But Wide 2009: Now he had a sizable sum to purchase a house for his new
wife; she did not want to live on a reservation.

(38) agent: people or person who verb-s

a. administration, cheat, congregation, consort, cook, delegation, federation, flirt,
following, prosecution, sneak, stray, help, snoop, sweep, tease
New York Times 2009: He said the administration had determined that it could not
b. 
stop the bonuses.

Washington Post 2008: The cook melts chocolate over a warm-water bath, mixes
c. 
water in, places the bowl over an ice bath and then whisks energetically until the
mixture firms up.

Triquarterly 1997: She was the girl dancing on the table at the party, the tease with
d. 
the gleam in her eye.

(39) measure: how much is verb-ed

a. deceleration, decrease, pinch, weight
Fantasy & Science Fiction 2009: He took a pinch of some sort of crumbled herb
b. 
from it and sprinkled it over the threshhold.

(40) path: the direction of verb-ing

a. ascent, ascendence, continuation, decline, descent, direction
Fantasy & Science Fiction 2009: The road, after several bends and a gradual ascent,
b. 
leveled off so suddenly he nearly lost his balance, caught by a forceful gust sweeping
up along mountain gulches from the unseen plains below.
Journal of International Affairs 2008: In 1924, the United States claimed the North
Pole was an underwater continuation of Alaska.

(41) p atient/theme: the thing verb‑ed, thing affected or moved (but not created) by verb-
ing
a. catch, nosh, assignment, convention, jam (paper), offer, pick, exhibit, buy, choice,
mount (horse), want, waste, purchase, washing, advance (money), acquisition,
appropriation (money), discovery, donation, inheritance, learning, reading, closure,
allocation, inclusion, investment, projection, repetition, submission, substitution,
teaching(s), hire, kill, tip, allowance, appendage, cull, eats, grant, hoard, imprint,
knowledge, marvel, belief, bequest, find, gain, payment, preference, yield, suspect
Good Housekeeping 2009: The Good Housekeeping Research Institute had a panel
b. 
of seniors test products that can help keep your parents safe; a small purchase now
might avoid a major medical expense later.
Triquarterly 2009: He was gobbling a fresh kill; his mouth and jaw were covered in blood.
212 Derived nouns: event, state, result

(42) state: the state of verb-ing or being verb-ed

a. admiration, alienation, annoyance, coexistence, comfort, contentment, dejection,
delight, disappointment, discombobulation, disdain, dislike, disquiet, existence,
exultation, fascination, fear, grief, hate, hope, perplexity, preoccupation, regret,
reverence, suffering, thriving, trust, wonder, worship
Newsweek 2009: The verse was written in the time of the shah, in the 1950s, when
b. 
Khamenei was a young, idealistic Shia cleric who shared with its hard-drinking
author a sense of claustrophobic alienation and deep frustration.
USA Today 2009: It’s a sly cautionary tag on a character study that reflects a
collective regret and despair in today’s uneasy world.

(43) instance: an instance of verb-ing

a. balk, bang, bark, beep, belch, bite, blast, blink, bounce, bump, climb, clunk, conflict,
cough, crack, crawl, cuddle, dip, dive, glint, grab, groan, guess, halt, heave, hit, howl,
hug, jab, jeer, jolt, jump, peek, ping, pitch, plop, prance, pucker, punch, push, rap,
rebuke, shiver, shriek, shrug, shudder, sip, slap, slurp, smirk, snooze, sob, throb,
throw, tickle, toss, try, twist, wag, wail, wait, win, yank
Houston Chronicle 1993: There had been four errors—all by the Rockies—and three
b. 
wild pitches. For good measure (or bad), there also were six walks and a balk.
U.S. Catholic 2006: Claretian Father Cyrus Banque is a jovial host, with warm
greetings for everyone and an engaging smile. But when the topic of conversation
turns to justice and reconciliation for the new nation of East Timor, the Filipino
missionary’s smile breaks into a frown and his brow wrinkles.

Among the non-eventive interpretations, the result interpretation is by far the most fre-
quent one exhibited, and should perhaps be considered the default non-eventive
interpretation.

10.3.2 Referencing of arguments or aspect

Key to the interpretation of the non-eventive nominalizations is the referencing of arguments
or aspectual characteristics (see Chapter 2.8.2).
In product or patient/theme nominalizations (35), (41) it is the object that is refer-
enced, albeit in slightly different ways. Embroidery is a product, the material thing that is
brought into existence by the action of embroidering. A purchase, in contrast, is not created
per se, but is rather the theme, that is, what is transferred by the action. Much less frequent is
the referencing of the subject of the verbal base, which is what we find in the agent nomi-
nalizations in (38). Some of these forms are what Melloni (2007) calls ‘agent collectives’: the
administration, for example, can be the people who administer something, and a following can
denote the people who follow someone (usually someone prominent). Conversion forms
like flirt, cook, tease tend to denote non-collective agents. In addition to forms that are clearly
 10.3 Semantic considerations 213

agentive, there are forms that are not directly agentive, but that nevertheless reference the
subject, frequently because the verbal base is intransitive. Among these are nominalizations
like: bloom, decay, decomposition, deterioration, drift, effervescence, emanation, precipitation,
ulceration, divergence, illumination, temptation, overhang, remains, sprout, spurt, affliction,
amazement, disturbance, flop, indication, occurrence. We will return to these forms in Chapter 11,
where we look more closely at personal nouns.
Also relatively infrequent is the referencing of adjuncts like location, path, or measure.
Location nominalizations of course denote the place where the action of a verb occurs: an
enclosure is the place where things are enclosed, and a dwelling the place where one dwells.
There are only a few nominalizations that might be classed as measure or path forms. A
pinch (as in a pinch of salt) is the amount which is pinched, the ascent is the trajectory of
ascending.
It should be noted that for agents, the suffix ‑er is far more typically used (see Chapter 11),
so it is not surprising that there are so few agent or instrument forms derived by conversion
or by affixes like ‑ation or ‑ment. On the other hand, although there is a dedicated affix ‑ee in
English for sentient patients, there exist no specific affixes in English that derive non-sentient
patients, either products or things transferred or affected, nor are there specific affixes denot-
ing location, means, measure, or path nouns. It is therefore not surprising that affixes like
‑ation, ‑ment, ‑al, and conversion are sometimes enlisted to derive such forms. We will return
to this issue in Chapter 11.
Non-eventive nominalizations do not always reference an argument of the verb. With
certain classes of verb (see Section 10.2.2) they may reference specific aspectual characteris-
tics of the verbal base, either a resulting state (admiration, depression, bewilderment, hate) or a
punctual instance of the action (yawn, kick, crack, chirp).

10.3.3 Predictability of nominal semantics

There is frequently a relationship between the semantic class of the base verb and the inter-
pretation of the nominalization of that verb. This is especially clear in the case of nominaliza-
tions that have either an instance or a state aspectual reading. The former nominalizations
derive from verbs of instantaneous contact (44), verbs of bodily motion (45), and verbs of
sound or light emission (46). State nominalizations most frequently derive from verbs of
psychological state (47). Verbs with inherently spatial denotations give rise to location nomi-
nalizations (48). Means nominalizations derive from verbs that denote actions that require
instruments of various sorts (49).

(44) verbs of instantaneous contact: beat, hit, bump, chop

(45) verbs of bodily motion: blink, chew, hop, lurch

214 Derived nouns: event, state, result

(46) verbs of sound or light emission: howl, click, groan, mew, gleam, glint

(47) verbs of psychological state: excite, exasperate, disgruntle, alienate

(48) inherently spatial verbs: rise, enter, surround, enclose

(49) verbs requiring instruments: swaddle, season, remunerate, equip

It is not always possible, however, to characterize the verbs that give rise to particular sorts of
nominalization.

10.3.4 Count versus mass interpretation

As Huddleston and Pullum (2002: 337) note, most nominalizations are not count or non-
count per se, but can have count and non-count readings, given specific contexts. So celebra‑
tion or change are possible on either interpretation:

(50) a. Some celebration is clearly called for.

They did not believe that there had been enough change.
b. We held two separate celebrations last week.
The committee recommended three changes.

Some nominalizations tend to favour one reading or the other, however. Although it is not
clear how systematic this phenomenon is, at least a few patterns can be discerned. For exam-
ple, conversion nominalizations that express instantaneous aspect always have count read-
ings, and typically do not also allow non-count readings.

(51) a. The monks began a chant.

We heard two burps.
b. *We want to hear some (/səm/) chant.
*We expected to hear some burp.
c. We wanted to hear some chanting.
We expected to hear some burping.

A non-count reading is frequently preferred for the corresponding ‑ing nominalization, as the
examples in (50c) illustrate.
We also observe that forms with lexicalized meanings are somewhat likelier than non-
lexicalized forms to have only the count reading. For example, semantically specialized
 10.3 Semantic considerations 215

nominalizations like assignation (‘meeting’), assignment (‘assigned task’), fixture (‘attached

piece of equipment’), proposal (‘a plan’), recital (‘musical performance’), or variance (‘official
dispensation’) exhibit only the count reading, while less (or un‑)lexicalized counterparts
favour the non-count reading (attention, composure, embitterment, importance, survival), or
allow either (accommodation, assessment, closure, disposal, disturbance, renewal).
 chapter 11

Derived nouns: personal

and participant

11.1 Prospectus
In this chapter we discuss morphological processes which derive personal or participant
nouns, that is, nouns denoting agents, patients, themes, instruments, inhabitants, locations,
and gendered forms. These are formed primarily by affixation in contemporary English, spe‑
cifically by the suffixes ‑an, ‑ant (and its spelling variant <ent>), ‑arian, ‑ee, ‑eer, ‑er (and its
variant <or>), ‑ese, ‑ess, ‑ette, ‑i, ‑ish, ‑ist, ‑ite, ‑meister, -nik, ‑ster, and ‑trix, as well as by the
prefixes grand‑, great‑, step‑, and vice‑. We will also revisit some of the nominalizing processes
discussed in Chapter 10, looking at the ways in which they overlap with the function of the
primary personal and participant affixes discussed here. Similarly, we will touch on a few of
the affixes that will be given a more detailed treatment in Chapter 12 as forming collective or
abstract nouns, but which double to some extent in creating participant nouns.
It should also be noted that some sets of the person noun derivatives to be discussed in the
present chapter are formally identical with pertinent adjectival derivatives. A majority of the
nominal derivatives in ‑an, ‑ant, ‑ist, and ‑arian can also be found as adjectives (e.g. American,
modernist, vegetarian), but the nature of the relationship between the nominal and adjectival
forms is not entirely straightforward. Thus, there are nominal cases that are not also adjec‑
tives (e.g. accountant, applicant, informant) and adjective cases that are not also nouns (e.g.
aberrant, pleasant, hesitant). While the fact that English allows prenominal modification by
nouns may explain the categorial variation, a theoretical account is nevertheless a challenge,
and there are in principle at least two different solutions available. The categorial variation
could be considered either a case of suffixal homophony or a case of conversion. We remain
agnostic with regard to the theoretical analysis and treat the pertinent derivatives in this
chapter and in the chapter on adjectival derivation (Chapter 14). The suffix ‑arian is dealt
with exclusively in this chapter, since the adjectives correspond straightforwardly to the
nouns.
The examples in this chapter are taken either from BNC or COCA, unless otherwise
specified.
 11.2 Formal considerations 217

11.2 Formal considerations

In this section we consider such formal characteristics of personal and participant affixes as
their selection of base types—whether they attach to native or non-native bases, bound or
free bases, what syntactic categories they select, whether they take compound or phrasal
bases, and so on—as well as their specific phonological restrictions and the extent to which
they display argument structure. In order to organize this section, a rough first pass at seman‑
tic categories seems necessary, even though we will put off a close semantic analysis until
Section 11.3. We will use five rough rubrics here, dividing affixes into those whose primary
readings are typically subject-referencing readings (‑an, ‑ant, ‑arian, ‑eer, ‑er, ‑ist, ‑meister,
‑ster), those which are typically object-referencing (‑ee), those which denote inhabitants or
languages (‑ite, ‑ese, ‑i, ‑ish), and those which denote feminine gender (‑ess, ‑ette, ‑trix). We
stress at the outset that these are only rough categories and that they pertain to primary read‑
ings; a great many of the affixes that we discuss in this chapter have multiple readings, a point
which will be taken up in detail in Section 11.3.

11.2.1 Subject-referencing affixes: ‑er, ‑ant, ‑an, ‑ist, ‑meister,

-eer, ‑ster, ‑nik, ‑arian
We divide this group up into those that attach primarily to verbs (‑er, ‑or, and ‑ant), those that
attach primarily to nouns (‑an, ‑ist), and the minor suffixes (‑meister, ‑ster, ‑eer, ‑nik, and
‑arian).

11.2.1.1 ‑er and ‑ant

Of this group of suffixes, by far the most productive is ‑er, which attaches to a wide array of
base types. It attaches indiscriminately to either native or non-native bases. It seems safe to
say that ‑er can be used on just about any transitive verb in English, and on a wide variety of
other verb types as well, as the examples in (1) illustrate.

(1) a. on transitive verbs: writer, owner, eater, attacker, employer

b. on di-transitive verbs: comparer, confounder, giver, trader, seller, offerer, placer,
sender
c. on prepositional object verbs: clinger, consulter, defaulter, disposer, escaper, harmonizer,
listener, looker, moocher
d. on unergative verbs: beeper, belcher, blinker, dancer, buzzer, cougher, drifter, fencer,
fiddler, sleeper, prancer, ranter, shrugger, skulker, womanizer
e. on unaccusative verbs: comer, descender, faller, perisher, riser, sloper
f. on causative/inchoative verbs: broiler, breaker, dropper, frier, hurter, leaker, rattler,
roller, suckler, increaser
218 Derived nouns: personal and participant

g. on verbs taking sentential complements: boaster, hoper, proclaimer, realizer, reckoner,

resolver, swearer, theorizer, thinker, reasoner, venturer, wonderer

It is also worthy of note that ‑er forms can be derived from particle verbs (pick up, punch out,
etc.) In NAmE this is generally done by adding ‑er to both elements, as shown in the COCA
examples given in (2). For these forms, only four competing forms with only one ‑er are
attested (putter-in, fixer-up, picker-up, loosen-upper). In BrE such ‑er derivatives of phrasal
verbs seem to be avoided: of the sample of bases shown in (2), only one has a derivative
attested in the BNC (cleaner-up). We find in the BNC, however, four words that do not occur
in COCA (roller-upper, washer-upper, roper-inner, stopper-inner).

(2) bringer-downer, cheerer-upper, cleaner-outer, cleaner-upper, cutter-upper, exciter upper,

filler-outer, filler-upper, fixer-upper, leaver-outer, loosener upper, patcher-upper, pepper-
upper, perker-upper, picker-upper, plane-checker-outer, puncher outer, pusher-upper,
putter-inner, roller-upper, roper-inner, setter-upper, sniffer-arounder, sopper-upper,
starter-upper, stopper-inner, taker-outer, thinker-upper, waker-upper, warmer-upper,
washer-upper, wire-puller-upper

Although some of the examples in (2) are intended to be humorous, not all are, presumably
as there is no plausible productive alternative that can be used to derive an agent noun for
this class of verb.
The suffix also appears on a wide variety of other categories. It is commonly found on
nouns, both proper and common.

(3) a. proper: Afrikaaner, Aucklander, Berliner, Dusseldorfer, Dubliner, Icelander, Londoner,

Lubavitcher, Montrealer
b. common: birder, bondager, caver, churcher, confectioner, falconer, forker, freighter,
islander, lifer, magicker, nutter, porker, ratter, roader, snouter, stoner

It can be found attached to nominal compounds and entire phrases as well.

(4) a. on nominal compounds: shirtwaister, wildcatter, backbencher, backpacker, freestyler,

householder, lowlander, outfielder, peasouper
b. on phrases: do-gooder, no hoper, hardliner, naked-heeler, spoken-worder, main-
chancer, all-nighter

And it is found at least occasionally on adverbial/prepositional bases and on cardinal

numbers.

(5) a. on adverbial/prepositional bases: insider, downer, outsider, southwester, uptowner

b. on cardinal numbers: fiver, niner, tenner, thousander
 11.2 Formal considerations 219

For the most part, ‑er attaches to free bases, although there are a number of neoclassical compounds
and other forms which end in ‑y when unsuffixed, but where we find loss of the final ‑y before ‑er:

(6) biographer, astronomer, philosopher, necromancer, adulterer, sorcerer, usurer

There are also several native forms (chandler, vintner, glazer) whose meaning suggests that they
are formed with these suffixes, but where there is no existing free base. There are several others
where an extending ‑i‑ exists, but which should probably be grouped with the previous three:
glazier, hosier. And the form collier features both an unusual base allomorph and the extender
‑i‑. It seems safe to say, however, that ‑er does not typically appear with bound bases.
The variant ‑or has a rather different formal profile. It attaches almost exclusively to non-
native bases, the vast majority of them verbs, and many of them verbs in ‑ate.

(7) a. on verbs in ‑ate: abdicator, accommodator, agglomerator, annihilator, captivator,

cremator, dehydrator, educator, humiliator, inseminator
b. on other non-native verbs: adjustor, assessor, attractor, circumcisor, condensor,
corrector, creditor, detector, enactor, governor, infector, licensor

The only form on a native base is sailor, which according to the OED is a nineteenth-century
respelling of earlier sailer.
Unlike ‑er, ‑or is found not infrequently on bound bases:

(8) ‑or on bound bases: acquisitor, admonitor, capacitor, cognitor, defensor, divisor, guarantor,
interventor, lector, mentor, seductor

As with the suffix ‑er, ‑or is possible with verbs of most types, although transitive verbs predominate.

(9) a. on transitive verbs: actuator, advisor, condensor, dedicator, emancipator, irrigator,

perpetrator
b. on verbs with prepositional objects: assimilator, commiserator, cooperator, interactor,
intervenor, participator
c. on unergative verbs: brachiator, hibernator, introspector, prevaricator
d. on unaccusative verbs: decelerator, mutator, regressor
e. on transitive/inchoative verbs: alternator, evaporator, levitator
f. on verbs taking sentential complements: postulator, prognosticator

Only a small handful of forms are found on nominal bases:

(10) debtor, senator, suitor, cryptor, duplexor

220 Derived nouns: personal and participant

Although it appears as ‑or in most forms, the suffix does occasionally exhibit the extenders
‑at‑ and ‑it‑, with the extender ‑c‑ showing up in verbs formed with the suffix ‑ify:

(11) a. ‑ator: declarator, continuator, conservator, configurator, citator, respirator,

comparator, conspirator, preparator, inhalator
b. ‑itor: competitor, servitor, conformitor, compositor
c. ‑cator: scarificator, purificator

The suffix ‑er is clearly productive on verbal and nominal bases. It attaches to just about any
newly formed verb to create the corresponding agent/instrument noun (googler, emailer),
and it appears on nouns as well (birther). Its non-native counterpart ‑or shows some degree
of productivity on non-native verbal bases as well, as evidenced by forms that appear in
COCA, but which are unattested in the OED:

(12) acculturator, adaptor, attributor, commentor, conformitor, cryptor, curruptor, debator,

deductor, depictor, discussor, duplexor, emanator, enculturator, exacerbator, fractionator,
frustrator, gestator, hereditator, implementor, improvisor, inflator, inflictor, intervenor,
iterator, manifestor, obviator, palpator, repulsor, requestor

Deverbal forms in ‑er are typically able to project the object argument of the base verb in the
form of a prepositional phrase headed by of: for example, the reader of the book; the writer of
the novel. Projection of prepositional objects is possible, as attested in formations like a lis‑
tener to classical music. Ditransitive verbs allow projection of the theme argument with the
preposition of (a seller of mangos), and less easily the goal argument with to (a seller to the
US), but simultaneous projection of both arguments appears to be unattested in the corpus
(e.g. a seller of mangos to the US). And with verbal bases taking sentential complements, pro‑
jection of the sentential complement seems impossible: *the realizer that the car was out of
gas; *a boaster that she could swim ten miles. A similar pattern obtains for the suffix ‑or.
The suffix ‑ant and its variant ‑ent are more restricted in their base types than ‑er and ‑or.
They attach predominantly to transitive verbs, with very few items in other verbal categories.
The formative ‑ant is found not only with nominal formations, but also with adjectival ones.
There is some overlap, in the sense that some forms (mutant, attendant, repellant) can be
construed as either nominal or adjectival in category depending on the syntactic context.
Not every adjectival form in ‑ant has a nominal correlate and vice versa, however; forms like
luxuriant, malignant, nurturant, abundant are used chiefly as adjectives, and forms like defend‑
ant or discussant as nouns.

(13) a. on transitive verbs: combatant, decongestant, defendant, deterrent, discussant,

inhabitant, protectant, suppressant
 11.2 Formal considerations 221

b. on prepositional object verbs: accountant, confidant, reactant, respondent

c. on causative/inchoative verbs: floatant
d. on unergative verbs: exhalant
e. on unaccusative verbs: descendant

Adams (2001: 31) notes that euphoriant is formed on an adjectival base, and to this we can add
sterilant. We know of no other examples in which ‑ant attaches to bases other than verbs or
bound bases.
This suffix occurs frequently on bound bases that have corresponding forms ending in
verbal ‑ate, or in other suffixes (‑ation, ‑able, or ‑ee, for example).

(14) a. on bound bases of verbs in ‑ate: asphyxiant, desiccant, fumigant, lubricant, migrant,
officiant, replicant
b. on other bound bases: applicant, appellant, occupant, reductant, surfactant

The suffix ‑ant attaches almost exclusively to non-native bases, the only apparent exceptions
being coolant, floatant, and healant.
The suffix exhibits a small degree of productivity in contemporary English, as evidenced
by the following examples, which can be found in BNC or COCA, but which do not appear
in the OED.

(15) actant, arrestant, calibrant, clonant, commutant, comparant, conductant, degreasant,

encapsulant, floatant, healant, innoculant, titrant

Forms derived with the suffix ‑ant seem to be more resistant to projecting the argument
structure of their verbal bases than are forms in ‑er. Phrases like an inhabitant of Canada or the
occupant of the White House seem natural enough, a discussant of issues somewhat less so,
although all three are attested in COCA. But many ‑ant forms derived from transitive verbs
sound odd with a projected argument, and are not attested in this pattern in the corpora.

11.2.1.2 ‑ist and ‑an

Although ‑ist and ‑an overlap in function with each other and with the suffixes in the previous
section, their formal profiles exhibit significant differences. The suffixes ‑ist and ‑an attach
primarily to non-verbal categories, in contrast to ‑er, ‑or, and ‑ant, which take verbs as their
bases far more often than nouns or adjectives. We consider each in turn.
The suffix ‑ist attaches to a wide range of nouns, both common and proper, as well as to
adjectives; it can be found only infrequently on verbs.

(16) a. ‑ist on common nouns: ageist, alarmist, artist, balloonist, banjoist, druggist, novelist,
racist, rockist, sexist, tourist
222 Derived nouns: personal and participant

b. ‑ist on proper nouns: Baathist, Dadaist, Darwinist, Decembrist, Francoist,

Mitterrandist, Stalinist, Zionist
c. ‑ist on adjectives: ableist, absurdist, dualist, extremist, leftist, modernist, purist,
realist
d. ‑ist on verbs: accompanist, conformist, shootist, splittist, transformist

The suffix also appears attached to bound bases.

(17) ‑ist on bound bases: agonist, altruist, animist, aphorist, baptist, catechist, chemist, deist,
dentist, evangelist, jurist, linguist

In some, but not all of these cases, the same bound base is also found with either ‑ize or ‑ism
or both (altruism ◆ altruist; agonism ◆ agonize ◆ agonist; baptism ◆ baptize ◆ baptist), suggesting
a pocket of English derivation with a paradigmatic flavour (see also Chapter 23).
Of all the agentive suffixes, ‑ist has the greatest propensity to appear on already derived
bases, both nominal (18) and adjectival (19). Interestingly, it appears at least sporadically on
other affixes that derive agents or instruments, as the examples in (18d–e) illustrate:

(18) ‑ist on other noun-forming suffixes

a. on ‑ion: abolitionist, adoptionist
b. on ‑age: assemblagist, marriagist
c. on ‑ment: movementist
d. on ‑an: Africanist, Algerianist
e. on ‑er/‑or: consumerist, redemptorist, detectorist, scooterist, settlerist, workerist

(19) ‑ist on adjective-forming suffixes

a. on ‑al: ambientalist, accidentalist
b. on ‑ic: aerodynamicist, aestheticist, historicist
c. on ‑ive: collectivist, objectivist, positivist, relativist
d. on ‑able (in its variant ‑ible): impossibilist, infallibilist
e. on ‑ant (and its variant ‑ent): indifferentist, obscurantist

It is not unusual to find ‑ist on multiply-affixed bases, as in examples like falsificationist, consti‑
tutionalist, congregationalist. Indeed, we sometimes find ‑ist in doublets like those in (20). In
some cases, such as (20a), there is no discernable semantic difference between the form
derived on the complex adjectival base and that on the complex nominal base. More often
than not, however, there is a distinction, as the examples in (20b) illustrate:
 11.2 Formal considerations 223

(20) a. Daily Telegraph 1992: Based on the teachings of A. S. Neill, the Scottish educationalist
and founder of Summerhill, Kinokuni is offering pupils an “alternative to the
production-line techniques” of the Japanese education system.
Parliamentary Affairs 1991: Two generations later one educationist, Geoff Whitty,
questioned the value of the political education movement precisely because of the
official support it had received.
b. The Art Newspaper 1992: Throughout his career at the library, James constantly
blocked efforts to improve security and cataloguing procedures, in particular those
proposed by the director, Islamicist Wilfrid Lockwood, who had formerly worked
with British Intelligence.
The Economist 1985–1994: It is trying to assert itself politically, offering money the
PLO can not match to young, educated Gazans to join the Islamist movement.

In the first example in (20b), we have a form that denotes someone who studies Islam, in the
second a form that refers to an advocate of a political system that is based on Islam.
In addition to appearing on bases formed by derivation, ‑ist can be found on compounds,
and occasionally on phrases.

(21) a. ‑ist on compounds: foot fetishist, folklorist, keyboardist, morphologist, touch-typist,

trade-unionist, watercolorist
b. ‑ist on phrases: gold-medalist, raw foodist, short-termist, white-supremacist

We find ‑ist predominantly on non-native bases and suffixes, including bases from outside the
classical languages or French (banjoist, Daoist, novelettist, Sikhist, tattooist). However, it does
occasionally take native bases and suffixes, as the examples in (22) illustrate.

(22) ‑ist on native bases: duckist, fattist, folklorist, harpist, hornist, keyboardist, landscapist, leftist,
rightist, scooterist, settlerist, shootist, splittist, stockist, welfarist, womanist

The suffix ‑ist does not generally affect base stress patterns (though note %obscurántist, cited
above) but it does sometimes induce elision of a root-final vowel, as illustrated in (23).

(23) a. elision of root-final /əʊ ǁ oʊ/: cellist, librettist, pianist, scenarist

b. elision of root-final /ə/: Buddhist, propagandist, Spinozist

Vowel elision tends, however, to be avoided when the result would be to reduce the base to
a single syllable. So we find hiatus in forms like copyist, Dadaist, flunkyist, hobbyist, lobbyist,
oboist, soloist, stuccoist, zeroist. Hiatus is also occasionally found in trisyllabic bases
(Nkrumahist, Pyreneeist). Forms with hiatus seem generally acceptable if their base-final
vowel is /u:/: fujitsuist, tatooist. An extender can be occasionally found (attitudinist, lutenist,
224 Derived nouns: personal and participant

platonist, tobacconist; egotist, programmatist, suprematist), but the appearance of these

extenders does not appear to be in any way systematic. The suffix ‑ist triggers velar softening,
as in cases like lyric ◆ lyricist, physics ◆ physicist.
The suffix ‑an forms agent nouns, nouns that denote inhabitants of a place or followers of
a person, among other things. We will return to its semantic properties in Section 11.3.
Formally ‑an, like ‑ist, attaches freely to nouns, both common and proper, and to bound
bases.

(24) a. ‑an on common nouns: clinician, comedian, grammarian, guardian, logician,

madrigalean, magician, musician, republican, vaudevillean
b. ‑an on proper nouns: Bolivian, Chicagoan, European, Fijian, Friedmanian, Gaussian,
Indonesian, Jungian, Leibnizian, Mozartian
c. ‑an on bound bases: amphibian, crustacean, equestrian, mortician, patrician,
quotidian, urban

We do not find ‑an on compounds or phrases.

Unlike ‑ist, ‑an rarely attaches to adjectives. There are no examples of this suffix attaching
to underived adjectives, and although there are a number of apparent forms that appear to be
derived on adjectives in ‑ic and ‑ary, it is unclear on semantic grounds whether these should
be analysed as true derivations from the adjectives. In forms like electrician or diagnostician,
derivation from the adjective seems plausible, but such an analysis seems less plausible for
theoretician (a theoretician is someone who does theory), for example. Also unclear is whether
forms like parliamentarian and disciplinarian are derived from adjectives parliamentary and
disciplinary or from the corresponding nouns parliament and discipline. To complicate matters
further, there is also the suffix ‑arian, which we will discuss further below.
Except when attached to proper nouns, ‑an takes almost exclusively non-native bases (but
see the exception elvan). It characteristically induces stress shift and palatalization of /t/, /s/,
/k/, in the appropriate environments, with primary stress appearing on the syllable preceding
the suffix—either the penultimate or antepenultimate, depending on whether the suffix is ‑an
or has an extender ‑i‑ or ‑e‑. For a more detailed discussion of the stress patterns of ‑an deriva‑
tives the reader is referred to the chapter on derived adjectives (Chapter 14), where the perti‑
nent properties are dealt with in the context of intricacies of related adjectival stress shifts.
The affix appears mostly as ‑ian, with ‑an occuring only rarely on common nouns (proto‑
zoan), bound bases (vegan), and personal names (Augustan, Copernican, Elizabethan,
Malartan, Petrarchan, Paulan). The variant without extender appears with some frequency
on place names, especially those ending in vowels to begin with (African, American, Anjouan,
Antiguan, Calcuttan, Chicagoan, Roman). Base-final schwa often coalesces with the schwa-
initial suffix, but it is not systematic where we find ‑an and where we find ‑ian on place names
ending in <a> or <o>; compare forms like Barcelonian and Canadian to American and
Calcuttan. Consonant-final place names tend to take ‑ian (Brazilian, Brightonian, Chaddian,
 11.2 Formal considerations 225

Egyptian, Palestinian), although forms like Roman also exist. Truncation of final segments can
also be seen in cases like Barbadian, Belgian, and Athenian.
Of the two suffixes ‑ist and ‑an, ‑ist is clearly the more productive one on common nouns,
with both suffixes displaying productivity on proper nouns. Humorous forms like fattist or
duckist (see example (22)) suggest that ‑ist is productive to some extent on the pattern of
‘person bearing a prejudice against X’, but it also seems likely that given a new activity, ‑ist
would be available simply for naming a practitioner of that activity; so someone engaging in
the relatively novel fitness activity called zumba might very likely be referred to as a zumbaist,
far less likely as a zumbian. On the other hand, it seems possible to attach ‑ist and ‑ian to just
about any personal name, deriving a noun denoting an adherent or follower: Bauerist,
Bauerian, Lieberist, Lieberian, Plagist, Plagian all seem likely creations, given the right
circumstances.

11.2.1.3 Minor suffixes: ‑eer, ‑ster, ‑meister, ‑nik, ‑arian

It should be said at the outset that ‑eer, ‑ster, and ‑meister are all surprisingly productive, per‑
haps more so in NAmE than BrE (‑meister seems to be NAmE only), although they are far
more limited in terms of register and pragmatics than the previously discussed agentives.
That is, the number of types to be found in the corpora is relatively small, but there are nev‑
ertheless quite a few low frequency and unfamiliar items among them that do not appear in
the OED. These suffixes are all agentive in flavour, but unlike the affixes discussed above, each
has an extra evaluative nuance along with agentivity. We will return to their semantic proper‑
ties in Section 11.3.1.
All three attach primarily to nouns, both simple and complex, although all three can also
be found occasionally on verbal bases, and ‑meister and ‑ster appear on a few adjectival bases.
Only ‑eer appears on bound bases, all of these being established examples borrowed directly
from French. In NAmE, ‑ster also attaches to proper names, where it is always accompanied
by the determiner the (that is, the Bingster, the Newtster):

(25) ‑eer
a. on simple nouns: budgeteer, cameleer, canyoneer, planeteer, summiteer
b. on complex nouns: allotmenteer, concessioneer, conventioneer, leafleteer
c. on phrases: free-marketeer
d. on verbs: envisioneer, imagineer, orienteer
e. on bound bases: buccaneer, pioneer, volunteer

(26) ‑ster
a. on simple nouns: anorakster, corkster, dealster, fraudster, funster, soupster, twerpster
b. on compounds: junk-bondster
226 Derived nouns: personal and participant

c. on proper nouns: Binkster, Budster, Chuckster, Newtster

d. on adjectives: hepster, oldster, swankster, youngster
e. on verbs: dumpster, dunkster, spinster, strumster

(27) ‑meister
a. on simple nouns: bagelmeister, cafemeister, peatmeister, ragemeister, trashmeister
b. on complex nouns: action-meister, blockbuster-meister, coalition-meister, licemeister,
mega-hitmeister
c. on adjectives: glibmeister
d. on verbs: boinkmeister, carvemeister, drill-meister, skimeister, talkmeister

These suffixes display no allomorphy, although apparent extenders appear in a few cases with
‑eer: funkateer, muleteer, mouseketeer, the latter clearly on analogy to musketeer. Although ‑eer
favours non-native bases, all three can be found on either native or non-native bases.
The suffix ‑nik was productive primarily in the 1960s and 1970s (see Bauer 1983: 256–66),
with denominal, deverbal, and deadjectival forms such as citynik, failnik, but also derivatives
based on phrases (e.g. holdupnik, no-goodnik, nuclear-freezenik, way-outnik). In COCA and the
BNC we also find very few more recent formations such as flopnik.
In many cases derivatives ending in the string <arian> can be analysed as containing the
adjectival or nominal suffix ‑ary followed by the nominal suffix ‑an (e.g. contr-ary-an, document-
ary-an, libr-ary-an). The reason for treating ‑arian as a suffix in its own right is that it attaches
(moderately productively) to many bases where no corresponding form in ‑ary (or ‑ar form)
is attested (e.g. conservatarian, fruitarian, trinitarian, vegetarian versus Nconservat-ary, Nfruit-
ary, Ntrinit-ary, Nveget-ary). In these cases, the derivatives in ‑arian regularly denote persons
who follow a doctrine or frame of thought related to the base. The suffix is auto-stressed with
main stress on the antepenultimate syllable (authoritárian, communitárian, uniformitárian). In
COCA as well as the BNC we find the suffix exclusively on non-native bases.

11.2.2 Object-referencing affixes: ‑ee

The suffix ‑ee is the only suffix in contemporary English that is primarily object-referencing,
although as we will see in Section 11.3.1, its semantics is fairly complex. It can be found on a
wide variety of verbal bases, and occasionally on nominal bases:

(28) ‑ee on verbs

a. on transitive verbs: arrestee, batteree, clonee, crushee, infectee, offendee, penetratee,
producee, spankee
 11.2 Formal considerations 227

b. on ditransitive verbs: addressee, allotee, assignee, consignee, dedicatee, givee, issuee,

offeree
c. on verbs with prepositional objects: complainee, conferee, insistee, lookee
d. on unergative verbs: bowlee, enlistee, snoree
e. on unaccusative verbs: fallee, standee

(29) ‑ee on nouns: bargee, biographee, executionee, haircutee, mastectomee, migrainee,

refugee

Among the verbal bases, forms derived on transitives and ditransitives are most common, as
would be expected for a suffix that is primarily object-referencing. It is interesting, however,
that intransitive verbs—both unaccusative and unergative—are also attested. The reasons for
this will be clarified in Section 11.3.1. Forms on nominal bases are rare, but they do occur.
The suffix ‑ee can be found on both native and non-native bases:

(30) a. ‑ee on native bases: askee, callee, chokee, cuddleee, drawee, fuckee, gossippee, helpee,
standee, tellee, washee, wishee
b. ‑ee on non-native bases: complainee, contactee, expellee, indictee, interpretee,
monitoree, observee, seducee

The suffix ‑ee prefers underived bases, but does occasionally appear on complex bases, as in
blackmailee, optionee, overlappee. For verbal bases in ‑ate, the suffix is sometimes truncated
(amelioree, nominee, rehabilitee, separee) and sometimes not (appropriatee, delegatee, enuncia‑
tee). Like ‑arian, ‑ee itself bears primary stress, with the usual stress alternation effects auto-
stressed suffixes bring with them (emplóy, but èmployée; see Chapter 9.4.3).
It is unclear whether argument structure can be projected with deverbal derivations in ‑ee.
It is easy enough to find phrases like the company’s employee attested, but more difficult to
argue that the company is a projection of the subject argument of the verb, as opposed to a
simple possessor. In the example from COCA in (31), however, the presence of the adjective
recent forces the subjective reading of the possessive phrase:

(31) Arab Studies Quarterly 2002: It is no surprise that both interim Afghan President Hamid
Karzai and Zalmay Khalilzad, Bush’s recent appointee as special envoy to Afghanistan,
were former consultants to UNOCAL, the United States oil company that has sought
to build an oil pipeline through Afghanistan.

Other convincing examples that are suggestive of the projection of argument structure come
from ditransitive verbs, where the ‑ee form is construed as goal, and the theme argument
appears in a prepositional phrase headed by of. Examples like the following can be found in
COCA.
228 Derived nouns: personal and participant

(32) Style 2002: The “gentle friend,” however, seems to disappear from the poem at its
conclusion, frozen out of the scene as the speaker turns definitively toward her
disembodied lover, the addressee of her final series of speech acts: “shall not I, too, be,
/My spirit-love! upborne to dwell with thee?”
The Salzburg Years 1991: Meanwhile Queen Charlotte had asked to be the dedicatee of
some of Mozart’s works which Leopold duly had engraved at his own expense.

Such examples are relatively infrequent, however, and we must conclude that at this point the
evidence in favour of projection of argument structure is suggestive but not conclusive.

11.2.3 Inhabitant and language: ‑ite, ‑ese, ‑ish, ‑i

The suffixes ‑ite, ‑ese, ‑ish, ‑i, and ‑an all form inhabitant and/or language nouns. We have
already discussed the formal characteristics of ‑an in Section 11.2.1. Here we will take up ‑ite,
‑ese, ‑ish, and ‑i. These four have in common their propensity to attach to proper nouns, espe‑
cially place names, and to designate inhabitants associated with those places, among other
things. We will go into their semantic characteristics in Section 11.3.2.
The suffix ‑ite occurs with some productivity on names of cities, counties, and states, as
well as on personal names, in which case it means ‘follower of ’, see Section 11.3.2.

(33) a. ‑ite on place names: Berkeleyite, Bethlehemite, Bombayite, Boulderite, Bronxite,

Brooklynite, Cannes‑ite, Dallasite, Hampshireite, Hamptonite, Harlemite, Hebronite,
Houston‑ite, Jerusalemite, Labradorite, Manhattanite, Memphisite, Muscovite,
Seattleite, Wisconsinite, Wyomingite
b. ‑ite on personal names: Austenite, Bakuninite, Benjaminite, Benthamite, Brezhnevite,
Buchananite, Carterite, Clarkite, Clintonite, Friedmanite, Mengele‑ite, Naderite,
Paulite, Reaganite, Saddamite, Thatcherite

Interestingly, it is infrequent on names of countries, although we do find Yemenite alongside

Yemeni. Some of the formations we find are Biblical in reference, among them Canaanite,
Gileadite, Israelite, Nazarite.
The suffix is rarely found on common nouns or adjectives, although there are a few well-
established forms like anchorite, laborite, socialite, stylite, and urbanite. More frequently we
find ‑ite on bound bases, where it forms chemical, geological, or biological terms.

(34) coprolite, hydrosulfite, merozoite, nitrite, stalagtite, trilobite

According to the OED, the chemical use is actually not the same suffix as ‑ite in the other
cases, but was created as ‘the systematic termination of the names of the salts of acids denom‑
inated by adjectives in ‑ous’, to differentiate such substances from salts corresponding to
adjectives in ‑ic, which receive the ending ‑ate.
 11.2 Formal considerations 229

The suffix ‑ese occurs frequently and productively on place names, on proper nouns other
than place names, and on common nouns. With place names it can designate either the
inhabitant noun or the noun naming the language, or both. With other bases, it designates
something pertaining to language; we will go into its nuances in such cases more deeply in
Section 11.3.2.

(35) a. ‑ese on place names: Burmese, Cantonese, Lebanese, Sudanese

b. ‑ese on proper nouns other than place names: Australopithese, Clintonese,
Greenspanese, Newsweek‑ese, Starbucks‑ese, Swinburnese, Whitehallese
c. ‑ese on common nouns: aboriginese, cat‑ese, computerese, dissertation‑ese, doctor‑ese,
funeralese, headlinese, motherese, servicese
d. ‑ese on adjectives or nouns: Americanese, evangelicalese, institutionalese, legalese,
medicalese, mid-Victorianese

On common nouns it does not discriminate between native and non-native bases.
On place names it frequently results in elision of a base final /ə/ (Burmese, Chinese,
Guyanese), and occasionally other segments (Lebanese, Portuguese), or insertion of an
extender which avoids vowel hiatus (Balinese, Congolese, Javanese, Rwandanese, Shanghainese,
Togolese), but examples like Faroese and Chicagoese suggest that this phenomenon is not sys‑
tematic. The suffix bears primary stress, so when it is attached to a base with stress on the final
syllable, stress shift occurs: Sudán ~ Sùdanése, Taiwán ~ Tàiwanése.
Neither the suffix ‑ish nor the suffix ‑i are productive in forming names of nationalities
(‑ish is productive elsewhere, see Chapter 14). Each is used in just a handful of cases to form
the noun referring to the inhabitants and/or language of a country or other geographic
designation.

(36) a. inhabitant names in ‑i: Adeni, Azerbaijani, Azeri, Baluchi, Bangladeshi, Bengali,
Bhutani, Bihari, Gujarati, Iraqi, Israeli, Kashmiri, Kuwaiti, Nepali, Pakistani, Panjabi,
Somali, Tajiki, Yemeni, Zanzibari
b. inhabitant names in ‑ish: Danish, English, Finnish, Flemish, Frankish, Gaulish, Irish,
Kentish, Kurdish, Pictish, Polish, Spanish, Swedish, Turkish

The suffix ‑i seems to be confined to inhabitant names in parts of the Middle and Far East.
Bauer (1983: 253‑5) raises the question whether it makes sense even to call ‑i a bona fide English
suffix, as many of the items in (36a) are borrowed, and limitations on available bases make the
creation of new forms unlikely. The suffix ‑ish seems to occur only on monosyllabic bases.
With regard to inhabitant names, it is generally hard to find clear patterns. Two potential
patterns are that place names ending in ton take ‑ian (Bostonian, Wellingtonian), and those
ending in ‑land tend to take ‑er (e.g. Aucklander, Icelander), but there are many instances of
unpredictable bases (Scotland ◆ Scot, Monaco ◆ Monegasque, Uzbekistan ◆ Uszbek).
230 Derived nouns: personal and participant

11.2.4 Gender: ‑ess, ‑ette, ‑trix

Of the three contemporary feminizing suffixes, ‑ess and ‑ette display modest productivity,
whereas few new forms in ‑trix can be found in the corpora. All three are confined to nominal
bases, either native or non-native:

(37) a. ‑ess: bartendress, bumess, composeress, dictatress, falconress, folkstress, funkstress,

hauntress, huckstress, kidnapperess, maestress, mentoress, monstress, punkstress,
vampiress, witchess
b. ‑ette: astronette, bumette, chefette, columnistette, conductorette, disciplette, dudette,
goblinette, gothette, hustlerette, idolette, nerdette, professorette, raiderette,
reporterette, santa‑ette, wenchette
c. ‑trix: auctionatrix, auxiliatrix, coredemptrix, punditrix, robotrixes

The only base allomorphy we find with ‑ess and ‑ette is the frequent elision of /ə/ in bases
ending with the suffix ‑er or ‑or. The form governess may be analysed as having base-final ‑or
elided, or as a rare case of ‑ess attached to a verb. The form governess could, however, also be
analysed as a case of (lexicalized) haplology, analogous to murderess and sorceress (Chapter
9.5, Plag 1998: 206).

11.2.5 Prefixes
There are also four person noun-forming prefixes grand‑, great‑, step‑, and vice‑, which all
attach exclusively to nouns (e.g. grandmother, grand-niece, step-daddy, stepfather, stepnephew,
vice-chair, vice-chancellor, vice-regent). The grand‑ and step‑ formations are main-stressed on
the prefix, while great‑ and vice‑ formations are regularly secondarily stressed on the prefix,
with occasional, lexicalized exceptions (e.g. víceroy).

11.3 Semantic considerations

What is most striking about the affixes discussed in this chapter is the degree of polysemy
that they exhibit, and the extent to which their domains overlap with one another. Although
one might expect that affixes would exhibit distinct and specialized semantic domains, this is
certainly not the case in contemporary English. At best we can speak of the predominant
semantic domains or functions of an affix, but virtually all the affixes we discuss in this chap‑
ter also have secondary domains or functions. Indeed, they overlap in some cases with the
noun-forming affixes discussed in Chapters 10 and 12.
Table 11.1 illustrates these points by giving an overview of the semantic domains in which
the personal/participant suffixes overlap with the event/state/result suffixes. The overlaps
 11.3 Semantic considerations 231

are restricted to those semantic domains that might be characterized as thematic domains,
that is, domains like agent, experiencer, instrument, patient, theme, goal, location, measure,
and means. Athematic domains (e.g. ‘follower’, ‘inhabitant’, or ‘chemical’) show no rivalry
between the suffixes of the two sets.
Table 11.1 shows that each of the event/state/result suffixes has a secondary usage as a
personal noun-forming suffix, and that these formations cover the whole range of the mean‑
ings that are conveyed by the personal suffixes, apart from the instrumental meaning, which
is reserved for the two suffixes ‑er and ‑ant. The suffix ‑er has the widest distribution and the
largest number of rival suffixes.
In what follows the semantic behaviour of the personal suffixes in Table 11.1 will be dis‑
cussed in more detail, beginning with the thematic domain, in which all derivatives denote
participants in events. We then go on to look at non-thematic domains including affixes that
form nouns denoting followers, inhabitants, names of languages, chemical, biological, or
geological terms, kinship terms, or terms denoting gender.

11.3.1 Thematic domains

The thematic domains we will explore here are those of agent (also experiencer, stimulus),
instrument, patient/theme, goal, location, measure, and means (see Chapter 2). English is

Table 11.1 Overlapping semantic domains: thematic readings of nominal suffixes

Agent/ exper./ Instrument Patient/ Location Measure Means
stimul. theme
Personal/ participant
‑er prim prim sec sec sec sec
‑ant prim prim
‑ist prim
‑ee sec prim
‑an prim
‑ster eval
‑eer eval
‑meister eval
Event/ state/ result
‑ing sec sec sec sec
‑ation sec sec sec sec sec
‑ment sec sec sec
‑ure sec sec
‑ance sec sec sec sec
‑age sec sec
‑ery sec sec
conversion sec sec sec sec

Note: prim = primary use, sec = secondary use, eval = with evaluative flavour
232 Derived nouns: personal and participant

rich in derivational affixes that form agent nouns, with agent-forming affixes frequently also
deriving instrument nouns. Interestingly, the language is far less rich in affixes that derive
other thematic domains. Indeed, what we find is that thematic domains other than agent and
instrument are generally covered by a number of affixes, with those affixes in turn often being
used primarily or partially for other purposes.

11.3.1.1 Agent, experiencer, instrument: ‑er

The primary affixes in English for deriving agent, experiencer, and instrument nouns are ‑er,
‑ant, ‑ist, and ‑an, with ‑er being by far the most productive. Although agent nouns formed
with the suffixes discussed in this section can denote professionals (e.g. accountant, acupunc‑
turist, biologist, breeder, clarinettist, consultant, grammarian, librettist, mortician, photographer,
pianist, taxidermist, trader, writer), they certainly need not; whether they do or not depends
to some extent on whether the verbal base involves an activity that can be done as an occupa‑
tion. Certainly, for example, mortician denotes a professional, but only because preparation
of the dead for burial is not something that is done casually. Similarly with acupuncturist. But
for clarinettist or photographer, the non-professional reading is certainly possible.
The vast majority of agent terms are derived with the suffix ‑er; indeed ‑er is probably the
default for novel agentive formations, as the low-frequency items below suggest.

(38) a. New England Review 2007: She didn’t remember him being such an inveterate
shrugger.
b. NPR Science 2009: That’s more straightforward. One of our Twitter commenters,
I guess a tweeter, did note that the promise of this is much better math ability.
c. Teaching Spelling 1988: Because the error often produces a different word altogether
the misspeller ought to recognise his mistake: so this again suggests poor visual
memory.

Whether these are novel forms or not is probably impossible to determine, as the suffix is so
productive that new forms derived with it are likely to go entirely unnoticed and therefore
uncaptured in dictionaries; none of these three appears in the OED as of the writing of this
chapter (tweeter occurs, but with an older instrumental reading).
Given the right kind of verbal base, forms in ‑er like hearer, seer, smeller or pleaser, screamer,
puzzler may be interpreted as experiencers or stimuli, rather than as agents per se:

(39) a. Odour nuisances and their control 1984: A higher test, namely that “nothing in the way
of smells was a nuisance to public health unless it so nauseated the smeller that he
vomited” was rejected.
b. Harvard Journal of Law and Public Policy 2005: The question whether identity is a
limitation or starting point for freedom may be a puzzler for twenty-first century
man, but it is an easier question when tossed the way of Publius.
 11.3 Semantic considerations 233

c. Entertainment Weekly 2001: An atypically slapsticky episode, but it works: Perry

and Schwimmer do the girlyman thing to spastic perfection, and the final diner
scene with Monica dancing to “YMCA” while wearing flame-retardant boobs is a
howler.

In other words, to the extent that thematic roles other than agent are subject-referencing, ‑er
can form participant nouns expressing those roles as well (see Rappaport Hovav and Levin
1992; Booij 1986; Lieber 2004).
Instrument nouns are also subject-referencing, and not surprisingly ‑er is used produc‑
tively in their derivation. We find forms like amplifier, atomizer, beeper, blender, coaster, con‑
verter, cruiser, freighter, mailer, recliner, revolver, sparkler which are lexicalized as instruments,
but also cases like browser, camper, kneeler, knocker, rocker, server, slaver, tumbler, walker,
whaler which can occur as agents or instruments depending on the animacy of the referent
and on specific context.
Where we find ‑er on nominal, adjectival, or adverbial bases, such forms can denote agents
or instruments, just as deverbal ‑er forms can. In such cases, the verbal relation is implied by
context or can be inferred from the nature of the non-verbal base. So for example, a confec‑
tioner is someone who makes or sells confections, a cricketer someone who plays cricket, and
a freighter something that carries freight. Such interpretations follow from the sort of activi‑
ties that the base nouns could conceivably be involved in.

11.3.1.2 Agent, experiencer, instrument: ‑ant

The suffix ‑ant has also formed a significant number of agent forms in English, but it appears
rarely to be productive in this function any more. Of the few apparent novel forms we find in
the corpora, only two – actant and commutant – are agents.

(40) a. Style 1994: Stanzel’s category, it will be remembered, is based on a binary opposition
of the (non) coincidence of “realms of existence,” whereas in Genette’s model the
defining criterion is whether or not the narrator is an actant on the story level of the
narrative.
b. Atlanta Journal Constitution 2005: Compared to my colleagues who struggle with
more challenging commutes, I’m a lucky man. . . . Trim 8-hour commute, then let’s
talk # You can learn a lot being a “commutant.”

This suffix has frequently derived instrument nouns, specifically those denoting chemi‑
cal agents: accelerant, adulterant, asphyxiant, coagulant, colorant, concealant, contaminant,
defoliant, denaturant, deodorant, depressant, desiccant, fumigant, inhalant, intoxicant, lubri‑
cant, oxidant, pollutant, propellant, relaxant, sealant, sterilant, stimulant. Among the appar‑
ently novel instrument forms found in the corpora, we find encapsulant, floatant, and
healant.
234 Derived nouns: personal and participant

11.3.1.3 Agent, experiencer, instrument: ‑ist

There are many forms in ‑ist that have a truly agentive flavour, but there are also many that
denote followers of a prominent person or doctrine rather than participants in an event. The
two readings shade off into each other with no distinct line between them; whether we find a
thematic agentive reading or an athematic non-agentive reading depends largely on the nature
of the base noun. Given a base noun that can be construed as a participant in an event, the
agentive reading becomes more plausible. So a picturist can be someone who makes or does
something with pictures (see (41d)). On bases that denote modes of thought, political or
social movements, and the like, the athematic reading is more likely to surface; a consumption‑
ist therefore need not be someone who consumes, but can rather be someone who believes in
some theory concerning consumption. Further, as Marchand (1969: 309) points out, any noun
designating a doctrine that can be formed using the suffix ‑ism can also have a corresponding
noun in ‑ist, denoting a proponent of that doctrine (see also Chapter 12 on ‑ism).
We will return to the athematic use of ‑ist, including some rather specialized ones, in
Section 11.3.2. In (41) we give some of the novel agentive forms in ‑ist from our data.

(41) a. affairist, aggressionist, ameliorist, assemblagist, batonist, celebrologist, concubinist,

confrontationist, dietologist, explorationist, expressivist, installationist, picturist
b. Innocent III 1993: The priest had to be denounced by his parishioners; many with
concubines would have escaped challenge. The married priest might be more
obvious to the archdeacon on his visitation or inspection and to the rural deans in
charge of the groups of parishes; the unmarried concubinist could well escape
justice.
c. The Art Newspaper (1985–1994): There he assisted him in the preparation of an
installation called “La Salle Blanche”. Broodthaers, who was an installationist in the
days when such work was still rather uncommon, became a guru to a select handful
of French and later American artists.
d. Conversation 1991: In the all play category the word card is shown to the picturist of
each team. The all play word is sketched simultaneously by picturists in their
respective teams at the start . . .

Unlike the other agent-forming suffixes we have discussed so far, ‑ist does not also form
instrument nouns.
It does have two more specialized functions, however. It is standardly used on names of
musical instruments to denote someone who plays that instrument (banjoist, hornist). Its
competitor ‑er in this domain seems restricted to only a handful of forms (e.g. drummer,
trumpeter). In at least one case, both can occur, with a potential for a semantic distinction
concerning the type of music or type of harp that they play (harper versus harpist). And in
recent years on the model of words like racist and sexist, it has come to attach to adjectives or
nouns to designate someone with a prejudice against whatever the base denotes (ableist,
 11.3 Semantic considerations 235

ageist, lookist). Some of the latter formations are intended as humorous or ironic (duckist,
fattist).

11.3.1.4 Agent, experiencer, instrument: ‑an

As with ‑ist, the suffix ‑an exhibits both thematic and athematic uses, deriving agent nouns,
but also nouns denoting inhabitants, followers or proponents of an individual or doctrine, as
well as several more specialized uses. And as with forms in ‑ist, the predominant reading of
forms in ‑an is to some extent determined by the nature of the bases to which it attaches.
Agentive forms in ‑an are generally established and relatively high frequency items (e.g.,
arithmetician, comedian, grammarian, historian, librarian, mortician, musician, physician, politi‑
cian, technician, theologian). We find few if any novel agentive forms in the corpora. This suffix
does not form instrument nouns. We will return to the athematic forms in ‑an in Section
11.3.2. As Bauer (2002) points out, forms in ‑ician (as opposed to ‑ian) are sometimes created
as a means of elevating the status of an occupation (e.g., aesthetician, beautician, mortician).

11.3.1.5 Agent, experiencer, instrument: ‑eer, ‑ster, ‑meister

The derivational patterns we have described thus far are all evaluatively neutral. We turn now
to the suffixes ‑eer, ‑ster, and ‑meister, which in contemporary English are agentive, but which
also convey special evaluative nuances. Perhaps because of already existing forms like musket‑
eer, buccaneer, and cannoneer, ‑eer has sometimes come to suggest someone who not only
performs an action, but does so in a warrior-like way:

(42) New York Times 1992: There is indeed a larger issue at work here, but it is not the one Mr.
O’Keefe mentioned. Mr. O’Keefe, a budgeteer who has yet to sit for Senate confirmation
of his post and who has never served in the military, decided after conferring with
Defense Secretary Dick Cheney, who likewise has never served, that he has the moral
authority to discredit the cultural ethos of the entire Navy based on the conduct of a
group of drunken aviators in a hotel suite.
Christian Science Monitor 2003: May 29, 1953 The first successful ascent, by New Zealander
Sir Edmund Hillary and Nepali Tenzing Norgay. # May 1, 1963 The first American
summiteer, Jim Whittaker. # May 20, 1965 The first person to summit twice, Nawang
Gombu Sherpa.

The suffix ‑ster is more colloquial than ‑er. It appears frequently in journalistic writing, often
with a jocular tone. In novel forms it often carries an undertone of admiration or approval.

(43) Forbes 1996: The remaining stake is owned by dealster Michael Dingman, now a
taxpatriate in the Bahamas (FORBES, Nov. 18), and by Unexim Bank, one of Russia’s
homegrown merchant banks.
Ebony 1995: Shaquille O’Neal, the 23-year-old Orlando Magic megastar center and
master dunkster, is tossing a miniature basketball through one of the many hoops
236 Derived nouns: personal and participant

positioned throughout his sprawling, two-story, $7 million home in an exclusive

Orlando, Fla., suburb.

The second use of ‑ster in contemporary NAmE is somewhat more complex in its semantics.
The suffix is found not infrequently in both colloquial written and spoken English attached to
a person’s name and preceded by the determiner the.

(44) a. Houston Chronicle 1997: You need to start hanging out with the Chuckster more. #
Somebody in the media crowd mentioned that by buddying up with the Chuckster,
Maloney could have had a guaranteed 6-for-18 shooting night, like Barkley had in
Game.
b. San Francisco Chronicle 1996: In the business world no self-respecting Republican
would buy the stock of a corporation whose execs were jumping ship. And why this
sudden flight? Could it have anything to do with the thoroughly inhospitable and
unproductive stance taken by the Newtster and his gang(sters)?

This use always conveys familiarity, and often approval of the person named, as is apparently
the case in (44a), but it can have a slightly derisive tone, as (44b) illustrates.
A third evaluative agentive is ‑meister, which also conveys both agency and a clear evalua‑
tive note.

(45) Atlanta Constitution 2005: Are we to believe that coalitionmeister Ralph Reed expects
Georgians to entrust him with the office of lieutenant governor? When swine take
flight!
San Francisco Chronicle 1990: Bay Area gossipmeister Herb Caen said, “I guess the best
gossip item of all time was the one about Adam and Eve and the apple, and it’s been
downhill every since. Maybe before I pass on, I’ll come up with something half as good.
[…]”
PBS_Newshour 1996: Sullivan was just the latest in a string of dieting cheerleaders.
We’ve had bouncy exercising, eating gurus like Richard Simmons and, of course, our
chief dietmeister these days, Oprah Winfrey, whose gains and losses we’ve tracked as if
they were our very own.

As with the forms in ‑ster, these derivations are most often found in an informal register,
sometimes conveying pure approval, but also sometimes slightly derisive or self-
deprecating.

11.3.1.6 Agent, experiencer, instrument: others

In addition to the suffixes that typically derive agents and instruments, we must also mention
other means of derivation that only occasionally give rise to agent nouns. Among these are
the rare cases of agent-nouns formed by suffixes like ‑ation or by conversion, both of which
more typically are deployed in event/state/result derivations (see Chapter 10):
 11.3 Semantic considerations 237

administration, cheat, congregation, consort, cook, delegation, federation, flirt, following, help,
prosecution, sneak, snoop, stray, sweep, tease. We also occasionally find agentive nouns derived
with the suffix ‑ee, which generally derives patient or goal nouns. We will return to the pri‑
mary use of ‑ee in the next section, but before we do, we briefly look at cases in which ‑ee has
an agentive flavour. COCA and BNC yield the examples in (46).

(46) a. attendee, bargee, departee, devotee, enlistee, enrollee, escapee, fallee, retiree, returnee,
standee
b. Houston Chronicle 1995: Center Ray Donaldson—a capable fill-in for free-agent
departee Mark Stepnoski—went down with a broken leg, leaving the job to an
overweight, out-of-shape Derek Kennard.
Outdoor Life 2007: For the observer, it may seem as if the falling person arrives at his
landing spot in a mere fraction of a second. For the fallee, however, there is a great
deal of time to think.
Christian Science Monitor 2009: “[. . .] There is a recently arrived group of excellent
scholars of international caliber and international standards who returned to China
because they saw its promise, and who are greatly improving academic standards,”
he wrote. # One such returnee is Rao Yi, who taught at Northwestern University in
Chicago before becoming dean of the College of Life Sciences at Peking University
just as Stearns began teaching there.
The Hudson Review 2005: Ann never left a choice place at the rail during intermission
from a conviction that another standee would take possession by the time she
returned.

While occasionally agentive forms in ‑ee are strongly agentive (bargee, for example, refers to
someone who works on a barge, devotee to someone who is devoted to someone else), forms
like departee, returnee, fallee, and standee convey a special nuance of weak agentivity, as
opposed to the perfectly conceivable, and actually attested departer, faller, returner, and
stander (indeed, all but departer are attested in COCA with far greater frequency than the ‑ee
forms; departer is also attested in the OED). That is, the ‑ee forms convey a nuance of non-
volitionality—a departee, fallee, returnee, or standee is not necessarily performing those actions
willingly, so the use of the normally patient-forming suffix ‑ee rather than ‑er serves to convey
that the doer of the action is less than fully agentive (e.g. Barker 1998).

11.3.1.7 Patient, theme, goal

The main affix that derives patient, theme, and goal nouns is ‑ee, although many of the nomi‑
nalization processes discussed in Chapter 10 also secondarily produce patient or theme
nouns, for reasons that will become apparent shortly. In the most thorough analysis of the
semantics of this suffix to date, Barker (1998) shows that ‑ee produces patient and goal nouns
that are sentient but lacking in volitionality, typically animate nouns, and more often than
not nouns which refer to humans. For transitive verbs whose object is potentially animate,
238 Derived nouns: personal and participant

affixation results in patient nouns. For ditransitives in which the object is generally inanimate,
the ‑ee form designates the goal:

(47) a. on transitives:
Houston Chronicle 1995: Blame it on child abuse from a cruel father. And, trust again,
even long before this misbegotten mystery flight lands in Bangor, you’ll figure for
yourself who was the abusee.
Today’s Parent 2000: Whether your child is the snubber or the “snubbee,” your
discreet guidance now is going to go a long way toward laying the groundwork for
the turbulent teens.
b. on ditransitives
ABC Nightline 1994: I have definitely been a ‘leakee’ on more occasion than one.
I won’t ask you whether you have been a ‘leaker’, but in principle, what is the
advantage to an attorney of getting things out in the media before it shows up in
court?
Boston College Environmental Affairs Law Review 1997: To be an intended beneficiary
of the agreement, a person would have to meet two requirements: she would have
to show that 1) “recognition of a right to performance in the beneficiary is
appropriate to effectuate the intention of the parties”; and 2) “the circumstances
indicate that the promisee intends to give the beneficiary the benefit of the
promised performance.”

Other derivational processes tend not to give rise to animate patient/theme terms; we find
examples like catch (as in he’s a real catch), or suspect, but few others. On the other hand, the
suffix ‑ee only rarely forms inanimate patient nouns:

(48) inanimate patient

Horticulture 1991: When a host shrub is not furnished with branches to the ground, or
when the chosen tree branch is out of easy reach, the gap between the climber and the
climbee must be artificially but unobtrusively bridged.

In the passage above, the word climbee apparently refers to a tree. Other inanimates in ‑ee
appear in specialized genres, such as linguistics or computer science, where forms such as
raisee and controlee have been coined for grammatical constituents that undergo processes
called ‘raising’ or ‘control’.
As was the case with denominal forms in ‑er, denominal forms in ‑ee can carry thematic
readings, even in the absence of an explicitly eventive or verbal base. In forms like biographee
or mastectomee the nature of the relevant event can be inferred from the nominal base itself,
with the ‑ee word denoting the patient/theme of that implied event. So a biographee is some‑
one about whom a biography is written, and a mastectomee someone who has undergone a
mastectomy.
 11.3 Semantic considerations 239

There are also several deverbal cases where an animate or sentient patient is not a direct
argument of the verb, but instead must be inferred from the nature of the event denoted by
the verb. Barker gives the example of amputee, where the theme argument of the verbal base
amputate is a limb, but the ‑ee form denotes the person whose limb has been removed. To this
we can add a form like snoree. Indeed, the verb snore is intransitive, and does not have a gram‑
matical patient argument at all, but the word snoree can nevertheless be created to denote
someone who suffers from the noise of the snorer:

(49) Redbook 1999: Buy Breathe Right nasal strips for the snorer and earplugs for the snoree,
or just reach over and pinch his nostrils shut when the rumbles begin.

While English has a specific affix that derives animate patient/theme nouns, it is quite strik‑
ing that it lacks specific affixal means to derive inanimate patient or theme nouns. What we
find, instead, is that a variety of affixes or derivational processes whose primary uses are else‑
where are deployed to produce such nouns. As noted in Chapter 10, we find forms in ‑ation,
‑ment, ‑al, ‑ing, etc. as well as forms derived by conversion that receive an inanimate patient or
theme interpretation, among them those in (50).

(50) acquisition, advance (‘money advanced’), allocation, allowance, appendage,

appropriation (‘money appropriated’), assignment, belief, bequest, buy, choice, closure,
convention, cull, discovery, donation, exhibit, find, gain, grant, hire, hoard, imprint,
inclusion, inheritance, investment, jam, kill, knowledge, learning, marvel, mount
(‘horse’), nosh, offer, payment, pick, preference, projection, purchase, reading,
repetition, submission, substitution, teaching(s), tip, want, washing, waste, yield

We also find the affix ‑er extended to form patient/theme nouns. As pointed out by Rappaport
Hovav and Levin (1992), this is not unusual on the assumption mentioned in the previous sec‑
tion that ‑er typically derives subject-referencing nouns, rather than specifically agent nouns. In
other words, verbs whose subjects are usually patients or themes would actually be predicted to
have patient forms in ‑er; examples would be the unaccusatives fall and sink, which give rise to
the patient nouns faller and sinker, two forms that exist alongside the corresponding ‑ee forma‑
tions. Similarly, we would expect ‑er to form patient nouns from verbs that can occur in a mid‑
dle construction, where the subject also bears the theme or patient role; examples would be a
verb like milk (this cow milks well ◆ a good milker) or look (she looks lovely ~ a real looker).
Not all cases of ‑er patient nouns can be accommodated by this analysis, however. We find
cases like loaner, which designates not the person lending, but the thing (a car or a piece of
equipment) that has been loaned; here we have a verb that does not take a patient/theme
subject and is not used in the middle construction, so the appearance of ‑er on the inanimate
patient noun can only be attributed to semantic extension. Similarly with the use of shooter
like the one in (51), where shooter clearly refers to the animal shot at, rather than the person
doing the shooting.
240 Derived nouns: personal and participant

(51) Outdoor Life 2005: I had taken bears before and had been hunting for several years for a
truly outstanding bear, and here one was standing broadside at 20 yards. I didn’t have to
think twice about this bear. It was a shooter.
Rolling Stone 2006: Black Thought sounds fierce on this banger about the world going
to hell in a handbasket

The word banger in the passage quoted above apparently refers to a song that is banged out,
rather than the person singing. Ryder (1999: 276) gives several other examples that have a
patient reading derived from verbs that do not take patient/theme subjects, among them
scratcher (‘a lottery ticket that is scratched to reveal the potentially winning patterns’) and
keeper (‘something that should be or will be kept’). To this we can add the word reader which
can refer not only to one who reads, but to a particular sort of thing that is read (a book of
stories or essays that is used for pedagogical purposes).

11.3.1.8 Location, measure, and means

English also lacks productive processes which systematically derive location, measure, or
means nouns, and therefore we also see extensions of various other derivational processes to
these domains. Location nouns are occasionally derived with suffixes like ‑er, ‑ery, ‑age, as
well as with the affixes and derivational processes discussed in Chapter 10.

(52) a. Location nouns in ‑er: diner, smoker, crapper (etymologically not an ‑er word to begin
with, but certainly perceived by native speakers as one now)
b. Location nouns in ‑ery: bakery, beanery, bootery, brewery, cannery, eatery, fishery,
grocery, hatchery, nunnery, nursery, refinery, rockery, swannery, tannery
c. Location nouns in ‑age: hermitage, moorage, orphanage, parsonage, vicarage
d. Location nouns in ‑ation, ‑ment, ‑al, etc.: camp, dump, dwelling, enclosure, entrance,
entry, establishment, exhibition, exit, haunt, lodge, mooring, perch, reservation,
residence, rise, roost, seat, store, surroundings, wallow

It is difficult to tell the extent to which these affixes are used productively in contemporary
English to form new location nouns.
An interesting development is the occasional borrowing or extension of the Spanish end‑
ing ‑ería which forms place nouns as one of its functions (carnicería ‘butcher shop’, papelería
‘stationery store’) to form English place-nouns. The word cafeteria has been attested at least
since 1839, according to the OED, which analyses the suffix as ‑teria. Some novel forms in
COCA and BNC still follow this pattern as examples like washateria, Caviarteria, Danceteria,
suggest. But it also appears that some recent creations are formed directly on analogy to the
Spanish affix: Condomeria, fruiteria, groceria.
Measure nouns are more often than not derived with the suffix ‑er, although, as we have
seen in Chapter 10, there are a few forms from ‑ation, ‑ance, and conversion (abundance,
 11.3 Semantic considerations 241

deceleration, decrease, pinch, weight). Sometimes it is the number itself that is affixed, as we
find in the example in (53).

(53) I remember, I remember 1990: Should the strings become entwined then the first
protagonist to shout “Strings!” had strike and so the conker bashing went on until one
split. The winner’s conker was then a oner, when it had disposed of two it became a
twoer, but if it should then beat a threeer it then became a fiver.

More frequently we find a number plus measure word affixed with ‑er, as in a four pounder, a
thirty-five footer, a twelve-reeler, and so on.
There is a fine line between a noun that expresses means and one that denotes an instru‑
ment. The distinction can be illustrated by comparing the words beeper and stroller. Whereas
a beeper is something that beeps, a stroller is not something that strolls, but something by
means of which one strolls (with a baby or toddler). There are a few nouns in ‑er that have
this sort of interpretation (stroller, walker, viewer), and it is no surprise that where such nouns
occur, the suffix ‑er is extended to them.

11.3.2 Athematic domains

Here we discuss derivational affixes that form nouns denoting inhabitants, followers of a
person, names of languages, biological, chemical, geological terms, feminine gendered terms,
or kinship terms. In other words, we cover here derivations that denote neither events nor
participants in events, but are also not purely abstract (the latter are covered in Chapter 12).
What we find in these domains is again a cluster of affixes with overlapping uses.
Nouns denoting inhabitants of a place may be formed with ‑er, ‑an, ‑ite, ‑ese, ‑i, and ‑ish.
Forms in ‑er and ‑ite refer exclusively to inhabitants, not languages. In contrast, the suffix ‑an
refers to language as well as inhabitant, wherever there is a language corresponding to the
place in question (for example, Italian or German), although in many cases there is no cor‑
responding language name (e.g. Salvadorean, Nigerian). The suffix ‑ese refers to both inhabit‑
ant and language as well, but it has also been extended beyond bases that denote place names
to other kinds of proper and common nouns, where it denotes kinds of language or ways of
speaking, often meant pejoratively: Americanese, cat-ese, Clintonese, computerese, funeralese,
Greenspanese, Swinburnese. We also find a few corresponding inhabitant and language names
in ‑i and ‑ish, but as mentioned in Section 11.2.3, these are not productive. In a few cases, lan‑
guage names do not have inhabitant names of the same form, for example, Icelandic versus
Icelander, Finnish versus Finn.
The suffixes ‑ist, ‑an, ‑arian, and ‑ite all designate names for followers of people or adher‑
ents of doctrines or cultural practices. They are often interchangeable:

(54) a. ‑ist: Buddhist, immersionist, inspirationalist, jihadist, Leninist, loyalist, Marxist,

Peronist, Trotskyist
242 Derived nouns: personal and participant

b. ‑an: Chomskian, Freudian, Friedmanian, Keynesian, Marxian, republican, utopian,

unitarian
c. ‑arian: conservatarian, fruitarian, realitarian
d. ‑ite: Gorbachevite, Friedmanite, Paulite, transvestite, Trotskyite, urbanite

Although individuals may on occasion try to assign different readings to the various affixes, it
is unclear whether there is a systematic distinction to be made. Cases are easily found where
there is no apparent difference in meaning, as in (55).

(55) Foreign Affairs 2009: During the nineteenth and twentieth centuries, a Marxist theory
of modernization proclaimed that the abolition of private property would put an end to
exploitation, inequality, and conflict.
Studies in Latin American Popular Culture 1996: A second reason for the crisis in the
theory of popular culture was its historic dependence on Marxian theory, even if that
theory was Gramscian and therefore suspect in much of the Marxist world.

Although Marxist may be used more frequently for a follower of Marx the person and Marxian
for a proponent of Marx’s doctrine, this is not necessarily the case, and therefore the distinc‑
tion is not to be attributed to the semantics of the suffixes.
In this cohort of suffixes, only ‑ite extends to scientific terminology, often forming terms
for chemical substances, geological formations, fossil forms, and the like.

(56) a. ‑ite forming chemical terms: hydrosulfite, nitrite, phosphite, phosphoramidite,

polysacrulite
b. ‑ite forming geological terms: meteorite, pegmatite, perlite, stalactite, stalagmite
c. ‑ite forming biological or paleontological terms: circumsporozoite, dendrite, merozoite,
trilobite

We turn finally to the suffixes which create feminine gendered terms in contemporary English.
As mentioned in Section 11.2.4, the suffixes ‑ess and ‑ette display modest productivity, a fact
which should be of some interest in the post-feminist world of the twenty-first century.
However, perhaps precisely because of strides in gender equality, such terms have now some‑
times come to have an evaluative nuance that they previously lacked. That is, where new
forms arise, they may sometimes be used either jocularly or pejoratively to call attention to
gender where their bases would typically not be taken as gendered (that is, to refer exclusively
to men).

(57) a. ‑ess
Paris Review 1995: Certainly, in an attaching-a-philosopher-to-a-radiator situation,
the handcuffs win; they are the supreme rhetorical device for attaining juxtaposition.
 11.3 Semantic considerations 243

If you must do this, I do endorse an attractive young woman as a kidnapperess,

though preferably one that doesn’t want you to write a book.
The Doors 1992: In the front of these three monoliths are about twenty beer bottles,
numerous Jack Daniels bottles emptied and a lesbian BARTENDRESS pouring up
a breakfast shot of bloody Marys . . .
b. ‑ette
Time 1990: “Some of the girls say I’m a nerd,” said smiling fifth-grader LaDonna
Wright, proudly wearing her jumper and blouse, “but I say, ‘I’m a radical dudette.’”
Ind_Limbaugh 1992: It doesn’t matter! We’re doing it the way the professorette in
Philadelphia wants and you need to learn affirmative action.
Sleepless in Seattle 1993: This kid calls up and says my dad needs a wife and I’m talking
to myself in the car saying, this is completely disgusting, you’re taking advantage of
a child, and then the father gets on and this shrinkette says, do you want to talk
about it?

The forms in ‑ess and ‑ette above range in connotation from self-deprecating and humorous
to downright nasty. These suffixes are rarely used in an evaluatively neutral fashion in con‑
temporary English. The suffix ‑ette of course also figures as a diminutive, where it is not always
pejorative. We return to this use in Chapter 18.
Let us turn to the prefixes. The prefixes great‑, grand‑, and step‑ all indicate specific types of
family relationships (and semantic extensions of this), which clearly differentiates them
semantically from the homophonous adjectives great and grand, and from the noun step. The
prefix grand‑ indicates a degree of ancestral or descendant relationship of two generations, as
in grand-child (‘the child of the child’) or grand-parents (‘the parents of parents or of one
parent’). More forms are given in (58), from COCA.

(58) granddad, grandmummy, grammy, grandfather, grand-poppa, grand-uberbabies, grand-

sire

The semantics of this prefix requires the selection of kinship terms as bases, but one can find
occasional examples that violate this restriction. However, these cases could be accounted for
by coercion or metonymy, as in the example given in (59).

(59) Commentary 2009: This open-ended definition of refugees applies for generations to
come. It bestows housing, utilities, health care, education, cash allowances, emergency
cash, credit, public works, and social services from cradle to grave, with many cradles
and grand-cradles along the way, to its beneficiaries.

The prefix great‑ indicates one degree of ancestral or descendant relationship more than
already indicated by its base. The prefix attaches to two kinds of bases, as shown in the COCA
244 Derived nouns: personal and participant

examples in (60a) through (60c). The first kind of base denotes an ancestral or descendant
relationship of two generations and thus usually involves bases that already have the prefix
grand‑ (60a), but words like ancestors seem to be able to be coerced into this meaning as well
(60b). The second kind of base is kinship terms that straddle only one generation but involve
a sibling on the next generation, as in (60c). The meaning ‘one generation in addition’ is,
however, constant across all derivatives, such that a great-grandfather is the father of a grand‑
father, and a great-aunt is the aunt of a parent. Kinship terms that straddle only one genera‑
tion with no sibling relationship involved, such as father or child are not eligible as bases
(*great-father, *great-child) and take grand‑ instead. Given its semantics of adding one genera‑
tion, it is not surprising to find derivatives with recursive use of great‑, as shown in (60d).

(60) a. great-grandfather, great-grandma, great-grandkid, greatgranny

b. Mother Jones 2001: And you could see in all these graveyard sites, all the bodies have
turned to dust. Our great-ancestors’ dust is right here.
c. great-aunt, great-auntie, great-nephew great-niece, great-uncle
d. great-greatgranddaddy, great-great-nephew, great-great-grandparents, great-great-
grandson, great-great-great-uncle

In present-day usage, formations with the prefix step‑ denote for the most part relatives
obtained through a later marriage. The prefix normally attaches to kinship terms (e.g. step-
dad, stepfamily, stepmother, stepniece), but it is sometimes extended to nouns denoting ani‑
mate beings or groups, or even inanimate things, as in stepboss, step-dog, stepfriend, and stepcar.
In such cases the reading is often extended to mean ‘X as taken over from an earlier relation‑
ship’, or ‘X as taken over from someone else’.

(61) Congrats on the apartment and your new stepdog! (http://petoftheday.com/talk/

archive/index.php/t-82025.html, 21 June 2011)
stepboss: Your spouse’s supervisor (http://www.urbandictionary.com/define.php?
term=stepboss, 21 June 2011)
stepcar: A vehicle owned by your spouse before you were married (http://www.
urbandictionary.com/define.php?term=stepcar)

The prefix vice‑ derives person nouns denoting someone who acts regularly in place of the
person denoted by the base. Bases generally refer to titles of offices, as in vice-governor, vice‑
mayor, vice-prefect, vice-president, vice-speaker.
 chapter 12

Derived nouns: quality,

collective, and other abstracts

12.1 Prospectus
In this chapter we look at affixes whose primary purpose is to derive nouns other than event/
state/result nouns or personal/participant nouns. Among the affixes we have in mind here
are those that create abstract and collective nouns of various sorts, including ‑ness, ‑ity, ‑dom,
‑ship, ‑hood, ‑ery, ‑ia, ‑y, ‑ana, and ‑age. We will also look at the suffix ‑ism, which has a more
specific meaning than the purely abstract or collective-forming affixes, but which neverthe‑
less forms nouns that denote neither events nor participants in events. Again, we begin with
formal considerations, and then move on to semantic considerations.

12.2 Formal considerations

The affixes listed above fall into four natural cohorts, which we will take up in turn. The suf‑
fixes ‑ness and ‑ity fall together as affixes that prefer (but are not confined to) adjectival bases.
The remaining suffixes, apart from ‑y, prefer (but again are often not confined to) nominal
bases. We will divide the latter group into those that typically form abstract nouns (‑dom,
‑ship, ‑hood), those that typically form collectives (‑ery, ‑age, ‑ana), and those that typically
denote fields of study or forms of doctrine (‑ism). The suffix ‑y will be treated separately.

12.2.1 The suffixes ‑ness and ‑ity

The vast majority of nouns in ‑ness and ‑ity are formed on adjectival bases. The suffix ‑ness
takes as its base both simple and derived native and non-native adjectives, and even adjectival
compounds, as the COCA examples in (1)–(3) illustrate:

(1) ‑ness on native adjectives: afraidness, aliveness, awakeness, badness, belovedness, bigness,
childishness, cleanness, closedness, coolness, deadness, deepness, fatherliness,
gloominess, largeness, lewdness, newness, wakefulness
246 Derived nouns: quality, collective, and other abstracts

(2) ‑ness on non-native adjectives: abjectness, abrasiveness, benignness, bizarreness,

ceremoniousness, combativeness, felicitousness, festiveness, globalness, inhibitedness,
massiveness, obtuseness, perfectness, ubiquitousness

(3) ‑ness on adjectival compounds: airheadedness, airtightness, bloodymindedness, bond-

indebtedness, boredom-proneness, choiceworthiness, church-relatedness, earth-
embeddedness, five-sidedness, practical-mindedness, running-friendliness, thong-readiness

This suffix is not confined to adjectival bases, however, as it is also found not infrequently on
nouns and nominal compounds, on phrases of various types, and on other categories as well,
as the following examples from COCA show (see also Williams 1965).

(4) ‑ness on nouns and nominal compounds: ageness, airness, appleness, babeness,

baseballness, birdness, celeb-ness, classness, cityness, cousin-ness, event-ness, factness,
fadness, goatness, nation-ness, Ohioness, couch-potatoness, cross-borderness, holy-
warness, homebodyness, hot-button-ness

(5) ‑ness on phrases: at-homeness, day-to-dayness, don’t-know-nothingness, down-to-

earthness, every-girlness, getting-on-ness, never-give-upness, not-quite-myselfness,
out-of-bodyness, in-chargeness, I-can-do-it-too-ness, take-it-for-grantedness, to-be-
looked-at-ness, you-are-thereness

(6) ‑ness on other categories: aboveness, aboutness, afterwardness, alwaysness, beforeness,

beingness, comingness, itness, there-ness

Indeed we would be hard-pressed to find any category with the exception of verbs and bound
bases to which ‑ness could not attach; ‑ness seems in effect to serve as a sort of default way of
forming abstract nouns from non-verbal categories in contemporary English.
The suffix ‑ity is also apparently productive, but more restricted in its application. It occurs
frequently on bound bases and adjectives of non-native origin, and as pointed out by
Marchand (1969), among others, it is frequent on bases that end in adjective-forming affixes
like ‑al, ‑ic, ‑ive, ‑ous, and especially ‑able.

(7) underived bases: aridity, density, gentility, liquidity, nullity, profanity, solidity, virility,
a. 
chastity, crudity
bound bases: amenity, celebrity, congruity, dexterity, disparity, mendacity, sagacity,
b. 
temerity

(8) a. on ‑able: addressability, arousability, catchability, changeability, damageability,

fishability, flexibility, impeccability, maintainability, merchantability, perceptibility,
runability, schmoozability, understandability, vulnerability
on ‑ive: aggressivity, ascriptivity, attractivity, captivity, collectivity, commutativity,
b. 
competitivity, conservativity, expansivity, inclusivity, massivity, predictivity,
reclusivity, successivity, transitivity
 12.2 Formal considerations 247

on ‑ic: analyticity, apostolicity, authenticity, canonicity, causticity, crypticity,

c. 
domesticity, endemicity, episodicity, formulaicity, hermeticity, messianicity,
rhetoricity, symbolicity, vorticity
d. on ‑al: cardinality, causality, centrality, centrifugality, comicality, communality,
conjugality, corpuscularity, figurality, filiality, historicality, ocularity, sentimentality,
structurality, verbality
on ‑ous: anfractuosity, curiousity, fabulosity, generosity, hideosity, homozygosity,
e. 
impetuosity, luminosity, miraculosity, preciosity, pomposity, religiosity, scrupulosity,
tuberosity, virtuosity

As Marchand and others have pointed out, ‑able attaches to native as well as to non-native
bases, and those ‑able derivatives that are formed on native bases can in turn be nominalized
with ‑ity just as any other ‑able adjective can.
The suffix ‑ity appears only infrequently on nominal bases. We find the following examples
in COCA.

(9) ‑ity on nouns: imbecility, moronity, nerdity, rascality, scholarity, spherity

Oddly, we also find one novel ‑ity noun in COCA formed on the Greek prefix/combining
form pseudo‑. This could also have been coined on the basis of BrE pseud, but this is unlikely
for the form in (10).

Style: 1996: Velickovic’s pseudo-diary thus triggers off yet another intertextual
(10) 
dialogue, one with the growing number of “true” or “documentary” war-stories. Its
“pseudity” implies an inborn inability in the museum refugees to discern clear and
unambiguous facts, to opt for a single cultural or ethnic notion or adhere to any
strong-thought ideology.

An indication of the continued vitality of this suffix is the appearance of forms that are item-
unfamiliar and either unrecorded in the OED, or listed there as rare or obsolete. Among the
forms of this sort in COCA are those in (11).

(11) Algerianity, Angolanity, Caribbeanity, cosmopolitanity, creolity, crypticity, expansivity,

miraculosity, nerdity, symbolicity, structurality, slackity

Also suggestive of the productivity of this suffix are forms that take ‑ity plus some part of a
preceding adjective-forming suffix (e.g. ‑osity, ‑ability, ‑ianity, ‑ativity) to derive mock learned
nouns. Some examples from COCA are given in (12).

(12) Americanosity, babe‑osity, craposity, fatassability, heaviosity, hornosity, inspirationosity,

kitschianity, spinosity, stickativity, suavosity, waterosity
248 Derived nouns: quality, collective, and other abstracts

The items in (12) are used in a jocular way, suggesting that ‑ity is often perceived as belonging
to a more formal register of English than ‑ness, and therefore that it carries a nuance of pom‑
posity. But the examples in (11) show that when it is used by itself, new derivations in ‑ity may
still be neutral in tone.
Note that ‑al before ‑ity is pronounced /æl/ (normality, principality), which is not what
happens before some other affixes such as ‑ia.
The suffix ‑ness is not sensitive to the phonological or prosodic structure of its bases, nor
does it affect either the segmental phonology or stress patterns of the bases it attaches to. The
suffix ‑ity, on the other hand shows both segmental and stress effects on its bases. Thus, bases
with the adjectival suffixes ‑able, ‑ous, ‑ic undergo phonological changes (and corresponding
orthographic changes) when suffixed with ‑ity. Derivatives in ‑ity are obligatorily stressed on
the antepenultimate syllable, no matter where the base is stressed (cf. ácid ◆ acídity, círcular ◆
circulárity, límpid ◆ limpídity). The usual process of secondary stress preservation applies (or
vowel reduction and distressing when adjacent to the derivative’s main stress—see Chapter
9 for details: ìnstruméntal ◆ ìnstrumentálity). Vocalic strengthening of schwa syllables in the
base, with corresponding orthographic changes occurs with ‑able and ‑ous (‑able ◆ ‑abílity,
‑ous ◆ ‑ósity). Trisyllabic laxing also applies.

(13) a. with trisyllabic laxing

obscene /əbˈsiːn/ ◆ obscenity /əbˈsenɪti/
divine /dəˈvaɪn/ ◆ divinity /dəˈvɪnɪti/
profane /prəˈfeɪn/ ◆ profanity /prəˈfænɪti/
profound /prəˈfaʊnd/ ◆ profundity /prəˈfʌndɪti/
b. without trisyllabic laxing
obese ◆ obesity

Finally, we find the effects of velar softening (see also Chapter 9) with this suffix: a base-final
/k/ becomes /s/ before ‑ity, so we find pairs like electric /əlektrɪk/ ◆ electricity /eləktrɪsɪti/,
opaque /əʊpeɪk ‖ oʊpeɪk/ ◆ opacity /əʊpæsɪti ‖ oʊpæsɪti/.

12.2.2 The suffixes ‑dom, ‑ship, and ‑hood

These suffixes form various sorts of abstract nouns, and formally they are quite similar. All
prefer nominal bases, and among the nominal bases those that denote persons (and some‑
times animals), although many other types of nominal bases are attested:

(14) ‑dom on nouns

persons and animals: afficionadodom, beardom, crackpotdom, daddydom, geekdom,
a. 
girldom, pariahdom, pheasantdom, squirreldom, stakeholderdom
 12.2 Formal considerations 249

other concrete nouns: autodom, bagdom, bananadom, breakfastdom, diaperdom,

b. 
entreedom, moviedom, muscledom, pre-fabdom, tacodom, twigdom, yachtdom
abstract nouns: cultdom, feardom, hackdom, marveldom, numberdom, slothdom,
c. 
symboldom, thriftdom

(15) ‑ship on nouns

persons: advisorship, beginnership, buddyship, caretakership, cockneyship,
a. 
consulship, fathership, guruship, inventorship, payeeship, queenship, roommateship,
speakership
other concrete nouns: bagship, braidship, carpetship, latticeship, laurelship, seedship,
b. 
shieldship
abstract nouns: camaraderieship, hateship, licenceship, loveship, tenureship, titleship
c. 

(16) ‑hood on nouns

persons and animals: anthood, babehood, buddyhood, daddyhood, geniushood,
a. 
girlhood, gorillahood, insiderhood, poethood, princesshood, wimphood
other concrete nouns: fountainhood, mountainhood, objecthood, planethood,
b. 
potatohood, shithood, thinghood
abstract nouns: articlehood, egohood, eventhood, genderhood, hellhood,
c. 
narrativehood, one-step-behindhood, legendhood, powerhood, spherehood,
storyhood, truthhood, unexistencehood

We also find these suffixes on bases other than nouns, although these forms are relatively infrequent.

(17) ‑dom on non-nominal bases

a. on verbs: boredom, rubdom, scrubdom, weepdom
b. on adjectives: elvendom, hipdom, nicedom, preppydom, singledom

(18) ‑ship on non-nominal bases

a. on verbs: courtship, settleship, viewship
b. on adjectives: hardship, entrepreneurialship

(19) ‑hood on non-nominal bases

a. on verbs: danglehood
b. on adjectives: foulhood, teenagehood, tribalhood, deafhood, falsehood, singlehood

All three of these suffixes can appear on compound and complex bases (crackpotdom, stake‑
holderdom, roommateship, insiderhood, unexistencehood). We have even found one instance of
250 Derived nouns: quality, collective, and other abstracts

a dephrasal formation, with -hood: low fathood. All three can take either native or non-native
bases, and do not appear on bound bases. They neither display allomorphy themselves, nor
induce allomorphy on their bases.
The suffixes ‑dom, ‑ship, and ‑hood all appear to be highly productive in contemporary
English. Their productivity is most apparent on nominal bases, but we do find the occasional
novel forms on verbal or adjectival bases, as well. Indeed, among the forms cited in (14)–(19),
only a few are item-familiar and of high frequency (boredom, courtship, hardship, falsehood);
the vast majority are not recorded in the OED, and are apparently novel forms. We repeat the
apparently novel forms in (20):

(20) a. afficionadodom, autodom, bagdoom, bananadom, breakfastdom, crackpotdom,

cultdom, daddydom, diaperdom, elvendom, entreedom, feardom, hackdom,
hipdom, marveldom, newdom, nicedom, numberdom, pheasantdom, pre-fabdom,
preppydom, rubdom, scrubdom, singledom, slothdom, squirreldom, stakeholderdom,
symboldom, tacodom, thriftdom, twigdom, weepdom
b. advisorship, bagship, beginnership, blendship, braidship, buddyship, camaraderieship,
caretakership, carpetship, entrepreneurialship, hateship, inventorship, latticeship,
licenceship, payeeship, roommateship, seedship, settleship, tenrueship, viewship
c. articlehood, babehood, buddyhood, daddyhood, danglehood, deafhood, eventhood,
fathood, flathood, foulhood, fountainhood, genderhood, geniushood, gorillahood,
insiderhood, legendhood, mountainhood, narrativehood, objecthood, potatohood,
powerhood, princesshood, shithood, spherehood, storyhood, teenagehood,
tribalhood, truthhood, unexistencehood, wimphood

Forms like these are sometimes meant to be playful, facetious, or humorous, but for the most
part they are evaluatively neutral.

12.2.3 The suffixes ‑ery, ‑age, ‑ana, and ‑ia

The suffixes ‑ery and ‑age both form nouns that denote collectives, locations, and nouns
denoting aspects of behaviour. The suffix ‑ana has a more specific meaning, with its deriva‑
tives denoting collections of items associated with the denotation of the base. The suffix ‑ia
also primarily denotes collectives and locations but diverse other meanings can also be found.
We will return to their range of polysemy in Section 12.3. Here we concentrate on their formal
properties, which are similar, although not identical.
The suffixes ‑ery and ‑age attach predominantly to nouns and verbs, although for many
forms derived with these suffixes, it is difficult to tell whether their bases are nominal or ver‑
bal, as they frequently attach to bases that have both nominal and verbal forms related by
conversion (Marchand 1969: 236 makes a similar observation).
 12.2 Formal considerations 251

(21) a. ‑age on nominal bases: antage, baronage, boobage, brokerage, clientage, cordage,
footage, gooseage, herbage, mileage, parsonage, patronage, porterage, porkage,
stumpage, symbolage, teacherage, veinage
b. ‑age on verbal bases: assemblage, cleavage, creepage, dilatage, eatage, fosterage,
hurlage, pilferage, readage, reapage, shrinkage, spoilage, stowage, tillage, windage
c. ‑age on indeterminate (N/V) bases: awardage, blockage, chattage, dosage, reportage,
coverage, cuttage, drippage, leakage, linkage, pumpage, sparkage, wrappage

(22) a. ‑ery on nominal bases: cakery, cheesery, crockery, duncery, geekery, grainery, knavery,
midwifery, museumery, nitwittery, nunnery, prudery, riflery, slavery, speechery,
webbery, wifery
b. ‑ery on verbal bases: begrudgery, cajolery, cookery, dazzlery, distillery, eatery, forgery,
forgettery, gawkery, hatchery, mockery, mopery, refinery, waggery
c. ‑ery on indeterminate (N/V) bases: boozery, cannery, dockery, drudgery, foolery,
bribery, fakery, framery, jokery, puffery, trimmery

Both suffixes appear infrequently on adjectival bases. In COCA we find commonage, roughage,
shortage; bravery, finery, drollery, greenery, slackery, snuggery, and only a few others. The suffix
‑age also occurs on two prepositions (outage, overage), but these seem to be isolated cases.
We do not encounter ‑age on bound bases. With ‑ery, on the other hand, we very occasion‑
ally find bound bases (electrocautery, monastery, lepidoptery, presbytery). One question that
arises is how to analyse forms like those in (23).

(23) a. butchery, butlery, grocery, mummery, sorcery,

b. delivery, discovery, flattery, pilfery

Here it would seem that we must choose between an analysis in which ‑ery is attached to bases
like groc‑ or butch‑, or one in which we have nominal or verbal bases that happen to end with
<er> to which a suffix ‑y has been added. Neither of these options seems appealing. As for the
first, the agentive suffix ‑er typically does not take bound bases, and such an analysis would not
work in any case for the examples in (23b), where the bases are clearly simplex. The second
option seems unlikely on semantic grounds, as these word show the same range of polysemy as
the other ‑ery forms, rather than the semantics of ‑y, which is either an adjective forming suffix
(e.g. meaty, sweaty; see Chapter 14), or a suffix forming diminutives (e.g. doggy; see Chapter 18),
or an abstract-noun forming suffix that also derives event nouns. A third possibility seems to be
more promising, namely that we have a case of haplology where the output forms lose one
syllable if the base ends in /ə(r)/. This is similar to other cases of haplology involving final
/ə(r)/, for example with the suffix ‑ess, as in murderer ◆ murderess, sorcerer ◆ sorceress (e.g. Plag
1998; Raffelsiefen 1999; see Chapter 9 for a more general discussion of haplology).
252 Derived nouns: quality, collective, and other abstracts

The suffix ‑age does not display allomorphy, nor does it induce allomorphy in its bases.
The suffix ‑ery induces the haplology effect just discussed and variation between two forms,
orthographically represented as <ry> and <ery>. In general, if the final letter of the base is
<e>, the suffix surfaces as <ry> (e.g. riflery, slavery, wifery). However, there are also words for
which this generalization does not hold, and the <ry> variant occurs in spite of the lack of a
base-final <e> (e.g. fetishry, mimicry, rivalry, victimry, all from COCA). The conditions for
this variation are not entirely clear, but it seems that the variation reflects the general ten‑
dency of this morphological category towards a dactyl-final output pattern.
Both ‑age and ‑ery accept either native or non-native bases. They prefer simplex bases,
although we do occasionally find ‑ery on compounds or phrases (asskickery, crackpottery,
head-squishery, tightwaddery, wedding-cakery, dogoodery, old-fartery, Pink Sugar Cupcakery)
and very occasionally on already affixed bases (confectionery, witchessery, interlacery are among
the very few we find in COCA).
Both ‑age and ‑ery display a reasonable degree of productivity in contemporary English.
Among the forms attested in COCA, we find quite a few forms that are not listed in the OED,
among them the ones in (24).

(24) a. antage, awardage, bonerage, boobage, brickage, bummage, chattage, derrickage,

dilatage, dupage, fillage, garterage, gooseage, hurlage, nappage, peakage, peonage,
pilferage, pissage, porkage, pupilage, readage, sandage, sleepage, smileage, sparkage,
stemmage, stumpage, womanage
b. 
ass-kickery, auntery, barkery, benchery, boozery, brickery, bunnery, crushery,
dazzlery, dishwashery, dockery, feedery, forgettery, gasbaggery, geekery, head-
squishery, juicery, kookery, meatery, museumery, pimpery, ragtaggery, roastery,
savery, signery, slackery, snittery, tightwaddery, tomcattery, vaguery, wedding-
cakery, white-trashery

The suffix ‑ana seems to be attached mostly to proper nouns (Americana, Africana, Nixoniana,
Shakespeariana, Victoriana), but forms based on common nouns can also be found, for example
cricketana (OED), hooliganiana, railwayana (both BNC), tobacciana). The suffix carries the main
stress of the derivative and thus belongs to the group of auto-stressed suffixes (see Chapter 9).
In most forms we find the extender ‑i‑, but it is unclear under what circumstances it occurs.
The low frequency forms attested in COCA all seem to have the extender, while only a small
minority of forms, all of them listed in the OED, occur without it (e.g. Americana, Africana).
Forms with the extender show variably a secondary stress and the full vowel /əʊ ‖ oʊ/ on the
pre-antepenultimate syllable (e.g. Nìxoniána ~ Nixòniána), especially if the base ends in <on>,
probably in analogy with pertinent ‑an derivatives (e.g. Thompsòniána, Washingtòniána).
We find quite a few low frequency and new formations in corpora, which suggests that the
suffix is productive (Arizoniana, Hollingworthiana, hooliganiana, Houdiniana, Lincolniana,
Nixoniana, railwayana, Thompsoniana, Washingtoniana).
 12.2 Formal considerations 253

The suffix ‑ia denotes two seemingly distant types of concept. On the one hand we find
clear collectives and locatives, on the other hand the suffix is used to create words with a
scientific appeal and many different potential readings, two of which seem to be especially
prominent. One of these readings is shown by the set of words ending in ‑ia that refer to a
disease or condition, the other reading concerns names of flowers and other plants. Apart
from a few established formations such as encyclopedia, fossilia, genitalia, marginalia, memora‑
bilia, suburbia, the collective and locative suffix occurs productively only in derivatives based
on nouns with the final element land, as in Fordlandia (the title of a book about Henry Ford),
gringolandia, Swamplandia (a novel title), New Zealandia, Radiolandia, TacoLandia (the name
of a taco stand). Other productive formations are much rarer, but can be found, and they may
also be based on adjectives (e.g. Americania, Bob-Dylan-Musicia, bizzaria, devotionalia, all
from COCA). Note that ‑al before ‑ia is pronounced /eɪl/ (marginalia, regalia), which does
not happen elsewhere (e.g. before ‑ity).
Names of diseases and conditions are usually based on bound bases (as in anorgasmia,
homophobia, hypochondria, insomnia), but occasionally other formations can be found, for
example disorientia. Floral names are based on proper nouns, as in Cooksonia, Dieffenbachia,
Jeffersonia, Mertensia, Stewartia, Sinningia (and many more). No matter which meaning is
expressed, stress is always found on the penultimate syllable of the derivative, with the usual
side-effects for the prosody of the base.

12.2.4 The suffix ‑ism

This suffix is highly productive, forming nouns referring to doctrines, kinds of speech, and
scientific (or pseudo-scientific or mock scientific) fields of study.
The suffix ‑ism is formally much like ‑ist (see Chapter 11). This abstract noun suffix attaches
to nouns, both proper and common, to adjectives, and to bound bases but only rarely to
other categories such as verbs and adverbs. Among nouns and adjectives it attaches to both
simple and complex forms, if anything taking an even wider range of already-suffixed bases
than ‑ist does.

(25) ‑ism on underived nouns

on proper nouns: Baathism, Bartokism, Bushism, Cezannism, Daltonism, Darwinism,
a. 
Freudism
on common nouns: ageism, alcoholism, tourism, bossism, boyism, careerism, clanism,
b. 
cougarism, dwarfism, faddism, foodism, motherism, salaryism

(26) ‑ism on complex nouns

a. on ‑ion: abnegationism, citationism, compressionism, diffusionism, inflationism
b. on ‑ite: Jacobitism, Mennonitism
254 Derived nouns: quality, collective, and other abstracts

c. on ‑ment: employmentism (http://wikibin.org/articles/employmentism.html)

d. on ‑an: Africanism, authoritarianism, centenarianism, Hegelianism, Heideggerianism
on ‑er: believerism, boosterism, computerism, consumerism, founderism, haterism,
e. 
insiderism, producerism, sleeperism, stakeholderism, strugglerism
f. on ‑ee: absenteeism, presenteeism, refugeeism, returneeism
g. on ‑ster: doomsterism, hipsterism, pranksterism, tricksterism
h. on ‑y: leftyism
i. on ‑ity: securityism

(27) ‑ism on simple adjectives: ableism, absurdism, evilism, goodism, mutism, oralism,
purism, realism, rowdyism, truism

(28) ‑ism on complex adjectives

a. on ‑al: accidentalism, bestialism, biologicalism, compartmentalism, herbalism
b. on ‑ic: academicism, aceticism, crypticism, didacticism, hermeticism, heroicism
c. on ‑ive: abstractivism, cognitivism, collectivism, comparativism, electivism
d. on ‑ate: corporatism, separatism
e. on ‑able: fallibilism, infallibilism, possibilism
f. on ‑ant: decadentism, emergentism, indifferentism, militantism, protestantism
g. on ‑ine: alpinism, Byzantinism, femininism
h. on ‑ish: Britishism, Irishism, Yiddishism
i. on ‑ile: infantilism

(29) ‑ism on bound bases: ameliorism, amorphism, aphorism, baptism, fascism, exterminism,
illuminism, judaism, Olympism, pessimism, scientism, theism

This suffix seems equally at home on native and non-native bases, although non-native pre‑
dominate. It is found only infrequently on verbs (donatism, splittism, distributism) and adverbs
(beyondism, evenism), but quite frequently on compounds, both neoclassical and native, and
on whole (largely lexicalized) phrases.

(30) ‑ism on compounds:

native compounds: bandwagonism, big-shotism, blue-bloodism, copycatism, deaf–
a. 
mutism, foot-ballism, highbrowism, landlordism, meatballism
b. neoclassical compounds: autobiographism, cryovolcanism, diaheliotropism
 12.2 Formal considerations 255

(31) ‑ism on phrases: can-doism, clean-shavenism, come-outerism, do-goodism, don’t-give-

a-damnism, get-even-with-themism, good-governmentism, hotshitism, me-tooism,
not-in-my-backyardism, one-worldism, red-tapism

As with ‑ist, it is not unusual to find ‑ism on multiply affixed bases, as we see in forms like
configurationalism, establishmentarianism, and gigantificationism.
The range of allomorphy we find with ‑ism is similar to that we find with ‑ist. We frequently
find elision of a base final /i/, /əʊ ǁ oʊ/, or /ə/.

elision of /i/: allegorism, conciliarism, exemplarism, Gandhism, peyotism,

(32) a. 
planetarism, therapism, zombism
b. elision of /əʊ ǁ oʊ/: albinism, caravaggism, pianism
c. elision of /ə/: Buddhism, Spinozism

Elision is by no means obligatory, however, especially where the final segment of the base is
stressed, or if elision would result in an inability to identify the base. In such cases, we find
vowel hiatus.

(33) hiatus with base-final /i/ or /iː/: absenteeism, Bennyism, flunkeyism, celebrityism
a. 
hiatus with base-final /əʊ ǁ oʊ/: Averroism, machoism, sopranoism, desperadoism, jingoism
b. 
c. hiatus with base-final /ə/: Bera‑ism, Cassandraism, Pollyannaism, Sinatraism

The phenomenon is by no means systematic, however, as evidenced by the existence of occa‑

sional doublets in COCA: tantrism ~ tantraism, vigilantism ~ vigilanteism, voluntarism ~ volun‑
taryism, yuppism ~ yuppieism. And as with ‑ist, we occasionally find extenders: egotism,
miasmatism, Platonism, rabbinism, conservatism, witticism, comparatism. The suffix ‑ism does
not trigger stress shift in the base, but it does result in velar softening with bases ending in ‑ic:
agnosticism, athleticism, criticism.

12.2.5 The suffix ‑y

The suffix ‑y derives event nominals (burglary, beggary, felony, gluttony) and abstract nouns of
various semantic types. For the event nominalizations see Chapter 10. We will deal here with
the abstract nouns; the reader should note, however that the distinction between the two
categories is not always easy to draw (cf. cruelty, which is paraphrased in the OED as ‘[t]he
quality of being cruel’ and as ‘a cruel deed’).
For a number of reasons, the suffix is somewhat elusive. First, many words of English end
in <y>, but their status as complex words may be doubtful (cf. amnesty, family, industry,
therapy). Second, there is unclear overlap in distribution, meaning and form with other
256 Derived nouns: quality, collective, and other abstracts

suffixes, such as -ery, and ‑ity. Third, the range of meanings is vast, covering practically the
whole range of nominal meanings discussed in this book, from event to abstract to more
specialized meanings. Fourth, for certain types of meaning, it is unclear whether they emerge
in interaction with their base or whether the pertinent reading should be considered to be a
property of the suffix. Finally, due to ‑y’s roots in Latin, Greek, and French very many perti‑
nent forms are either semantically lexicalized and somewhat opaque, or they are based on
bound roots whose meaning independent of the suffix is hard to determine. This generally
makes it hard to discern the semantic contribution of the suffix. Our treatment of this suffix
is therefore quite tentative.
One group among these are abstract nouns derived from adjectives (difficulty, honesty,
modesty), but this usage is largely unproductive. It is also not clear whether, for this type of ‑y
formation, one should also postulate another, separate suffix ‑ty (in cases like certainty, loy‑
alty, entirety, safety), whether the ‑t‑ is an extender, whether ‑ty is an allomorph of ‑y, or
whether both ‑y and ‑ty are allomorphs of ‑ity. In view of forms such as bankruptcy one might
even posit another allomorph, ‑cy. Given the non-productivity of the de-adjectival pattern,
these potential problems will be of no further concern.
There are also a lot of formations on the basis of nouns, such as autarchy, baronetcy, barony,
bastardy, chaplaincy, chieftaincy, idiocy, matriarchy. Some of these have an extender ‑c‑, whose
distribution seems lexically governed. Even more numerous are derivatives involving neo‑
classical formations (e.g. androgyny, endoscopy, isochrony, optometry, philosophy, photography,
polygamy, telepathy), many of them having a bound base.
What is remarkable about the neoclassical forms is the fact that they show stress shift
(where necessary) to ensure a dactylic pattern at the right edge of the word (e.g. phótograph
◆ photógraphy). Other morpho-phonological alternations occurring with ‑y derivatives con‑
cern final consonants, with the regular alternation of ‑ate ◆ ‑acy, as in diplomat ◆ diplomacy,
adequate ◆ adequacy.
Bound bases are also frequent among bases that are not neoclassical, as in family, therapy,
with some sets of forms inviting analyses in terms of base truncation, for example of the base
suffixes ‑ous or ‑ic, as in lunatic ◆ lunacy, anomalous ◆ anomaly.

12.3 Semantic considerations

Here we look at the complex web of polysemy displayed by the suffixes ‑ness, ‑ity, ‑dom, ‑ship,
‑hood, ‑ery, ‑age, ‑ism, and ‑y. Of the suffixes discussed in this chapter, the majority—‑ness, ‑ity,
‑dom, ‑ship, ‑hood, ‑y—form abstract nouns, although we will need to unpack what we mean
by ‘abstract’ here, as the concept is not as straightforward as it may seem. The suffixes ‑ery,
‑age, and ‑y cover a constellation of meanings that include collectives, locations, and modes
of behaviour, but to some extent these can also overlap with the other five suffixes that form
abstracts. Our last suffix ‑ism forms nouns that denote modes of belief or action, forms of
speech, and fields of study.
 12.3 Semantic considerations 257

12.3.1 Abstracts
English has a number of ways of forming abstract nouns depending on the sort of base
preferred by particular suffixes. As mentioned in Section 12.2, ‑ness and ‑ity are most often
found on adjectival bases, and ‑dom, ‑hood, and ‑ship on nominal bases. Interestingly, how‑
ever, since it appears that we also find ‑ness on nouns with some frequency, and at least
occasionally find ‑dom and ‑hood on adjectives, we will explore here whether the semantic
contributions of these affixes are basically the same, with differences following from the
category of the base, or if their contributions are actually somewhat different in nature. We
will begin with the case of ‑ness, ‑ity, and ‑y on adjectives, next look at ‑dom, ‑hood, ‑ship, and
‑y on nouns, and finally look at the cases where we find the sort of cross-over in base type
mentioned above.
The first question to be settled is what ‑ness and ‑ity mean, and whether they mean the
same thing. Broadly speaking, forms in ‑ity and ‑ness denote the abstract quality or state con‑
veyed by the base adjective. So happiness is the abstract quality or state of being happy, purity
the abstract quality or state of being pure. Riddle (1985), however, argues for a subtle differ‑
ence between the two suffixes: according to her, whereas ‑ness attaches to an adjective that
denotes an ‘embodied trait’, ‑ity creates forms denoting abstract or concrete entities. What
she means by the distinction is not entirely clear, but it seems to hinge on the degree of reifi‑
cation of the quality in question: ‑ness forms abstract nouns with a lower degree of reification
than ‑ity. Riddle (1985: 437–8) cites a number of reasonably persuasive minimal pairs to sup‑
port this hypothesis. For example, whereas hyperactivity names a diagnosable condition,
hyperactiveness denotes only a property or set of properties that can be displayed by a particu‑
lar individual at a particular moment. Similarly, ethnicity is a reified state of belonging to a
particular group, whereas ethnicness denotes only a quality or property that describes a par‑
ticular group or location (Riddle 1985: 440).
Persuasive though these specific examples are, not all possible ‑ness and ‑ity pairs seem to
display a clear difference. Purity and pureness, for example, are used interchangeably, as are
exclusivity and exclusiveness, passivity and passiveness, and many forms ending in ‑able-ity and
‑able-ness. Consider the examples from COCA in (34).

(34) a. purity versus pureness

Bicycling 2009: The purity of having climbed Galibier is almost too much to bear—so

raw, so honest that we need to step away, walk off and again convince ourselves that
our jobs are important and our world depends, after all, on paying our phone bills
and washing our cars and answering our e-mails.
New York Times 1998: And when the show is over, even if some of us think it

maybe isn’t so hot that night, there is Harvey clapping and cheering like a fan.
He is a fan, and it is the pureness and passion of his enthusiasm that has defined
his accomplishment.
258 Derived nouns: quality, collective, and other abstracts

b. passivity versus passiveness

Commentary 2009: His father cleared his throat and Dan realized that the subject at

hand was painful for him. Dan’s passivity was no surprise to himself, but seemed to
be hurting his father.
The Untamed One 2006: The woman’s passiveness angered him. Where were her

instincts for survival? Where was her rage that she had been given a life different
from everyone else’s? Why did she offer him her throat when she should be fighting
him to the bitter end?
c. exclusivity versus exclusiveness
 Denver Post 2009: And, for a few years anyway, filmmakers passionate about telling
stories, true ones, held their collective breath and wondered whether they were
going to push outside the exclusivity of the film-festival circuit and the Sundance
or Discovery Channel into the mainstream movie-watching world.
Journal of International Affairs 2006: This would mean acknowledging the limits of

pluralism by accepting the fact that all differences can not be accepted and through
devising criteria to determine what is admissible and what is not. Mostly, it means
sapping the cultural exclusiveness of our schools, offices, clubs, associations and
political parties.

It would seem that either the ‑ness or the ‑ity form could be used in the contexts above with
no apparent difference in meaning.
To the extent that Riddle’s observation seems correct with respect to her own examples,
we attribute it to the greater propensity of forms in ‑ity to be high frequency established
forms and to have lexicalized meanings. Lexicalized forms can denote reified concepts or
concrete objects. And indeed many ‑ity forms have such reified denotations. Indicative of this
reification is the fact that some ‑ity nouns have become count nouns on specific readings,
among them the examples in (35).

(35) 
activity, atrocity, authority, capacity, community, curiosity, deformity, disability,
eccentricity, enormity, facility, falsity, familiarity, festivity, formality, generality,
imbecility, immunity, liability, locality, majority, mediocrity, minority, monstrosity,
nationality, necessity, obscenity, oddity, particularity, peculiarity, personality, possibility,
principality, probability, profanity, profundity, publicity, rarity, regularity, responsibility,
triviality, utility, vulgarity

In addition we find other readings, and often one form can have more than one lexicalized
reading. There are eventive readings (activity, atrocity, festivity), collectives (Christianity,
humanity, laity, majority, minority, nationality, nobility, security), many forms denote concrete
things (curiosity, deformity, extremity, humidity, monstrosity), people (fatality), or places
(locality, principality). A few have specialized to refer to types of speech (imbecility, obscenity,
profanity, vulgarity) or have an otherwise not fully transparent meaning, as in publicity. In all
 12.3 Semantic considerations 259

cases, the bases of these words can be suffixed with ‑ness to form nouns with the non-lexical‑
ized quality or state meaning. We should point out, however, that although ‑ity forms are far
more prone to lexicalization than forms in ‑ness, it nevertheless appears that the vast majority
of ‑ity forms attested in COCA are in fact not idiomatized. Adjectives suffixed with ‑ness, on
the other hand, are only occasionally prone to idiomatization (as in highness or illness), seem‑
ingly always conveying the quality or state reading.
De-adjectival formations in ‑y are quality nouns very similar in sense to derivatives in ‑ness
and ‑ity, as in certainty, difficulty, entirety, honesty, loyalty, modesty, or safety. Many ‑y forma‑
tions based on nouns express abstract meanings that denote various kinds of abstract notions
related to the respective base (e.g. anarchy, allergy, dynasty, epilepsy, lethargy, philosophy).
With pertinent bases, ‑y formations can also denote a kind of behaviour, an action or the
result of action (e.g. burglary, beggary). The reader is referred to Chapter 10 for the latter
formations.
The suffixes ‑y, ‑dom, ‑ship, and ‑hood can also form abstracts many of which can be roughly
characterized as meaning ‘the state or condition of being X’. As we have seen, their bases are
typically nouns, rather than adjectives. So we have forms like geekdom, buddyship, poethood,
or baronetcy that mean ‘the condition or state of being a geek, buddy, poet, baronet’.
The suffixes ‑dom, ‑ship, and ‑hood are native in origin, and as Trips (2009) shows, all three
started out in Old English (or pre-Old English) as free bases that entered frequently into
compounds, gradually becoming grammaticalized to suffixes. The process of grammaticaliza‑
tion has largely bleached them of any original independent meaning, however, and as Lieber
(2010) argues, any subtle differences in meaning between forms with the three suffixes appear
to be attributable to the semantics of their bases rather than to the affixes themselves. Given
specific kinds of bases, we can find extended meanings for all three. So with a base denoting
a kind of ruler, forms in ‑dom can be extended to mean ‘territory governed by X’, as we find
in kingdom or sheikhdom. And given a base that denotes an activity requiring a specialized
skill, we find forms in ‑ship denoting the skill itself (e.g. musicianship, sportsmanship). We will
see that these suffixes can also be extended to form collectives, but we will return to that
aspect of their polysemy in Section 12.3.2.
Nevertheless, in many cases we find doublets or triplets on the same base which appear to
be virtually indistinguishable in meaning, as the examples in (36) from COCA illustrate.

(36) a. base: guru

Newsweek 2003: But despite the lure of full-time gurudom, Christensen says he

finds B-school life too invigorating to give up.
People 1996: Now the doc is approaching guruhood as another chubster, Bill

Clinton, signed on, and Hillary asked him to rework the White House menu.
Atlanta Journal Constitution 1993: That’s Charles. He’ll go out of his way to do

something for a friend, even a stranger. Still, a busy life and impending guruship do
have a downside.
260 Derived nouns: quality, collective, and other abstracts

b. base: student
Atlanta Journal Constitution 1991: I re-evaluate my situation in the fall, rather than

the New Year. It’s that mind frame of perpetual studentdom.
Commentary 2005: On the surface, Hyde Park appeared to be just another Chicago

neighborhood, maybe a little more varied in its architecture—fewer bungalows,
almost no two-flats–but the spirit of the place was conferred by the abundance of
graduate students and the many hangers-on who, long after departing the university,
remained in this enclave in a state of suspended studenthood.
Physical Educator 2003: Studentship can consist of an array of behaviours (Oleson

& Whittaker, 1968). An example might be to tailor their journal entries to what they
feel the teacher educator wanted to hear.

A similar case can be made for ‑y, where synonymous doublets like baronetcy, baronetship,
baronethood can be easily found. We conclude that for the most part these suffixes serve to
create abstract nouns from concrete bases, with little or no semantic distinction among the
four suffixes.
Interestingly, we sometimes find these suffixes attaching to nominal bases that themselves
already denote abstract nouns, and we might wonder why this would happen. Indeed, in
some cases, it appears that the suffix is gratuitous, adding nothing to the base in terms of
meaning.

Fox_HC 2007: Well, when I saw the movie at the Washington premier a couple weeks,
(37) 
there was nothing new there for me because all of the same myths that he is peddling
have been peddled for years by the advocates of single-payer healthcare here in the
United States. So there was nothing really new there. As far as truthhood […], it is
kind of hard to find there in this film.
Christian Science Monitor 1991: Of the 28,000 ejidos, only 3,000 communities have legal

titles. # Even critics long to see the titleship problem resolved.
Prevention 2006: Whether your locale has frigid winters or simmering summers, the

weather needn’t be a one-way ticket to slothdom.

It is unclear whether there is any meaningful difference between truth and truthhood, title and
titleship, or sloth and slothdom in these examples. None of the three examples in (37) appears
to be used in a jocular fashion or to convey mock pomposity—they seem quite neutral in
tone. It is possible that the suffixes enforce a mass or collective reading in these cases.
Having looked at the semantic range of the affixes ‑ness, ‑ity, ‑dom, ‑ship, and ‑hood when
they occur on their preferred category of bases, we can now look at what these suffixes do on
atypical categories of bases, that is, nouns for ‑ness and ‑ity, adjectives for ‑dom, ‑ship, and
‑hood. In doing so, we can get a clue as to the similarities and differences in the actual seman‑
tic contribution of the affixes.
 12.3 Semantic considerations 261

It appears that when ‑ness attaches to nouns, the derived forms are actually rather different
in meaning than the same nominal base with ‑dom, ‑hood, or ‑ship. In its capacity to attach to
nouns or to whole phrases or sentences, ‑ness does not so much denote the state or condition
of being the kind of entity denoted by the noun; rather it highlights or picks out the signifi‑
cant characteristics that make that entity what it is, denoting the abstract quality or state of
those characteristics.

Cosmopolitan 2007: With that goal in mind, Cosmo asked a variety of different
(38) a. 
experts to offer simple tips for making the most of four common physiques. Just ID
your body type, discover the best ways to work your quirks, and then get ready to
radiate some major babeness.
Raritan 1995: Directing a “sorrowful” look towards her potential killer, the bird is, in
b. 
fact, mimicking, playing her own birdness, at least the kind of birdness Audubon
wants for his lifelike, life-sized paintings.
Christian Science Monitor 2005: Darrah challenges stereotypes that portray working
c. 
women as doing all the household chores while husbands lounge on the sofa watching
football games. “We found very few occasions of true couch-potato‑ness,” he says.
Houston Chronicle 1998: This is what it’s all about—this sense of empowerment, of
d. 
I-can-do-it-too‑ness. # “It’s really important to me that these kids, these individuals,
get the opportunity to be all they can be, to go out into the community and be
useful,” she says.

So babeness highlights the qualities of whatever it is that makes one a babe, birdness the essence
of birds, and couch-potato-ness the essential qualities that define a person as a couch potato, and
so on. In effect, ‑ness homes in on quintessential properties of its bases, whether they are adjec‑
tives or nouns. The suffixes ‑dom, ‑ship, and ‑hood, in contrast, have a more holistic denotation;
so babedom or babehood denotes a status rather than a set of qualities. Although there are only a
few cases of ‑ity attached to nouns (imbecility, nerdity see above, Section 12.2.1), we find that this
suffix has the same propensity as ‑ness to highlight qualities that define the nominal base.
In contrast, when we find ‑hood or ‑dom on adjectives, the derived forms seem all but
identical in meaning to forms derived with ‑ness.

Rolling Stone 2000: I’m not trying to profess nicedom to the world, but we’re nice
(39) a. 
guys.
Radical Society 2003: This whole time I let my life revolve around a secret foulhood,
b. 
and liked it and let it make me think I was a special person, when really I was just a
person who let her life revolve around a secret foulhood.

In other words, were we to substitute niceness or foulness for nicedom and foulhood in these
examples it seems unlikely that any meaning would be lost or changed.
262 Derived nouns: quality, collective, and other abstracts

12.3.2 Collectives and location nouns

Our second broad category comprises ‑age, ‑ery, ‑ana, and -ia. The suffixes ‑age and ‑ery are
most frequently used to form collective and location nouns. These suffixes are only par‑
tially overlapping, however, as ‑age also sometimes forms event, state, or result nouns,
whereas ‑ery sometimes forms nouns that designate a kind of behaviour associated with a
collective or group. Some forms with these suffixes are lexicalized with one or another of
the various meanings (e.g. nunnery is a place noun, rather than a collective or behaviour
noun; buffoonery is a behaviour noun, but not a place or collective), others can vary in
reading depending on context, as can be seen with the examples of rockery and fernery
below.

(40) collective or measure

a. ‑age
Atlanta Journal Constitution 2009: Smith, the UGA economist, expects Georgia

farmers to plant 20 percent less acreage in peanuts this year and prices to be down
40 percent.
American Craft 1998: Although Di Mare rejected the fisherman’s occupation, he

reenacts his father’s activities daiLy [sic] in art making—tying, knotting, twisting
elements into cordage, carving small Lures, casting a sheet of paper, working with
the materials of wood, bone, horn, feathers and thread.
Town and Country 2009: The shop is stocked with kitchen and tabletop accessories,

furniture, gifts, wine and foodstuffs, all hand selected—and exuberantly described
in accompanying signage—by the telegenic Emmy Award-winning chef and author
Michael Chiarello.
NPR_ATC: When hurricanes pass fairly close here, you get enough water dumpage

where I’ve seen a three-foot wall of water come down and just take trees and
everything out and throw them all in the water here.
b. ‑ery
Analogue Science Fiction & Fact 2004: Charlie arose, brushed off rusty sand, and

surveyed the broken rockery stretching to the horizon, listening to the wind keen
as Phobos and Deimos gleamed in a navy blue sky.
New England Review 2002: On either side were banks knee-deep in lush fernery. At

the top of the ridge, the road followed cleared pastures—mathers wondered who
farmed them now—to more woods, even thicker.
Smithsonian 1993: In the fairly new, immense underground-entrance space

below I. M. Pei’s celebrated glass Pyramid (p. 53), dedicated shoppers roamed
through boutiques crammed with expensive knickknackery and shawls priced
at $300.
 12.3 Semantic considerations 263

(41) location
a. ‑age
Smithsonian 1990: The architects will be further discouraged to learn that the

addition provides harborage for other pests besides my family: to wit, starlings.
Motor Boating 2005: Not far away, a Carlos ’n Charlies restaurant (the first in the

U.S.) is perched on the shoreline, providing convenient dockage as well as a point
where non-boaters can join in the fun.
Enlightenment for Idiots 2008: The student of hatha yoga should practice in a solitary

place, in a temple or a hermitage, an arrow-shot away from rocks, water, and fire.
The hermitage should have a small door and no windows.
b. ‑ery
USA Today 2009: Loren bagged dried milk and ran a forklift at Lake Norden’s

creamery but moved to Willow Lake to farm so work didn’t conflict with his
daughters activities.
The Crow 1994: MICKEY is the grease-aproned entrepreneur of MAXI DOGS, a

steamy open-front fast foodery.
Country Living 2008: A fern jungle thrives within the moist glass walls of a restored

circa-1870 Scottish fernery. The Victorians embraced ferns for their exoticism, but
also because their reproductive process is strictly behind the scenes. A fernery (a
conservatory filled with ferns) was the ultimate jolly green indulgence.
Houston Chronicle 1996: When using stone in a garden, it’s important to understand

the relationship between rocks and plants—especially if you are going to create a
rockery or a garden in which plants are interspersed among rocks.

(42) behaviour (‑ery only)

Fortune 1992: Got a sneaking suspicion that your own company, or boss, may be

reverting? Think you may see a few bristles erupting on his or her forehead, reminiscent
of Lon Chaney Jr. in the grip of werewolfery as the moon waxes full?
American Spectator 1991: So, while the questions of the appropriate size of government,

the appropriate reach of U.S. foreign policy, and the appropriate rate of change in the
ethnocultural character of the United States all matter, the question that matters most
in the recurring Buchanan controversies is an easier one: How deeply into kookery can
a man who claims to speak for conservatism go before other conservatives are obliged
to repudiate him?
San Francisco Chronicle 1999: Robin Williams is perfectly cast in this story based on a

real doctor who uses clownery to reach out to patients.
 Military History 2009: One only has to read accounts of the Royal Navy’s efforts to
defend its ships from attacking Argentine aircraft during the Falklands War to see how
feeble shipboard gunnery has become.
264 Derived nouns: quality, collective, and other abstracts

(43) event/state/result (‑age only; see Chapter 11 for ‑ery)

Cross Currents 1993: Local communities in the biblical tradition can only achieve their

fullest identity through deep rootage in this historical community of pilgrim people;
otherwise they become isolated, quirky, self-indulgent, and anxious seekers after
continual emotional highs from groovy worship or instant gurus.
Theological Studies 2008: Among the consummationists, for instance, a serious problem

proved to be how to distinguish marriage from concubinage, which was lawful in
Roman jurisprudence and had been tolerated in the church for centuries.
International Journal of African Historical Studies 2005: The bonds between grandparents

and grandchildren within the context of clans persisted even under colonialism,
enhanced by clans’ ongoing resilience as social networks in crises, renewed on a regular
basis by norms of inheritance and fosterage.

In many cases the readings that we find with these suffixes follow from the types of base they
attach to. Where they attach to concrete count nouns, the collective reading is frequently
available (jewelry, signage); similarly, with ‑age, a collective reading is predominant for bases
that denote measures (mileage, kilowattage). There are also other meaning extensions such as
‘payment for’ (postage, stampage, tonnage). The location reading seems productive with bases
denoting food items (creamery, meatery) or types of animals or people (swannery, hermitage).
Behaviour nouns in ‑ery tend to be formed on bases that can denote agents or instruments
(clownery, gunnery). And event/result/state nouns in ‑age are frequently formed on bases
that themselves imply events or activities of some sort (fosterage, rootage).
The constellation of collective, measure, location, and behaviour readings seems to be
quite a natural one. If we take the collective meaning as central, it is a logical extension from
a collective of people or things to the place where those people or things are located, and the
modes of behaviour that they exhibit. Lieber (2004) points out that it is a simple metonymy
to go from a collective to a description of the behaviour of the collective (thus midwifery is
what the collective of midwives do). And others have pointed out that place names are often
extended to mean the group of people living in that place (e.g. Seattle voted Democratic), so
the extension in the opposite direction from the collective to the location noun would be
perfectly natural as well.
We might speculate, however, on why ‑age can derive event/result/state nouns, in addi‑
tion to collective or place nouns, as this does not seem to be a simple logical extension of the
cluster of meanings discussed in the previous paragraph. As we saw in Chapter 10, the major‑
ity of affixes that form event/state/result nouns take verbs as their bases. The suffix ‑age can
also take verbs as its base, as well as nouns or bases that are indeterminate between the two,
and we find it forming event/state/result nouns where alternatives with ‑ation, ‑ment, etc. do
not exist: coinage, coverage, fillage, linkage, among others. As we saw in Chapter 10, of the typi‑
cal affixes that create nominalizations from verbs, only ‑ation displays any degree of produc‑
tivity, and that only on non-native verbs, mostly those formed with the suffixes ‑ize and ‑ify.
 12.3 Semantic considerations 265

We might speculate that ‑age is marginally able to produce event/state/result nominaliza‑

tions on verbs for which none of the other affixal means of creating nominalizations are avail‑
able. Similarly, we have few options for creating event/state/result nouns from other nouns,
and ‑age seems to provide a marginally productive means of doing so. As we saw above, the
affixes ‑dom, ‑hood, and ‑ship, which take nominal bases, make a somewhat different semantic
contribution to their bases: so concubinedom or concubinehood would profile the status of
being a concubine rather than the activity of a concubine.
Interestingly, although ‑dom, ‑hood, and ‑ship do not form eventive nouns, they do some‑
times form nouns denoting groups or collectives. Forms such as brotherhood and membership
are often used as collectives, and quite a few ‑dom forms can have the collective interpretation
as well.

Mother Jones 1990: McNeilly’s favorite example of well-executed peacetime conversion

(44) 
is taking place in St. Paul, Minnesota, where the International Brotherhood of Electrical
Workers has taken the lead in planning for conversion of a Pentagon-dependent Unisys
plant.
American Heritage 1990: The Longmeadow Country Club is the most prestigious, and

its membership is mostly WASP.
MSNBC_Carlson 2006: Surely they may be the envy of mere mortals, but when

celebritydom’s worshipped couples have a nasty falling out, few of us would ever wish
to bask in their limelight.

The extension from an abstract specifically denoting a status to a collective of individuals

holding that status seems a natural one, and often it is unclear in context whether the status
or collective reading is intended.
The suffix ‑ana is more specific in meaning and less polysemous. Pertinent derivatives
denote collections of items that are associated with the denotation of the base. The kinds of
items referred to are often anecdotes, sayings, or quotations (as in Shakesperiana), but may
refer to all kinds of (often memorable) things or activities connected with the base (e.g.
Arizoniana, Hollingworthiana, Houdiniana).
The suffix ‑ia is quite diverse in its potential meanings. The interpretation of many deriva‑
tives is not very clear. For example, the OED gives the meaning of suburbia as ‘A quasi-proper
name for: The suburbs (esp. of London)’. This suggests a collective reading, which we take to
be the core meaning of the suffix in present-day English, as we find it in fossilia, genitalia,
marginalia, memorabilia, or militia. Pertinent forms that lend themselves to a locative inter‑
pretation (often alongside a collective interpretation) can also easily be found, especially
with bases ending in ‑land (Fordlandia, Dalilandia, Disneylandia, gringolandia, sugarlandia).
Established names of countries or regions could also be included in this category (Australia,
Melanesia, Polynesia, Rhodesia). Apart from collectives and locatives, names of diseases and
conditions are quite common (anorgasmia, disorientia, homophobia, hypochondria, insomnia,
266 Derived nouns: quality, collective, and other abstracts

mania), and other areas of application are names of flowers (Cooksonia, Dieffenbachia,
Jeffersonia, Mertensia, Stewartia) and alkaloids (ammonia, strychnia).

12.3.3 System of belief, action, or scientific study

The suffix ‑ism forms abstract nouns, specifically ones that designate systems of belief or
action, but it has also given rise to several sense extensions from its core meaning.
For the most part, nouns formed with ‑ism denote systems of belief or doctrine, but they
also sometimes denote patterns of action associated with a base. So whereas Confucianism is
the system of belief or doctrine associated with Confucians, cannibalism is not, however, a
doctrine or system of belief associated with cannibals, but rather is a pattern of action attrib‑
uted to cannibals: cannibalism is the eating of one’s own kind. Further examples of ‑ism
nouns that exemplify the belief and action readings are given in (45):

system of belief or doctrine: abolitionism, Anglicanism, atheism, Baathism,

(45) a. 
bandwagonism, Chassidism, creationism, egotism, essentialism, wiccanism
pattern of action: absenteeism, tourism, athleticism, barbarianism, boosterism,
b. 
carnivorism, charlatanism, despotism, escapism, hooliganism, mesmerism, nudism

In some cases, a word can be interpreted indifferently either as a system of belief or as a pat‑
tern of action; fanaticism and extremism, for example, could be read either way, depending on
context.
The suffix ‑ism has three extended meanings as well, forming nouns denoting peculiarities
of language, kinds of prejudicial belief, or types of disease.

peculiarities of language: aphorism, archaism, Britishism, Bushism, colloquialism,

(46) a. 
neologism, malapropism, witticism, Yiddishism
kinds of prejudice: ableism, ageism, heightism, looksism, racism, sexism, speciesism,
b. 
weightism
types of disease: alcoholism, astigmatism, autism, dwarfism, mutism, parkinsonism,
c. 
retrognathism

Use of ‑ism is relatively productive in creating words referring to peculiarities of language, as

it can be attached to proper nouns for people or places to denote kinds of language peculiar
to that person or place. An Obamaism, for example, would be a turn of phrase characteristic
of Barack Obama. In forming names for kinds of prejudice, ‑ism is productive only to a small
degree, probably because the range of bases that would be appropriate for this kind of deriva‑
tion is somewhat limited. As for types of disease, ‑ism seems not to be productive in this
domain.
 chapter 13

Derived verbs

13.1 Prospectus
The derivation of verbs is treated not only in this chapter, but also in other chapters through-
out this book. The reason for this somewhat scattered discussion of complex verbs is, first,
that a number of prefixes derive words of various syntactic categories, including verbs. These
prefixes are treated in Chapters 16 through 18, that is, the chapters focusing on the expres-
sion of locatives of time and space, of negation, and of size, attitude, and quantity. The verb-
deriving prefixes are, for example, de‑ (dethrone), dis‑ (disrobe), down‑ (downsize),
fore‑ (forewarn), hyper‑ (hyperventilate), out‑ (outdo), re‑ (retry), up‑ (upgrade), or un‑
(unsaddle), and the reader is referred to the pertinent chapters for discussion. Second,
complex verbs may arise through clipping or blending, and these processes are discussed in
Chapters 18, 19, and 20.
In this chapter we will concentrate on those morphological categories that are not dealt
with in one of the abovementioned chapters. Section 13.2 will explore formal characteristics,
and Section 13.3 the semantic properties of these affixes.

13.2 Formal considerations

In this section we will first discuss the verb-forming prefixes that are not covered in other
chapters. This is followed by a description of verbal suffixation, conversion into verbs and the
treatment of back-formation, and other processes. In the final subsection we discuss the com-
petition among the different verb-deriving processes.

13.2.1 Prefixation
Apart from the prefixes dealt with in the chapters mentioned above, there are three other
prefixes that derive verbs: a‑, be‑, for‑, en‑. They are all unstressed and none of them is of
noteworthy productivity in the present-day language; we will therefore not go into very
much detail. Some examples of verbs derived with these prefixes are given in (1).
268 Derived verbs

(1) a. aggrieve, allay, amend, arise, ascend, aspire, awake

b. bebless, beblood, beglamour, befoul, bejewel, belabor, beleper, bepraise, beseech,
bespeak, bestill, bethank, betrench
c. encash, encircle, encode, encrypt, endistance, enfeeble, enhat, enlarge, enqueue,
entomb, envision

There are not many derivatives in a‑ and the OED lists no new formations since the year 1900.
Further, where we do find existing verbs prefixed with a‑ it is not clear that the examples even
constitute a single morphological category. Verbs that begin with a‑ have different historical
origins, some being native (awake) and others non-native (ascend), and the meanings con-
veyed by the prefix are disparate, ranging from, for example, expressing upward movement
(ascend) to bringing into a particular state (allay).
The prefix be‑ occurs on nouns (beblood), adjectives (befoul), and verbs (bebless) of native
provenance. According to the OED, the prefix be‑ enjoyed its last neologism in 1926 (beglam‑
our). However, COCA gives evidence of a small amount of continued productivity in the
past participle form expressing an ornative meaning; novel formations based on nouns
(bedenimed, bejowled, besmogged, bestuccoed) are not difficult to find; in such forms no dis-
tinction seems to be made between native and non-native bases (see also Chapter 21). It does
not appear, however, that other verb forms are being produced in the contemporary
language.
The prefix en‑ occurs on nominal (enhat), adjectival (enfeeble), and verbal (encapsulate)
bases of both native and non-native origins, as well as on bases whose category is indeter-
minate between noun and verb (encash, enqueue). It has an allomorph with a bilabial nasal
/ɪm/ (spelled <em>) before bases with initial /b/,/p/, or /m/ (embroil, empower,
emmesh), although this allomorphy appears not to be strict (for example, both emmesh
and enmesh are attested in the OED). There are also spelling variants <in> or <im>, which
can sometimes even occur with the same base, as in <enwrap> ~ <inwrap> providing
synonymous outputs in most cases (though note lexicalized non-synonymous ensure
versus insure).
The prefix gives rise to occasional neologisms, with six of them having been recorded in
the OED for the past 100 years (encode, encrypt, endistance, enhat, enqueue, envision). The BNC
adds enmire to this list. COCA has several forms not found in the OED: encapture, enchamber,
enchange, enclod, encloth, enculturate, enculture, enfield, enflood, enfog, enmagick, enmire, enpool,
ensorcer, enspirit, entextualize, entragick, envenomate, envoice, enweb. Notably, many of these
more recent formations are denominal in origin, which may count as an argument that this
prefix is category-changing, similar to de‑ (see Chapter 17), but unlike the majority of prefixes
in English.
Finally, the prefix for‑ occurs in a handful of verbs, mostly inherited from Old English
(forbear with a spelling variant <forebear>, forbid, forbode, forget, forgive, forsake, forswear).
These forms are all lexicalized, and the semantic contribution of the prefix is unclear.
 13.2 Formal considerations 269

13.2.2 Suffixation
13.2.2.1 The suffixes ‑ize and ‑ify
We will treat the suffixes ‑ize and ‑ify together as they are intricately related. Their meanings,
which we will discuss in Section 13.3.2, are very similar, if not identical, and their phonologi-
cal restrictions lead to an almost complementary distribution with regard to the kinds of
bases the two suffixes can attach to.
Bases may be either native or non-native, although non-native predominate.

(2) a. on non-native bases

alphabetize, dynamize, elasticize, opinionize
classify, diversify, humidify, pacify
b. on native bases
awfulize, glitterize, leatherize, womanize
falsify, prettify, webify, youthify

The two suffixes attach to nouns, adjectives, and bound bases, as the examples in (3)–(5)
illustrate.

(3) on nouns
a. alphabetize, Beethovenize, cannibalize, dieselize, hospitalize, idolize, magnetize,
opinionize, paganize, routinize, satirize, sectorize, sloganize, standardize, terrorize,
tokenize, vaporize, victimize, womanize
b. artify, bourgeoisify, citify, classify, codify, countrify, gasify, gentrify, karstify, mythify,
nazify, notify, plastify, tubify, youthify, yuppify, zombify

(4) on adjectives
a. 
elasticize, equalize, eroticize, fertilize, glottalize, laminarize, nativize, passivize,
paternalize, radicalize, randomize, reflexivize, rigidize, virilize
b. aridify, densify, diversify, falsify, humidify, justify, prettify, purify, scarify, simplify, solidify

(5) on bound bases

a. baptize, dynamize, exorcize, erotize, feminize, harmonize, maximize, mathematize,
mechanize, monetize, ostracize, pancreatize, quantize, temporize
b. calcify, certify, crucify, deify, electrify, gratify, horrify, identify, magnify, mortify,
pacify, quantify, ramify, ratify, reify, rectify, specify, technify, terrify

The analysis of the cases in (5a) as involving bound bases may sometimes compete with an
analysis that posits predictable phonological adjustments, as in the case of feminize ◆ feminine
or harmonize ◆ harmony (see below for discussion, and Chapter 9).
270 Derived verbs

Very occasionally, one can even find phrases as bases as shown in (6). Note that the final
element of the phrase is a phonologically legitimitate base for ‑ify (see below).

American Spectator 1998: After all, in today’s Ben-and-Jerrified Vermont, the only
(6) 
Republicans you ever see on bumper stickers are confined to sentiments like “Newt
Happens.”

Forms based on verbs are also rare (schmoozify, blackenize) as are those on compounds (hip-
hopify). Proper nouns are well-attested as bases (Bobbitize, Finlandize, Haussmannize,
Chuckify, Kurdify, Lilliputify).
Quite often, we also find the presence of extenders, as shown in (7), where examples are
taken both from COCA and from the OED. With some of the forms the morphological
analysis involving an extender may be controversial and depends on what is regarded as the
base. For example, mediocritize could be argued to be either derived from the adjective
mediocre, with extender, or from the noun mediocrity, with regular deletion of the final vowel
(see below).

(7) AnnKleinicize, attitudinize, cinematize, dogmatize, fetishicize, lemmatize, mediocritize,

negritize, platitudinize, problematized, rhematize, thematize

Although the overall number of verbs formed with these suffixes might be low in comparison
to that of productive adjectival or, especially, nominal suffixes (e.g. Baayen and Lieber 1991),
the suffixes ‑ize and ‑ify can nevertheless be considered highly productive, as can be seen
from three kinds of evidence. First, there are many recent neologisms in COCA, as well as in
the OED. Plag (1999: 115) found 346 new ‑ize formations and 30 new ‑ify formations for the
twentieth century in the second edition of the OED (with the most recent online version of
the third edition of the OED having even more, e.g. therapize). This productivity is corrobo-
rated by the corpus data, where we find a large proportion of low frequency forms. For exam-
ple, the 20-million-word Cobuild corpus (Sinclair 1987) investigated by Plag (1999: 115)
contains 80 ‑ize hapaxes and 18 ‑ify hapaxes. Data from COCA reveal even stronger produc-
tivity: a check of 200 item-unfamiliar ‑ize words turns up over 100 that are not recorded in the
OED. Similarly, of 77 item-unfamiliar words in ‑ify, 52 are not listed in the OED. The exam-
ples in (8) illustrates neologisms that can be found in COCA and the BNC, as well as in
Wordspy.

(8) a. COCA
‑ize: algebraicize, batteryize, blenderize, clientelize, Enronize, felonize, gadgetize,
glitterize, imagize, lichenize, margarinize, mestizoize
‑ify: blottify, bumify, bumpify, Chuckify, dizzify, extensify, freakify, gourmetify,
hippify, Kurdify, Lilliputify, silkify, startify, utopify
 13.2 Formal considerations 271

b. BNC
‑ize: awfulize, Beethovenize, corpusize, Dylanize, Haydnize, Hawaiianize, Marleyize,
Ohmannize, otherize, productize, Rwandaize, saxophonize, serverize, Tanalize,
taslanize, Tillerize, Wymanize
‑ify: fuzzify, geriatrify, gypsify, faintify, Jessify, Lewisify, moodify, sickify
c. Wordspy
‑ize: dollarize, velocitize
‑ify: gamify, sonify, wikify

Both suffixes are subject to strong phonological restrictions. Neither suffix can justifiably be
characterized as stress-shifting, but there are other prosodic constraints at work. The suffix
‑ize generally selects bases that consist of a trochee (e.g. rándomize) or a dactyl (hóspitalize),
while ‑ify selects monosyllabic bases or bases that consist of an iambic foot (ártify, bourgéoisify).
In output-oriented terms, ‑ify derivatives must have a trochee before the last syllable of the
derivative, while ‑ize words allow both trochees and dactyls (and under certain conditions
even secondarily-stressed syllables, see below).
A systematic overlap of the domains of the two suffixes concerns trochaic bases ending in
a high front vowel, such as dandy, Dolby, Disney, orgy, Nazi, Turkey, wiki. These forms can
take both suffixes, as the base-final vowel and the initial vowel of ‑ify coalesce in these cases,
such that forms like dandify, Nazify, wikify fulfil the output-restrictions on ‑ify derivatives.
COCA gives us Dolbyize, Disneyize, orgyize, and Turkeyize as examples of pertinent ‑ize
derivatives. Where different base forms are available that may satisfy the respective con-
straints, one can find doublets, as with Russianize (with trochaic base) and Russify (with a
monosyllabic base).
But things are much more complicated than that. Let us first look at stress shift and stress
reduction, then at other cases of base allomorphy. Although both suffixes typically satisfy the
prosodic output conditions of their category by selecting the right kind of base instead of
effecting a stress shift upon their bases (cf. *randómify, *ídentize) there are quite a few forms
that appear to go against this generalization. These forms are of at least four kinds.
The first group of violators consists of cases of long established formations which do not
necessarily conform to the present-day restrictions, such as baptize (first attestation: 1297).
Second, there are formations that are most probably based on other complex words
(instead of a simplex base) and preserve the foot structure of the complex base words they
are derived from (e.g. Jàpanése ◆ Jápanìze, subjéctive ◆ subjéctify ‘make subjective’); we might
cite these as examples of analogical or paradigmatic word-formation (see Chapter 23). A sub-
group of this set consists of ‑ify words whose bases are adjectives ending in ‑id (e.g. arídify,
humídify, solídify, valídify). All of these have related ‑ity derivatives with the same main-
stressed syllable (e.g. humídity), and these ‑ity forms are all attested much earlier than the
corresponding ‑ify derivatives (see the respective entries in the OED). Given the uniform
272 Derived verbs

behaviour of this set of words, this creates another domain of overlap between ‑ize and ‑ify, as
we would expect that, without stress shift, these trochaic bases would also take ‑ize. And
indeed, such doublets can be found (e.g. flúidize ◆ fluídify, rígidize ◆ rigídify), but there are also
cases where only one of the two possible formations are attested, for example, líquidize ◆
Nliquídify and Nhúmidize ◆ humídify.
Third, very occasionally there are formations that simply fail to follow the clear trend set
by the vast majority of more recent forms, as the examples from COCA in (9) illustrate:

(9) a. ‑ize: blandize, cartoonize, cementize, discretize, event‑ize, Francize, Iraqize,

McKeachize
b. ‑ify: devil‑ify, hip-hopify, Lewisify (BNC), lichenify, parentify

A few of the forms in (9b) (devilify, Lewisify, parentify), in order to have the normal stress
pattern of ‑ify derivatives, would have to be stressed on the antepenult, inducing a stress shift.
One reason for such aberrant behaviour may be that these forms usually occur in the written
mode and it appears that the prosodic restrictions are not as forceful in the written as in the
spoken language (cf. Raffelsiefen 2004; see Hanssen et al. to appear for empirical evidence
for such an effect).
Finally, in the remaining cases, the English forms are modelled on similar words from
other languages (e.g. stylize coined on German stilisieren, OED s.v. stylize, first attestation
1904).
Overall, however, our discussion of potential stress-shift violators should not obscure the
fact that both the newly derived forms, as collected in (8), and the large set of existing deriva-
tives appear to overwhelmingly observe the prosodic restrictions set out above (see also Plag
1999: ch. 6.2; Raffelsiefen 2004: 111).
Another stress-related phenomenon is stress reduction, with concomitant vowel reduc-
tion. This alternation generally occurs with bases that end in a main-stressed syllable followed
by a secondarily-stressed syllable. Plag (1999: 166) gives the following list of examples.

(10) ánòde ◆ ánodìze /əʊ/ ~ /ə/

pódzòl ◆ pódzolìze /ɒ/ ~ /ə/
sílàne ◆ sílanìze /eɪ/ ~ /ə/
strýchnìne ◆ strýchninìze (/i/ ~ /ɪ/) ~ /ɪ/

Note that such words would not satisfy the input-oriented restriction formulated above
that ‑ize attaches to trochees and dactyls. From an output-oriented perspective, however,
the stress reduction makes these derivatives fit the general rhythmic pattern of ‑ize
derivatives.
 13.2 Formal considerations 273

As observed by Raffelsiefen (2004: 138–9), stress- and vowel-reduction is restricted to the

environment with two adjacent stresses and generally does not occur with bases where an
unstressed syllable intervenes between main stress and secondary stress. Some of Raffelsiefen’s
examples are listed in (11).

(11) áerosòlìze, cátalòguìze, épidòtìze, prótocòlìze, vágabòndìze

We note, however, that láterìte ◆ láteritìze (/aɪ/ ~ /ɪ/) is an exception.

Apart from stress-related effects, derivatives in ‑ize are subject to further phonological
restrictions that manifest themselves in intricate allomorphy patterns. To begin with, we have
to distinguish between dactylic bases that end in a consonant and those that end in a vowel.
Those that end in a consonant show no phonological effects (see (12a)) unless the attach-
ment of ‑ize leads to the repetition of identical phonological constituents. These haplology
effects occur either with bases whose final syllable starts and ends with the same consonant,
as in (12b), or with bases that end in a sibilant (12c). In all these cases, we see the loss of the
base-final rhyme, which can be either interpreted as phonological adjustment or stem selec-
tion (see also Chapter 9).

(12) a. hospital ◆ hospitalize ◆ Nhospitize

federal ◆ federalize ◆ Nfederize
relative ◆ relativize ◆ Nrelatize
b. Nfemininize ◆ feminize
Nminimumize ◆ minimize
Nmetathesisize ◆ metathesize
c. NSocratesize ◆ Socratize
Nsyphilisize ◆ syphilize
Ntetanusize ◆ tetanize

Curiously enough, disyllabic bases are subject to haplology to a lesser extent, as they allow
bases to stay intact even when their final syllable starts and ends with the same consonant, as
in (13a). No forms with disyllabic bases can be found, however, if the two consonants are
both alveolar sibilants (13b).

(13) a. mirrorize, strychninize, terrorize

b. Ncrisisize, Noasisize

The absence of forms such as those in (12c) and (13b) has led to claims that there are lexical
gaps that arise in those cases where it is impossible to observe important conflicting
274 Derived verbs

restrictions (e.g. Raffelsiefen 2004). For instance, the deletion of the final rhyme in the words
listed in (13b), as in Ncrísisìze would lead to a violation of the requirement that ‑ize does not
attach to main-stressed syllables. At an empirical level, however, lexical gaps are hard to
establish firmly, as the number of words with the pertinent context are not numerous and the
evidence is thus somewhat equivocal. An example is Raffelsiefen’s (2004: 122–3) proposed
haplology-induced gap with bases ending in /st/ (e.g. Nearnestize, Nforestize, Nmajestize),
which is called into question by words such as Eucharistize, fascistize, reforestize, travestize, all
listed in the OED, and modestize, attested in COCA.
Let us now turn to dactylic bases that end in a vowel, that is schwa and /i/ (e.g. patina,
summary). They regularly lose the final vowel, as shown in (14a). In some cases one also finds
the presence of extenders, as shown in (14b). Occasional counterexamples can be found,
however (batteryize, Coventryize, demythologyize, all from COCA).

(14) a. epitomize, fantasize, empathize, iotize, jeopardize, memorize, patinize, summarize,

vasectomize
b. cinematize, operatize

With disyllabic vowel-final bases, there are three patterns observable. Base-final schwas can
alternate with a full vowel (15a) or, much more commonly, may take an extender (or a differ-
ent base allomorph, depending on the analysis), see (15b). Other vowels survive intact, no
matter whether the final syllable is unstressed (15c) or carries secondary stress (15d). Longer
words that end in a secondarily-stressed vowel are also possible bases for ‑ize, with no specific
phonological effects observable (15e).

(15) a. Juda ◆ Jud[eɪ]ize, Mithra ◆ Mithr[eɪ]ize

b. asiatize, dogmatize, lemmatize, migmatize, stigmatize, traumatize
c. cockneyize, dandyize, Harveyize, Toryize, trolleyize
d. ghettoize, heroize, jumboize, Maoize, zeroize
e. radioize, scenarioize

To summarize, there are intricate phonological restrictions and mechanisms at work that regulate
both the distribution of the two suffixes and the segmental and rhythmic shape of their derivatives.
As is the case with other competing affixes in English derivation (see the other chapters in Part III
of this volume for details), overlaps of the two structurally defined domains can be found, and in
these domains we may find either one of the two possible forms attested, or indeed doublets.

13.2.2.2 The suffix ‑ate

As has been suggested by several authors in the past (e.g. Marchand 1969: 258; Plank 1981: 214),
‑ate may be a good indicator of verbhood, but its analysis is complex and fraught, as derivatives
 13.2 Formal considerations 275

ending in this string are formally and semantically very diverse. It is often not clear which words
should belong to this morphological category. However, there is a consensus that there is at least
a rather clear pattern of productive denominal derivatives such as hyphenate that exhibit some
regularities. We include in our discussion here all verbs that end in the string ‑ate and that have a
base that is also found in other morphological contexts. Our examples in this section come
largely from COCA and from Plag (1999), which includes a detailed discussion of this suffix.
The clearest cases of verbs in ‑ate are those that have nouns and adjectives as bases, with
adjectival bases being less frequent than nominal ones.

(16) a. ‑ate on nouns: amalgamate, assassinate, caffeinate, decorate, factionate, halogenate,

machinate, nodulate, ozonate, pixelate, urinate
b. ‑ate on adjectives: activate, authenticate, captivate, differentiate, domesticate, estivate,
explicitate, objectivate, passivate, rusticate, serpentinate, sublimate

Bases may be morphologically complex, as for example in activate, differentiate, but there are
no compounds or phrases attested as bases.
In (17) we present some of the very many derivatives with bound roots.

(17) allocate, conjugate, copulate, defecate, eradicate, fraternate, hibernate, irrigate, isolate,
lactate, maximate, migrate, nutate, perforate, rotate, saturate, ventilate

We also find forms in ‑ate that appear to be conversion pairs, either verb ◆ noun or verb
◆ adjective.

(18) a. verb ◆ noun: affricate, aspirate, coagulate, collocate, concentrate, conglomerate,

correlate, duplicate, estimate, geminate, graduate, initiate, laminate, predicate,
syndicate
b. verb ◆ adjective: appropriate, approximate, consummate, deliberate, elaborate,
intimate, moderate, separate

In such cases we find placement of secondary stress on the final syllable of the verb (córrelàte,
dúplicàte, gráduàte, etc.) so that there is no true conversion, the adjectives and nouns having
unstressed final syllables. This type of relationship is also frequently observable in nouns
ending in the suffix ‑ate that are used in chemistry to denote certain substances, as in, for
example, súlphonàteV ◆ súlphonateN (see Chapter 14).
Finally, there are quite a few forms that appear to be back-formations from nouns in ‑ation.
COCA gives us the examples in (19).

(19) administrate, allegate, combinate, commentate, commutate, conversate, degradate,

deportate, deprivate, documentate, excitate, exclamate, explorate, exultate, flotate,
imaginate, limitate, maturate, multiplicate, orientate, pronunciate, recitate
276 Derived verbs

In cases such as these, given the existence of established verbs without the suffix, it may
appear synchronically that ‑ate has attached to a verb (e.g. imagine with ‑ate, pronounce with
‑ate). Forms such as these are sometimes frowned upon by prescriptivists, but as the exam-
ples in (19) suggest, novel forms are not uncommon.
In some cases, due to the prevalence of base truncation (see below) and non-suffixal
mechanisms of deriving verbs in ‑ate, it may not be clear whether a given derivative is noun-
based or adjective-based. For example, federate may have been coined on the basis of federa‑
tion or on the basis of federal (or both).
As with the suffixes ‑ize and ‑ify, it is not always obvious how to differentiate base trunca-
tion from suffixation to bound roots, that is the selection of a particular base allomorph. For
example, fraternate may be analysed as being coined on the basis of the adjective fraternal
under loss of the final rhyme, or directly on the bound base fratern‑, which is found also with
other derivatives (fraternize, fraternism, fraternity). Extenders may also be present in some
forms, as in affectuate, alginate, authoricate, differentiate, pathosticate.
The suffix was extremely productive in the nineteenth century with hundreds of forms
listed in the OED, but, as shown in Plag (1999), it is less productive than ‑ize and ‑ify in con-
temporary English.
Phonologically, ‑ate derivatives are very similar to ‑ize derivatives in that the suffix nor-
mally does not induce a stress shift and prefers bases that end in a trochee. Thus, there seems
to be a clear output restriction to derivatives that end in a trochee followed by the secondar-
ily stressed final syllable (e.g. artículàte, cávitàte, mótivàte). To enforce this prosodic pattern,
more severe phonological adjustments are observable with ‑ate than with ‑ize. Thus, ‑ate is
generally intolerant concerning dactylic bases and induces truncation across the board, that
is with both vowel-final and consonant-final bases. The examples in (20) illustrate this.

(20) a. Vowel-final dactylic bases: cavity ◆ cavitate, mercury ◆ mercurate

b. Consonant-final dactylic bases: alluvium ◆ alluviate ◆ Nalluviumate, nitrosyl
◆ nitrosate ◆ Nnitrosylate, residual ◆ residuate ◆ Nresidualate

One finds, however, many exceptions to the pattern just proposed. Apart from idiosyncratic
formations (see Plag 1999: 216 for discussion), there are two kinds of systematic exception.
First, there is a group of forms whose phonology is a direct reflection of their morphological
relatedness, that is, they are prosodically faithful to their base word. For example, Plag (1999:
217) lists the forms shown in (21) as ‑ate derivatives that show the same stress patterns as the
action nouns in ‑ation from which they are back-derived.

(21) formáte, geláte, lènáte, notáte (BrE), oláte, predáte, soláte, solváte

The second class of exceptions contains derivatives with main stress on the penultimate
­syllable. These are formed on the basis of nouns ending in the nominal suffix ‑ate (which
 13.2 Formal considerations 277

denotes particular kinds of salt). As noted by Plag (1999: 217) the corresponding ‑ate verbs
are derived by conversion, as evidenced by cítràte, hýdròbóràte, phósphàte, and xánthàte.

13.2.3 Conversion
For ease of reference in this section we will use the term ‘converted verb’ for verbs derived by
conversion. For a discussion of the theoretical problems involved with conversion as a mor-
phological process, see Chapter 25.
Conversion into verbs is a highly productive process in English, although it is not as easy
to document productivity for conversion as it is for affixal derivation, since it is difficult to
extract pertinent forms from corpora using automatic or semi-automatic procedures.
Nevertheless, we have made an attempt to do so. Plag (1999) uses dictionary-based quantita-
tive measures, and finds that conversion is the most productive process deriving verbs from
bases of other categories. Our results generally support this conclusion, but it is difficult to
compare the productivity of conversion with that of the suffixes ‑ize and ‑ify, as the methods
used for obtaining data are necessarily quite different.
Here, we describe the procedure we followed. We first extracted the 1000 most frequent
singular common nouns from the 500,000 word COCA frequency-list. We counted all forms
with item-familiar converted verbs (N = 473), and then checked the remaining nouns against
both COCA and the OED for corresponding converted verbs. To make sure that we had only
genuine cases of conversion we searched in COCA for forms in ‑ing or ‑ed that were attested in
clearly verbal contexts (that is, when following auxiliary be, but not when prenominal). This
procedure is quite conservative, and if anything leads us to undercount cases of conversion.
Using this procedure, we found a further 229 converted verbs that were attested either in
the OED or in COCA or both, and that were not item-familiar. There were 24 verbs that were
attested only in COCA that we can safely assume to be neologisms. We list them together
with some of their contexts in (22) to illustrate their usage.

(22) a. animal, army, baseball, bathroom, career ‘to have a career’, category, chest (attested
in OED, but with different meaning), computer, crisis, decision, direction, door,
driver, furniture, governor, manager, prayer, presence, response, song, theme,
therapy, version, youth
CNN_Presents 2006: There’s times that he doesn’t remember, but when I was racing, and
b. 
I was probably more of a no-name, he doored me and knocked me out of the way.
African American Review 2004: You’ll be preparing our meals, bathrooming us,
doing laundry, lifting if needed.
Sports Illustrated 1992: When Holmes decisioned Ray Mercer recently, it took a
trained eye to locate Smith amid the assemblage.
New Yorker 2004: Buck would come chesting in, he would give us all a cobra smile
with real white teeth.
278 Derived verbs

Conversion into verbs allows a great diversity of bases, including simplex, derived, or compound
nouns and adjectives, onomatopoeic expressions, and phrases. This is illustrated in (23) to (28)
with a subset of the 488 twentieth-century neologisms in Plag (1999: Appendix 1) gleaned from
the OED, and additional examples from Wordspy. The data in (23) and (24) clearly demonstrate
that suffixed nouns and adjectives can indeed be bases of conversion into verbs, contrary to claims
by Marchand (1969: 372), Gussmann (1987: 82), and Nagano (2008: 13). Examples in (25) and
(26) show converted verbs based on onomatopoeic expressions and phrases, and (27) contains
examples based on prepositions or adverbs, all of which predate the twentieth century. This last
class of words shows many lexical gaps (e.g. Nabove, Nbelow, Nfrom), which, together with the
lack of more recent neologisms and the fact that many of the attested de-adverbial converted verbs
are obsolete (e.g. about, below, under, see OED), may suggest that it is an unproductive class.

(23) Nouns as bases

a. simplex: angel, apex, archive, ball, caddy, dingo, doll, eel, fountain, garage, gel, goal,
ham, jazz, lagoon, marmalade, matrix, muscle, nanny, napalm, orbit, petal, radio,
vacuum, video, weasel, witch
b. derived: birdie, clipper, interface, junction, larder, option, racketeer, sleeper, zipper
compound: blockhouse, catfoot, eyeball, filmset, finger-post, highlight, keyboard,
c. 
limehouse, newspaper, phagocyte, postcard, pot-hole, time-table
d. proper: Amazon, Dell, MacGyver, Nasdaq
e. acronym: RIF

(24) Adjectives as bases

a. simplex: camp, cruel, dual, hip, main, multiple, phoney, pretty, young
b. derived: filthy, lethal, premature, romantic, skinny
c. compound: polychrome, rustproof

(25) Onomatopoeic expressions as bases

burp, chuff, clink-clank, oink, ooh, plink, poof, pong, pring, quiff, razz, tick-tock,
whomp, wham, whooom, zizz, zonk

(26) Phrases as bases

blind-side, cold-call, cold-cream, hands-up, wrong-foot, wrong-slot

(27) Adverbs or prepositions as bases

backward, down, in, out, up

Finally, other types of expression such as interjections or conjunctions can be converted into
verbs, too, as shown in (28).
 13.2 Formal considerations 279

(28) Other expressions

but, oh, no, oh-oh, yes

It is possible to find conversion to verb from a suffixed noun even where the base noun is not
semantically opaque, contrary to the claim by Marchand (1969: 372–3), Bauer (1983: 226),
and Nagano (2008: 13) on the basis of examples like birth, conference, motion, pleasure, pres‑
sure, question. Where this happens, the new verb is not always synonymous with the base
noun, though it may be, as is illustrated by the following examples from COCA.

Christian Science Monitor 1993: By this week’s end we will have dinnered eight guests,
(29) 
two at a time.
CBS_Morning 1993: Britain has something like 3,000 troops on the ground in central

Bosnia and we have drivered, I think, 40,000 tons of food.
PBS Newshour 1997: But what we find is that it’s a virus that does not match any strain

of influenza virus isolated since, but it is most related to the kind of influenzas that
infect swine, suggesting that this influenza entered the human population after being
passaged through pigs.

In the first case, it does not seem strange to find the verb dinner with the meaning ‘provide
dinner to’, as this is clearly distinct from the meaning of dine. However, the two following
converted verbs are more surprising, since the base verbs drive and pass would seem to fit the
contexts above just as well as the converted verbs.
Given a homophonous pair of words, one of them being a verb, and the other of some other
category, it is not always clear whether the verb should be analysed as the derived word, but it
seems that in most cases it is possible to take that decision based on a number of criteria (e.g.
semantic complexity, semantic relatedness, frequency, age, etc.). This problem is more gener-
ally discussed in Chapter 25 (see also Nagano 2008 for a recent treatment and Bram 2011 for
some empirical investigation into the degree of correlation between the different criteria).
There is a very large group of noun–verb pairs in English in which the two words differ in
their stress pattern, but which are otherwise (apart from stress-induced differences in vowel
quality) segmentally identical (e.g. dígest ◆ digést, pérmit ◆ permít, tórment ◆ tormént, see
Chapter 10 for many more examples). Independent of the decision whether these pairs
should be considered cases of conversion or not (see Chapter 25 for discussion), a closer look
at the data reveals that in the vast majority of cases the verb is the base and the noun is
derived. For this reason this set of forms is discussed in the chapter on event nominalizations
(Chapter 10). However, sometimes we find competing stress patterns with pertinent verbs
and it is those cases that are of interest in the context of this chapter. For example, the verbs
protest or refund can be found with main stress on the first or on the second syllable. In such
cases, the verb with the initial main stress is most probably derived by conversion from the
derived noun and thus maintains the stress pattern of the noun, as all other converted verbs
do (protéstv ◆ prótestn ◆ prótestv).
280 Derived verbs

Converted verbs may be transitive or intransitive (with transitive usages being slightly
more common, see Nagano 2008: 126), but it has frequently been observed (e.g. by
Jespersen 1942: 108) that the intransitive converted verbs are sometimes used in quasi-
transitive constructions with a semantically empty object it, as exemplified in (30), from
COCA (see, for example, Nagano 2008, for a collection of pertinent examples). It is unclear,
however, whether converted verbs are more susceptible to this kind of behaviour than
other verbs.

ABA Journal 2004: In the midst of this, I get a call from another client who wants me to
(30) 
grab the next plane to Ohio to take a look at three Midwestern industrial plants, located
about four hours’ drive from one another. Afterwards, I’ll have to hot foot it to Chicago
to give some talks at the ABA’s headquarters.

Quite often, we find a base noun being converted to a verb in the presence of a particle, as
shown in (31), from COCA.

Walk to the Full Moon 2002: Yes, we could fool about with bits and pieces of the genome
(31) 
and clone the occasional sheep in the 1990s, but not create a new race—or should I say
re-create an old one.
Mens Health 1995: Don’t expect to appreciate the flavor of a good, home-cooked meal if

you’re going to wolf it down.
A week in Texas 2010: I could tell she thought I’d chicken out.


13.2.4 Back-formation and other processes

We have already seen in Section 13.2.2.2 that some verbs in ‑ate appear to be formed by back-
formation. There are a great many other verbs that are derived by back-formation as well.
Nagano (2008: ch. 6) lists some recent examples, a selection of which are given in (32), with
those in (32b) representing examples of technical terms in linguistics.

(32) a. attrition ◆ attrit, breathalyzer ◆ breathalyze, demerger ◆ demerge, emotion ◆ emote,

hijacker ◆ hijack, television ◆ televise
b. 
affix-hopping ◆ affix-hop, antecedent-government ◆ antecedent-govern, back-
formation ◆ back-form, Chomsky-adjunction ◆ Chomsky-adjoin, pied-piping ◆
pied-pipe

Varying evidence can be brought to support the analysis of such verbs as back-formations.
The most obvious sort of evidence comes from date of attestation, with the noun being
attested before the verb. As Nagano (2008: 162) points out, however, sometimes the evidence
is semantic, in that the meaning of the noun is included in the meaning of the verb (for
 13.3 Semantic considerations 281

example, attrit means ‘to undergo attrition’). Sometimes evidence can come from frequency,
with the longer form having a higher frequency than the back-formed one. And sometimes
the evidence comes from allomorphy. Bauer (1983) gives the example of the verb contracept
from contraception; nouns with the formative cept + ion generally correspond to verbs with
the Latinate formative ceive (reception ◆ receive, perception ◆ perceive), so cept in a verb suggests
back-formation from the noun.
The kinds of expressions that can serve as bases of back-formed verbs are quite diverse.
Complex nouns as in (32a) are very common, as are compounds, as in (32b). The back-
formed verbs arising from compounds are discussed in more detail in the chapters on com-
pounding (see Chapters 19 and 20). Interestingly, bases for the back-formation of verbs need
not be morphologically complex, as the examples in (33), taken from Nagano (2008: 172,
originally from the OED) and our own collection, show.

(33) Bolshevik ◆ bolsh, liason ◆ liase, lysis ◆ lyse, propaganda ◆ propagand, rotisserie ◆ rotisse,
tiffin ◆ tiff, jacuzzi ◆ jacuze

Such forms seriously challenge the traditional definition of back-formation as the deletion of an
affix, and indeed it seems that analogy at various levels of abstraction can make better sense of
how these verbs are derived. For example, Nagano lists numerous nouns ending in ‑is that have
a back-formed corresponding verb without this ending (e.g. lysis ◆ lyse, phagocytosis ◆ phagocy‑
tose, pinocycosis ◆ pynocycose, sonolysis ◆ sonolyse, symbiosis ◆ symbiose, thrombosis ◆ thrombose).
These formations can be accounted for by analogy with the pair analysis ◆ analyze, or by making
reference to the fact that even if final <is> may not be a suffix, it is segmentable from the bound
roots that can be found with words ending in ‑is, as in emphasis ◆ emphasize, hypothesis
◆ hypothesize.
Some forms look like clippings that have not preserved the syntactic category of their
nominal bases, that is, bolsh is presumed to be a verb clipped from the noun Bolshevik.

(34) bolsh ◆ Bolshevik, hush ◆ husht, lap ◆ lapcock, loll ◆ Lollard, rattle ◆ ratline, tiff ◆ tiffin

However, the OED also lists the verbs bolshevize, husht, ratline, and tiffin, which suggests to us
that the derivatives in (34) are more convincingly analysed as clipped from verbs. The verbs
lap (first attestation 1839) and loll (given as obsolete by the OED, first attestation 1394) would
also be structurally regular clippings, but their potential bases are not attested as verbs.

13.3 Semantic considerations

There is a great deal of overlap in the semantic categories exhibited in derived verbs, with
conversion showing the widest range of available readings and the prefixes the narrowest
282 Derived verbs

range. The range of meanings generally correlates with the degree of productivity of the verb-
deriving process: the more productive the process, the wider the range of readings that seems
to be available. We begin with prefixes in Section 13.3.1, then treat suffixes (in Section 13.3.2),
conversion (in Section 13.3.3) and back-formation (in Section 13.3.4).

13.3.1 Prefixes
The dubious status of a‑ as a verb-forming prefix makes it nearly impossible to say anything
coherent about its semantics and we will not attempt to do so here. A similar case is for‑.
Verbs with this prefix are mostly opaque and it is not clear what the semantic contribution of
for‑ is. The prefix be‑ is easier, although very many of the item-familiar forms in be‑ have highly
lexicalized meanings (for example, befall, beget, beseech, beset, bestow). Where meanings are
transparent we see several prominent senses. Many of the transparent forms are ornative
(that is, ‘provide with X’) as with beblood, beglamour, bejewel, beleper, although interestingly,
there is at least one verb in be‑ that is privative (behead). Frequently the ornative forms carry
a nuance of intensification or wide coverage (‘provide with X all over’, as in beshit, bethank),
especially where they occur as past participles (bejowled, bespeckled). Some forms are causa-
tive (befoul, bestill) and a few simply transitive (belabor, bespeak, bemoan).
The prefix en‑ is most frequently found with a locative or directional sense (that is ‘put
something into X’), as in encoffin, encapsule, entomb. In other cases it can be paraphrased as,
for example, ‘turn into’ (encash, encode, encrypt) or ‘provide with’ (encrust, enhat, enshroud).
On adjectival bases en‑ has a causative meaning (enable, enfeeble, endear, enlarge, ensure).

13.3.2 Suffixes
The semantics of ‑ize and ‑ify has been explored in considerable depth in Lieber (1998, 2004)
and Plag (1999). The two suffixes are generally considered to share the same range of
polysemy. This range is large but not unrestricted. Table 13.1 summarizes the different mean-
ings proposed in the aforementioned literature.
In general, any given derivative is potentially ambiguous within the range of meanings that
can reasonably be construed on the basis of the semantic potential of the two suffixes and the
meaning of the base.
As shown in Plag (1999) and Lieber (2004) the different readings can be accounted for by
rather straightforward meaning extensions of a basic causative/locative meaning. For example,
inchoative and causative/resultative interpretations depend on the transitivity of the verb, that
is on the presence of an additional argument. The ornative and locative meanings are reverse
images of each other (that is, ‘cause X to go into/onto something’ or ‘cause something to go
into/onto X’) that can even occur with the same base word. Thus, computerize may refer to an
event in which data are put into a computer (ornative), or it can refer to an event in which
 13.3 Semantic considerations 283

Table 13.1 The semantics of ‑ize and ‑ify

Category label Paraphrase Examples
inchoative ‘become X’ oxidize, aerosolize, acidify, calcify
causative ‘make X, cause to become X,’ standardize, velarize, diversify,
acidify
resultative ‘make into X’ crystallize, unionize, teddify,
yuppify
ornative ‘make X go to/in/on something, accessorize, texturize, glorify,
provide with’ youthify
locative ‘make something go to/in/on X’ hospitalize, containerize, syllabify,
codify
similative ‘do/act/make/ in the manner of or Boswellize, despotize, (Lewisify)
like X’
performative ‘do X’ philosophize, economize speechify,
boozify

computers are put somewhere (locative), for example into an office. Furthermore, the inchoa-
tive/causative/resultative meanings on the one hand and the ornative/locative on the other
are related to each other as the latter can be analysed as change-of-place verbs, while the former
are change-of-state verbs. A change of state can thus be interpreted as a metaphorical change
of place (see Plag 1999: ch. 6 and Lieber 2004: ch. 3 for discussion and formal modelling).
The similative and performative meanings arise with nominal bases denoting types of
persons (hooligan, despot) or with proper names (Boswell, Marx), although not all derivatives
with proper names show similative or performative meanings. Consider, for example, (35)
from the BNC, in which the derivative shows an ornative meaning.

CBC W_newsp_other_social 1992: Agh! Get away from her! You’ll be Wymanised!
(35) 
I never knew what they meant until I read Bill Wyman’s story about his failed marriage
to child bride Mandy Smith. To be Wymanised is to be overcome by a feeling of
nausea when an excess of Wyman is forced upon you.

Metonymic readings of names may also lead to non-similative and non-performative read-
ings. The verb Marxize, for example, can have an ornative meaning in the attested example
The socialists Marxize the West, once the base Marx is interpreted metonymically as ‘the ideas
of Marx’ (Plag 1999: 139).
For each of the meaning categories, we can find comparable numbers of examples with both
‑ize and ‑ify derivatives, which speaks strongly for the synonymy of the two suffixes. The only
meaning that seems largely restricted to ‑ize is the similative one. Even with ‑ize there are not
many examples with this meaning, but with ‑ify they seem not to exist at all. The only example
that we were able to find in the corpora as a potential candidate for this category is given in (36).
284 Derived verbs

A7C W_biography 1990: It brings together Lewis the scholar, Lewis the voracious
(36) 
reader . . . Lewis the failed Romantic poet. . . . It is the story of a philologist (said to be
loosely based on Tolkien, but in fact fairly unlike him: Tolkien recognized some of his
own opinions and ideas Lewisified in the character) who, by a series of mishaps on a
walking tour, comes to a house where two sinister scientists, Weston and Devine, are
planning a visit to outer space.

Perhaps we should not be surprised, however, that examples of similatives are rare or non-
existent for ‑ize and ‑ify, as the core senses of these suffixes are clearly causative, with simila-
tives and performatives being sense extensions. As we will see below, it is more common for
non-causative verbal meanings to be expressed through conversion.
Semantically, verbal derivatives in ‑ate are also quite heterogeneous, and much of that
heterogeneity can be attributed to various mechanisms that may lead to new verbs in ‑ate.
Based on the analysis of twentieth-century neologisms, Plag (1999: ch. 7) arrives at the con-
clusion that only one-third of the neologisms display the expected ornative or resultative
meanings, chiefly with bases denoting chemical substances. We repeat some of his examples
from these two categories in (37a) and (37b).

(37) a. ornative
 alluviate, fluorinate, fluoridate, formylate, iodinate, mercurate, nitrogenate,
phosphate, protonate
b. resultative
gelate, methanate, pupariate, phosphate, sulphonylate

De-adjectival derivatives typically show causative readings but, as mentioned earlier,

­de-adjectival derivation seems not to be a productive pattern apart from isolated analogical
formations such as the twentieth-century coinage passivate (on the basis of activate), listed in
the OED.
Derivatives that arise through back-formation typically exhibit a reading dependent on
the meaning of the respective base word. As shown by Plag (1999: 207), a large proportion of
twentieth-century neologisms in ‑ate are back-derived from complex nouns and their mean-
ing reflects that of their nominal base. An illustrative example is escalate, which is ambiguous
between two readings that directly reflect the respective derivational history. One reading is
paraphrased as ‘To climb or reach by means of an escalator . . . To travel on an escalator’ by the
OED, and is thus obviously modelled on the noun escalator. The other reading is synony-
mous with ‘increase in intensity’, which is derived from escalation (see Plag 1999 for more
examples and discussion).
There are, in addition, very many derivatives whose meaning appears to be highly idiosyn-
cratic and heavily dependent on the context, as in dissonate ‘be dissonant’, fidate ‘to give
immunity from capture’, or vagulate ‘to wander in a vague manner’ (all from the OED).
 13.3 Semantic considerations 285

13.3.3 Conversion
The range of readings exhibited by converted verbs is larger than that of the verbs derived by affixa-
tion and we find semantic patterns that are quite different from the other verb-deriving categories.
As Plag (1999) and Lieber (2004) show, quite a few verbs converted from nouns or adjectives
fall into the semantic categories described in the preceding sections. Plag’s examples (all twenti-
eth-century neologisms from the OED) include locative archive, ornative marmalade, causative
rustproof, resultative package, inchoative gel, performative tango, and similative chauffeur. However,
we can find many other examples, for example the highly frequent instrumental ‘use X, perform
an action usually performed with X’ (hammer, glue, staple), privative ‘remove X’ (bark, skin), sta-
tive ‘be X’ (bay, landmark). As observed by Lieber (2004), among converted verbs, but not
among affixed verbs, one finds many with different motional meanings as in cartwheel, fishtail
(‘move in X manner’), jet, lorry, jeep (‘move using X’), quarterdeck (‘move at X location’). There
are no stative or motional affixed forms, while among the conversion verbs there are far more
performative and similative verbs than in any of the other categories. Furthermore, there are the
many converted verbs that are motivated onomatopoeically and which mostly denote the mak-
ing of the sound which the pronounced base is supposed to stand for (e.g. plink, zizz). In addi-
tion, one can find many highly idiosyncratic meanings that are largely dependent on the
surrounding discourse context. Such formations have even been awarded special terminological
status as ‘contextuals’ (Clark and Clark 1979; Aronoff 1980, see also Chapter 25).
Given this large variety of meanings and readings, it seems that the interpretation of newly
converted verbs is highly underdetermined. The best we can say is that these verbs generally
denote an event that is conventionally or contextually associated with the base. As might be
expected, this underspecified semantics can lead to widely differing readings for a single verb.
Example (38) illustrates this for summer, and there is an abundance of such cases. The para-
phrases are quoted from the OED.

(38) summer
‘To pass or spend the summer’
‘To keep or maintain during summer’
‘To sun oneself ’
‘To make summer-like’

With some verbs, one can even find completely opposite interpretations, depending on the
context. For instance, the verbs bark and bug are each attested with both privative and orna-
tive meanings, as shown in (39), with paraphrases from the OED.

(39) bark
‘To strip off the bark from (a tree)’
‘To enclose with or as with bark’
286 Derived verbs

bug
‘To clear (plants, etc.) of insects’
‘To equip with an alarm system or a concealed microphone’

This extreme semantic flexibility raises the question whether there are any restrictions as to
the semantic well-formedness of potential converted verbs. The answer to this question
seems to be a qualified ‘yes’. It follows from our above description of the semantics of conver-
sion into verbs that readings seem unlikely which do not arise from a conventional associa-
tion of the denotation of the base and some event. Thus, one can skin an animal, but legging
a table, that is removing a leg from a table, is something we normally do not do, and this leads
to problems in the interpretation of the potential verb leg as a privative verb in the absence of
strong contextual cues. This, however, is more a fact about the world than about the English
language.
Nagano (2008: 77–8) makes another claim with regard to restrictions on the semantic
range of conversion, suggesting that only those converted verbs seem possible as privative
verbs whose interpretation involves a part–whole relationship (for example, bone, husk, juice,
milk, pit); in contexts where such a relationship is not given, the privative relation needs to be
expressed through affixation (for example debug, disarm, unmask). However, although
Nagano lists many illustrative examples of either kind, it is not too hard to find counterexam-
ples to her claim. Thus, the verb fish is privative, even though fish are not a part of the water
from which they are removed, and even bug can, according to the OED, mean ‘remove bugs
(from plants)’, although there is no part–whole relation between the bugs and the plants
from which the bugs are removed. Given that very many privative events involve part–whole
relations anyway, further empirical research seems necessary to firmly establish this potential
restriction.

13.3.4 Back-formation
Like converted verbs, verbs derived by back-formation are dependent on the meaning of
their respective base word and denote the event conventionally associated with the base.
One illustrative example already cited in Section 13.3.2. is escalate, which has two different
readings, each of which can be straightforwardly related to the respective base (escalator or
escalation). Nagano (2008: 163) gives many examples that suggest that an especially promi-
nent meaning associated with back-formed verbs is the performative one: for example, to
burgle is ‘to do what a burglar does’, to buttle is ‘to do what a butler does’. Also prominent are
instrumental readings: to breathalyze is ‘to use a breathalyzer’, to jacuze is ‘to use the jacuzzi’
(the last is from our own examples).
 13.4 Competition among verb-deriving processes 287

13.4 Competition among verb-deriving

processes
The wealth of different morphological mechanisms available to derive verbs calls for an analy-
sis of their potential competition, and we have already hinted at potential competition in vari-
ous places throughout this chapter.
Some of the restrictions we mentioned have the consequence of a rather neat distribution
of labour in some domains. For example, we saw for ‑ize and ‑ify that the two suffixes are
almost in complementary distribution due to their phonological restrictions. Or the fact that
the verbal suffixes do not normally allow compounds or phrases as bases restricts such forms
to conversion, back-formation, or clipping. Furthermore, since the semantics of the affixes is
not identical to that of conversion, it should be expected that some processes are not eligible
if a certain meaning is to be conveyed by the verb. Thus, of the 488 converted verbs in Plag’s
study, 409 express meanings that could not be denoted by the suffixation of ‑ize, ‑ify, or ‑ate
(Plag 1999: 232).
But what about those domains where the categorial, morphological, or phonological
restrictions, as well as the semantics, allow for more than one morphological process to
derive a verb from a given base? A systematic look at such cases reveals that, given these cir-
cumstances, one can indeed often find doublets.
Plag (1999), for example, lists a fair number of synonymous pairs of converted verbs and
suffixed or back-derived verbs, such as carbon ◆ carbonize, dual ◆ dualize, indemn ◆ indemnify,
gel ◆ gelate, pressure ◆ pressurize. In Nagano’s (2008: 98) collection of derived verbs, roughly
every seventh verb derived with the suffix ‑ify or ‑ize has a converted verb counterpart. This
is similar to what we found with other rival processes (see Chapters 9 through 18) and can be
taken as another indication that blocking is not a categorical mechanism (see Chapter 26).
 chapter 14

Derived adjectives

14.1 Prospectus
In this chapter we consider the full range of affixes that derive adjectives in contemporary
English, looking at both their formal characteristics and their semantic range. Of all the areas
of English derivation, it seems safe to say that adjectival derivation has received the least
scholarly attention, although there are several interesting descriptive and theoretical ques-
tions that arise with respect to the pertinent derivatives. We will take up the issue of stress
placement with this cohort of affixes, the extent to which the various affixes are productive,
and the extent to which some of them are semantically interchangeable. We will also consider
whether such categories as gradable versus non-gradable or relational versus qualitative are a
function of the specific affix, the base, or the context in which individual items appear. The
examples in this chapter are taken from COCA, unless otherwise specified.
Some of the adjectival morphological categories to be discussed in the present chapter have
numerous members that are formally identical with homophonous nominal derivatives. A
great many adjectival derivatives in ‑an, ‑ant, ‑ist, ‑arian, and ‑oid have nominal counterparts
(e.g. American, pollutant, modernist, vegetarian, humanoid). The relationship between the nomi-
nal and the adjectival categories is not entirely straightforward. There are nominal derivatives
that have no adjectival counterpart (e.g. accountant, applicant, informant, prefixoid) and adjec-
tival derivatives that are without nominal counterparts (e.g. aberrant, pleasant, hesitant,
Picassoid). The fact that English allows prenominal modification by nouns may explain the
categorial variation. A theoretical account is nevertheless a challenge, and there are in principle
at least two different solutions available. The categorial variation could be considered a case of
affix homophony or a case of conversion (see Chapter 25 for a general discussion of these
notions). We remain agnostic with regard to the theoretical analysis and treat the pertinent
derivatives in this chapter and, if also relevant, in the chapter on the derivation of person nouns
(Chapter 11). The suffix ‑arian is dealt with exclusively in the noun chapter, since the nouns
involved have a very specific meaning and the adjectives largely correspond in meaning. The
suffix ‑oid is treated only in this chapter as the nominal derivatives seem to be dependent on
the similative meaning that seems to be extended from the adjectival suffix.
Note that there is also a class of Latinate adjectives, so-called ‘collateral’ adjectives, which
are semantically, but crucially not phonologically, closely related to corresponding nouns
 14.2 Formal considerations 289

(e.g. brachial to arm, canine to dog, paternal to father, vernal to spring, lupine to wolf). We
consider these relationships lexical and not morphological (cf. Koshiishi 2002).

14.2 Formal considerations

We begin with an overview of the formal characteristics of the adjective-forming affixes,
which are summarized in Table 14.1. In this table we list in the top row the kinds of bases the
suffixes attach to and, in the final column, whether base allomorphy is observed with the
respective suffix.
Adjective-forming affixes can be grouped according to their etymology: those that are of
non-native origin, which are frequently associated with both base allomorphy and stress
shift, and which often attach to bound bases, and those of native origin, which attach to free
bases of any provenance, are inert with respect to stress, and are not associated with base
allomorphy. The vast majority of adjective-forming affixes in English are non-native, specifi-
cally ‑able, ‑al, ‑ant, ‑ary, ‑esque, ‑an, ‑ible, ‑ic, ‑ical, ‑ine, ‑ive, ‑oid, ‑ory, ‑ous. Native affixes are
‑ful, ‑ing, ‑ish, ‑ly, ‑some, ‑y. The privative adjective-forming suffix ‑less is also native, but it will
be dealt with in Chapter 16 with the other negative and privative affixes. We begin our discus-
sion with the non-native affixes.
Before we do so, three remarks are necessary. We have chosen to treat ‑ical as a single suf-
fix, rather than as a sequence of ‑ic and ‑al for two reasons. First, there are a fair number of
forms in ‑ical that do not have corresponding bases in ‑ic, for example, practical, vertical, bibli‑
cal, commonsensical, indexical, quizzical, and especially medical terms like colovesical, surgical,
and the like. Second, for many forms in ‑ical, even where there is a corresponding ‑ic form
attested, there is no sense in which the two suffixes are semantically additive. We will take up
the relationship between ‑ic and ‑ical in more detail in Section 14.3.3. Some forms in ‑ist‑ic
may suggest a similar analysis as one suffix. See Section 14.3.4 for discussion.
A second point to be made at the outset is that we will treat the suffixes ‑able and ‑ible as
distinct for the purposes of formal considerations, as they pattern somewhat differently, but
will treat them as one suffix from the point of view of semantics, where there is no reason to
distinguish them.
Third, there is the status of the element ‑like. In Chapter 16 we argue with Dalton-Puffer
and Plag (2000) that a bound formative that is homophonous to a free form should only be
considered to be an affix (as opposed to an element of a compound) if it shows a different
semantic behaviour than the free form. This is not the case with ‑like, which should therefore
be considered a compound element (and not even a ‘semi-suffix’, as in Marchand 1969: 356).
We nevertheless treat ‑like here on a par with suffixes for practical reasons. Together with ‑ish,
‑y, ‑esque, and ‑oid it forms a set of closely related rival formatives that all derive words
expressing a similative meaning. It is therefore useful to investigate this set together, irrespec-
tive of the categorial status one of the formatives might have or not have.
Table 14.1 Formal characteristics of adjective-forming suffixes
Suffix N A V bb Compound Phrase Native Non-native Stress shift Base allomorphy
‑able (✓) ✓ ✓ (✓) ✓ ✓ ✓ (✓)
‑al ✓ (✓) (✓) ✓ * ✓ ✓ (✓)
‑ant ✓ ✓ * ✓
‑ary ✓ (✓) (✓) ✓ ✓ ✓ (✓)
‑esque ✓2 (✓) * (✓) ✓ ✓
‑ful ✓ ✓ ✓ ✓ ✓
‑an ✓2 (✓) ✓ * * ✓ ✓ (✓)
‑ible ✓ ✓ ✓ ✓
‑ic ✓ * (✓) ✓ * * ✓ ✓ ✓
‑ical ✓ ✓ * * ✓ ✓ ✓
‑ine ✓1 ✓ ✓
‑ing ✓ ✓ ✓
‑ish3 ✓1 ✓ (✓) ✓ ✓ ✓ ✓
‑ive (✓) (✓) ✓ ✓ * ✓ ✓
‑like4 ✓1 (✓) ✓ ✓ ✓ ✓
‑ly ✓ ✓ ✓
‑oid ✓1 (✓) ✓ ✓1 ✓
‑ory * ✓ ✓ ✓ ✓
‑ous ✓ (✓) (✓) ✓ * ✓ ✓
‑some ✓ ✓ ✓ (✓) ✓ ✓
‑y ✓ (✓) ✓ ✓ ✓ ✓
past pple ✓ ✓ ✓ ✓ (✓)

✓ well-attested, (✓) infrequently attested, * isolated examples

1
includes proper nouns,
2
mostly proper nouns,
3
also occasionally pronouns and numbers,
4
also occasionally pronouns
 14.2 Formal considerations 291

14.2.1 Non-native affixes

14.2.1.1 Kinds of bases
We can divide the non-native affixes into two cohorts, those that attach primarily to nouns
and those that attach primarily to verbs.

(1) a. Attaching primarily to nouns: ‑al, ‑ary, ‑esque, ‑ian, ‑ic, ‑ical, ‑ine, ‑oid, ‑ous
b. Attaching primarily to verbs: ‑able, ‑ant, ‑ive, ‑ory

We give examples of words derived with these affixes in (2) and (3).

(2) a. ‑al: adenoidal, behavioral, chromosomal, deicidal, elemental, familial, germicidal,

hexagonal, incestual, junctural, lectoral, manorial, natural, obituarial, paroxysmal,
quadrupedal, referential, sacrificial, textual, urethral, verbal, zonal

‑ary: alimentary, budgetary, customary, dietary, evidentiary, fusionary,

b. 
insurrectionary, legendary, momentary, nobiliary, originary, planetary, revisionary,
supplementary, transitionary, urinary, visionary

‑esque: ayatollaesque, bimboesque, cougar‑esque, divaesque, futuresque,

c. 
gazeboesque, hippie‑esque, island‑esque, kindergartenesque, lionesque,
minivanesque, novelesque, onionesque, painteresque, rubble‑esque, shamanesque,
traveloguesque, violinesque, zoo‑esque

d. ‑an: animalian, brontosaurian, centaurian, diocesan, equatorian, herbivorean,

jackassian, mammalian, oceanian, piscivorian, republican, suburban, tragedian,
utopian, zootopian

‑ic: alcoholic, basaltic, cyclonic, diadic, ectomorphic, fumarolic, genomic, halalic,

e. 
imbecilic, jihadic, kleptocratic, lethargic, melancholic, nomadic, ozonic,
palindromic, quietistic, rhapsodic, satiric, thoracic, urologic, vampiric, warrioristic,
xerographic, yogic, zoophilic

‑ical: alphabetical, brahminical, charlatanical, demiurgical, farcical, indexical,

f. 
nonsensical, paradoxical, theatrical, typical, whimsical

g. ‑ine: alkaline, alpine, crystalline, dinosaurine, elephantine, estuarine, infantine,

insectine, labyrinthine, pantherine, riverine, serpentine, tridentine, vulturine

h. ‑oid: albinoid, beastoid, bungaloid, ellipsoid, fibrinoid, orchidoid, parasitoid,

plasmacytoid, reptiloid, rubberoid, salmonoid, suffixoid, targetoid, walrusoid

‑ous: acronymous, bulbous, cadaverous, dolorous, eczematous, feverous, gelatinous,

i. 
hazardous, injurious, lecherous, mountainous, nervous, penurious, rancorous,
scandalous, tempestuous, usurious, victorious, wondrous, zealous
292 Derived adjectives

(3) a. ‑able: abradable, blendable, cancelable, deployable, enactable, feelable, grazable,

holdable, imbibable, jumpable, kneadable, listenable, matchable, networkable,
ordainable, parsable, quashable, registerable, settleable, throwable, utterable, visitable,
wadeable, zoomable
‑ant: ascendant, combinant, defiant, excitant, floatant, guidant, inhalant,
b. 
manifestant, nurturant, observant, pollutant, relaxant, securant, transformant
c. ‑ible: admissible, combustible, defensible, extensible, perceptible, recognizable
d. ‑ive: adaptive, coercive, decisive, emanative, formulative, gestative, immersive,
mitigative, negotiative, oppressive, pervasive, receptive, seductive, transgressive,
ulcerative, vituperative
‑ory: applicatory, combinatory, defecatory, executory, facilitatory, genuflectory,
e. 
hallucinatory, illusory, justificatory, legitimatory, mediatory, obfuscatory, possessory,
redemptory, satisfactory, titillatory, undulatory, ventilatory

While the vast majority of deverbal adjectives of this type have transitive verbs as bases, all of
them show some variety in the types of verbal bases they select. The suffix ‑able, for example,
shows a marked preference for transitive bases, although it does occur to some extent on
ditransitive bases and base verbs that take prepositional objects. There are only a few plausi-
ble examples on unergative or unaccusative bases.

(4) a. ‑able on ditransitive bases: addable, addressable, equatable, faxable, placeable, tellable
b. ‑able on bases with prepositional objects: accountable, atoneable, commentable,
conformable, listenable, panderable, wishable
c. ‑able on unergative bases: stutterable
d. ‑able on unaccusative bases: abateable, eruptable, flowable, perishable, rottable

We can find ‑ant on transitive verbs and verbs with prepositional objects, but there are almost
no formations on other types of verb.

(5) a. ‑ant on transitive verbs: arrestant, combinant, considerant, excitant, executant,

expectant, observant, presentant, recombinant, transcendant
b. ‑ant on bases with prepositional objects: accordant, actant, aspirant, convergant,
conversant
c. ‑ant on unergative bases: respirant
d. ‑ant on unaccusative bases: ascendant

The suffixes ‑ive and ‑ory are similar to ‑ant.

 14.2 Formal considerations 293

(6) a. ‑ive on ditransitive bases: ascriptive, associative, donative, inductive

b. ‑ive on bases with prepositional objects: collaborative, commiserative, elaborative, reactive
c. ‑ive on unergative bases: genuflective, gesticulative, perseverative
d. ‑ive on unaccusative bases: accelerative, degenerative, putrefactive, regressive

(7) a. ‑ory on ditransitive bases: applicatory, compensatory, explanatory

b. ‑ory on bases with prepositional objects: collaboratory, contributory, discriminatory,
participatory
c. ‑ory on unergative bases: genuflectory, hallucinatory, ovulatory, vomitory
d. ‑ory on unaccusative bases: acceleratory, escalatory

Many of these affixes attach to bound bases, and to varying extents to other categories.

(8) Adjective-forming affixes on bound bases

a. ‑able: communicable, educable, isolable, regulable
b. ‑al: apical, carnal, decimal, fluvial, mortal
c. ‑ant: clairvoyant, distant, malfeasant, renaissant, spirant
d. ‑ary: ancilliary, binary, culinary, liminary, pulmonary
e. ‑an: circadian, equestrian, median, metropolitan, sylvan
f. ‑ic: aquatic, barbaric, eclectic, hedonic, memetic
g. ‑ical: amical, identical, practical, radical, surgical
h. ‑ine: calcarine, equine, saline, uterine, vulpine
i. ‑ive: amative, captive, diminutive, fictive, native
j. ‑oid: android, asteroid, benzoid, Caucasoid, silicoid
k. ‑ory: amatory, gustatory, nugatory, perfunctory, valedictory
l. ‑ous: abstemious, bibulous, copious, fastidious, hydrous

The suffix ‑able shows the peculiarity that it frequently prefers to be attached to the bound
base on which a verb in ‑ate is formed rather than to the full form of the verb, although both
are possible (for example, aggregatable, activatable). The suffix ‑esque can be found in only a
few cases on bound bases (grotesque, picaresque), although these are clearly early borrowings
from French (see below) and not English derivations. The suffix ‑oid seems to be at least as
often found on bound bases as on free bases.
Of the affixes that typically attach to nouns, the following can also be found on verbal
bases.
294 Derived adjectives

(9) a. ‑al: configural, continual, excretal

b. ‑ary: deputary, expeditiary, imaginary
c. ‑ic: beatific, beautific, encryptic, integratic
d. ‑ous: continuous, covetous, infectious, prosperous, usurpatious

And similarly, some of the affixes that typically take verbal bases can also be found on nomi-
nal bases.

(10) a. ‑able: avalanchable, braillable, PBRFable (Middleton 2006), knowledgeable,

merchantable
b. ‑ive: agentive, contemptive, qualitative, quantitative, sportive
c. ‑ory: statutory, preceptory

A surprising number of these affixes can be found at least occasionally on adjectival bases.

(11) a. ‑al: accusatorial, inclemental, monetarial

b. ‑ant: benignant, contrariant, malignant
c. ‑ary: mobiliary, sanguinary
d. ‑esque: africanesque, brutalesque, globalesque, lunaresque, suburbanesque
e. ‑ic: astralic, civilic, sanguinic
f. ‑ive: abruptive, distinctive, diversive, profusive
g. ‑oid: fantastic‑oid, modernoid, pinkoid, simploid
h. ‑ous: complicitous, duplicitous, sanguineous, triumphalous

The non-native adjective-forming affixes attach almost exclusively to non-native bases,

although there are isolated examples of attachment to native bases for a few of them. In (12)
we present a relatively exhaustive list of the pertinent examples from COCA. Some of these
are clearly intended as jocular (crotchital, freshmanic, raspberrical, walkative), but the majority
are well-established and item-familiar.

(12) a. ‑al: bridal, tidal, crotchital, queerial

b. ‑ant: floatant
c. ‑an: elvan
d. ‑ic: apostolic, councilmanic, folkloric, freshmanic, runic, skaldic, Icelandic,
Greenlandic (and other forms ending in ‑landic)
 14.2 Formal considerations 295

e. ‑ical: churchical, coxcombical, folklorical, raspberrical

f. ‑ive: talkative, walkative
g. ‑ous: heathenous, lumpous, plunderous, righteous, scabrous, slaughterous,
slumberous, thunderous, wondrous

The three non-native affixes that attach freely to native bases are ‑oid, ‑esque, and ‑able. That
‑esque attaches to native bases is not altogether surprising considering that this suffix has a
rather different history than the others; it is a borrowing from French, rather than Latin, and
comes into English a bit later than the others (Marchand 1969 cites the first borrowings from
the late sixteenth century). Why ‑able has taken so readily to native bases is less clear, although
we might speculate that it has filled a semantic niche that is not covered by any native adjec-
tive-forming affix (see Section 14.2.2.1 below). The suffix ‑oid, while of Greek origin, also
attaches quite freely to native bases.

(13) a. ‑esque on native bases: girlesque, hell‑esque, holly‑esque, hood‑esque, island‑esque,

rubble‑esque, slackeresque, snowbirdesque, wagonesque
b. ‑able on native bases: bakeable, bearable, bite‑able, doable, feedable, fishable, hangable,
meltable, nameable, paintable, ringable, settable, singable
c. ‑oid: bluesoid, craboid, flintoid, freakoid, rubberoid, walrusoid

The suffixes ‑esque and ‑oid are also interesting in that they attach also to names and other
proper nouns, and indeed ‑esque shows a marked preference for them.

(14) ‑esque on proper nouns: Aspen‑esque, Barbieesque, Brokeback‑esque, Coplandesque,

Dadaesque, Gatsbyesque, Halloweenesque, Nazi‑esque, Pygmalianesque, Vegas‑esque

(15) ‑oid on proper nouns: Dawkins‑oid, Jacksonoid, Kuiperoid, Plutoid, Tayloroid

Also showing a marked preference for proper nouns is the suffix ‑an.

(16) ‑an on proper nouns: Alaskan, Aristotelian, Beethovenian, Burgundian, Caddoan,

Djiboutian, Godzillian, Kazakhstanian

There is no doubt a connection between these adjectival forms and the personal nouns in ‑an
discussed in Chapter 11.
Of the non-native adjective-forming affixes, ‑esque is also the only one that attaches with
some frequency to compounds. Occasional examples on compound bases can be found for
‑an, ‑ic, and ‑ical, and somewhat more than occasional for ‑able, as the examples in (17) illus-
trate, but ‑esque seems quite productive on compound bases (18).
296 Derived adjectives

(17) a. ‑able: backpackable, bushwackable, copyrightable, downloadable, lockoutable,

networkable, spinoffable
b. ‑an: fisticuffian, jackassian
c. ‑ic: aldermanic, folkloric, freshmanic, freemasonic, councilmanic
d. ‑ical: coxcombical, folklorical, raspberrical

(18) ‑esque on compound bases: appleseed-esque, Coldplay-esque, dot-comesque, hum­

mingbird-esque, merchant-ivoryesque, prom-queenesque, ragtime-esque, snowboard-
esque, strobe-lightesque

None of the non-native adjective-forming affixes in English can be found on phrasal bases.
The suffixes ‑ive and ‑ory are selective about the phonological form of their bases. Specifically,
they prefer bases that end in /t/ or /s/, often requiring a /t/ or /s/‑final allomorph where the
base shows allomorphy (see Section 14.2.1.2). So, for example, we find many forms like impres‑
sive, dismissive, conductive, associative, assistive, or promissory, dispossessory, locomotory, vomitory,
where verbal bases already end in /t/ or /s/, as well as forms like ascriptive, abrasive, repulsive,
and conjunctive, intercessory, delusory, where base allomorphs ending in /t/ or /s/ are available.
Where there are no such allomorphs, the suffix shows up with the extenders ‑at‑ or ‑it‑ that
allow the phonological requirement to be met, as we find in examples like informative, determi‑
native, definitive, explorative, purgative, punitive, condolatory, applicatory.
Adjectival suffixes frequently attach to bases that are already suffixed, and the reader is
referred to Chapter 26 for some discussion.

14.2.1.2 Base and affixal allomorphy

As is well-known from the literature on generative phonology (Chomsky and Halle 1968),
Metrical Phonology (Hayes 1982), and Lexical Phonology and Morphology (Siegel 1974;
Allen 1978; Kiparsky 1982a; Mohanan 1982; Giegerich 1999; among many others), some of the
non-native affixes that form adjectives trigger allomorphy or stress shift or both on their
bases. We can divide suffixes into those that do not affect stress, those that themselves bear
stress, and those that can trigger stress shift. As we will show below, the patterns involving
stress shift can be quite intricate and are sometimes variable.
It is clear that stress assignment to derived words in English in general often depends on
the structure of the syllables involved. One crucial distinction to be made here is the one
between heavy and light syllables. Another important aspect of our description is that we will
focus on output patterns rather than rule-based input-oriented generalizations, as such an
output-oriented approach has been shown often to be more adequate for the description of
morpho-phonological alternations (e.g. Plag 1999; Lappe 2007, see also Chapter 9).
The group of non-stress-shifting adjectival suffixes can be illustrated with the non-native
suffix ‑ine. This suffix has secondary stress, but is associated with neither stress shift nor base
allomorphy (although elephántine might be an exception in BrE). The suffix ‑esque is an
 14.2 Formal considerations 297

example of a suffix that bears primary stress; affixation of ‑esque does not change the underly-
ing stress contour of the base except to reduce a primary stress to a secondary one. Most
adjective-forming suffixes in English do, however, trigger some changes to the stress patterns
of bases and are associated with some degree of base allomorphy.
The suffix ‑able typically does not alter the stress pattern of its bases if they are only one or
two syllables. It can be found without stress shift on monosyllabic bases, iambic and trochaic
bases, as well as disyllabic bases that end in a secondarily stressed syllable.

(19) a. monosyllabic: askable, bakeable, wantable

b. iambic: abridgeable, abusable, besmirchable
c. trochaic: alterable, answerable, budgetable, balanceable, challengeable
d. base ending in a secondarily stressed syllable: archivable

The only exceptions we find are isolated forms that are highly lexicalized (ádmirable,
préferable, révocable) or the rare monosyllabic bound base to which ‑ize has attached
(baptízable).
Words in ‑able on bases that are longer than two syllables show three patterns. If the ante-
penultimate syllable of the derivative is heavy, the stress falls on the antepenultimate syllable,
which consistently leads to stress shift with one subset of these forms, illustrated in (20a).
Another subset of derivatives with heavy antepenult show variable stress shift. Some forms of
this group are given in (20b). If, in contrast, the antepenult is light (i.e. contains only a short
vowel or syllabic consonant), the stress is not shifted, as shown in (20c).

(20) heavy antepenultimate syllable, non-variable: al.lo.cá.ta.ble, a.ro.ma.tí.za.ble, ar.ti.cu.lá.

a. 
ta.ble, ca.te.go.rí.za.ble, com.mer.cia.lí.za.ble, cul.ti.vá.ta.ble, di.ver.si.fí.a.ble, do.cu.mén.
ta.ble, e.xe.cú.ta.ble, ex.tra.dí.table, ma.ni.pu.lá.ta.ble, per.so.ni.fý.able, pre.di.cá.ta.ble
b. heavy antepenultimate syllable, variable: á.na.ly.za.ble ~ a.na.lý.za.ble, cér.ti.fy.a.ble ~
cer.ti.fý.a.ble, drá.ma.ti.za.ble ~ dra.ma.tí.za.ble, i.dén.ti.fi.a.ble ~ i.den.ti.fí.a.ble, í.te.
mi.za.ble ~ i.te.mì.za.ble, ló.ca.li.za.ble ~ lo.ca.lí.za.ble
c. light antepenultimate syllable: jét.ti.so.na.ble, mó.ni.to.ra.ble, ré.gis.te.ra.ble

The suffix ‑able is not associated with base allomorphy, but its orthographic variant ‑ible fre-
quently is, as is illustrated by examples like admissible, permissible, transmissible, responsible,
comprehensible, defensible, divisible, extensible, perceptible, destructible, submersible, and offensible.
The suffix ‑al also occurs in the forms ‑ial, ‑ual, and ‑ar. The choice among ‑al, ‑ial, and ‑ual
seems not to be rule-governed, but largely based on the respective Latin allomorphy patterns.
For example, we find doublets in COCA like factoral alongside factorial or monarchal alongside
monarchial, and near doublets and triplets like accentual/agential/cliental, baronial/cantonal,
habitual/digital. The extender ‑u‑ is typically found on bases ending in clusters of obstruents
298 Derived adjectives

(aspect‑u‑al, concept‑u‑al, context‑u‑al) or /n/ + obstruent (consens‑u‑al, accent‑u‑al), but at

least in the latter case, it is not the only possibility (cf. agent‑i‑al and cliental).
On the other hand, the choice of the allomorph ‑ar seems to be more or less rule-governed.
We find it with some consistency on bases that end in /l/ (tonsillar, polar, capsular) or a con-
sonant cluster containing /l/ (bulbar, vulgar), and rarely on cases in which there is an /l/ in a
preceding syllable (linear, lumbar, lunar, columnar, plantar, planar are the examples to be found
in COCA). The single form laminar is notable in that the /l/ is in fact not even in a syllable
adjacent to the suffix. But /l/ in a preceding syllable does not automatically trigger the ‑ar
allomorph, as we also find forms like clausal, colonial, influential, larval, millennial, and the like.
When a base ends in syllabic [l̩], the suffix appears as ‑ular (clavicular, carbuncular, vehicular).
As Bauer and Huddleston (2002: 1708) point out, there are a few bases in which we find
semantically distinct forms in both ‑ar and ‑al (linear versus lineal; familiar versus familial).
With ‑al, we observe that the suffix is prosodically highly integrated in that it triggers resyl-
labification, and stress shifts where necessary to achieve a certain output pattern that character-
izes this morphological category. The words of this category are generally stressed like English
monomorphemic nouns, imposing a stress shift on the base where necessary. Thus, if the penul-
timate syllable is light, we find stress on the antepenult, as shown in (21a). If the penult is heavy,
the penult is stressed, illustrated in (21b). The stress shift cases are among those in (21b).

(21) a. penult is light: aboríginal, áctional, clósural, dígital, épochal, fáctoral, lúmenal,
márginal, monárchial
b. penult is heavy: affínal, ancéstral, archíval, collóidal, collégial, concéptual, decádal,
ethérial, fragméntal, hormónal, mollúscal, monsóonal

There are, however, also (apparently rare) cases of derivatives that show variable stress behav-
ior, such as cómmunal versus commúnal, where both stress patterns are in use.
The suffix ‑al and its variants involving extenders occasionally trigger allomorphy on their
bases. We find frequent deletion of a base-final schwa vowel (aortal, malarial), which may be
an instance of a more general constraint against identical vowels across morpheme bounda-
ries. Only occasionally do we find the avoidance of vowel hiatus by means of the insertion of
an extending consonant (embryo‑n‑al). Palatalization or spirantization of a base-final conso-
nant is frequent (agential, appendiceal, consensual, facial, laryngeal), but voicing of a base-final
consonant without palatalization is rare, perhaps confined to the example basal. We also find
occasional examples in which trisyllabic laxing has occurred (residual, criminal).
With the suffix ‑ant, we again observe the strong tendency toward an output pattern in
which the penult is stressed if heavy, and the antepenult is stressed if the penult is light. Thus,
with bases like triumph or dilate we find stress shift (tri.úm.phant, di.lá.tant), while có.lo.rant
and núrtu.rant show no stress shift. Longer base words also behave in accordance with the
general output condition (a.na.lý.sant, e.xé.cu.tant, ma.ni.fés.tant). This suffix only occasionally
results in base allomorphy. For verbal bases in ‑ate, the ‑ant adjective is invariably formed on
 14.2 Formal considerations 299

the bound base without ‑ate rather than on the full form of the verb (congreg-ant, devi-ant,
etc.). Otherwise, we find divisant from divide, and where a final /n/ is resyllabified with the
suffix and underlying /ɡ/ surfaces (benignant, malignant).
With respect to stress, the suffix ‑ary shows variable behaviour. The majority of the forms do
not show stress shift, so that we get derivatives that stress the pre-antepenult or even the fifth
syllable from the right, as in de.pó.si.ta.ry, dí.sci.pli.na.ry (NAmE), and ac.cré.tio.na.ry. The few
forms that do show stress shift have stress on the antepenult or pre-antepenult: a.li.mén.ta.ry,
di.sci.plí.na.ry (BrE, with /ɪ/ in the antepenult), e.vi.dén.cia.ry, ves.tí.bu.la.ry. We find only small
amounts of segmental base allomorphy with forms in ‑ary: occasional cases of palatalization of
base-final consonants (beneficiary, evidenciary) and deletion of base-final vowels (medullary).
The suffix ‑an has variants with extenders ‑e‑ and ‑i‑. There seems to be no principled reason
for the appearance of either variant, but forms with extenders and forms without seem to be
numerically quite similar. We find occasional doublets like Arizónan/Arizónian. The suffix is
capable of shifting the stress to ensure penultimate or ante-penultimate stress. Derivatives
with the variant ‑an are variably stressed on the penult or antepenult (e.g. Arizónan, carnivóran,
diócesan), while ‑ean and ‑ian derivatives are consistently stressed on the antepenult (see, for
example, actórian, bacchanálian, barónian, Baudeláirian, discórdian, equatórian, micróbian, ani‑
málian, avesáurian, cantelóupean, repúblican, and the examples in the next paragraph).
We find only a small amount of segmental base allomorphy with ‑an. There is occasional
palatalization of final /t/, as in Aleutian, but this is not a regular phonological process, espe-
cially where ‑an is attached to proper nouns (e.g. Beckéttian). Base-final vowels are often
deleted (ambrósian, volcánean), but not always. With final /əʊ ‖ oʊ/ we sometimes find vowel
hiatus (Chicágoan, Cáddoan, Chácoan) and sometimes the addition of an extender like /n/
(Apollónian, Buffalónian). Base-final /s/ sometimes deletes before ‑an (Ándean, Athénian,
Azórean, Cervántean) but again not always (compare Bráhmsian with the previous examples).
The suffixes ‑ic and ‑ical appear to be textbook cases of stress shift: whatever the stress
pattern of the base, we find stress on the syllable before the suffix. Thus, words in ‑ic almost
invariably end in a trochee (agéntic, ballétic, halálic, jihádic, occúltic, pirátic, poetic; exceptions
are, for example, the lexicalized forms árabic, cátholic, chívalric). Derivatives in ‑ical invariably
stress the antepenult, also inducing stress shifts where necessary (apocalýptical, eléctrical,
meteorítical, rádical, theátrical).
This suffix also exhibits various sorts of base allomorphy that we have seen with other non-
native adjective-forming suffixes. We find deletion of base-final vowels (historic, allergic, aortic,
tornadic), although we also find insertion of extenders ‑t‑ or ‑n‑, which prevents vowel hiatus
(aromatic, asthmatic, empathetic; draconic, embryonic, messianic). There does not seem to be
anything systematic that determines which we find; indeed in COCA we find doublets like
empathic versus empathetic or diarrheic versus diarrhetic. For a few cases, we find addition of
‑at‑ or ‑et‑ before the suffix (axiomatic, emblematic, enzymatic, genetic). Only in the case of heroic
do we find vowel hiatus. Bases that end in /s/ exhibit /t/ before the suffix, as in apocalyptic,
chaotic, pelvic, synaptic, syntactic, syphilitic. Base-final sonorants resyllabify with the suffix as in
cyclic, calendric, chasmic, rhythmic. Cases of trisyllabic laxing can be found (meteoritical).
300 Derived adjectives

The suffix ‑ive has three additional variants with extenders, ‑at‑ive, ‑it‑ive, and ‑ut‑ive. The
last of these appears only on verbs containing the formatives solve and volve (dissolutive, evolu-
tive), and ‑itive on verbs with the formatives quire (inquisitive, acquisitive), pos (oppositive), or
pet (competitive, repetitive). Otherwise we find it only in the forms additive and sensitive. The
variant ‑ative is found more frequently, but it is not clear that there is anything systematic
about its appearance, as opposed to the appearance of the simple ‑ive variant. Indeed, in
COCA we find at least a few doublets like augmentative/augmentive and adaptative/adaptive,
and many near doublets like revulsive versus pulsative. That said, verbal bases that end in /r/
or non-coronal consonants like /m/ or /k/ appear to prefer the ‑ative variant (comparative,
explorative, confirmative, informative, evocative). The string ‑ative also arises through the com-
bination of the verbal suffix ‑ate with our adjectival suffix ‑ive. In the vast majority of cases this
leads to destressing and vowel reduction to schwa of the secondarily stressed verbal suffix (cf.
íteràte versus íter[ə]tive). Very rarely, we find two forms alongside each other (íntegràtive and
íntegr[ə]tive, íllustràtive and illústr[ə]tive).
With regard to stress placement, ‑ive rarely has any effect on its base. Consider the many
non-stress-shifting examples with penultimate stress (extrospéctive, apprehénsive, corréctive,
creátive), with antepenultimate stress (dónative, attríbutive, depósitive), and even pre-antepe-
nultimate stress (dédicative, íntegrative, íterative, spéculative). There are several highly lexical-
ized derivatives in ‑ive that show stress shift, in which case stress is placed on the
antepenultimate syllable (contémplative, corrélative, demónstrative, exécutive, négative, rélative,
sédative, remónstrative). Notably, to achieve stress on the antepenult, the stress is sometimes
moved forward (córrelàte ◆ corrélative), sometimes backward (reláte ◆ rélative). If stress is
shifted in a verbal base ending in ‑ate, vowel reduction in the verbal suffix is obligatory, as
shown in the variably shifting adjective illustrative. In case of shift, this word does not sec-
ondarily stress the verbal suffix (e.g. illústr[ə]tive), thus avoiding a stress clash.
As ‑ive favours non-native verbal bases, it displays the full range of base allomorphy that we
find in Latinate verbs. The range of base allomorphy that we find is illustrated in (22).

(22) adhere ◆ adhesive

appeal ◆ appellative
ascribe ◆ ascriptive
assume ◆ assumptive
cohere ◆ cohesive
compel ◆ compulsive
conjoin ◆ conjunctive
induce ◆ inductive
inquire ◆ inquisitive
perceive ◆ perceptive
submit ◆ submissive
succeed ◆ successive
 14.2 Formal considerations 301

Note that with some of the forms involving <s> before the suffix, we find variation between
voiced and voiceless realizations. In addition, we find occasional vowel laxing (interventive)
and trisyllabic laxing (definitive, declarative, competitive), as well as occasional devoicing of a
base-final obstruent (infusive, attentive, disclosive). Base-final /d/ sometimes becomes /s/, or
sometimes /z/, in the derived form (obtrusive, responsive, abrasive, conclusive).
The suffix ‑oid does not induce stress shifts, but often attaches to bound bases that, given
the denominal character of the suffix, could be analysed as truncated nouns, for example
bungalow ◆ bungaloid, magma ◆ magmoid, negro ◆ negroid, Picasso ◆ Picassoid, Pluto ◆ plutoid
(sic), rhombus ◆ rhomboid, sigma ◆ sigmoid, volcano ◆ vulcanoid. Such base truncation seems to
be especially frequent when base-final vowel and suffix-initial vowel are the same.
The profile of ‑ory is very similar to that of ‑ive. It exhibits two variants with extenders,
‑it‑ory and ‑at‑ory, which appear under circumstances similar to the allomorphs of ‑ive. The
variant ‑itory is infrequent, appearing on bases with the Latinate formatives pos and pet
(expository, competitory) as well as in rare cases like definitory and divinitory. The ‑atory vari-
ant appears in words like applicatory, comparatory, perspiratory, condolatory, amendatory, and
combinatory, with no particular pattern discernable. Again, at least one doublet improvisa‑
tory/improvisory can be found in COCA.
As with ‑ive, the stress pattern of the base is almost always maintained with ‑ory. On disyl-
labic bases, this seems invariably to be the case for trochees (mandatory, gyratory, migratory,
vibratory), and typically the case for iambs (conclusory, advisory, observatory, amendatory)
although there are a few high frequency or lexicalized cases where stress shifts to the initial
syllable of the base (respiratory, combinatory, preparatory, revelatory). For trisyllabic bases,
amphibrachs and anapests exhibit maintenance of base stress (contributory, interjectory, loco‑
motory). Dactylic bases typically maintain stress (postulatory, dedicatory, copulatory) except
in a few cases where their penultimate syllables are heavy (compensatory, obfuscatory, satisfac‑
tory). For bases with more than four syllables, we only occasionally find stress retraction
(compare for example, prognosticatory versus interrogatory).
With regard to base allomorphy, ‑ory triggers allomorphy on Latinate verbs, as ‑ive does, so
we find forms like compulsory, introductory, redemptory, satisfactory, and intercessory. Base-
final /d/ becomes /s/ before ‑ory, as in conclusory, delusory, derisory. We also find frequent
laxing of a vowel in the final syllable of the base (declamatory, declaratory, definatory, divina‑
tory, explanatory).
The suffix ‑ous appears with a wide variety of extenders: ‑iti‑ous (excrementitious), ‑ati‑ous
(flirtatious, usurpatious, vexatious), ‑t‑ous (edematous, eczematous), ‑in‑ous (altitudinous,
cartilaginous, leguminous, voluminous), ‑u‑ous (contempuous, sensuous, spirituous), ‑i‑ous
(capricious, censorious, felonious), and ‑e‑ous (consanguineous, nectareous). The last two vari-
ants are probably just spelling variants. As for the rest, it is hard to tell if there are any regu-
larities to be found. We find doublets in COCA like carbonous and carbonious, spiritous and
spirituous, as well as near doublets like treasonous versus felonious, humorous versus
censorious.
302 Derived adjectives

The words with the suffix ‑ous are generally stressed on the penult if the penult is heavy, and
on the antepenult if the penult is light. The suffix triggers stress shift where it is necessary to
achieve the pertinent stress pattern. The former pattern is shown by fi.la.mén.tous, ca.ta.rác.tous,
ca.prí.cious, de.sí.rous, in.féc.tious, the latter pattern by a.cró.ny.mous, béau.te.ous, ca.dá.ve.rous,
cár.bo.nous, ce.re.mó.ni.ous, com.plí.ci.tous, co.ní.fe.rous, con.tí.nu.ous, có.ve.tous, he.te.ró.ny.mous,
me.ló.di.ous. The highly lexicalized forms ad.van.tá.geous, cou.rá.geous, out.rá.geous are peculiar
in their behaviour. They totally conform to the antepenult stress pattern with heavy penult, but
they do so by showing stress shift and vowel tensing. This is somewhat strange, however, since
there is no necessity for this. The unattested, non-shifting potential derivative, for example
*advántageous, would be in accordance with the required prosody for penultimate stress.
We find a modest amount of segmental base allomorphy with forms in ‑ous. There is occa-
sional vowel laxing (capricious, gangrenous, zealous) and variety-dependent palatalization of a
base-final /s/ (capricious, gaseous, gracious, sensuous). There is also sporadic resyllabification
of base-final schwa syllables with loss of the schwa (disastrous, lustrous), but this seems not to
be obligatory as we find monsterous alongside and synonymous with monstrous.

14.2.1.3 Productivity
The non-native adjective-forming suffixes show a range of productivity, varying from the
fully productive to the nearly unproductive. Among them, it seems clear that ‑able and ‑esque
are the most productive; indeed they appear to be completely productive. The suffix ‑able can
be found on just about any transitive verb (and some intransitives, as we have seen), and
‑esque on any personal name. Since transitive verbs can be formed productively (see Chapter
13), and proper nouns are virtually inexhaustible, we seem to have an ever-expanding pool of
bases to which these affixes can attach. The suffix ‑oid is also highly productive, even if some-
what restricted due to its scientific flavour.
At the other end of the scale we find the suffixes ‑ant, ‑ary, and ‑ine all of which exhibit rela-
tively few types in COCA with almost no forms that appear to be novel (although possible
novel forms for ‑ant and ‑ary are averrant and testimonary, neither of which are found in the
OED). The suffix ‑ary is found on quite a few forms in ‑ion, but rarely on bases derived with
the productive variant of the suffix, ‑ation.
The suffixes ‑ine and ‑ive also exhibit few apparently novel forms. With ‑ine we find dinosau‑
rine, cervantine, sigmodontine none of which are attested in the OED. For ‑ive we find assortive,
contestive, precessive, situative, and walkative, the last of which is clearly intended to be jocular.
It is of course not surprising that we should find so few novel forms with ‑ive, since as we have
seen above this suffix places fairly strict phonological conditions on its bases.
The other non-native adjective-forming suffixes fall somewhere in between these two
poles. The suffix ‑al, for example, attaches quite productively to a base ending in the suffixes
‑oid, ‑ation, ‑or, or ‑ure, or neo-classical combining forms like ‑cide or ‑some (as in chromo‑
some). There are occasional novel forms on underived bases, as for example, altazimuthal,
appendiceal, pelletal, silical, summital, tapestrial, and trombonal, all attested in COCA.
 14.2 Formal considerations 303

The suffix ‑ic also shows some productivity. In addition to attaching to a wide range of
neo-classical combining elements like meter, naut, phile, phobe, logy, graph, path, gen, morph,
scope, crat, nym, phage, soph, sphere, bot, cyt, and so on, it also attaches quite freely to forms
derived on the suffixes ‑ite and ‑ist. To the extent then that neoclassical compounds and
derived forms in ‑ist are themselves productive, ‑ic is as well. The suffix ‑ic also appears on
proper nouns where it is potentially productive, although it is not clear how many adjecti-
valizations of names are formed this way (Saddamic is surely a recent example though).
And it is not difficult to find low frequency, apparently novel forms in COCA, among
which are the forms in (23).

(23) civilic, agentic, anodynic, armageddonic, aureolic, banjoic, bimbonic, boronic, bosonic,
Boswellic, codic, dharmic, dolphinic, golemic, halalic, hurricanic, hybridic, koanic,
legerdemainic, pheromonic, potentatic, rosemarinic, schadenfreudic, strychninic,
tantrumic, tortic, tsunamic

The suffix ‑ical also attaches to neoclassical combining forms (especially those ending in
‑ology) and derived words in ‑ist but otherwise seems not to be nearly as productive as ‑ic.
A few apparent neologisms can be found (chessical, churchical, existorical), but these seem to
convey a self-conscious or ironic nuance that the neologisms in ‑ic generally lack.
The suffixes ‑ory and ‑ous are modestly productive. The former appears frequently on verbs
in ‑ate. COCA yields apparently novel forms like cavitatory, comparatory, congregatory, contes‑
tatory, defecatory, divinatory, enduratory, immolatory, incarnatory, inhalatory, isulatory, litiga‑
tory, productory, and tonsilory. The suffix ‑ous is partial to neoclassical combining forms like
vore, nym, mat, morph, phone, phage, phil, etc., and frequently gives rise to jocular or mock-
scholarly forms like ginormous, grandilomentitudinous, redunculous, rumbustious, up-tighteous,
mind-bendous, bimbocious, and crapulous. In COCA we find other novel forms as well, includ-
ing: attitudinous, curtainous, debaucherous, desertous, exotendinous, fidelitous, gargantuous, hir‑
sutulous, inquirious, razorous, revengeous, rippulous, rosivorous, squanderous, theonomous,
tonerous, usurpatious, and vibrous.

14.2.2 Native affixes

English has a wealth of native adjective-forming affixes alongside the non-native ones we have
treated above. We will begin with the clearly derivational suffixes ‑ed, ‑en, ‑ern, ‑ful, ‑ish, ‑ly,
‑some, and ‑y, followed by the prefix a‑ and take up the participial adjectives in Section 14.2.3.

14.2.2.1 Kinds of bases

The adjective-forming suffixes ‑ful, ‑ish, ‑like, ‑some, and ‑y attach to a range of categories
(24)–(27).
304 Derived adjectives

(24) a. ‑ful on nouns: deceitful, faithful, gleeful, lawful, mindful, purposeful, slothful
b. ‑ful on verbs: reflectful, resentful, vengeful, wakeful, watchful
c. ‑ful on adjectives: gladful, proudful, rightful, wrongful

(25) a. ‑ish on nouns: babyish, clannish, doggish, gluish, iconish, lardish, nannyish
b. ‑ish on verbs: skitterish, sloggish, snappish, ticklish, whimperish
c. ‑ish on adjectives: angry‑ish, brownish, freeish, goodish, modernish, narrowish,
roughish
d. ‑ish on numerals: fivish, twentyish

(26) a. ‑some on nouns: adventuresome, fruitsome, healthsome, joysome, mirthsome,

riddlesome
b. ‑some on verbs: frolicsome, irksome, meddlesome, vexsome, wrigglesome
c. ‑some on adjectives: blithesome, darksome, lithesome, lonesome, wearisome

(27) a. ‑y on nouns: arty, bitchy, coppery, feathery, girly, houndy, nebbishy, perfumy, rusty
b. ‑y on verbs: blowy, choosey, droopy, grabby, moochy, picky, scrabbly, teetery, wilty
c. ‑y on adjectives: crispy, dreary, greeny, moderny, swanky, yellowy

The suffix ‑ly occurs on nouns and noun phrases (28a), and ‑ed can be found on nouns
(including nominal compounds) and noun phrases (28b).

(28) a. ‑ly: actorly, beggarly, comradely, innerworldly, fleshly, laggardly, nightly, readerly
b. ‑ed: bearded, blue-eyed, broad-minded, empty-headed, four-footed, three-wheeled,
glass-roofed, wooded

As some of the examples show, plural inflection does not surface on base noun phrases, in
spite of a plural interpretation (see also four-holed, four-leafed, four-petaled). The two suffixes
‑en and ‑ern are no longer productive, with ‑ern appearing only on nouns denoting geographic
directions (e.g. northern, western), and ‑en deriving adjectives from nouns denoting materials
or substances (earthen, silken, wooden). These unproductive suffixes will not be treated in any
more detail.
Note that there are isolated cases of a native extender with ‑ly, when it attaches to nouns
de­noting geographic directions (south‑er‑ly, north‑er‑ly). Native adjective-forming affixes as a
rule do not attach to bound bases, with the exception of a few archaic or obsolete bases on
which we find ‑some, for example gruesome, handsome, lissome, noisome, winsome. As is typical of
native affixes generally, the adjective-forming suffixes are equally comfortable on native or non-
native bases:
 14.2 Formal considerations 305

(29) a. ‑ful native: heavenful, deathful

non-native: deceitful, beautiful
b. ‑ish native: doomish, freeish
non-native: caricaturish, modernish
c. ‑ly native: deathly, nightly
non-native: musicianly, spectatorly
d. ‑some native: mirthsome, wholesome
non-native: adventursome, joysome
e. ‑y native: handy, cheesy
non-native: actressy, chocolatey

Of the native suffixes, ‑ish and ‑y appear on compound bases, and at least occasionally on
phrasal bases.

compounds: homeboyish, punk-rockish, show-offish, trailer-parkish

(30) a. ‑ish 
phrases: dog-in-the-mangerish, feelgoodish, up-your-buttish
b. ‑y compounds: hot-tubby, piss-stinky, woodwork
phrases: secret-agenty

As is typical of native affixes, this cohort is associated with neither stress shift nor allomorphy.
The prefix a‑, which derives what is frequently called an adjective but could often be an
adverb, only very occasionally leads to new formations (such as a‑move, a‑pant, asmirk (all
from the OED), acrawl, and, from Barnhart et al. (1990), aclutter, aglaze, asquish, awhir), all
apparently based on verbal bases that are monosyllabic.

14.2.2.2 Productivity
Of these suffixes, clearly the most productive are ‑ish, and ‑y, which, as we have seen, attach
to virtually any kind of available base, including compounds and phrases; indeed, it seems
safe to say that they are among the most productive suffixes of contemporary English. They
are most productive, of course, on nominal bases, but there are novel forms on adjectives for
‑ish, and on verbs for ‑y. Apparently novel forms from COCA include: beginnerish, dungeon‑
ish, gloomish, bloodyish, familiarish, modernish; kitteny, lummoxy, spoofy, flinchy, moochy, revolvy.
It is not difficult to add examples to these.
The suffix ‑ful is somewhat less productive. It exhibits fewer types overall, and its bases
seem to cluster around a specific semantic category that poses natural limitations for poten-
tial novel forms. To the extent that new forms can be derived, the suffix ‑ful favours bases that
denote psychological states or processes; verveful, angstful, groundful, importful, learningful,
306 Derived adjectives

driveful, swoonful, despiseful, and provokeful are novel forms attested in COCA. There are only
a few other novel ‑ful forms: fluteful, ghostful, girthful, limbful. Also only modestly productive
is ‑some, for which there are at best a handful of apparently novel forms attested in COCA:
fiercesome, fruitsome, hunksome, problemsome, whoresome, clattersome, droolsome, jinglesome,
coosome, rilesome.
One can very easily find new ornative adjectives with the suffix ‑ed, with new formations
having a tendency to have phrases as bases, especially those containing a numeral in the first
position of the phrase (one-horned, two-tailed, three-masted, four-engined, five-sided, and so
on).
Adjectival ‑ly is barely productive, with new forms largely confined to person nouns
(demonly, dudely, speakerly, spectatorly) and places (neighborhoodly). There are relatively few
types with this suffix, and most of them are item-familiar. The suffixes ‑en and ‑ern are unpro-
ductive, and the prefix a‑ seems to be available for new coinages only to a very limited degree,
chiefly in poetic or literary discourse.

14.2.3 Participial adjectives

Both the ‑ing form of verbs and past participle forms are frequently used as premodifiers to
nouns, and have sometimes been argued to be categorially adjectival, as evidenced by their
frequent ability to accept prefixation with negative un‑, to form the comparative and superla-
tive, sometimes even the morphological comparative and superlative (see Chapter 6), and to
be sub-modified by very, so, etc. Formally, the participial adjectives are identical to the present
and past participles, and as the formal characteristics of the latter have already been covered
in Chapter 5, we will not revisit them here. There is one point that should be highlighted,
though, which is the tendency of adjectives formed from past participles to retain archaic
strong forms where weak past participles have arisen and are more often used in verbal con-
texts. Among these are burnt, cloven, bereft, mown, shorn, shod, smitten, spilt, spoilt, and perhaps
sewn.

14.3 Semantic considerations

The semantic interpretation of adjectival derivation in contemporary English has received far
less attention than the semantics of nominalization or verbalization, but it has been touched
upon at least by Ljung (1970), Beard (1995), Hamawand (2007), and Kaunisto (2007). In
Section 14.3.1 we consider various aspects of the semantics of adjective-forming affixes: what,
if any, specific semantic content is carried by various affixes, whether there is a fundamental
difference between so called ‘relational’ and ‘qualitative’ adjectives, and whether specific
affixes can be designated as gradable or non-gradable.
 14.3 Semantic considerations 307

14.3.1 Semantic content

The semantic characteristics of derived adjectives to some extent depend on what sort of
base they are attached to. When attached to verbal bases, adjectival affixes typically reference
either an argument of the base or the event itself. That is, for a transitive verb verb whose
arguments are X and Y (in other words, X verbs Y), subject-referencing adjectives can be
predicated of X, object-referencing adjectives can be predicated of Y, and event-referencing
adjectives can be construed as ‘characterized by verb’. Some deverbal adjectives also have
aspectual nuances, as we will see. Except in the case of eventive nouns, adjectives derived
from nominal bases do not reference arguments, and generally exhibit more loosely relational
meanings. A rather specific non-argumental meaning, however, is associated with ornative
‑ed. We will first discuss deverbal adjectives and then denominal (or de-adjectival)
adjectives.

14.3.1.1 Argument referencing

As noted above, those adjective-forming affixes that accept verbal bases typically prefer either
a subject-referencing or an object-referencing interpretation. The participial ‑ing is somewhat
more complex, so we will treat it separately.

Object-referencing
We will begin with the typically object-referencing affixes, as they are less numerous. The
most important of these is the suffix ‑able (by which we now mean both the ‑able and the
‑ible allomorphs). As indicated in Section 14.2.1.1, ‑able prefers transitive bases, and adjec-
tives derived with this suffix on transitive bases always reference the object argument. So for
a verb like conceal, the derived adjective concealable refers to the theme, the thing concealed,
rather than the agent of the action. Similarly for other typical examples like adaptable or
grillable. In the relatively rare cases where we find ‑able on unaccusative verbs, the ‑able
adjective appears to be subject-referencing, for example perishable. However, it has been
argued in the framework of generative grammar that unaccusative verbs should actually be
analysed as having an underlying object, and not an underlying subject argument, in which
case we might assume that ‑able is object-referencing here too. Regardless of syntactic analy-
sis, the referent of ‑able in these cases bears the same thematic relation as in the case of the
transitives; so forms like perishable or rottable reference the patient/theme on both transi-
tive and unaccusative bases.
We do find ‑able on the occasional unergative verb, as stutterable, attested in COCA.
Interestingly, however in this case the ‑able adjective does not reference the subject, the only
available argument, but rather a contextually or pragmatically determined referent.

Psychology Today 1998: Over that time, he strived for one main goal at local gatherings:
(31) 
“making the atmosphere stutterable.” He wanted people to have permission to
stammer openly; after all, he says, “stuttering isn’t something you get over in a week.”
308 Derived adjectives

In the few other cases where ‑able appears to attach to an unergative verb, the context indi-
cates that those verbs have been coerced into a transitive reading, as for example, in the case
of perspirable in (32).

Saturday Evening Post 2003: To become sensible of this by an experiment, let a person
(32) 
keep his position in the bed, but throw off the bed-clothes, and suffer fresh air to
approach the part uncovered of his body; he will then feel that part suddenly refreshed;
for the air will immediately relieve the skin, by receiving, licking up, and carrying off,
the load of perspirable matter that incommoded it.

An ‑able adjective is conceivable for the verb perspire as long as it is possible to imagine some
output or product of perspiring.
Where we find ‑able on nominal bases, these bases are typically amenable to an eventive or
stative interpretation. In such cases, ‑able seems able to reference an implied argument or
refer to a participant in the associated event. If the participant is an argument, that argument
can be either an implied subject or object. In cases like knowledgeable or pleasurable we have
a subject-referencing interpretation, whereas in marriageable, braillable, or impressionable we
have an object-referencing interpretation. A non-argument is focused in ferryable, which can
refer to the path of the ferrying event, as in ferryable river.
The suffix ‑able of course has semantic content beyond its object-referentiality. It is clearly
modal in nature, indicating capacity to undergo the action denoted by the verb (washable,
heatable), or occasionally inclination to undergo the action denoted by the verb (perishable,
agreeable) (Bauer and Huddleston 2002: 1707). Bauer and Huddleston point out as well that
in particular forms with the ‘possibility’ reading may shade off into other modalities, as in the
‘necessity’ reading of answerable (as in to be answerable to someone) or the ‘permissibility’
reading of photocopiable (as in ‘legal to photocopy’).
The other typically object-referencing derived adjective is the past participial adjec-
tive, which also has an aspectual nuance in keeping with its participial nature. Specifically,
the adjectival past participle typically has a completive interpretation. So in phrases like
an eaten apple or filtered water, the premodified noun refers to the object of the verb, and
the participial adjective implies a completed action. Transitives predominate (33) and
unaccusatives are possible (34), but unergatives are rare. Where we do find apparently
unergative verbs, they have been coerced to a transitive reading, as (35) illustrates:

New York Times 2008: To better understand their lifestyle and behavior, the Wildlife
(33) 
Conservation Society sent specially trained dogs into the piney woods here recently,
not in search of actual moose, but their scat, or excrement.

Health & Social Work 1993: The findings indicate that a strong sense of personal efficacy
(34) 
was associated with fewer symptoms of depression. However, current refugee
resettlement models have not yet developed training programs to enhance and improve
personal efficacy among recently arrived refugees.
 14.3 Semantic considerations 309

Harpers Magazine 1998: The mirrors’ remains of sebum and pus and sneezed detritus.
(35) 

The only other affix that bears mention in this section is ‑ary, which only occasionally attaches
to verbs. Most of the resulting adjectives are too lexicalized in meaning to merit systematic
analysis, but in the few cases of transparent derivations the resulting words are object-refer-
encing. This is the case, for example with words like documentary (which applies to the thing
documented rather than the documenter); similarly with words like imaginary, salutary, and
perhaps honorary.

Subject-referencing
The affixes ‑ant, ‑ive, and ‑ory, all permit a subject-referencing interpretation, as do ‑al, ‑ous,
and ‑y when they attach to verbal bases, but these affixes seem more fluid in interpretation
than the object-referencing cases discussed above. That is, words derived with these affixes
typically do not have fixed interpretations, but can vary between subject-referencing and
eventive interpretations depending on context. We will return to the eventive interpretation
in the next section.
Examples (36–41) show words derived with these affixes that receive a subject-referencing
interpretation.

Psychology Today 2010: A woman maybe high functioning as mother to a 4-year-old

(36) 
who remains compliant and eager to please, but difficult to a 14-year-old exercising
a teen’s capacity for criticism and opposition—and yet mellow to a 40-year-old,
when her anxieties about a child’s independence or difference may have finally been
resolved.

New York Times 2010: I don’t think the American people last year voted for higher taxes,
(37) 
higher deficits and a more intrusive government.

Analog Fiction & Fact 2001: It’s a wasting away, and microscopic vibratory forms of life
(38) 
can do that just as well as us big beings.

Journal of Drug Issues 2010: Additional research should examine whether the causal
(39) 
factors influencing arrest trajectories differ by gender.

Cosmopolitan 2010: After weeks of hoping Rob would make a move now that he knew
(40) 
how I felt and his alternating between acting awkward and being flirtatious toward
me, I finally got fed up and found another job.

Paul is Undead 2010: That said, George is nearly as honest as John Lennon, albeit he’s
(41) 
less chatty and has a significantly lower level of angst.

It is not difficult to find other examples that have this subject-referencing interpretation,
especially for the suffix ‑ive.
310 Derived adjectives

Participial ‑ing
Participial ‑ing is strongly subject-referencing, although with an interesting twist. On transi-
tive bases like annoy, the adjectival ‑ing participle typically refers to the subject; so an annoy‑
ing neighbor is a neighbour who annoys (as opposed to an annoyed neighbor). Similarly with
unergative or unaccusative verbs, the ‑ing participle is predicated on the subject, as in for
example, a sneezing parakeet (unergative sneeze) or the falling leaf (unaccusative fall). What is
interesting, however, is that with verbs that vary between transitive/causative uses and unac-
cusative/inchoative uses—that is, verbs like open, close, grow, boil, cook—the ‑ing participle is
much likelier to reference the subject of the unaccusative form of the verb, which corresponds
to the object of the transitive form of the verb. So we easily find examples like boiling water or
growing weeds, but the sort of example illustrated in (42) is much harder to come by:

The Fireman’s Fair 1991: She was not a cooking woman, but she liked to be complimented
(42) 
on her shrimp.

The adjectival ‑ing participle has aspectual meaning as well, conveying an ongoing or habitual
event.

14.3.1.2 Event-referencing
Many of the affixes discussed in Section 14.3.1.1, including participial ‑ing, can also receive an
eventive reading, which is to say that adjectives derived with them need not refer to one
argument or the other, but can alternatively focus on the nature of the event denoted by the
verbal base. Put more simply, such derived adjectives may receive interpretations paraphras-
able as ‘characterized by verbing’ or ‘pertaining to verbing’ or ‘involved in verbing.’ (43)–(48)
are characteristic examples.

Professional School Counseling 2008: Traumatized children had more intense play, play
(43) 
disruptions, repetitive play, avoidant play behavior, and negative affect.

Roeper Review 2007: His teacher embedded a variety of explorative activities into her
(44) 
classroom.

Antioch Review 2006: Their machines hum a vibratory song like a gather of insects in
(45) 
unison, the monitors clacking and chirring at the bedsides.

North American Journal of Psychology 2004: Given that no decrement in performance is

(46) 
evident, this suggests that the configural aspects of the other-race faces are not taken
advantage of when the faces are processed.

Kenyon Review 1997: Her eyes, the color of lime slices, always gazed at me with
(47) 
predaceous intensity.
 14.3 Semantic considerations 311

The American Spectator 2009: And growing up, I would hear in him the breezy, chatty
(48) 
style that he must have decided would help him with his customers.

In a sense, the eventive interpretation is very similar to the range of interpretation we find for
adjectives derived on nominal bases, to which we turn in Section 14.3.1.3.

14.3.1.3 Non-argumental
Adjective-deriving affixes that attach predominantly or exclusively to nominal and adjectival
bases are semantically rather different from affixes that take verbal bases, as their meanings
obviously cannot depend on the referencing of base arguments. The affixes we have in mind
here are ‑al, ‑ary, ‑ed, ‑esque, ‑ful, ‑an, ‑ic, ‑ical, ‑ine, ‑ish, ‑ly, ‑oid, ‑ous, ‑some, and ‑y. What is
critical is to determine to what extent specific meanings are to be associated with particular
affixes and to what extent various meanings are dependent on the semantic characteristics of
the base rather than the semantics of the affix. We will begin with the adjective-forming suf-
fixes that are more contentful, specifically ‑esque, ‑ful, ‑ish, and ‑oid, and then look at the
remaining affixes, which are closer to being purely transpositional.
The suffix ‑ful is perhaps the easiest to deal with, as it comes the closest of all these affixes to
having a lexical meaning. Often the simple glosses ‘full of’ or ‘having’ are appropriate; so sorrow‑
ful is ‘full of sorrow’ and insightful is ‘having insight’. Other ‑ful words are also arguably compo-
sitional, although their interpretation is slightly different. For example, forms like dutiful and
effortful are more aptly glossed as ‘displaying duty’ or ‘expending effort’. The difference in inter-
pretation is, however, possibly attributable to the difference in bases, the former two being
formed on nouns denoting psychological states and the latter on more volitional nouns.
The cohort of formatives ‑esque, ‑ish, ‑like, and ‑oid are also relatively contentful, all being what
might be termed similative, often paraphrased as ‘like X, in the shape of X, in the style of X,
resembling X’, depending on the kind of base. The affixes ‑ly and ‑y which sometimes have simila-
tive meanings are dealt with elsewhere. Comparing triplets and doublets with these forms is
instructive. In many cases, there seems to be little or no difference in their semantic content, as
illustrated by the doublets and triplets shown in (49), from COCA, and (50), from the BNC.

(49) a. Washington Post 1991: Babysitter Courtney and Babysitter Skipper are two Barbie‑ish
dolls that come with infants and Walkmans.
Fortune 2003: Most branches include a WaMu “store” that sells Barbieesque teller
dolls for kids and personal[ ]finance books for adults.
Esquire 1998: The Karen Carpenter Story (which is enacted entirely by Barbie‑like
dolls), Poison, and Safe-Haynes is, like all of his movies, smart and good-
looking.

b. Men’s Health 2004: And second, it comes with an iPod‑esque remote, complete
with a color LCD screen, that sends your tunes to any “pod” (shown under remote)
throughout the house.
312 Derived adjectives

USA Today 2005: Receptacle on the dashboard of most versions accommodates

iPod‑ish music machines so you can play your music library through the car
stereo.
Time 2005: The Digital Music System is the sleekest solution available— complete with
a cool iPod‑like handheld controller that allows you to play different songs simultane-
ously in different zones all around your house.
Town and Country 2009: Since opening in December, Benjy’s has been strictly SRO.
c. 
The modernesque look sets the stage for boisterous fun in the upstairs lounge,
where you can enjoy the signature blood-orange margaritas, and downstairs, where
the best choices include tuna tartare on a tempura edamame roll, […]
San Francisco Chronicle 1996: “I always like to do Shakespeare in modernish dress
because I want to put people into clothes rather than costumes,” he explains.
Public Interest 1992: [O]verall, these designs are refreshingly free from formula: no
modernoid lunar landscapes with lollipop lightbulbs, no “Olmstoid” caricatures of
Frederick Law Olmsted’s pioneering work of a century ago.

(50) Bad dreams: Its fur was arranged in punkish spikes.

New Musical Express: […] Gary Lee Conner is the band’s only obvious showman, a
tangle of sweat, hair and punkoid Townshend windmill routines […]

Although, as suggested by these examples, there appears to be no intrinsic semantic differ-

ence in these affixes, this is not to say that individual doublets or triplets cannot display dif-
ferent meanings. Such differences often appear with pairs of ‑ish and ‑like words. We find, for
example, cases like amberish versus amber-like, where the ‑ish adjective refers to the color, and
the ‑like adjective to the substance in the examples below.

Sky and Telescope 2000: With patience you might see that the yellow looks more
(51) 
specifically amberish or apricot orange and that the blue is closest to the ashy blue of
the Atlantic Ocean.
Analog Science Fiction & Fact 2005: Then she turned back suddenly, her hand reaching
into her jacket pocket and emerging with the amber‑like keepsake.

Similarly, it would seem that in a pair like blondish versus blonde-like, the former would refer
to a color, and the latter to a type of person, or in the pair dwarfish versus dwarflike we might
suppose that the former is likelier to pick out the quality of small size, and the latter to refer
to a kind of person. It might seem then that ‑ish is more likely to denote similarity to indi-
vidual salient qualities, and ‑like similarity to a whole. The pair childish versus child-like is
another case of two different meanings having developed for the two forms.
We must reiterate, however, that this is no more than a tendency. There are other ‑ish and ‑like
pairs that do not appear to differ at all, as the examples babyish and baby-like in (52) suggest:
 14.3 Semantic considerations 313

Callaloo 2005: In appearance, Marty is babyish in his features, a sixteen-year-old

(52) 
chubby-cheeked foul-mouthed infant looking out scornfully at the world from a
bonnet of black lustrous hair.
Literary Review 1994: If he made the effort to turn his head toward her he could see her
pudgy hand with its baby-like dimples and three silver rings, the kind they used to sell
at the Central Bus Station, and which Avner used to buy Rachel by weight.

There seems to be even less difference between ‑like and ‑esque in terms of pure denotation.

Inc. 2005: Ostensibly, our destination was Rancho Parsones, as Parsons bemusedly
(53) 
called the chateauesque house he was building, but first we had a few errands to run.
Washington Post 1990: The Trupin house is unmistakably a symbol, sitting vacant on
the beach in Southampton, its chateaulike turrets still encased in scaffolding, but a
symbol of what?

At best we might say that ‑like is the more neutral in connotation than ‑esque, which is some-
what more elevated or academic in style. The suffix ‑oid frequently has a scientific flavour to
it, but otherwise seems semantically not clearly distinguishable from the other similative
suffixes.
We should also say a word about these affixes on adjectival bases. We might be tempted to
attribute a slightly different semantics to ‑esque, ‑ish, ‑like, and ‑oid when we find them attached
to adjectives, as they seem to mean not so much ‘similar to X’ but ‘approximating X’. We
would argue, however, that the first meaning is derived by inference from the second. If we
say something is similar to dull, baptismal, lunar, or modern, the inference is drawn that we
cannot mean exactly dull, baptismal, lunar, or modern but rather must mean something not
exactly the same as those qualities, that is, approximating those qualities. Compare the exam-
ples in (54)–(56), and again (49c) above.

The Southern Review 2004: Lee Ann remembered watching her parents dance at her and
(54) 
Ted’s wedding; they were good, dullish, ordinary people, but they were great dancers.

Houston Chronicle 1995: Eggs rolled over the body to absorb evil, hands placed over the
(55) 
afflicted area, scented and blessed oils, herbs, and a pool of water for baptismal-like
cures are a small sampling of the famous folk healer’s “remedios.”

Omni 1992: Devastation Trail is a boardwalk laid out on glistening jet cinders in a
(56) 
setting that looks lunaresque—if the moon had burned.

From this inference we derive the approximative sense of the similative affixes.
The suffix ‑ed has an ornative sense and derives adjectives with the general meaning ‘hav-
ing X, being provided with X’. Examples abound, and we list only a few here: low-prized,
large-sized, mud-walled, propertied, short-legged, white-robed.
314 Derived adjectives

14.3.1.4 Transpositional
The remaining affixes that attach predominantly to nouns or other adjectives are ‑al, ‑an, ‑ic,
‑ical, ‑ine, ‑ly, ‑ous, ‑some, and ‑y. These are as close as affixes come in English to being purely
transpositional, in the sense that they appear to add no specific meaning beyond what would
be attributed to their categorial status as adjectives. This is not to say that individual derived
adjectives do not exhibit more specific relational meanings, especially in context. As Ljung
(1970) points out, specific derived adjectives might mean (given the right context) ‘covered
in’ (a muddy floor), ‘affected with’ (the furious teacher), or ‘in accordance with’ (a normal tem‑
perature), and the like; indeed, Ljung isolates close to twenty general meanings that can be
expressed by denominal adjectives.
It is clear, however, both from his data and from examples gleaned from COCA that spe-
cific meanings arise from a combination of the base, the affix, and most importantly the
context in which a form appears. The affixes themselves appear to have no fixed meaning
beyond the most general meaning of ‘characterized by’, ‘pertaining to’, ‘relating to’. In this
sense their semantic behaviour is very similar to that displayed by non-argumental com-
pounds, where the relationship between first and second elements is constrained only by the
sort of reasonable relationships that might be imagined between one lexical base and another
(see Chapter 20).
We will illustrate this point in two ways. First we will look at individual forms that may have
several interpretations depending on the contexts in which they occur, thus suggesting that
specific relationships expressed in the derived adjective are to some extent driven by context.
Secondly, we will look at sets of forms that have the same nominal bases, but different adjec-
tive-forming affixes, and that nevertheless, in similar contexts appear to mean the same thing.
We concentrate here on the affixes ‑al, ‑an, ‑ic, ‑ous, and ‑y, since the affixes ‑ine and ‑some
are only marginally productive. We will return to the suffix ‑ical in Section 14.3.3. The affixes
‑al, ‑an, ‑ic, ‑ous, and ‑y and their variants with extenders are all broadly relational in meaning.
As we said above, how we gloss them in individual forms depends heavily on both the nature
of the base and the context in which the derived word finds itself.
We will illustrate first with the native affix ‑y. This suffix is unique among the adjective-
forming suffixes in preferring mono- or disyllabic bases, often concrete mass or count nouns.
In a word like brothy, we might be tempted to gloss the suffix as ‘containing’, for example, but
examples from COCA show that a wider range of interpretations is available in context.

(57) a. San Francisco Chronicle 2005: Much of it is meant to appeal to a more American
palate, but no one would complain over a hot, brothy bowl of vegetarian pho with
tons of still-crisp vegetables.
b. San Francisco Chronicle 2009: A great version of siu mai $3.20/4 features tender
wrappers enclosing a well-seasoned filling with brothy undertones.
The Others 2009: He pushed the bowl toward her, letting her smell the brothy
c. 
steam.
 14.3 Semantic considerations 315

So in (57a) we get the expected meaning ‘containing broth’, but in (57b, c) we get something
more like ‘tasting like broth’ or even ‘smelling like broth’. It seems more sensible to say that
this affix merely transposes the noun to an adjective in which any salient quality of the noun
relevant to a particular context can be focused.
Similarly, we can take a word like rabbinic, and find it used in a number of distinct senses:

(58) a. Lilith 2005: The law says count all the days of your menstruation, then check your
underwear to make sure there are no more leaks, and then count seven more days
till immersion. It’s the rabbinic way, so we do it, no questions asked.
b. Joy Comes in the Morning 2005: In rabbinic school there had always been students
who wrestled with praise and took a[n] attitude toward God, an attitude of human
entitlement and anger.
Houston Chronicle 2004: He knew it well, having been born in 1904 into a rabbinic
c. 
family in Leoncin, one of those close-knit villages—shtetls—in which age-old
strictures held sway over life even as the Enlightenment, which Singer was to
embrace passionately, threatened them.

The word rabbinic can be used to describe the practices of rabbis (58a), a school for rabbis
(58b), and a family composed of rabbis (58c). Clearly we cannot attribute a specific gloss to ‑ic
in this word (much less in others), as the gloss of the derived word is so highly dependent on
context. We could provide similar examples for the other adjective-deriving suffixes under
consideration in this section.
So on the one hand an individual affix can give rise to many different nuances depending
on the base it attaches to and the context in which the derived word is used. On the other
hand, different affixes on the same base can give rise to identical meanings; that is to say that
with this cohort of affixes we find doublets (and occasionally triplets) that do not seem to
differ in meaning.

(59) ‑al and ‑ic

a. NPR_Fresh Air 2004: And the more I started to look into it, the more it made
absolute sense in a real epiphanal, you know, sort of way; I mean, really hitting you
like a ton of bricks.
b. Michigan Quarterly Review 2007: But I don’t want this to be one of those epiphanic
moments when we see ourselves in each other and reconcile.

(60) ‑an and ‑al

a. NPR_Science 2004: Case in point: the complete sequences for microbian genomes,
such as smallpox, anthrax, SARS and Ebola.
b. Science News 2009: Do-it-yourself DNA Making a complete microbial genome from
scratch by assembling the individual letters of its genetic code paves the way for
making synthetic microbes
316 Derived adjectives

(61) ‑ic and ‑an

a. Paris Review 1999: All except Mrs. Minna, who was at her stove, piling together a
cornucopic holiday plate for her older son.
b. Raritan 1995: I was looking for olives that afternoon, which, I had not realized until
then, come in a panoply of sizes and colors, seasoned with a cornucopian array of
flavorings and herbs, and preserved through any number of methods.

(62) ‑ous and ‑al

a. Bring Me the Head of Prince Charming 1991: There, in a shop dedicated to
necrophilious memorabilia, he was fortunate enough to find a pickpocket’s hand
for Princess Scarlet.
b. Habeas Corpses 2005: “It’s a war story, Bubba, not necrophilial porn.”

(63) ‑ic and ‑ous

a. Current Psychology 2002: Despite the employment of antonymic relationships in
previous work (Brownell, Potter, and Michelow, 1984), it is possible that the use of
synonyms might be more appropriate in the present context.
b. Atlanta Journal Constitution 1993: In your letters section of Jan. 6, Mr. Franklin Burke
McMahan seeks a well-mannered word that is antonymous to misogynist.

(64) ‑ic and ‑y

a. Dark City 1998: The two men walk cautiously through the echoic interior.
b. The Answer is Always Yes 2008: He was a few paces past when his voice returned,
borne up from the echoey corridor.

(65) ‑an and ‑y

a. PBS_Newshour 1990: In the center of Leningrad in a red brick dungeon nearly a
century old is the labyrinthy maze of cells and towers known as “The Crosses”, a
famous and infamous landmark never before open to foreign television cameras.
b. Chicago Review 1992: My mother told me that convicts and orphans were once sent
from the mainland to the island to work, and most times die, in the labyrinthian
mines.

It is in fact possible to go pairwise through the suffixes we are concerned with here and in
nearly every case find doublets in which the individual forms have no discernable difference
in meaning, at least in some context, although we will not be exhaustive here.
Again, this is not to say that pairs cannot differ in meaning. Many of them do: as Hamawand
(2007) has shown, there are any number of examples in which the members of pairs have
been lexicalized with specific and very different meanings, as for example in industrious and
 14.3 Semantic considerations 317

industrial, ceremonial and ceremonious, or thunderous and thundery. But these differences in
meaning cannot reasonably be attributed to any semantic contrast between affixes, and there
appears to be nothing systematic about the kinds of differences that get lexicalized or what
gets lexicalized with a specific affix.
Our conclusion is not uncontroversial, however. Indeed, Hamawand (2007) contends the
opposite, on the basis of an analysis of lexicalized words. We do not find his claims convinc-
ing, since we have seen that for unlexicalized, low-frequency forms we frequently do find
synonymy, in spite of what happens in lexicalized derivatives.

14.3.2 Relational versus qualitative, gradable versus

non-gradable
The distinction is sometimes made in the linguistic literature between relational and qualita-
tive adjectives (Beard 1995; Plag 2003, as well as Huddleston and Pullum 2002, who use the
terms associative and ascriptive adjectives rather than relational and qualitative). Relational
adjectives are ones which can appear prenominally, but cannot be modified with a degree
modifier, and cannot be used predicatively. They are non-gradable, and therefore do not
occur in comparative or superlative forms. For example, it is usually said that with relational
adjectives we can have a phrase like a nuclear reactor, but not *a very nuclear reactor, or a sen-
tence like *that reactor is nuclear or *this reactor is more nuclear than that one. Qualitative adjec-
tives, on the other hand, can occur premodifying nouns, but can also themselves take a
degree modifier, can occur predicatively, and can be gradable. It is worth asking, then,
whether adjective-forming affixes might be divided into two cohorts—those that form quali-
tative/gradable adjectives and those that form relational/non-gradable adjectives.
Examination of data available from the corpora, however, suggests that the situation is
somewhat more complex. While it seems safe to say that some of the affixes considered in
this chapter always derive qualitative adjectives, it does not seem to be possible to designate
any affix as deriving exclusively relational adjectives. For example, the similative affixes are
reliably qualitative, occurring easily in predicative position, or modified by degree adverbs, or
in the comparative or superlative:

(66) a. Men’s Health 2001: Also, be aware that your fashion-forwardness is indicated by three
things: the number of buttons (two is traditional, three is contemporary, one is
hipsterish).
b. Critical Matrix 1994: Oscar Wilde, whose argument it is, goes on to say with very
un-Dombeyesque irony that “egotism itself, which is so necessary to a proper sense
of human dignity, is entirely the result of indoor life.”
BNC.The Fraxilly fracas: The spike-skinned exter who had been following me was
c. 
coming through the door, in company with several other exters, none of them very
humanoid.
318 Derived adjectives

Although it has been suggested that some affixes specifically derive relational adjectives (e.g.
Plag 2003: 95–6 for ‑al, ‑ary, and ‑ic), evidence from COCA suggests that this is not invariably
the case. Neither specific affixes, nor even particular words derived with those affixes are
fixed once and for all as relational. Consider, for example, the adjective nuclear, which would
seem at first glance to be solidly relational. COCA yields examples like the following:

(67) a. Newsweek 1997: France is second— 75 percent of French electricity is nuclear,

which has reduced French air pollution fivefold—followed by Russia and Japan.
USA Today 2005: The outspoken Texas conservative, who displays the Ten
b. 
Commandments in his office but admits he has a hard time loving his enemies, declined
to run for House speaker in 1998 because he considered himself “too nuclear.”

These examples suggest that it is possible to coerce just about any relational adjective into a
qualitative reading. This is especially true when we focus on specific salient characteristics
associated with that adjective, as in (67b) where the sense of nuclear that is intended is ‘giving
off dangerous radiation’. We provide additional examples in (68), with the adjectives aborigi‑
nal and dynastic, which would also seem at least superficially to be solidly relational.

(68) a. Washington Post 1996: Consider shamanism, “probably the most aboriginal form of
meditation, and most closely associated with nature,” said Finbarr Lismore, a
Harvard Divinity PhD who spent a summer as the Omega Institute’s “answer man,”
explaining the catalogue to callers.
NPR_TalkNation 2002: And there were the Tafts of Ohio, the Browns of California, the
b. 
Longs of Louisiana, and perhaps the most dynastic, the Roosevelts and the Kennedys.

We conclude that while some adjective-forming affixes can be classified as qualitative, and
others may tend towards deriving relational adjectives, no relational affix is immune to being
coerced to a qualitative reading.

14.3.3 The suffixes ‑ic and ‑ical

Given the general neglect in the linguistic literature of adjective-forming derivational affixes,
there is a surprisingly large literature devoted to determining whether the suffixes ‑ic and
‑ical can be distinguished in any systematic way. Kaunisto (2007) reviews most of this litera-
ture, first noting as we have done above, that there are many bases that show forms in both
‑ic and ‑ical (comic ~ comical, periodic ~ periodical), as well as many bases that only evidence
one or the other (public, atomic; radical, theatrical) (Kaunisto 2007: 4–5). In his review of
the literature, he suggests that much of the discussion of this pair, from Marchand (1969), to
more recent studies like Marsden (1985) and Gries (2001, 2003) has taken pains to find
systematic distinctions, but has not been terribly successful. Interestingly, all of these
 14.3 Semantic considerations 319

studies, including Kaunisto’s, have concentrated on a limited number of mostly item-familiar

and high frequency pairs (classic versus classical, economic versus economical, historic versus
historical, and the like); the search has been more to see whether consistent patterns of lexi-
calization can be found than to see whether there is any intrinsic semantic difference
between the two suffixes.
The latter is, of course, the question that more interests us. COCA gives us ample oppor-
tunity to consider this issue. It is possible to find many examples in the corpus of ‑ic and ‑ical
pairs where neither form is item-familiar. For example, with pairs like calendric and calendri‑
cal, cryptologic and cryptological, hematologic and hematological, herpetologic and herpetological,
toponymic and toponymical, neither item is particularly frequent, and the contexts in which
they are used suggest that there is no semantic distinction between them.

(69) a. Geographical Review 1991: Moreover, they are built on complex calendric systems
analogous to those of neighboring regions.
b. Natural History 2000: You see, poor Dionysius Exiguus (Dennis the Short), the
sixth-century monk who devised the B.[C].–A.D. calendrical system, made a little
error in setting Christ’s birth.

(70) a. Foreign Affairs 1991: Though the cryptanalysts followed these developments in the
newspapers, their inspiration came less from the pressure of events than from
fascination with the cryptologic problem.
b. Mercury 1999: Astrophysical coding means that a message is designed to yield
information through an astrophysical-type analysis of the signal, rather than a
cryptological analysis.

(71) a. ENT: Ear, Nose & Throat Journal 2009: While the diagnosis of this tumor has become
more accurate, especially in patients who have a history of hematologic
abnormalities, it is important to keep this rare entity in mind when faced with
poorly differentiated neoplasms of the head and neck in order to provide prompt
and appropriate therapy.
b. Internet Journal of Rheumatology 2007: Hematological abnormalities included
eosinophils of 7.6% (laboratory normal 0 to 5%).

(72) a. Washington Post 2000: Deaths from venom are rare even in the herpetologic
wonderland of Arizona, where about a dozen people have been killed by rattlesnakes
in McNally’s two decades at the poison center.
b. Houston Chronicle 2001: The sprawl of Greater Houston encompasses much of this
herpetological Eden.

(73) Geographic Review 2001: The standard biographical accounts are those of T. L. Venkatarama
a. 
Aiyar (1968) and R. Raghavan (1975), who also provide toponymic information.
320 Derived adjectives

b. Hispanic Review 1996: Rey understands the toponymical misnomers not as lies but
as rhetorical ironies precisely because he can see for himself that what they name,
describe, or represent is far from their literal meaning.

We might conclude from such examples that whatever their historical source, in contempo-
rary English ‑ic and ‑ical are alternate realizations of the same affix. Some bases will appear
exclusively with one or the other of the suffixes. Where either suffix is to be found on a
single base, lexicalization may of course take place, although the studies cited above suggest
that there is nothing systematic in this process. But it is also possible for pairs to co-exist—
especially for low-frequency forms—with no differentiation.

14.3.4 Multiple affixes and semantic interpretation

We return finally to a curious observation that we made in Section 14.2.1.1, namely that it is
sometimes possible to find forms which exhibit more than one adjective-forming suffix on a
single base; for example, we can find items with the sequences ‑or‑ial (accusatorial, combina‑
torial, gustatorial, improvisatorial), ‑oid‑al (arachnoidal, cuboidal, ovoidal, planetoidal), ‑iv‑al
(conjunctival, gerundival), ‑al‑esque (brutalesque, industrialesque, lunaresque), ‑an‑esque (afri‑
canesque, americanesque, suburbanesque), ‑ar‑ian (equalitarian, millenarian, sectarian, uniformi‑
tarian). For some of these combinations there seems to be a semantic distinction: the
adjective industrial, for example, merely means ‘related to industry’, whereas industrialesque
seems to mean something like ‘industrial in style’.

Washington Post 1994: Actually I, too, sat there, ravenously hungry—on my own
(74) 
unbelievably uncomfortable, spasm-inducing metallic industrialesque chair—taking
teeny-weeny itsy-bitsy bites and, every now and again, nodding sagely.

A similar contrast can be found with forms like suburban versus suburbanesque. But there
seems to be no discernable difference between, for example, spheroid and spheroidal, or gusta‑
tory and gustatorial in the following examples.

Associated Press 2009: It has been known through astronomy that spiral disk galaxies are
(75) 
enclosed by a spheroid halo of dark matter with a mass about ten times greater than the
ordinary mass of the spiral galaxy and also that a weak magnetic field pervades the universe.
Astronomy 2008: Milkomeda’s spheroidal shape is not unusual, as it characterizes a
major class of objects called elliptical galaxies.

San Francisco Chronicle 1996: Prepared by the people who put on the Peninsula’s bimonthly
(76) 
Decadent Dinner Parties, EEET!, the so-called Environmental Eating Education Team,
this gustatory experience will be a veritable vegetarian pig-out, to mix metaphors.
 14.3 Semantic considerations 321

Washington Post 1996: But in gustatorial matters, there’s really only one bottom line:
Maybe they just taste good.

This is not to say that there is never a difference between a form in ‑ory and a corresponding
form in ‑orial, or between one in ‑oid and ‑oidal, but merely that there often is not. Why we
get this iteration of affixes is an open question, however.
Another frequent combination of affixes that we should point out is ‑istic. Here we appear
to have nothing unusual in terms of formal considerations: ‑ist forms nouns, and ‑ic attaches
to nouns to form adjectives. Indeed, some forms in ‑istic are also unremarkable in terms of
semantics, for example, atheistic and pacifistic, which could plausibly be interpreted as ‘per-
taining to atheists’, ‘pertaining to pacifists’. But there are many forms in ‑istic in which the
suffix ‑ist appears to make no semantic contribution to the derived word at all. Consider, for
example, forms like cannibalistic or charlatanistic, which must mean ‘pertaining to cannibals’
or ‘pertaining to charlatans’, there being no plausible corresponding forms in ‑ist. Marchand
(1969: 295) suggests that such forms are actually derived from forms like cannibalism or char‑
latanism, but given other examples of semantically empty derivations like those above, this
analysis does not seem completely convincing to us, especially as there are also forms like
analogistic that lack both the agentive ‑ist form and the corresponding ‑ism form.
 chapter 15

Derived adverbs

15.1 Prospectus
In this chapter we look at the ways in which adverbs are formed in English. The very fact of
this chapter, not to mention its title, implies at least two theoretical assumptions, namely that
adverbs are a separate category from adjectives, and then that the affixation involved in creat-
ing adverbs is a matter of derivation rather than inflection. Both of these are controversial.
We will take up the categorial issue first, and then consider the specifics of adverb-forming
processes in English. Although the formation of adverbs is far less central to English word-
formation than the other major types we have considered, there are nevertheless interesting
questions which are raised in this chapter, especially considering the phonological condition-
ing of the ‑ly affix, the analysis of forms like ‑ways, and the status and semantics of ‑wise.

15.2 Are adverbs and adjectives

the same category?
Scholars have long debated the categorial status of adjectives and adverbs. Payne et al. (2010)
trace the idea that adverbs and adjectives constitute a single syntactic category to Kuryłowicz
(1936), with Lyons (1966) being the first to advocate for a single category for English. Debate
has continued in the work of Emonds (1976), Jackendoff (1977), Radford (1988), Bybee
(1985), Zwicky (1995), Plag (2003), Baker (2003), and Giegerich (2012) with cogent argu-
ments both for and against a distinction between the two categories. A further, and to some
extent orthogonal issue is whether the adverb-forming affix ‑ly is to be classed as inflectional
or derivational, an issue to which we will return in Section 15.3.1.
The classic argument for treating adjectives and adverbs as a single category in English is that
they appear to occur in complementary distribution: adjectives occur as modifiers to nouns,
adverbs as modifiers to other categories. The argument goes that if there is no overlap in distribu-
tion, adjectives and adverbs can be treated as positional variants of a single category. Consequently,
Payne et al. (2010) argue at length for a categorial distinction in English claiming that adjectives
and adverbs do overlap in several environments (for example, post-head nominal modifiers,
premodifiers to adjectives), and therefore are not strictly in ­complementary distribution.
 15.3 Suffixation 323

Giegerich (2012), however, points out that distributional overlap can sometimes be inter-
preted as free variation rather than complementary distribution: just as allophones of a
phoneme can sometimes be in free variation, adjectives and adverbs might occasionally be
found in the same environment, as with the examples in (1) from Payne et al. (2010: 52),
without requiring us to postulate distinct categories.

(1) a. shortages both nationally and internationally of these metals

b. shortages both national and international of these metals

Giegerich also argues against overlapping distribution in the preadjectival context on the
grounds that items like blindly drunk, madly keen are phrasal, whereas blind drunk or mad
keen are compounds. If so, the distributional context is not in fact the same. Giegerich
himself offers morphological and phonological reasons for treating adverbs—specifi-
cally ‑ly adverbs—as inflectional variants of adjectives, a subject to which we will return
shortly.
We will remain agnostic on the subject here, as we do not find the arguments conclusive
one way or the other. Our decision to treat adverbial derivation separately from adjectival
derivation should therefore be taken as a practical if not a theoretical move.

15.3 Suffixation
15.3.1 The suffix ‑ly
The suffix ‑ly is one of the most productive suffixes in English, added more or less freely to
any adjective to create a corresponding adverb. Historically, the suffix ‑ly appears to be the
same form as the adjective-forming suffix in words like neighborly, motherly, discussed in
Chapter 14. In most instances this common origin is no more than accidental homophony in
contemporary English.
Adverbial ‑ly has sometimes been argued to be inflectional rather than derivational.
Haspelmath (1996) treats ‑ly as category-changing inflection, largely on the grounds of its
productivity and regularity. This is not necessarily a strong argument, though, as there are
other affixes that are highly productive and semantically transparent (for example ‑ness) that
are never claimed to be inflectional.
Giegerich (2012) gives several arguments that ‑ly adverbs are inflected forms of the corre-
sponding adjective. First, he notes that the ‑ly affix in such adverbs cannot be followed by any
other affixes, either derivational or inflectional. It is characteristic of inflectional endings
that they do not allow further affixation in English. Giegerich draws special attention to the
fact that comparative ‑er and ‑est do not occur on ‑ly adverbs, although they can occur on
adjectives that end in ‑ly (likelier, likeliest) or on other adverbs (sooner, soonest). Picking up an
324 Derived adverbs

argument from Plag (2003: 195), Giegerich also suggests that the absence of any lexical
­meaning for ‑ly supports its status as inflection.
Giegerich’s first argument is the more persuasive of the two; even derivational affixes like
‑dom that do not allow other derivational affixes to attach to them do allow the appropriate
inflection at least (kingdoms, for example). But we do not find the argument decisive. Note
that although ‑ly adverbs cannot take the inflectional suffixes ‑er and ‑est, they do form peri-
phrastic comparatives and superlatives (more quickly, most swiftly). The comparative and
superlative are among the few inflectional categories in English that alternate with a peri-
phrastic form (the genitive is the other), and as we have seen in Chapter 6, the conditions on
the use of the affixal versus periphrastic forms are intricate, so the inability of ‑ly adverbs to
take the affixal comparative and superlative might be attributed to other factors. Indeed,
Zwicky (1989: 157) points out that there are a number of adjective-forming affixes that
together with a monosyllabic base would fit the phonological profile necessary for the affixal
comparative and superlative, that nevertheless do not allow ‑er and ‑est: ‑ive (*activer, more
active), ‑al (*brutaler, more brutal), and for the most part ‑ish (*greenisher, more greenish;
COCA provides one exception foolisher). If in contemporary English the adverb-forming ‑ly
has become a distinct affix from the adjective forming ‑ly (as implied by the existence of two
separate entries for ‑ly in the OED), there would be no problem in conjecturing that adjective-
forming ‑ly patterns like ‑ive, ‑ic, and ‑ish, whereas adverb-forming ‑ly does not.
Giegerich’s second argument is also unconvincing: there are many derivational affixes that
are close to purely transpositional and therefore cannot be said to carry lexical meaning (e.g.
the ‑al that forms adjectives from nouns, see Chapter 14, or nominalizing suffixes like ‑ation
or ‑ment, see Chapter 10).
A final reason to be sceptical of the inflectional analysis of ‑ly is one that we will return to
shortly, namely that there are other affixes that derive adverbs and that adverbs are derived
from other categories than adjectives, contrary to what Giegerich (2012: 8–9) suggests (see
Sections 15.3.2–15.3.5). But there is no reasonable evidence that these might be inflectional as
well. In other words, whatever our ultimate decision about the status of ‑ly, we must still
acknowledge the existence of derived adverbs in English.
As we have mentioned, ‑ly is highly productive, although there are two major ways in
which this productivity is constrained, one apparently phonological, one that looks like
blocking. In the first instance, there is a great reluctance to add ‑ly to any base which already
ends in <ly>. There are several subcases to be distinguished. The first is where the <ly> is
itself a suffix; here we can distinguish between the ‑ly which is added to nouns denoting time
periods (daily, monthly) and the ‑ly which creates adjectives from person nouns (friendly,
womanly) or other bases (courtly, leisurely, comely). Second is the case where the final <l> of
the base is followed by a suffixal ‑y (oily, steely, woolly). Third is the case where the base itself
ends in the orthographic sequence <ly>, but this has no morphemic value (holy, silly).
The time-period ‑ly does not allow a subsequent ‑ly; there is no *hourlily, *dailily, *weeklily,
*fortnightlily, *monthlily, *yearlily. The OED allows the base forms as adjectives or adverbs:
 15.3 Suffixation 325

hourly, daily, weekly, fortnightly, monthly, yearly, and we can find examples of such forms in the
corpora.

Bioscience 2011: Pattern of carbon dioxide production and retention is similar in adult
(2) 
pigs when fed hourly, but not when fed a single meal.
Popular Mechanics 2011: I take the inspection cover off my gas furnace yearly and look

inside.

The adjectival suffix ‑ly, which attaches to person nouns, does not generally allow a subsequent
‑ly: *gentlemanlily, *motherlily, *neighbourlily, *womanlily. The OED does list friendlily, with cita-
tions from the nineteenth century; it appears once in COCA and not at all in the BNC.

New Yorker 2006: “It’s not an intrusion I hope?” she apologized when she called on them,
(3) 
and was friendlily reassured.

Turning to other adjectival formation in ‑ly, we find a variety of bases of sometimes unclear
provenance (e.g. courtly, leisurely, comely, earthly, gainly, kindly, lonely, seemly, worldly). The
OED lists some of these with an adverbial ‑ly: kindlily, lonelily, and worldlily (this last with a
twentieth-century citation) and seemlily as obsolete. In many cases, adjectival derivatives in
‑ly are used exclusively as adjectives by some speakers and as adverbs (possibly in addition to
their use as adjectives) by others, as illustrated in (4), from COCA.

Jenny Shank, 1976: the Police Athletic League was to introduce neighborhood children
(4) 
to what decent and friendly guys police officers could be
Literary Review 2004: They talked so friendly, he couldn’t believe them
Fantasy and Science Fiction 2010: the boys picked up their speed, their water buffalo

following them at a more leisurely pace.
Brett Battles 2011: Kolya drove the sedan leisurely down the street.

Another set of potentially problematic words is adjectival formations in ‑y whose bases end
in /l/ (such as crumbly, curly, freckly, frilly, growly, hilly, oily, smelly, spindly, squally, treacly,
wobbly). Generally there is no adverb form, either with or without the ‑ly.
To items like holy and silly we can add early, jolly, (un)ruly, surly, ugly. These have entries for
the adverb in ‑ly in the OED, though earlily and (un)rulily are marked as obsolete. The others
have nineteenth-century citations.
From all such possible words, the BNC has attestations only of the following (and then
with few attestations): beastlily, ghostlily, jollily, oilily, seemlily, surlily, uglily, and COCA adds
only friendlily, holily, sillily.
In other words, it appears that users tend to avoid ‑lily endings, whatever their source. The
repetition of a ‑ly affix is the least favoured form, the addition of ‑ly to a monomorphemic
326 Derived adverbs

base seems to be the type which is easiest to use. The fact that we can distinguish the various
subcases seems to indicate that speakers have, at some stage or other, made use of the differ-
ences in morphological structure, but today the phonological structure seems to be the big-
gest determinant of the constraint on these forms. The phenomenon could thus be viewed as
a case of haplology (see Chapter 9.5).
The second constraint on the productivity of adverbial ‑ly might look, at first glance, like
blocking, although closer scrutiny argues against this position. Here we have in mind the
simple adjectives which form adverbs by conversion (see Chapter 24). In at least some of
these cases the converted adverb has all the functions that a ‑ly adverb would be expected to
have. For instance, for many speakers corresponding to a fast car we have he drove fast, but
there is no *fastly. There is no strict blocking here, however: the OED, COCA, and the BNC
all illustrate fastly, as in the following example from COCA.

FIC_Cullum, Paul 1994: “My point,” said Poppy, fastly becoming exasperated, “is that he
(5) 
who hesitates is lost.”

In the case of several other simple adjectives that can undergo conversion, there is wide-
spread free variation between the adverb with and without ‑ly; among these are fair, loud,
quick, slow, soft.
We find a few other cases in which there might appear to be blocking. In some, there is an ‑ly
adverb but it is not synonymous (hard and hardly) or does not reflect the central meaning of
the base (bare and barely); potentially transparent forms appear to be avoided in these cases.
Apart from these two constraints, ‑ly is widely productive, occurring on simple bases,
native and foreign, suffixed bases, native and foreign, and compound bases. A few examples
from the hundreds in COCA are given in (6).

(6) 
abdominally, abjectly, accordingly, airily, angelically, anthropologically, artfully,
barefootedly, believably, belittlingly, bloodthirstily, bluntly, boisterously, boldly, callously,
complexly, compliantly, coquettishly, dearly, drolly, entirely, fluorescently, Freudianly,
goodnaturedly, gruesomely, hagiographically, hamhandedly, indicatively, insularly

It is perhaps worth commenting specifically on the combination of ‑ic and ‑ly. Although the
preferred orthographic pattern is to use ‑ically rather than ‑icly, the corpora show that this is
not always done, and the OED provides plenty of historical support for ‑icly forms. The
only word in which ‑icly is commonly found in COCA and the BNC is publicly. Most
others could, from the point of view of the standard, be viewed as errors, but there are
sufficient examples to suggest that such a dismissal would oversimplify the actual ­situation.
The pronunciation /ɪkli/ is clearly common, even when the <ically> spelling is used.
In terms of semantics, the suffix ‑ly covers a broad range of meanings. As modifiers of VPs,
‑ly adverbs can express manner, degree, means, duration, frequency, or other temporal rela-
tion (using the categories suggested in Pullum and Huddleston 2002: 576), the precise
 15.3 Suffixation 327

reading of the adverb to a large extent depending on the meaning of the base to which it is
attached.

(7) a. Manner:
 The Cereal Murders 1993: Her golden hair curled angelically around her diminutive
heart-shaped face.
b. Means:
 Technology Teacher 2007: PGD is the process by which an embryo is microscopically
examined for signs of genetic disorders.
c. Degree:
 Good Housekeeping 2011: Paltrow is the first to admit she doesn’t have all the answers
on how her doubly famous marriage is affecting the children.
d. Duration:
 Humanist 2011: Retrospectivists see the past as a golden age to be eternally
reverenced rather than improved upon.
e. Other temporal:
 Science News 2010: A study of people who habitually run barefoot shows that these
runners’ feet strike the ground in a way that tempers impact forces and smooths the
running movement [. . .]

As clausal modifiers, ‑ly adverbs can designate a domain or a modality, provide evaluative
information, or designate a particular kind of speech act (categories again from Pullum and
Huddleston 2002: 576).

(8) a. Domain:
 To Climb a Flat Mountain 2009: “Geologically speaking,” she said, “we’re on a lava
field between two volcanoes.”
b. Modality:
 New York Times 2011: This is possibly the next phase of evangelical Christianity’s
muscle flexing.
c. Evaluation:
 Academic Questions 2002: Evidently, and hearteningly, not everyone feels that the
significance of an honorary doctorate has been irredeemably devalued, or that the
situation is beyond redemption.
d. Speech act related:
 Blood: Butchers Tale 2007: Confidentially, I think Tony is behind this stealing
spree.
328 Derived adverbs

15.3.2 The suffix ‑s

The suffix ‑s forms adverbs from a wide variety of bases, including apparently bound bases, as
indicated in (9). The lack of any strict pattern here probably reflects a lack of productivity.

(9) besides, betimes, downstairs, hereabouts, indoors, midships, nowadays, ofttimes,

outdoors, overseas, perhaps, sometimes, thereabouts, unawares, upstairs, whereabouts,
widdershins

Examples of ‑s in ‑wards and ‑ways have been omitted in (9), and will be considered below.
This ‑s suffix appears to be totally unproductive now unless part of ‑wards or ‑ways.

15.3.3 The suffix ‑ward(s)

The suffix ‑wards is used productively to create directional adverbs from prepositions, direc-
tional adjectives, and nouns. In (10) standard adverbs of this type are listed.

(10) a. prepositional bases: afterwards, backwards, downwards, inwards, onwards,

outwards, upwards
b. directional adjective bases: eastwards, leftwards, north-westwards, northwards,
rearwards, etc.
c. nominal bases: homewards, landwards, leewards, sidewards, windwards

The forms towards and forwards appear to be formed on prepositional bases, although they
are semantically opaque. The former is also rather odd, being a preposition more than an
adverb. Strikingly, each of these words has a potential synonymous counterpart in ‑ward
without the final ‑s. Mostly, the ‑ward form can also be an adjective (a backward glance, a
downward motion, the inward eye, a westward direction). Afterward is unusual in not having this
adjectival use.
The suffix is quite productive on nominal bases, as the following less established forms
from the BNC suggest.

(11) 
bedwards, ceilingwards, choruswards, citywards, deathwards, earthwards,
equatorwards, floorwards, goalwards, heavenwards, hubwards, Islingtonwards,
officewards, polewards, rimwards, roofwards, seawards, shorewards, skywards,
sternwards, Tyne-wards

Interestingly, while ‑ward(s) typically attaches to nouns that express locations, it can turn just
about any noun into a directional adverb: in addition to deathwards in (11) we find examples
like ankleward and bloodward in COCA.
 15.3 Suffixation 329

The difference between the ‑ward form and the ‑wards form is sometimes assumed to be
dialectal, aligning with the difference between NAmE and BrE. This is an oversimplification.
Not only are the ‑wards forms used in NAmE, but the ‑ward forms are also used in BrE. We
can only say that there is variation between the forms.

15.3.4 The suffix ‑ways

In (12) we list adverbs in ‑ways from the BNC.

(12) always, anyways, arseways, cornerways, edgeways, leastways, lengthways, longways,

sideways, widthways

COCA adds the words in (13).

(13) aways, backways, contrariways, crabways, crossways, elseways, endways, frontways,

mostways, noways, slantways, someways

Most of these words are established, but it seems there is some limited possibility of produc-
tive usage; crabways, elseways, and mostways, for example, are not recorded in the OED.

Fantasy & Science Fiction 2010: His hair has long gone, he must wear sunglasses to
(14) 
protect his bleary eyes, and he shuffles hunched and crabways.
Analog Science Fiction & Fact 2004: Of course their god exists, elseways the world

wouldn’t be here, so the fact that the world is here proves that their god exists.
Red River 2008: I was there, watching, like all the women done, up close some of the

time but mostways from a distance.

The last of these seems to be an attempt to imitate some kind of dialect, but the others are
not.
Of these anyways is unusual in having a corresponding anyway. While there is a no way, it
is a phrase and not synonymous with noways; away is not clearly semantically related to
aways. Some of these words can be used as adjectives (a sideways glance), but the adverbial
use appears to be primary. This means that the ‑ways affix is probably not analysable, as the
‑wards one is, and the final <s> has no morphemic value of its own.

15.3.5 The suffix ‑wise

The suffix ‑wise is very productive where there is a noun base (15), less so with other types of
base (16). The examples here come from COCA.
330 Derived adverbs

(15) 
accountingwise, actorswise, agewise, appearancewise, babywise, chocolatewise,
clockwise, columnwise, computerwise, crabwise, engineeringwise, exercisewise,
facultywise, featurewise, flavoringwise, informationwise, lookswise, marketwise,
minoritywise, moneywise, perceptionwise, plantwise, presswise, pricewise, profitwise,
ratewise, spacewise, specwise, speedwise, trafficwise, weightwise, workwise

(16) 
anywise, contrarywise, defensivewise, fitwise, leastwise, likewise, mentallywise,
otherwise, palewise, scientificwise, socialwise

Some of the ‑wise forms seem to form direct competitors to forms in ‑ways, including the
following from COCA:

(17) 
anywise, contrariwise, cornerwise, edgewise, endwise, lengthwise, sidewise,
slantwise

In some cases the forms in ‑ways and ‑wise seem to be all but interchangeable.

National Review 2002: Contrariwise, suppose you are right. What do we have to lose
(18) 
by failing to reform?
Cut Quick 1993: Contrariways, he’s always begging off folks or pestering them to buy

his wares.

It is notable that bases may be monomorphemic, as most of the examples above, or inflected
(actorswise, numberswise), derived (defensivewise, employmentwise) or neoclassical com-
pounds (technologywise). We have no examples with native compounds or phrases, although
it is not clear that this is a principled exclusion.
Adverbs in ‑wise can have a number of basic meanings, according to Dalton-Puffer and
Plag (2000: 238–9). On the one hand, ‑wise forms what they call manner/dimension adverbs
that can roughly be glossed as ‘in the manner of X’ or ‘in an X spatial relation to’. We give
examples of the former in (19a) and the latter in (19b).

(19) a. American Scholar 2004: The man clung to a spur of rock near him and moved slowly
frogwise his green legs in the deep jelly of the water.
African Arts 2004: And sure enough, when girls are given by their fathers at the

Ebuhleni citadel’s festival betrothals of affianced couples, bride confers on groom
the token of her virginity, the Nazarite version of this ucu, a long twined double-
string of pure white new beads, worn wound several times around the neck, then
sashwise down to a folded handkerchief pendant on the hip or the injobo.
b. Art Bulletin 1992: The artist’s conception of the event, spread beltwise across the
Sistine Ceiling, anticipates by some eighty years a major theme of late Renaissance
and Baroque poetry.
 15.4 The prefix a‑ 331

Hostage Zero 2010: Rough-finished beige bricks dominated the walls, arranged edgewise

in horizontal courses that rose from the brown tiled floor to the acoustic tiled ceiling.

A more common reading is what they call the ‘viewpoint’ reading, which can be glossed
roughly as ‘from the point of view of X’.

Bazaar 2005: Fashionwise, that decade was heralded by the explosion of such designers
(20) 
as Helmut Lang and Martin Margiela and Jil Sander.
San Francisco Chronicle 2009: “I may be the one with the offbeat career, but

personalitywise, it was the eldest of our five siblings who were the most rebellious,”
he says.

Dalton-Puffer and Plag (2000) suggest that the viewpoint ‑wise adverbs, although displaying
a smaller number of tokens in the BNC than the manner/dimension ones, appear to provide
the larger number of hapaxes, which suggests that this type of formation is actually the more
productive of the two. Our scrutiny of the COCA data corroborates this suggestion.
The morpheme wise can also be used in forming adjectival compounds, where it has
roughly its expected lexical meaning; examples like streetwise in the sense ‘being aware of
how to behave in a city’ are not hard to find. Businesses frequently capitalize on the ambigu-
ity between the adjectival compound form and the viewpoint ‑wise adverb in creating names
for services or products. For example, pricewise is most likely to mean ‘in relation to price’ but
could mean ‘wise about prices’. COCA gives us examples of this type like Travelwise (pre-
sumably a travel agent), Medwise (a chain of medical clinics), Moneywise (the name of a televi-
sion programme), and Worldwise (a company that makes recycled products).

15.4 The prefix a‑

The prefix a‑ is used to form some adverbs. Historically this is often a reduced form of the
prefix on‑, although on‑ is no longer used in this sense. Given its origins, it is not surprising
that this prefix is most often locational in sense, although it sometimes adds an aspectual
flavour of ongoing action, as is illustrated in the COCA example in (21).

Washington Post 1996: On one court, Marla O’Hara, 35, kicks the sand and stalks the
(21) 
court in a leonine fashion, a long mane of brunette hair afly.

Many of the forms in a‑, however, are highly lexicalized in meaning, as exemplified in (22).

(22) aback, abeam, abed, aboard, about(s), abreast, abroad, abuzz, adrift, afar, afield, afire,
afloat, afly, afresh, agape, aglow, agog, aground, ahead, ahold, ajar, akin, alike, alive,
332 Derived adverbs

aloft, alone, along, aloof, aloud, amid, amidships, amiss, anew, apace, apart, apiece,
aquiver, aright, around, ashore, aside, askance, askew, aslant, asleep, asprawl, astern,
astir, astray, atop, away, aweigh, awhile (often written as two words), awry

As a prefix, a‑ may have some residual productivity on nouns and verbs, but only in poetic
styles. The OED cites amove, apant, asmirk; COCA gives us afly, exemplified above, which is
not also attested in the OED.
In some instances the same form may be an adjective and an adverb: abroad, acrawl, afloat,
away, etc. In some instances it may be difficult to tell whether the word is an adverb or an
adjective, even in context.

15.5 Conversion
As discussed in Chapter 24 and repeated above, there are instances of conversion between
adjective and adverb.
In cases like fast, hard, loud, quick, soft it seems that the adjective is the base and the adverb
is the converted form. One argument for assuming that conversion goes from adjective to
adverb, rather than the other way around, is that all adverbial forms of this sort have corre-
sponding adjectives, but it is not the case for every adjective that there is a corresponding
adverb formed by conversion. See Chapter 24 for further discussion.
In instances with ‑ward it is less clear which form is the base and which is the derivative.
However, we can build up a case that the adverb is primary. In etymological terms, the adverb
comes first. At least afterward is an adverb but not an adjective; it is not clear that there are
any the other way round. The adverbs appear to be more common than their corresponding
adjectives in the BNC. The OED does not list the ‑wards forms as adjectives, but they seem
to have such uses, as, for instance, in (23) from the BNC. This implies that such extensions
are recent.

(23) W_fict_prose 1991/2: you go faster and faster in an upwards direction

The instances with the prefix a‑ and the suffix ‑ways seem to be primarily adverbs with the
adjective as the converted form.
There are also instances where there is conversion involving prepositions. This is perhaps
not surprising, since there is a class of adverbs which is sometimes treated as intransitive
prepositions (Huddleston 2002a: 272). Examples include abreast, amid, around, besides,
toward(s) and the OED lists several other ‑ward forms (downward, inward, upward, etc.)
which used to be prepositions. There does not appear to be any productive process involved
here.
 chapter 16

Locatives of time and space

16.1 Prospectus
In this chapter we survey the affixes used in English to express spatial and temporal
concepts. All such affixes are prefixal, and many of them do double duty with spatial or
temporal readings depending on context. English has two cohorts of such prefixes, non-
native ones and native ones, and there are generally pairs (or sets) that overlap in mean-
ing, with only subtle differences apparent. Several of the prefixes (super‑, sur‑, extra‑,
over‑, under‑) mix spatial readings with quantitative (or at least scalar) readings, a con-
nection we will explore in Section 16.3 (see also Chapter 18). With few exceptions
(Williams 1992; Southerland 1994; Lehrer 1995; Bauer 2003a; Börger 2007), these pre-
fixes have largely been ignored in the literature on English word-formation. We begin in
Section 16.2 with their formal characteristics and go on to semantic considerations in
Section 16.3.

16.2 Formal considerations

The non-native prefixes that convey spatial and temporal meanings in English are ante‑,
arch‑, circum‑, cis‑, co‑, contra‑, counter‑, epi‑, ex‑, extra‑, inter‑, intra‑, meta‑, neo‑, para‑,
peri‑, post‑, pre‑, proto‑, retro‑, sub‑, super‑, supra‑, sur‑, and trans‑. The native locatives are
largely coextensive with prepositions, and we will discuss our reasons for treating them
as prefixes rather than as the first elements in compounds in Section 16.2.2. These include
after‑, back‑, by‑, down‑, fore‑, in‑, mid‑, off‑, on‑, out‑, over‑, and under‑. Table 16.1 sum-
marizes pertinent formal information including the syntactic categories of base the pre-
fix attaches to (N = noun, V = verb, A = adjective, ✓ indicates attachment to the relevant
category), whether the affix attaches to bound bases, to native or non-native bases, and
to compounds or phrases. In the following sections we will discuss each prefix in more
detail.
334 Locatives of time and space

Table 16.1 Formal characteristics of locative prefixes

N A V Bound base Native Non-native Compound Phrase

after‑ ✓ ✓ ✓ ✓
ante‑ ✓ ✓ ✓ ✓ ✓ ✓
arch‑ ✓ ✓ ✓ ✓ ✓
back‑ ✓ ✓ ✓ ✓
by‑ ✓ ✓ ✓ ✓ ✓
circum‑ ✓ ✓ ✓ ✓ ✓
cis‑ ✓ ✓ ✓
co‑ ✓ ✓ ✓ ✓ ✓ ✓ ✓
contra‑ ✓ ✓ ✓ ✓ ✓ ✓
counter‑ ✓ ✓ ✓ ✓ ✓ ✓
down‑ ✓ ✓ ✓ ✓ ✓
epi‑ ✓ ✓ ✓ ✓
ex‑ ✓ ✓ ✓ ✓ ✓ ✓
extra‑ ✓ ✓ ✓ ✓ ✓
fore‑ ✓ ✓ ✓ ✓
in‑ ✓ ✓ ✓ ✓ ✓
inter‑ ✓ ✓ ✓ ✓ ✓ ✓
intra‑ ✓ ✓ ✓ ✓
meta‑ ✓ ✓ ✓ ✓ ✓ ✓
mid‑ ✓ ✓ ✓ ✓ ✓ ✓
neo‑ ✓ ✓ ✓ ✓ ✓ ✓
off‑ ✓ ✓ ✓ ✓ ✓
on‑ ✓ ✓ ✓ ✓ ✓ ✓
out‑ ✓ ✓ ✓ ✓ ✓ ✓
over‑ ✓ ✓ ✓ ✓ ✓
para‑ ✓ ✓ ✓ ✓ ✓ ✓
peri‑ ✓ ✓ ✓ ✓ ✓
post‑ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
pre‑ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
proto‑ ✓ ✓ ✓ ✓ ✓ ✓
retro‑ ✓ ✓ ✓ ✓ ✓ ✓ ✓
sub‑ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
super‑ ✓ ✓ ✓ ✓ ✓ ✓
supra‑ ✓ ✓ ✓ ✓
sur‑ ✓ ✓ ✓ ✓ ✓ ✓
trans‑ ✓ ✓ ✓ ✓ ✓ ✓ ✓
under‑ ✓ ✓ ✓ ✓ ✓
up‑ ✓ ✓ ✓ ✓ ✓ ✓

16.2.1 Non-native prefixes

All of the non-native prefixes can be found attached to nouns, almost all to adjectives,
and many to verbs. In (1) we find examples from COCA in which the prefix does not
change the category of the base.
 16.2 Formal considerations 335

(1) a. ante‑ On N: ante‑building, antechamber, antechoir, anteroom

On A: antenatal, antenuptial, antetypal
On V: antedate
b. arch‑ On N: archabbey, archangel, archbetrayer, archlute
On N: circumstance
c. circum‑ 
On A: circumambient, circum‑Atlantic, circumpolar
On V: circumambulate, circumlocute, circumnavigate
d. cis‑ On N: cisplatinum
On A: cisalpine, cislunar, cismoral
e. co‑ On N: co‑agent, co‑biographer, co‑compost, co‑denizens
On A: co‑conscious, co‑ethnic
On V: co‑adjust, co‑adopt, co‑construct, co‑sleep
f. contra‑ On N: contra‑bass, contra‑dance, contraflow, contra‑perfume
On A: contra‑causal, contra‑factual, contra‑lateral
On V: contra‑distinguish, contra‑indicate
g. counter‑ On N: counteragent, counterbeliefs, counterchagrin, counter‑odor
On A: counter‑incentive, counter‑pastoral
On V: counteract, counterbalance, counterrotate
h. epi‑ On N: epi‑cells, epi‑bacteria, epicenter, epiphenomena
On A: epicritical, epideictic, epinucleic
On N: ex‑addict, ex‑billionaire, ex‑cat
i. ex‑ 
On N: extra‑mission, extranet
j. extra‑ 
On A: extrabiblical, extracorporeal
k. inter‑ On N: intercolitis, intercooler, internet, intersex
On A: interabdominal, interaural, intercanine, interracial
On V: interact, interbreed, interchange, intermarry
On N: intramarriage, intrameasure, intranet, intraspace
l. intra‑ 
On A: intraabdominal, intracoastal, intracutaneous, intraracial
m. meta‑ On N: meta‑account, meta‑computer, meta‑drama, meta‑glass
On A: metacarpal, meta‑ironic, metaliterary, metaweird
On V: meta‑analyze, meta‑reflect
336 Locatives of time and space

n. neo‑ On N: neo‑allergens, neo‑beatnik, neoether, neo‑folkies

On A: neo‑abstract, neo‑cool, neo‑feudal, neo‑Norse
o. para‑ On N: para‑athletes, paracone, paraeducator
On A: paraclinical, para‑formal, para‑Islamic, para‑limbic
On V: paraglide, parasail
p. peri‑ On N: periapse, periclaw, perilymph, perimenopause
On A: perianal, periglacial, perinatal, periurban
q. post‑ On N: postadolescents, postblackness, postdoc, postfeminist, posttest
On A: postapostolic, postblack, post‑conservative, post‑digestive, post‑glam
On V: postdate
On N: preadult, precalculus, predead, preembryo
r. pre‑ 
On A: preacademic, precanonical, predark, pre‑egoic
On V: preacquire, preapprove, prebake, preclear
On N: proto‑adult, proto‑chordate, proto‑elephant, protoheart
s. proto‑ 
On A: proto‑bourgeois, proto‑classic, proto‑feminine, protoglyphous
On N: retro‑chiffon, retro‑comedy, retrohead, retro‑logic
t. retro‑ 
On A: retroactive, retrobulbar, retro‑campy, retro‑chic
On V: retrofit
u. sub‑ On N: subaccount, subarea, subcamp, subclause
On A: subabusive, subalpine, sub‑brown, sub‑cheap
On V: subcategorize, subcontract, subdivide, subgroup
On N: superaddressee, superblock, supercabinet, supercavity, superego
v. super‑ 
On A: superconscious, superfamilial, supergalactic, superlunary
On V: superintend
w. supra‑ On N: supracommunities, supraconductivity, supra‑identity, supra‑molecule,
supraspan
On A: supraannual, supra‑basal, supraethnic, suprahuman
x. sur‑ On N: surcharge, surcoat, surfiction, surname
On A: surreal
On V: surcharge, surmount, surpass
On N: transaxle, trans‑children, transfat, transgenes
y. trans‑ 
 16.2 Formal considerations 337

On A: trans‑Baltic, trans‑cervical, trans‑ethnic, transglobal

On V: transact, transfigure, transform, transload

Items derived with these prefixes can be spelled with or without hyphen. Here we have pre-
served the spelling found in COCA.
In (2) we give examples of the prefixes that can be found on bound bases. Many of the
examples we give here might not be considered complex words by contemporary speakers,
but based on our considerations in Chapter 3 we include them here.

(2) a. ante‑ antecede, antediluvian

b. circum‑ circumcise, circumference, circumspect
c. contra‑ contraceptive, contradict, contravene
d. counter‑ counterfeit, countermand, countervail
e. epi‑ epicanthic, epidemic, epigram
f. ex‑ exceed, expect, expose, extend
g. inter‑ intercept, intercede, interdict, interpellate
h. meta‑ metamorphosis
i. neo‑ neophyte, neoteny, neoteric
j. para‑ parabellum, parabrachial, paraphilic
k. peri‑ perianth, pericardium, periphery
l. post‑ postdict, postlude, postpone
m. pre‑ precede, preclude, predict, preempt
n. retro‑ retrocede, retrodict, retroflex, retrogress
o. sub‑ subduct, subjugate, subject, submit
p. super‑ supercede, supercilious, superficial, superlative
q. sur‑ surface, surfeit, surmise, survive
r. trans‑ transceiver, transcend, transduce, transfer

Several of the locative prefixes can be found attached to compounds or phrases.

(3) a. arch‑ arch‑Clinton loyalists, arch‑comical designs, arch‑dipshit, arch‑liberal

columnist
b. co‑ co‑artistic director, co‑boat-owner, co‑cave-dweller, co‑medical director,
co‑topdog
338 Locatives of time and space

c. ex‑ ex‑aerial gunner, ex‑ballplayer, ex‑battleship, ex‑big band, ex‑boomtown,

ex‑French colony
d. extra‑ extra‑Chinatown
e. neo‑ neo‑baby boom, neo‑bullshit, neo‑deathwish, neo‑hunter‑gatherer
f. post‑ post‑affirmative-action, post‑babyboom, post‑baby-on-board,
post‑baseball, post‑budget-cut, post‑communist-era, post‑Connery-as-
Bond, post‑death-of-God
g. pre‑ pre‑air-conditioned, pre‑baseball, pre‑complaint-filing, pre‑Cold War,
pre‑first-grade, pre‑leveraged-buyout, pre‑main-sequence, pre‑Redford-
and-Hoffman
h. proto‑ proto‑daily-paper, protohighlands
i. sub‑ sub‑cabinet-level, sub‑five-second, sub‑freshman, sub‑headline,
sub‑lightspeed
j. trans‑ trans‑lightspeed, trans‑windshield

Interestingly, quite a few of the locative prefixes appear to be category-changing. While the
examples in (1) and (2) are category-preserving, the examples we present in (4) raise the
issue of whether locative prefixes can effect category change. We find abundant examples in
which nouns with locative prefixes appear as pre-modifiers to other nouns.

(4) a. nominal bases

Anthropological Quarterly 2007: But though these extra‑household activities may

extract energy and time from women’s focus on immediate domestic duties, they are
seldom demonized in the same way that the career women’s working is.
Atlanta Journal Constitution 2000: United Airlines and Air Canada began offering an

inter‑airline electronic-ticketing service, which links the computer reservation
systems of both carriers.
Current Psychology 1996: Because the study was designed to explore how males and

females would evaluate the trait-pairs with and without a biologically relevant “starting
point,” intragender comparisons among Conditions 1 and 2 were performed.
Inc. 1999: we should expect only that these complaints will get to be higher and higher

complaints; that they will move from lower-grumble levels to higher-grumble levels
and finally to metagrumble levels . . .
Washington Post 1992: And who could have imagined a poetry reading inside the 15

Min. nightclub on 15th Street between K and L, a sullen little nook with creepy
button-tufted couches and a high-grunge, proto-funk, black-light decor with
neo‑brothel red walls and ridiculous cupids.
Denver Post 2003: “That’s the reality of the world we now live in, post-Sept. 11 and

post‑anthrax letters,” said Harris, a staffer with the National Security Council from
1993 to 2001.
 16.2 Formal considerations 339

Night Embrace 2003: But the retro‑beatnik Coffee Stain had nice artwork on the

walls and her friends seemed partial to drinking gallons of the tar-liquid.
PSA Journal 2003: Other than a very old female, and a sub‑adult male, the adults were

gone.
CNN_On Story 2004: The only recommendation that I’ve heard that has sort of

gotten out there is, OK, we need a superagency, a supercabinet position, to overlook
intelligence, to bring it all together.
America 1994: In many ways, the liberal or mainline churches and their suprachurch

organizations are intolerant of religious plurality and particularity, even if their
vocabulary is more open.
Geographical Review 2008: It was close to the bus and train station‑‑an ideal location

for serving transborder shoppers.
b. verbal bases
Houston Chronicle 1996: A post‑emerge herbicide is needed to treat a perennial

broad-leaved weed such as dichondra.

Some theorists might want to take examples such as these as evidence that the locative
prefixes can change nouns to adjectives, but given the existence of monomorphemic nouns
occurring as premodifiers of other nouns (for example, stone wall), we are somewhat reluc-
tant to do so (for further discussion see Chapter 19). The example in (4b) is, however,
different as monomorphemic verbs typically do not occur as premodifiers to nouns;
although examples like this appear to be quite rare, they suggest that further study would
be useful here.
For the most part, non-native locative prefixes neither undergo allomorphy themselves
nor induce allomorphy on their bases. Placement of stress seems not to be systematic. While
there is a tendency for stress to be on the prefix in nouns and on the base in verbs, Wells
(2008) shows both patterns for the verb antedate, for example, stress on the prefix for the
verb countersign, and stress on the base for the noun counterirritant. We also find stress on the
second syllable of the prefix, for example in forms like extrapolate, interpret, interpolate,
periphery, superfluous, superlative. Most of these can be accounted for by the general prosodic
constraints imposed on these derivatives by the suffixes in these words. For example, verbs in
‑ate as well as ‑y formations on the basis of neoclassical formations are characterized by ante-
penultimate main stress (see Chapter 13.2.2.2 and Chapter 12.2.5, respectively).
The non-native prefixes vary in productivity. The prefix cis‑ is the least productive, with
very few types occurring in COCA. The prefixes ante‑, circum‑, contra‑, epi‑, para‑, peri‑, supra‑,
and sur‑ show moderate productivity, as do extra‑ and super‑ with their spatial readings; these
last two are wildly productive, however, in their quantitative readings. The remaining prefixes
(arch‑, co‑, counter‑, ex‑, inter‑, intra‑, meta‑, neo‑, post‑, pre‑, proto‑, retro‑, sub‑, and trans‑) all
show a high level of productivity.
340 Locatives of time and space

16.2.2 Native prefixes

The first formal issue we must raise here is why we choose to treat the native morphemes
after‑, back‑, by‑, down‑, fore‑, in‑, mid‑, off‑, on‑, out‑, over‑, under‑, and up‑ as prefixes rather
than as the first elements of compounds, as some scholars do. Marchand (1969) treats fore‑
and mid‑ as prefixes, but the others are covered in his section on compounds as what he calls
‘combinations with locative particles as first elements’, on the grounds that the former are no
longer free morphemes whereas the latter are. Although neither treats the whole range of
locatives we consider here, Bauer (1983) and Adams (2001) both consider the non-native
elements as prefixes, and the native ones as the first elements in compounds. Lehrer (1995),
however, claims that distributional evidence justifies treating both native and non-native ele-
ments as prefixes, although again she does not discuss the entire range of locatives.
We have several reasons to adopt Lehrer’s position, one practical and two theoretical. Our
practical reason is that both formally and semantically the native forms show parallels to the
non-native ones; treating them in the same chapter allows us to make comparisons. Our
theoretical reasons are perhaps more convincing. First, a few of the native morphemes have
effects on the argument structure of verbal bases, something that we would not expect to see
if they were the first elements of compounds. We return to this point in Section 16.3.3.
We take our second theoretical argument from Dalton-Puffer and Plag (2000), who argue that
a formative can be considered to be an affix (as opposed to an element of a compound) to the
extent that it shows a coherent semantics that is at least partially distinct from a homophonous
free form. The native prefixes after‑, by‑, down‑, in‑, off‑, on‑, out‑, over‑, under‑, and up‑ are obvi-
ously homophonous with prepositions in English and we find when we compare their semantics
that their meanings are not always perfectly coextensive. On the one hand, the prepositions often
show a wider range of meanings than the prefixes, but we also find meanings for the prefixes that
do not occur in the free forms.
For example, the prefix and preposition after both express core meanings like ‘subsequent in
time’ (aftershock, after the earthquake), ‘posterior to’ (afterdeck, to follow after), but the preposition
also has specialized uses like ‘in imitation of ’ (a painting after Michelangelo) or ‘in accordance with’
(after my own heart). Similarly, the prefix and preposition by share the core meaning ‘alongside’
(bystander, by the river), but the preposition also indicates means or instrument (by foot) or, in a
construction with a reflexive, ‘alone’ (e.g. by myself).
There are also senses that can be found in the prefixes but not in the prepositions. With
by‑, for example, the prefix sometimes conveys the sense of ‘not major’ (byroad, bylaw,
byproduct), which in turn leads to more specialized meanings like ‘accidence’ (by-catch, by-kill),
or meanings that are not typically found with the preposition. Sometimes a meaning can be
found with both, but is much more prominent in the prefix than with the preposition, as with
the ‘excess’ meaning of over‑. This sort of non-overlap in meaning provides another piece of
evidence for treating the native formatives as prefixes.
Like the non-native prefixes surveyed in Section 16.2.1, the native locatives typically attach
to nominal bases, and sometimes to adjectival and verbal bases as well. Either native or
 16.2 Formal considerations 341

n­ on-native bases seem to be possible, although native bases are more frequent with this
cohort than with the non-native prefixes. Because of the extent of conversion in English, it is
not always possible to identify the category of the base uniquely, and some of these forms
may have more than one analysis.

(5) a. after‑ On N: afterbirth, afterdeck, afterglow, afterscent

On A: after‑shocking, after‑sweet

b. back‑ On N: back‑action, back‑country, backhoe, backporch
On V: back‑annotate, back‑announce, back‑comb, backfire

On N: by‑blow, byhikers, bykill, byproduct
c. by‑ 
On V: bygone, bypass

d. down‑ On N: downbeat, downblast, down‑gusts
On A: down‑deep, downlow, downright

On V: download, downsell, downshift

On N: forearm, forebrain, forefin, forenoon
e. fore‑ 
On V: forebear, forebode, foreclose, forejudge

On N: in‑basket, in‑break, in‑crowd, infield
f. in‑ 
On A: inborn, inbuilt, ingrown, in‑marrying

On V: indwell, ingather, inmix

g. mid‑ On N: midabdomen, midbite, midchannel, midengine
On A: midaxillary, mid‑Cambrian, mid‑distal, mid‑gibbous

h. off‑ On N: off‑flavor, off‑month, off‑ramp, off‑taste
On A: off‑blond, off‑diagonal, off‑gray, off‑width

On V: off‑load

On N: on‑passing, onrush, onset
i. on‑ 
On A: onmoving, onsweeping

On V: on‑load, onlook

On N: out‑agent, outcamp, outfield, outhouse
j. out‑ 
On V: outdate, outgrow, outlay, outsource

k. over‑ On N: overbag, over‑bite, overboot, overdog
On A: (quantitative/scalar reading only): over‑academic, over‑arty, overbold,

overdark
On V: overarch, overfly, overhang, oversee

342 Locatives of time and space

l. under‑ On N: underarm, underbark, underbody, underbrush

On A: under‑diagonal, underlit, underslung, underthrown

On V: undercut, undergo, underlie, underline

m. up‑ On N: up‑angle, up‑chute, up‑cropping, updraft
On A: upclose, upcoming, updrawn, upright

On V: upbring, upconvert, update, upgrade


As with the non-native locative prefixes, forms may be spelled either with or without
hyphen.
Not surprisingly, the native prefixes do not appear on bound bases (the form
onslaught is perhaps an exception, if the obsolete slaught is to be treated as bound at this
point). Most of them do not appear on either compounds or phrases, the exception
being mid‑.

(6) On compounds: mid‑dogfight, mid‑freefall, mid‑grain-filling, mid‑photo-spread

mid‑ 
On phrases: mid‑five‑figure, mid‑single‑digit, mid‑wanna-escape

The prefixes over‑ and out‑ can be found on compounds and phrases, but only with a quanti-
tative/scalar (as opposed to purely locative) meaning (over-aircondition, out-fund-raising,
out-redneck); see Section 16.3.1 for a full treatment of the semantic distinction.
As with the non-native locatives, we find many cases in which the prefixed items occur as
premodifiers to nouns and raise the same theoretical issue with regard to syntactic category
specification that we discussed with regard to the examples in (4).

(7) a. nominal base

Todays Parent 2004: Take that sensation a step further by smoothing on an after‑bath

gel after you dry off.
Backpacker 2011: The best backcountry meals don’t come from chichi cookbooks

that use words like “nap” to refer to a sauce coating instead of a siesta.
Atlantic Monthly 2007: These ranged from those that confront all senators—how

should they tend to the interests of downstate constituents?—to those that confront
only a few.
College Student Journal 2011: This can take the form of group projects, before or after

class interactions, office hour visits, and in‑class discussions that require students to
interact with each other.
Ethnology 1993: Vetale vaLLi and nuulam paTai are found from altitudes of about

1,250 meters down to 300 meters; that is, from the forested flanks and midaltitude
plateaus of the mountains down to the very edge of the agricultural plain.
 16.2 Formal considerations 343

PC World 2000: The machine you get is an offbrand computer, and you have to

purchase the monitor separately.
CSM 2009: One was the development of the American crawl, an overarm thrashing

stroke known today as freestyle.
PBS Newshour 2011: The prime minister of a major European economy, not just

accused, but now formally charged with underage prostitution and abuse of power.
NPR Science 2011: Tennessee has a lot more of upland limestone, and it’s pretty

riddled.
b. verbal base
Environment 1991: Groins and jetties built perpendicular to the shore intercept shifting

sand on the updrift side, but deny sand to the opposite, or downdrift, shoreline.

Again, it is not clear whether this is to be counted as a category-changing phenomenon, but

we do not rule this analysis out.
There is one case, however, where out‑ really does seem to be category-changing. Consider
the examples in (8).

(8) a. adjective to verb

USA Today 2000: He would do one pose, and I would try to out‑absurd him.

Chicago Sun Times 2008: A critical part of McCall’s spring charge will be to merge

smash-mouth, forward football into an offense that far too frequently tried to
‘outcute’ brute opponents.
Houston Chronicle 2003: He didn’t fight it. He didn’t outsmart himself, didn’t outdumb

himself, didn’t get too greedy or too cautious.
b. noun to verb
Journal of International Affairs 2002: Serbia: If You Can’t Out‑Birth Them, Make

Them Run Away
Houston Chronicle 2002: Both candidates are already using television advertisements,

and are trying to “out‑Bubba” one another for the rural audience.
CBS_Sixty Minutes 2004: Fuller first gained notoriety in 1997 when she was named

editor-in-chief of the racy “Cosmopolitan” magazine, where she soon out‑Cosmoed
even “Cosmo.”
San Francisco Chronicle 1991: And when money buys the senator’s vote . . . we’re

outnumbered and outdollared.

Here we have out‑ occurring on either adjectival (8a) or nominal (8b) bases. This pattern is
highly productive in contemporary English and raises an interesting theoretical issue. For the
most part, the examples with the nominal forms in out‑ are purely locative (outhouse, outfield),
344 Locatives of time and space

and the verbal forms on verbal bases are sometimes locative (outsource, outbreed ‘breed out-
side of a particular population’), whereas the verbs formed on nouns and adjectives always
seem to have a comparative meaning. This raises the question of whether we have here one
polysemous prefix or two homophonous prefixes. We will return to this issue in Section
16.3.1.
One way in which the native prefixes differ from the non-native ones is that the former
occasionally appear to function as heads of their words. That is, at least the prefixes down‑,
up‑, in‑, and out‑ can derive words that are distributionally identical to prepositions or prepo-
sitional phrases (or adverbs, in traditional terms), for example, forms like up-river, down‑
stream, in-house, and outside. The prefixes down‑ and up‑ appear to be productive in this
function, at least with topographical nouns (up-canyon, downbeach), but in‑ and out‑ are not,
being confined, as far as we can see, to a number of item-familiar and high frequency bases.
The native locative prefixes neither induce allomorphy in their bases nor undergo allo-
morphy themselves. As was the case with the non-native prefixes, nouns derived with the
native prefixes are typically stressed on the prefix. Derived verbs and adjectives seem to be
somewhat variable in stress pattern. With respect to verbs, Wells (2008), for example, gives
both pronunciations to download and offload, stress on the prefix for backcomb and bypass,
and stress on the base for indwell. The verb outgrow is stressed on the base, but outsource on
the prefix. Verbs in fore‑ are typically stressed on the base.
As with the non-native prefixes, the native locatives vary widely in productivity. The pre-
fixes fore‑, by‑, and on‑ are perhaps the least productive. Somewhat more productive are after‑,
back‑, down‑, in‑, off‑, and up‑, and most productive are mid‑, out‑, over‑, and under‑. It should
be kept in mind, however, that the prefixes out‑, over‑, and under‑ are a great deal more pro-
ductive with their scalar/quantitative readings than with their locative readings.

16.3 Semantic considerations

Locative prefixes are those that denote position with respect to either a spatial or a temporal
referent, or both; we include in this cohort the prefixes arch‑ and meta‑ which are metaphori-
cally but not literally locative. In this section we will first consider basic meanings of the loca-
tive prefixes (Section 16.3.1), then compare denotationally similar native and non-native pairs
(or triples) (Section 16.3.2). Finally, we will turn to the native prefixes which have valency-
changing effects when they occur on verbal bases (Section 16.3.3).

16.3.1 Core meanings

We can distinguish the semantic features of locative prefixes along several dimensions. We
will first look at the general semantic characteristics of the locative prefixes, distinguishing
those prefixes that are primarily spatial, those that are primarily (or exclusively) temporal,
 16.3 Semantic considerations 345

those that do double duty as both temporal and spatial locatives, those that have both spatial
and quantitative meanings, and finally those that are primarily metaphorical in nature.
Like locative prepositions, locative prefixes can be characterized semantically according to
the implied geometric characteristics of their referents. By ‘referents’ we mean the objects of
those prepositions, or the bases of the prefixes. Some prefixes make reference to a one-
dimensional object, that is, to a line, and more specifically to a line with either a horizontal or
a vertical orientation. The prefixes up‑, over‑, super‑, supra‑, sur‑, sub‑, down‑, and under‑, for
example, make reference to a vertical axis of orientation, whereas ante‑, fore‑, pre‑, after‑, and
post‑ make reference to a horizontal axis. We can further divide these into those prefixes that
refer to the positive end of the axis (up‑, over‑, super‑, sur‑, meta‑, arch‑; after‑, post‑) and those
that make reference to the negative pole (sub‑, down‑; ante‑, fore‑, pre‑). The prefix mid‑ can
refer to either axis but to an area between the two poles.

(9) up-, over-, super-, supra-, sur-, meta-, arch-

ante-, fore-, pre-, retro- − + after-, post-

−
sub-, down-, under-

Other prefixes are best described with reference to a two- or three-dimensional space (back‑,
circum‑, para‑, peri‑, by‑, off‑, on‑, out‑, in‑, epi‑, extra‑):

(10) off- epi-, on-

in-, intra- by-, circum-, para-, peri-

out-, extra-
346 Locatives of time and space

For the latter group, in‑ and intra‑ make reference to the interior of the referent, on‑, and epi‑
to a surface, extra‑, off‑, and out‑ to the area exterior to the referent, and circum‑, peri‑, by‑, and
para‑ to the edge or border. The latter four are distinguished in that circum‑ makes reference
to the whole of the periphery, whereas by‑, peri‑, and para‑ merely reference the vicinity of the
referent. The prefix back‑ refers to the rear surface of a three-dimensional referent, or the area
behind that surface.
Finally, several prefixes can be described as relating one point of reference to another (cis‑,
trans‑, inter‑, co‑, counter‑, contra‑).

(11) a.
cis-, trans-
b.
inter-
c.
co-
d.
counter-, contra-

The prefix trans‑ may focus on the path (going through or across something), or on the final
destination, as being beyond something from the speaker’s perspective (trans-Alpine Gaul).
In (12) we group the locative prefixes according to their main semantic functions:

(12) a. Primarily spatial: by‑, circum‑, cis‑, contra‑, counter‑, down‑, epi‑, in‑, off‑, on‑, para‑,
sub‑, supra‑, trans‑, up‑
b. Exclusively temporal: ex‑, neo‑, proto‑
Both spatial and temporal: after‑, ante‑, back‑, co‑, fore‑, inter‑, intra‑, mid‑, peri‑,
c. 
post‑, pre‑, retro‑
d. Both spatial and quantitative: extra‑, out‑, over‑, super‑, sur‑, under‑
e. Metaphorical: arch‑, meta‑

Simple temporal reference to a point on a timeline is generally accomplished by a prefix that does
double duty as a spatial prefix. So we have spatial/temporal pairs such as the following.

(13) Spatial Temporal

afterdeck aftercare
antechamber antedate
backcountry backdate
co‑author co‑consider
 16.3 Semantic considerations 347

forearm foresight
intercampus intermenstrual
intragroup intrasemester
midbrain midargument
perianal perimenopause
postauricle post‑acne
preschool preadolescent
retrocurved retro‑date

We find spatial readings more frequently than temporal for back‑ and intra‑, and temporal
readings more frequently for post‑, pre‑, and retro‑, but the majority allow both readings with
relatively equal frequency. Where the spatial readings are dominant, it is often necessary for
the base to be explicitly temporal in reference (intra-annual) or to be readily interpreted as
eventive (intra-semester) to get the temporal reading from the prefix. Where the temporal
reading is dominant, even completely non-eventive bases can be interpreted temporally, as
we find when post‑ is attached to any proper noun (for example, post‑Clinton).
Several of the locative prefixes have come to combine both locative and quantitative
meanings. These are typically prefixes that express endpoints on an axis, usually the vertical
axis (over‑, super‑, sur‑, under‑; extra‑ and out‑ are the exceptions here). The semantic exten-
sion takes us from the spatial endpoint of a scale to beyond that endpoint, thus inviting the
inference of excess (for extra‑, out‑, over‑, super‑, sur‑) or dearth (under‑) depending on
whether reference is to the positive or negative pole. The prefix out‑ is somewhat different in
that its quantitative/scalar reading on derived verbs is comparative in nature.
The prefix out‑ raises an interesting theoretical issue: as mentioned in Section 16.2.2, it is
typically locative in meaning when it derives nouns (outfield, outhouse), and when it derives
verbs on verbal bases it can bear either a locative meaning (outsource, outbreed ‘breed outside
of a particular population’) or a comparative meaning (outdance, outcompete), but in its
­category-changing guise (out-absurd, out-Bubba) it is exclusively comparative in meaning. On
the one hand, the semantic uniformity and robust productivity of the category-changing
­version of the prefix might be taken as evidence that out‑ has evolved into two distinct
homophonous affixes. On the other hand, the existence of an overlap between the two mean-
ings in forms derived from verbs might argue for a polysemy analysis. We will not decide
between these two possibilities here.
English has three prefixes that seem to be exclusively temporal: ex‑, neo‑, and proto‑. Neo‑
and proto‑ are fairly straightforward, the former denoting ‘new’ (neo-family) or ‘revived from
an older form’ (neo-conservative), and the latter ‘earlier’ or ‘original’ (proto-Indo-European).
The prefix ex‑ is somewhat more complex. Meaning ‘former’, it occurs more comfortably on
stage-level nouns, that is, nouns like pedestrian or musician that denote temporary states, than
348 Locatives of time and space

on individual level nouns like car or horse that denote permanent states of being (we take
Monty Python’s ex-parrot to be an instance of coercion). The prefix ex‑ can occur on indi-
vidual level nouns if they are preceded by a possessive, in which case the prefix actually takes
scope over the possessive rather than the noun to which it is attached; in other words, my
ex-car normally refers to a car that no longer belongs to me, as opposed to something which
is no longer a car. With more complex noun phrases, the prefix ex‑ can take wide scope over
an entire phrase as well as narrow scope over the base to which it is phonologically attached.
So if we have a phrase like an ex-ballplayer for the Red Sox, the person in question might no
longer be a ballplayer (narrow scope), but might equally well still be a ballplayer, but for
another team (wide scope).
The final two prefixes we consider here are arch‑ and meta‑. These differ from the rest of
the locatives in that their spatial sense is purely metaphorical. The prefix arch‑ is typically
used on personal nouns (arch-critic, arch-dipshit, archsorcerer) to indicate higher status. The
prefix meta‑ favours abstract nouns (belief, pathology, theory) and relational adjectives
­(aesthetic, cognitive) as bases and denotes an over-arching or a higher-level form of the base.
They may both be considered figurative analogues of prefixes like over‑ or super‑ in their
­spatial (as opposed to quantitative) senses.

16.3.2 Native and non-native cohorts: a comparison

As the diagrams in examples (9), (10), and (11) suggest, some of the spatial relationships we have
looked at are associated with both a native prefix and a non-native one (and sometimes with
more than one non-native one). The question therefore naturally arises whether the cohorts of
prefixes can be distinguished semantically in any way. This is the question we turn to here.

16.3.2.1 over‑ versus super‑, supra‑, sur‑

Despite the similarity in core denotation, these prefixes are not identical either in productiv-
ity or in semantic nuance. The prefix sur‑ is at best minimally productive, although the term
surfiction (found in COCA) does seem to be a neologism—albeit an extremely self-conscious
one. The other non-native prefixes are more productive, but only on nouns and adjectives.
The native over‑ is the only one of this cohort that can be freely used to coin verbs.
While both over‑ and super‑ frequently have a nuance of ‘excess’, with over‑ this nuance is
typically intended in a negative sense and with super‑ often (but not always) in a positive
sense, as the examples in (14) illustrate.

(14) a. Field and Stream 2005: Audubon Pennsylvania agrees with the overwhelming
scientific evidence that overabundant whitetail deer have seriously degraded
habitats throughout the commonwealth.
b. Skiing 1995: Targhee was a super place to test fat skis, with its abundant terrain and
super-abundant snow (the area’s slogan is “Snow from heaven, not hoses”).
 16.3 Semantic considerations 349

The prefix supra‑, in contrast to super‑, does not have the nuance of ‘excess’ and is sometimes
used in distinction to super‑ to convey a spatial sense.

(15) a. America 1994: In many ways, the liberal or mainline churches and their suprachurch
organizations are intolerant of religious plurality and particularity [. . .]
b. New York Times 1995: “[. . .] I do think that’s a danger you face in the superchurch,”
said Dr. Gregory, who described his experience in a book, “Too Great a Temptation”
“[. . .] A megachurch minister can,” he said, “get promoted right out of reality.”

16.3.2.2 under‑ versus sub‑

These prefixes sometimes overlap in sense. Both can be straightforwardly spatial, as we find
in the examples in (16) and (17).

(16) a. Esquire 2004: Everybody at Sony Pictures says they’re happy—the top boss, the
various sub‑bosses, the would‑be bosses.
b. Nightline 2010: FBI wiretaps have captured this underboss of the Colombos
bragging about the dozens of men he’s killed and describing his favorite way to
dispose of bodies, using knives, a kiddy pool and a trash compactor.

(17) a. Airforce One 1997: Korshunov pushes Alice down the stairs to the underdeck.
b. Skiing 1994: The wooden deck, which is shaped like a skateboard, is attached to a
subdeck with a plastic rail along the length of the bottom.

While under‑ frequently has a quantitative nuance, the prefix sub‑ only occasionally does. Where
sub‑ does carry a quantitative nuance, it seems generally more neutral in tone than under‑.

(18) a. Mechanical Engineering 2007: Boeing has the new subjumbo 787, designed to serve
what the company sees as the future demand of air travel, as well as a “new”
superjumbo 747‑8 family.
b. Education 2000: Swart (1987:163) maintains that too little stress may even lead to
underaverage achievement.

The pejorative reading is particularly prominent when under‑ is attached to verbal bases
(underachieve, underbake, underproduce). The prefix sub‑ has one nuance that under‑ lacks.
Forms with sub‑ on nominal or verbal bases frequently carry a meaning of partition or hierar-
chy, rather than a purely spatial meaning, as we find in subclause or subdivide, for example.

16.3.2.3 fore‑ versus ante‑, pre‑

These three prefixes appear to be denotatively similar, although of the three, only pre‑ shows
any substantial degree of productivity. We only rarely find doublets. In (19) the forms are
350 Locatives of time and space

denotatively identical, although the form in fore‑ is meant to convey a self-consciously archaic
flavour.

(19) a. Fantasy & Science Fiction 2007: When I set out in the fore‑dawn, methought I wore
a shadow of subtle tints and colorations invisible in dawnlight.
b. Fantasy & Science Fiction 2010: Now, in the predawn, I arrived at an almost empty
skyscraper, signed in with security in a silent lobby, went up to the shabby station
where an engineer and a couple of other guys sat around gossiping.

In (20) we have an apparent neologism forecheck, but on a sense of check that is peculiar to
certain team sports (ice hockey, field hockey, etc.).

(20) a. USA 2009: “It’s just key to have people who can forecheck and possess the puck
and still maintain that scoring ability and poise once they get it,” Cahow said.
b. PSA Journal 1996: But many times, the presenter did not precheck the equipment
or was not familiar with the local setup.

16.3.2.4 after‑ versus post‑

This pair is not only similar in denotation, but also in connotation, neither having a particu-
larly evaluative flavour. The prefix after‑ occurs productively only on nouns, although the
prefixed forms, as mentioned in Section 16.2.2, frequently occur as premodifiers of other
nouns. The prefix post‑ is productive on both nouns and adjectives, and on nouns can also
form premodifiers. As the doublet in example (21) illustrates, the two prefixes appear to be
semantic equivalents.

(21) a. Ebony 2004: The typical teen, experts say, could spend in excess of $1,000 for the
prom ticket, limo rental, photos, corsage, spectacular one-of-a-kind dress and
matching garter belt, tuxedo rental (or purchase) personal grooming and, of course,
the after‑prom celebration which typically caps off the evening.
b. Atlanta Journal Constitution 2002: As the PTA president at Harrison High School in
Kennesaw, she has hosted a post‑prom party for several years at U.S. Play near
Town Center mall and planned to carry it on this spring.

Neither prefix attaches freely to verbs, although we do find the occasional exception such as
post-date.

16.3.2.5 by‑ versus peri‑

These two are perhaps not really parallel in the way that the previously discussed pairs are:
the prefix by‑ can be glossed as ‘beside’, peri‑ as ‘round about’, but pragmatically this often
amounts to the same thing. The prefix by‑ is at best minimally productive. We find the form
by-kill attested in COCA but not in the OED; most forms with this prefix are high-frequency
 16.3 Semantic considerations 351

and well-established. The prefix peri‑ is somewhat more productive, but is most frequently
used in learned and especially medical contexts (pericranial, peri-tumoral).

16.3.2.6 out‑ versus extra‑

Although these prefixes share a core meaning ‘external to’, they in fact do not overlap. The
prefix extra‑ is primarily productive on adjectives, and where it occurs on nominal bases the
derived forms typically function as premodifiers of other nouns. The prefix out‑ does not
attach to adjectives or form premodifiers, and in fact is most productive on verbal bases,
where it has a clearly comparative sense.

16.3.2.7 in‑ versus intra‑

Denotatively, these two prefixes are quite close, although in terms of potential bases there is
relatively little overlap. Of the two, only in‑ can attach to verbs, and on nominal and adjectival
bases intra‑ tends to favour non-native and especially learned bases, whereas in‑ favours native
bases. There are small areas of overlap, however, which illustrate the denotative similarity.

(22) a. Denver Post 1996: “I think the rivalry is more intra‑division, rather than
cross‑division,” Neuheisel said.
b. San Francisco Chronicle 1990: If you’re a Giants fan, you can say good-by to the West
for a while, catch your breath and hope things are still as interesting when the
traditional round of in‑division games start in September.

Neither of these prefixes has a particularly evaluative force.

16.3.3 Argument structure effects

Non-native prefixes do not affect argument structure when they attach to verbal bases. With
native prefixes, however, we sometimes do find changes to verbal diathesis that seem to be
induced by the prefix. This is especially true of the native prefixes over‑ and out‑, occasionally
true of under‑, and only sporadically true of the other native prefixes that can be found
attached to verbs. This phenomenon has occasionally been noted in the literature (Fraser
1974; Bresnan 1982; Williams 1992; Lieber 2004) at least for the first two prefixes, although it
has not to our knowledge been studied systematically for all locative prefixes.
With over‑, two sorts of argument structure effect can be found. On verbs of change of
state or directed motion like flow, fly, grow, or leap, the affixation of over‑ allows the addition
of a locative NP argument, as the examples in (23) illustrate.

(23) a. Christian Science Monitor 2006: Without the grazing species, the seaweeds these
small organisms ate soon overgrew the seafloor, crowding out other species and
making it much harder for already stressed coral reefs to regenerate.
352 Locatives of time and space

b. Futurist 2003: Recycling has delayed the “garbage glut” that threatened to overflow
the world’s landfills, but the threat has not passed simply because it has not yet
arrived.
Natural History 2003: Wallace’s Line is now understood to overlie a region
c. 
incorporating three major tectonic plates and several smaller ones.

In these cases, over‑ often has its locative rather than qualitative reading. On other verbs
(often, but not always activity verbs) the prefixation of over‑ allows for the deletion of an
argument, so that a normally transitive verb can be used intransitively.

(24) a. ABA Journal 1994: Solving problems others created or succeeding for someone else’s
benefit develops stress or a compulsion to overachieve.
b. Atlanta Journal Constitution 2008: Pros are strategic in what storage containers they
have clients purchase, whereas someone desperate to declutter in a single weekend
will tend to overbuy.
Christian Science Monitor 2009: Lithium-ion batteries using cobalt chemistry,
c. 
popularly used in laptop computers and cellphones, have in the past shown a
propensity to overheat, resulting in a few laptops going up in flames.

In these cases, over‑ tends to display its quantitative/scalar reading.

The prefix under‑ is much less productive in forming verbs than over‑ is, but where it does
occur on verbal bases, it shows some of the same argument structure effects. We find verbs
like underachieve and underearn that are predominantly used intransitively, as in (25a) and
(25b); there is at least one verb, bid, that normally selects a prepositional object when unpre-
fixed, but which, with the prefix under‑, can appear with an NP object (25c).

(25) a. Community College Review 2009: These results suggest that there might be an
interaction with income to explain why White students with Pell grants and Stafford
loans underachieve compared to African American and Native American students
without these forms of aid.
b. PBS_Tavis 2006: Cause I ain’t never met nobody, including Bill Gates, who thinks
he underearns.
USA Today 2006: The Revolutionary Guards’ preferential access to state financing
c. 
allows them to underbid private firms, says Bijan Khajehpour, chairman of Tehran-
based Atieh Group, which advises foreign firms on the Iranian market.

The prefix out‑ quite regularly derives transitive verbs, something which makes sense, given its
comparative semantics. If the verbal base is transitive to begin with, there is of course no change
in verbal diathesis, although the semantic category of the NP object may be different. That is,
whereas a verb like eat normally takes food items as its object (Fenster ate a pickle/*Raoul), when
prefixed with out‑, the object must be another animate NP (Fenster outate *pickles/Raoul).
 16.3 Semantic considerations 353

If the verb is intransitive to begin with (for example, fart, guffaw, linger, sleep), or takes a
prepositional (stare, yell) or clausal object (think), the prefixation of out‑ triggers the addition
of an NP argument, as the examples in (26) illustrate.

(26) a. Cosmopolitan 2003: On and off camera, more girls are dishing about discharge,
outfarting their friends, and taking part in other beyond-ribald behavior.
b. Christian Science Monitor 1993: Jim Mezon inhabits the role of Simon with an
uncanny feeling for the speech and habits of a man who hasn’t forgotten the folks
from the old neighborhood, but who can outwit and out‑guffaw the sleaziest
politician.
Radiance 1999: In spite of her tenderness, Aunt Jenny could outshout, outgesture,
c. 
and outstare the loudest of my Italian relatives.
d. Boys Life 2007: As an Eagle Scout in Mesa, Beck learned the importance of being
prepared. His favorite activity was fly-fishing, a sport in which you have to study the
behavior of fish, then try to outthink them.

Where out‑ derives verbs from adjectives or nouns, the result is typically transitive as well.

(27) a. USA Today 2000: He would do one pose, and I would try to out‑absurd him.
b. Atlanta Journal Consitution 1996: Hammerin’ Hank did not out‑Babe the Babe!

In all of these cases, out‑ has its scalar or quantitative reading, rather than a purely locative
one.
The only other native locative prefixes that can be found on verbs are back‑, down‑, and up‑,
although none of these is particularly productive. Generally back‑ has no effect on the argu-
ment structure of the base verb, although we do find the form backtalk in COCA used with
an NP object rather than a prepositional object. Similarly, where down‑ occurs on verbal
bases, it generally leaves verbal diathesis untouched, exceptions being downcut and downtalk,
both attested in COCA.

(28) a. Killing Mister Watson 1990: And she said, Don’t you backtalk me, you are my child!
b. Bioscience 1993: In addition, channels may erode, or downcut, triggering rejuvenation
of tributaries, which themselves begin eroding and migrating headward.
Analogue Science Fiction & Fact 2010: I suspected the boy had the right of it, but it
c. 
does no good to downtalk leaders.

It is difficult to say whether there is anything systematic about the effects illustrated in (28),
as they are limited to these isolated examples.
 chapter 17

Negatives

17.1 Prospectus
In this chapter we will examine affixal negation in English. We use a broad definition of nega‑
tion here, including along with standard negatives (‘not X’) also privatives (‘without X’ or
‘remove X from’), reversatives (‘reverse action of verbing’), and verbal pejoratives (‘do X
wrongly, badly’). The affixes we will examine are in‑, un‑, non‑, a‑, dis‑, de‑, ‑less, mis‑, and mal‑.
The semantically related formations involving ‑free and ill‑ are sometimes treated as affixes or
affixoids, but we consider formations with these elements as compounds, on the grounds
that there are no compelling arguments for their status as affixes or affixoids (see Chapter 16
and Dalton-Puffer and Plag (2000) for pertinent tests). At least the first four of these affixes
have received extensive discussion in the literature, including Jespersen (1917), Zimmer
(1964), Marchand (1969), Bauer (1983), Horn (1989, 2002), Bauer and Huddleston (2002),
Plag (2003, 2004), Lieber (2004), and Hamawand (2009).
The corpus data indicate that these affixes are frequently polysemous, displaying more
than one of the meanings/functions mentioned above and suggesting that to some extent the
semantic interpretation of the derived words is not solely a function of the meaning of
the affix, but is the consequence of combining the broadly negative content of the affix with
the meaning of the base; that is, straightforward negatives tend to occur with certain kinds of
bases, privatives with others, and so on. In this conclusion, we depart from much of the
above-mentioned research. From a different, base-driven, perspective, we note that one base
can take more than one of these affixes, sometimes with the formally different derivatives
having the same meaning, sometimes having different meanings.
We begin with formal considerations in Section 17.2 and continue with semantic considera‑
tions in Section 17.3. Examples are taken from COCA, unless otherwise specified.

17.2 Formal considerations

17.2.1 Basic characteristics
Table 17.1 sets out the basic formal characteristics of this cohort of affixes and affixoids, exam‑
ples of which are then given in (1). Clearly, there is a great deal of overlap in formal properties
among the negative and privative affixes of English.
Table 17.1 Formal properties of negative and privative affixes (native affixes are italicized). Parenthesized items are rare
Input category Output category Category Native bases Non-native
changing bases
A V N bound bases A V N
in‑ ✓ ✓ (✓) (✓) ✓ ✓ (✓) sometimes ✓ ✓
un‑ ✓ (✓) ✓ ✓ N → A, ✓
occasionally
N→V
non‑ ✓ ✓ ✓ ✓ sometimes ✓ ✓
N→A
dis‑ (✓) ✓ ✓ (✓) (✓) ✓ ✓ very (✓) ✓
occasionally
de‑ ✓ ✓ (✓) ✓ N→V ✓ ✓
a(n)‑ ✓ ✓ (✓) ✓
mis‑ ✓ (✓) (✓) ✓ ✓ ✓ ✓
mal‑ ✓ (✓) ✓ (✓) ✓ ✓ ✓ ✓
‑less ✓ ? ✓ N→A ✓ ✓
356 Negatives

The following examples list pertinent derivatives for illustration. Note that since dis‑ can
safely be considered non-category-changing, potentially ambiguous examples (e.g. discom‑
fort) have been analysed with this assumption in mind.

(1) a. in‑
on adjectives: inaccurate, inappropriate, inartistic, incoherent, incorrect, indifferent,
ineffective, infertile, inhumane, inorganic, insane, intemperate, invariable,
involuntary
on nouns: indecorum, indigestion, inefficacy, inexperience, inutility
on bound bases: inane, incognito, incommunicado, inept, inert, innocent, inoxia
b. un‑
on adjectives: unable, unactorish, unbandaged, unbegrudging, uncatlike, un‑devout,
unemotionless, unfair, unfishy, unghosted, ungreen, unimpressive, unlevel,
unmanifest, unorthodox, unpoisonable, un‑rebuilt
on nouns: unbelief, unburger, unconcern, undeath, unhero, un‑tundra, un‑chicken,
unchurch, uncircumcision, un‑commercial, uncorrection
on verbs: unbelt, unbraid, uncheck, undecorate, unenter, unfold, unglue, unharness,
unimagine, unlearn, unleash, unpeel, unpry, un‑relax
c. non-
on adjectives: non‑anxious, nonartsy, nonbasic, noncelibate, non‑deceptive,
non‑editorial, non‑fearful, non‑geeky, nonimpaired, non‑kosher, non‑manual,
non‑palatal, nonrandomized, nonsanitary, non‑soulful
on nouns: non‑astronaut, non‑baseball, noncatch, nonconformity, non‑dancer,
nonfeminist, nonguest, nonhighway, non‑latino, non‑marriage, nonnun, non‑peanut,
non‑question, nonresident, non‑speech
d. dis-
on adjectives: disconjugate, dishistorical, dishonest, disloyal, disordinal, displastic,
dissimilar, distemperate
on nouns: disanalogy, disarray, disbenefit, disbelief, discomfort, discontinuum,
disequilibrium, disharmony, disidentity, disorder, disquiet, disrespect, distaste,
distolerance, disunion
on verbs: disavow, disband, disengage, disinherit, dislocate, dislodge, dislove,
dismount, disoblige, dispossess, dispraise, disqualify, disreport, dissave, disthrone,
disunite, disvalue
on bound bases: discarnate, discrepancy, dispel, dissent, disphonia, distort, distract
e. de‑
on verbs: deacidify, decaffeinate, decarbonize, decommit, de‑erotify, de‑fry, deglamorize,
de‑merge, demodulate, demythify, depenalize, deracinate, dereify, desaturate
on nouns: debarb, de‑bling, debulk, decrud, defat, defrock, degender, delead, demoss,
depretzel, desilk, dethrone, detoad
 17.2 Formal considerations 357

on bound bases: deflect, deflate, deplenish, deplete, deport, depose, deprive, detract,
devolve
f. a‑ (with its allomorph an-)
on adjectives: acaucasian, acultural, adeictic, anestrous, arhetorical, asemantic, apolar,
aseismic, asexual, athematic, acephalous
on bound bases: agenic, aphelic, acephal, amorphous, anaemia, anarchic, anomic
g. mis‑
on verbs: misact, misallocate, misbehave, misbutton, miscall, misclassify, misdefine,
misestimate, misfuel, mishandle, mishit, misinvest, mislead, misname, mispredict,
misreport, misscore, mis‑tally, mistype, misu[z]e, misvote
on nouns: misadventure, misanalogy, misbelief, misbirth, mischance, misconceit,
misconduct, misdeed, misfortune, misjustice, mistruth, misu[s]e
on bound bases: misanthrope, miscreance, misfeasance, misnomer, misogyny
h. mal‑
on adjectives: maladjusted, malaligned, malcontent, maleducated, malgoverned,
mal‑hygienic, malplaced
on verbs: malabsorb, malform, malnourish, maltreat
on nouns: malfunction, mal‑intent, malnutrition, malodor, malunion
on bound bases: malediction, malevolence, malversation
i. -less
on nouns: ageless, airless, antless, batteryless, bosomless, captionless, dairyless,
flangeless, hasteless, ladderless, mooseless, pastorless, roseless, secretless, tableless,
visionless
on verbs: doteless, resistless, quenchless
on bound bases: feckless, gormless, reckless

The neo-classical element dys-, etymologically Greek and used mainly on Greek bases can be
treated as an orthographic variant of dis-. Examples, mostly from the area of medicine, include
dysarthia, dysfunction, dysgraphic, dyslexia, dyspeptic, dysplasia, dysrhythmia, dystrophy.
Of the affixes that we consider in this chapter, only un‑, mis‑, and ‑less are native in origin.
The prefix mis‑ is a difficult case, however, as it is both native and non-native in origin. Old
English had many forms with this element, to which later some French loans with the corre‑
sponding French prefix were added. How these French forms influenced the later develop‑
ment of the prefix is unclear.
Apart from un‑, mis‑, and ‑less all other negative affixes are non-native. The prefixal nega‑
tives almost always attach to more than one category of base. Only a‑ is restricted to a
single category, adjective. The prefixes un‑, dis‑, and mal‑ all attach to all three lexical cat‑
egories, in‑ and non‑ to adjectives and nouns, de‑, mis‑, and ‑less to verbs and nouns. With
the exception of de‑, the other prefixes usually do not change category, but occasional
358 Negatives

category changes with dis‑ (disrobe) and un‑ (unbosom, unearth) can be found. Nouns
negated with un‑ are sometimes used, adverbially modified as in (2a–c), as premodifiers,
occasionally even predicatively as in (2d), and could thus be argued to have become
adjectives.

(2) a. Backpacker 1994: the most unbackcountry part of the Florida trail
San Francisco Chronicle 1993: Admittedly the goal is a notoriously un‑Berkeley one:
b. 
profit.
c. Esquire 2000: A most un‑Boomer sentiment, indeed.
d. The Big Lebowski 1998: Come on. You’re being very unDude.

Like un‑, non‑ also sometimes appears to form adjectives, or at least nominal-premodifiers,
from nouns:

CNN_SunMorn 2006: So you might have a candy, for example, that is processed on a
(3) 
machinery that had peanuts on it right before they manufactured your non‑peanut
candy, or vice versa.
San Francisco Chronicle 2002: From Hollywood, visitors can easily take a 20-minute train
ride to Universal Studios or get in the car and visit Beverly Hills, about 25 painless,
non‑highway minutes away.

The negative prefixes in‑, mis‑, and mal‑ appear to be uniformly non-category-changing.
Of the negative prefixes, only a‑ is found with some frequency on bound bases. This is not
surprising, as many of the words found with this affix were borrowed directly from the clas‑
sical languages. The suffixes in‑, de‑, dis‑, mis‑, and mal‑ are sometimes found on bound bases,
although it seems safe to say that these have come into English as borrowings and are not
productively derived in contemporary English. It is thus doubtful that native speakers per‑
ceive the in‑ in innocent, the de‑ in detract, the dis‑ in distort, or the mal‑ in malediction as being
prefixes at all, or as being the same prefixes as found in intolerable, destabilize, disable, or mal‑
treat. One obvious difference is the fact that the former forms are not so easily segmented,
since they involve bound bases, whose meaning is not easily recoverable (‑nocent, ‑tract,
‑tort), or show a phonology different from that of the productive prefix (e.g. unstressed versus
secondarily stressed: detráct versus dèstábilize).
All of these prefixes can be found on non-native bases, in‑, a‑, and mal‑ exclusively so. Only
a small minority of bases found with dis‑ are native (disbelief, disbelieve, disband, disbond, dis‑
hear, dishearten, dislike, dislove, disown, disfathered). Note that the last of these is to be found
in COCA but not in the OED, suggesting that it is still marginally possible to coin new dis‑
words on native bases. The remaining negative prefixes are indifferent to the origin of their
bases, being formed with equal ease on native or non-native bases.
 17.2 Formal considerations 359

Of all negative affixes, only ‑less is suffixal. It is category-changing, forming adjectives

largely from nouns. We do, however, occasionally find both morphemes on bases of other
categories. The suffix ‑less, for example, is found on several bases that seem unequivocally to
be verb stems:

Psychology Today 1996: We need a clearinghouse for grown-ups who want to put
(4) 
themselves up for adoption—there are many of them but I’ve seen only grief for
everyone when men kick their own children out of the nest and replace them with some
doteless damsel.
Field and Stream 1995: Men may never have been killed by these flies as they have been
by the fever-carrying mosquitoes of Southern latitudes, but the poisonous afteraffects
[sic] of the gnat’s sting produce a resistless drowsiness amounting to stupor, or coma,
and is very slow to work off.
National Review 2001: The great advertising industry has helped by producing a
quenchless flow of effective, if sometimes crude, propaganda.

In the words feckless, gormless, and reckless, the suffix is found with three native bases that
might conceivably be analysed as bound, although their status is unclear. The bases gorm and
reck were historically or dialectally free morphemes, but arguably are no longer so in contem‑
porary English, and feck is conjectured by the OED to be either a clipping of effeck, an aphetic
pronunciation of effect, or a backformation from feckless.

17.2.2 Phonological and orthographic properties

Of the negative affixes, in‑, a‑, and arguably un‑ exhibit allomorphy. The prefix in‑ undergoes
assimilation to the point of articulation of a following stop or nasal, and is deleted before
liquid-initial bases, without gemination (see Chapter 9). Example (5) illustrates these
effects.

(5) inalienable /ɪn'eɪliənəbl/

implausible /ɪm'plɔːzəbl ‖ ɪm'plɑːzəbl/
immortal /ɪ'mɔːtl ‖ ɪ'mɔːrtl/
intolerable /ɪn'tɒlərəbl ‖ ɪn'tɑːlərəbl/
innumerable /ɪ'njuːmərəbl ‖ ɪ'nuːmərəbl/
incongruous /ɪŋ'kɒŋɡruəs ‖ ɪŋ'kɑːŋɡruəs/
irrational /ɪ'ræʃnəl/
illegal /ɪ'liːɡl/

The assimilation is orthographically sanctioned—except for <n> pronounced /ŋ/ before

velars—and is largely automatic (see Chapter 9). However, we find a few instances in COCA
360 Negatives

such as inbalance, inbearable, inbelievable, inperfections, inpracticable which suggest that the
process may not be fully automatic. In these instances, there is an alternative explanation in
an <n> for <m> typographical error.
The prefix in‑ occasionally attracts stress, but this does not seem to be a systematic process.
We find stress on the prefix in such words as ínfidel, íngrate, ínnocent, ínclement (although
Wells (2008) considers this pronunciation incorrect), ínfinite, ímpotent, and ínvalid (when it
means ‘ill’, but not when it means ‘not valid’). Jespersen (1917: 139) mentions impious as
another word in which the prefix has attracted stress, but this pronunciation is becoming less
common in BrE according to Wells (2008), and is not found in NAmE. Notably, stress on the
prefix is often accompanied by semantic opacity, as in impotent, infamous, and invalid, or hap‑
pens with bound bases (e.g. infidel).
The prefix a‑ has an allomorph an‑ which appears before bases that are vowel-initial or that
begin with /h/. The selection of the allomorph is somewhat variable (across and within
types) preceding /h/.

(6) a. before vowel-initial bases: anemia, anerobic, anesthesia, analphabetic, anastigmatic,

anechoic, anencephalic, anorexia, anorgasmia, anovulant, anoxia, anuria
before bases beginning with /h/: anharmonic, anhedral, anhedonia, anhydride,
b. 
anhydrous (but cf. ahistorical ~ anhistorical)
before consonant-initial bases: abiotic, acalculia, achromatic, agraphia, akinesia, alogical,
c. 
amitotic, amoral, aperiodic, apolitical, asexual, asocial, asymmetry, asymptomatic,
atheism, azoic

The a‑ allomorph is usually pronounced /eɪ/ (abiotic, achromatic, asexual), but is occasion‑
ally pronounced /æ/ (apathy, asymptote, %asphyxiate) or reduced to /ə/ (anomaly, amor‑
phous, aphasia, %asphyxiate). The distinction in pronunciation is not entirely predictable, but
it does seem to depend to some extent on the degree to which the form has been lexicalized,
with a higher degree of lexicalization correlating with reduction of the vowel. The form an‑ is
always /æn/.
The prefix un‑ is orthographically invariant, but at least in rapid, informal speech can
undergo assimilation of /n/ to the point of articulation of a following stop. See Chapter 9 for
details.
The remaining negative and privative affixes are largely invariant, apart from two phenom‑
ena. There are a number of forms like deport, depose, deprive, detract, devolve that have no
secondary stress on <de>. In fact, it is not even clear whether these forms are morphologi‑
cally complex as their meaning cannot be derived on the basis of the would-be prefix and the
base. Second, there is a small handful of forms with dis‑ that take stress on the prefix, rather
than on the base: disparate, dissonant, discord, discount, disjunct, dissolute, dissident, dissipate. It is
not clear, however, that there is anything systematic about this stress shift, especially since
we, again, have to deal with bound bases and semantic opacity in most cases. In a few cases
 17.2 Formal considerations 361

(e.g. discord) it may be a matter of the sort of stress shift we find with other Latinate noun–
verb pairs (e.g. pérmitN versus permítV), but this does not account for all of the cases.

17.2.3 Productivity
It is usually said that of the negative prefixes in‑, un‑, a‑, dis‑, and non‑ only un‑ and non‑ are
robustly productive in contemporary English. Bauer and Huddleston (2002: 1688), for exam‑
ple, echo earlier literature (e.g. Jespersen 1917; Zimmer 1964) in saying that ‘Negative in‑ is
no longer productive, while a‑ and dis‑ are now rarely used to create new words.’ The corpora
suggest that this is not quite right, however. While it is certainly true that un‑ and non‑ are
the most productive of this cohort of affixes, it is not entirely clear that in‑, a‑, and dis‑ are
completely unproductive. This productivity is shown in new words illustrating the use of
in‑ (7a), words where we find in‑ where some other allomorph might be expected (7b), and
words where we find in‑ even though a different negative prefix is established (7c).

(7) a. new words

immedical, inactual, inadult, inattentional, incompoundable, inconservative,
indescriptible, indominable, inexorcizable, inexplicatable, inextractable, injuvenile,
intesticular
b. words with an unexpected allomorph
inbalance, inbearable, inbelievable, inperfections, inpracticable, inprecision,
inprescribable
c. words where another prefix is established
inbalance, inbearable, inbelievable, inconservative, incordial, indisciplined,
induplicable, inerect, inesthetisable, infungibility (no infungible in OED),
ingovernability (no ingovernable in OED), inintelligent, inobjective, inspecific,
insupportive, immorbid, impleasant, irreal, irrelational, irrhythmic, incertainty (last
OED attestation 1792), imprompt (last OED attestation 1759)

Although Zimmer (1964) claims that a‑ is unproductive in contemporary English, and Bauer
(1983: 218) that the affix is only ‘marginally productive’, we find to the contrary that this prefix
is still surprisingly productive in specific registers like medical or academic writing. Indeed
we find quite a few forms in a‑ or an‑ in COCA that either do not occur in the OED or were
last attested in the seventeenth or eighteenth centuries, suggesting them to be novel coinages
in their recent appearances:

(8) a. abiotrophic, acaucasian, achronic, acontextual, acultural, adeictic, adimensional,

agametic, agenic, ageotropic, aglycal, alanate, amelanotic, ameloblastic, amenorrheic,
amillennial, amodal, amotivational, anarchival, anestrous, anhedonistic, anhepatic,
anhydrobiotic, anhypothetical, anisogamic, anisogamous, anparasitaemic, apoetic,
362 Negatives

areligious, arhetorical, arrhythmogenic, asectic, asemantic, astereotypical, astrategic,

ateleological, athermal, avenal, averbal
Style 1990: The male’s “castration anxiety” is just as “anhypothetical” […] a
b. 
hypothesis as the woman’s “penis envy.”
Houston Chronicle 1999: The pope and his lieutenants emphasize that their
millennium program is not done with an eye toward the apocalypse or Jesus’ second
coming. Catholic “amillennial” teaching, they say, is that the end of time is a
historical process that has already begun.
Good Housekeeping 2008: But not all melanomas conform. One type, amelanotic
melanoma, for example, has no brown or black.

The prefix dis‑ is at least marginally productive on bases of all kinds (contrary to, e.g., Zimmer
1964). We find apparent new formations with dis‑ such as those in (9) from COCA:

a. on adjectives: disconjugate, dishistorical, disordinal, displastic, discheerful,

(9)
dis‑aggrieved, disalienated, disfathered, disglobalized, disnarrated, disorbited,
disshelved, distopic
on nouns: discensus, disidentity, distolerance, disbond, disgodding, disrealness
 on verbs: dis‑align, disattend, discomplicate, disconstrue, dis‑elect, disenroll,
disenthrall, disimpact, dispray, dissanction, disadjectivize
Art Bulletin 1994: In terms of its qualities of displacedness, dispersal, and
b. 
distribution, however, networked history might more appropriately be called
dishistorical.
 American Studies International 1993: Only by the play of discensus—not
consensus—can we anticipate change and live together heterogeneously?
 Atlantic Monthly 1997: a taxi took him . . . to Posse Motors, where lax salesmen drifted
like disorbited satellites and where he bought a second hand Cadillac.
 PBS Newshour 1996: It unelected Bill Clinton’s Democrats in 1994, it elected a
Republican Congress. It’s about to dis-elect a Republican Congress in ’96.

Although the three prefixes already discussed are generally more productive than might have
been thought, it is nevertheless quite clear that the most productive of the negative prefixes
are un‑ and non‑. Un‑ is the default negative prefix for adjectives and for verbs that have the
requisite semantic characteristics to give rise to a reversative reading (see Section 17.3.2). We
find un‑ used on a wide range of monomorphemic adjectives, including ones which are them‑
selves negative in tone (unsordid, unpoor, unugly) and ones that have obvious monomorphe‑
mic antonyms (unrich, unlarge, unheavy); see Section 17.3.5 for further discussion. It is
especially productive on adjectival participles of various sorts and on adjectives derived with
such suffixes as ‑able, ‑ous, ‑ic, ‑al, and ‑ive. Non‑ is also fully productive on adjectives, and is
the preferred negative prefix on nouns.
 17.2 Formal considerations 363

What is perhaps surprising, however, is the degree to which un‑ has become productive on
nominal bases. This is pointed out by Horn (2002), and is corroborated by the data we find
in COCA. A few of these are item-familiar (unbirthday, unperson), but quite a few are novel
forms.

(10) un‑advice, un‑ass, unbachelor, un‑baguettes, unbirthday, unblessings, un‑boomer,

un‑brat, un‑car, un‑chicken, unchurch, uncircumcision, un‑commercial, uncorrection,
undeath, undiet, undinner, un‑difference, un‑diva, undude, unevent, unexistencehood,
un‑fact, unfaith, unfreedom, ungame, unhealth, unhero, un‑hollywood, unlife, unlogic,
un‑music, unperson, unquality, unrock’n’roll, unsleepover, unsuccess, unsun, unsystem,
un‑tundra, un‑turkey, unveracity, un‑war

As pointed out in Section 17.2.1, some of these appear in the corpus functioning as premodi‑
fiers, allowing for the interpretation that un‑ can be category-changing when attached to a
noun.
Un‑ also exhibits a small degree of productivity with compound and phrasal bases. We find
several examples like those in (11).

(11) unblockbusterish, un‑boondocksy, uncomputer-savvy, undoublespace, un‑bad-guy-

like, un‑deep-friable, undinosaur-looking, un‑fall-in-love, un‑hello-kitty-like

The privatives de‑ and ‑less are highly productive in contemporary English. De‑ regularly forms
new verbs both from nominal bases and verbal bases, especially those derived with the suffixes
‑ize (dehazardize, depenalize) and ‑ify (de-erotify, denitrify), but also with verbs in ‑ate (deaer‑
ate, degranulate), and less frequently with underived verbs (deblur, declutch). The suffix ‑less
attaches to nouns with great freedom; only rare examples can be found formed on verbs.
Both of the pejorative prefixes exhibit some degree of productivity. The prefix mis‑ is prob‑
ably the more productive of the two, with the following examples appearing in COCA that
are not recorded in the OED.

(12) a. 
mis‑activate, misarticulate, misassemble, misbalance, misbase, mis‑beginnings,
miscarved, miscognition, misconfusion, miscover, miscrawl, misdropped, misflying,
misforgotten, misfuel, misgrab, misgruntlement, misgust, misintuit, misjustice,
mismix, misnavigate, misnomial, mis‑overheard, misplot, mispreread, mis‑raise,
misreputation, misshipment, mis‑species, mis‑spin, mis‑story, mis‑sync,
misunderattribute, misunderestimate, miswant
CNN_Sonya 1993: But then we get to the justice system, and you and I first met
b. 
many years ago dealing with the misjustice of the justice system with your own
daughter, Dominick.
Psychology Today 2002: In reality, note Gilbert and his colleague Timothy Wilson,

Ph.D., of the University of Virginia, we often “miswant.” People who dream of a
364 Negatives

holiday on a deserted tropical island may in fact be disappointed when they see “how
much they require daily structure, intellectual stimulation or regular infusions of Pop
Tarts.”

We find apparently novel formations on nouns (misjustice, mis-story), and verbs (miswant,
miscarved, miscovered). The existence of complex formations like mispreread and misunderat‑
tribute shows that mis‑ is also productive before other prefixes.
There are also a few apparently novel forms with the prefix mal‑, although not nearly as
many as we find with mis‑, but perhaps because this prefix in general exhibits fewer types in
COCA than the others.

(13) a. 
malaligned, malapportion, malgoverned, mal‑hygienic, malknotted, malplaced,
mal‑socialized, malunited, maldigestion, mal‑intent, malplacement, malunion
Internet Journal of Gastroenterology 2007: Infection in humans arouse from ingesting
b. 
water plants (eg. wild watercress) with metacercaria in mal‑hygienic conditions.
Kenyon Review 1998: It limps when it walks, one leg bent out at an angle, like a

malplaced cane.
Orthopoedic Nursing 2006: Acute compartment syndrome, infection, and malunion

or nonunion of a fracture can lead to loss of a limb or impaired limb functioning.

17.3 Semantic considerations

17.3.1 Types of negativity
In order to understand the semantic range of this cohort of affixes, we must first clarify some
terms. At various points in this chapter we have used terms like ‘negative’, ‘reversative’, ‘priva‑
tive’, and ‘pejorative’ in an intuitive way. Here we first try to make these terms precise and add
some other necessary terminology, and then examine the meanings of complex formations in
light of them.
We can use the general term ‘negative’ as a default for readings that are neither reversative
nor privative nor pejorative, but in some cases it will be useful to further distinguish among
several sorts of negativity. Before turning to these distinctions we need to mention a negative
interpretation that is very similar to clausal negation. This type of negation can only be
expressed by a subset of verbs prefixed with dis‑, namely those expressing psychological states
(e.g. disagree, dislike, disobey).
Of the other types of negation, the contrast between so-called ‘contrary’ and ‘contradic‑
tory’ negatives is well-described in the literature ( Jespersen 1917; Zimmer 1964; Horn 1989).
The distinction is predicated on the relation obtaining between the negative ‘not-X’ and its
corresponding positive ‘X’. In contrary negatives ‘X’ and ‘not-X’ are terminal points on a
 17.3 Semantic considerations 365

gradable scale, such that there are conceivable points between them. For example, between
happy and unhappy there are intermediate states such that someone can be neither happy
nor unhappy. With contradictory negation, something is either ‘X’ or ‘not-X’; there is, for
example, no intermediate state between animate and inanimate or between attached and
unattached. Contrary and contradictory negative readings correlate with the gradability of
their bases; to the extent that we can conceive of a given base as denoting a scale, a contrary
reading will be possible. If the two polar opposites exhaust all possibilities, the negation will
be contradictory.
It may be worth digressing for a moment to note the difficulty in determining unequivo‑
cally whether a given adjective is gradable or not (see Chapters 6 and 14 for further discus‑
sion). Judgements on gradability are notoriously labile, differing from speaker to speaker, and
from one context to another. Another problem is that even adjectives that would seem to be
clearly ungradable frequently have possible interpretations that are gradable. This phenome‑
non is also known under the label of coercion (see Chapter 6.2, Chapter 25), that is the con‑
textual interpretation of an expression that would otherwise not show this type of polysemy
(Swart 2011). This said, it is possible that readers will sometimes disagree with our judge‑
ments on whether a given word has a contrary or contradictory reading.
To the relatively well-known distinction between contrary and contradictory negation, we
might add two further distinctions. One of them will be useful in describing the semantics of
the negative affix a‑. That is, there is a mode of negation which presupposes a scale or a pair
of polar opposites, but in which what is denoted is not one pole in the opposition or one part
of the scale, but the complete irrelevance of the scale or polar opposition in question. In this
kind of negation what is highlighted is a semantic space entirely external to the scale. So
something which is amelodic is neither melodic nor unmelodic (or anything in between), but is
outside the realm of melody. For lack of an established term we will call this sort of negation
scale-external negation.
The other term that we will need to add is ‘stereotype negation’. In cases of stereotype
negation, a noun is taken to denote a bundle of characteristics or qualities {x, y, z,…}.
When certain affixes are attached, what is negated is not the meaning of the noun as a whole,
but a number of its semantic characteristics or qualities. The resulting derivative still gener‑
ally denotes the same entity or something close to it, but one that is missing several key
characteristics. In effect, the noun denotes a non-stereotypical exemplar of its category. We
will return to this kind of negation below when we discuss non‑ and un‑ on nominal bases.
We will use the term reversative for verbs like untie which can be roughly paraphrased
as ‘reverse the action of tying’. ‘Reversative’ is not a primitive notion, however. Rather, for
verbs whose meaning implies some sort of result, reversatives occur when an affix negates the
result, as has been pointed out by Marchand (1969) and Horn (1989, 2002). So with a verb
like zip whose result is that the object is ‘zipped’, a negative prefix will denote an action such
that the endpoint is an object which is ‘not-zipped’. The derived verb is thus interpreted as a
reversal of the action.
366 Negatives

The term privative is used for verbs like debug or demilitarize, or adjectives like shoeless.
So at the core of a privative based on ‘X’, we have ‘no X’. With an affix whose output category
is verb, the output normally has both a causative and a negative meaning, even though the
base of the verb does not have that causative meaning. The causative sense typically accom‑
panies the negative meaning ‘no X’, giving ‘cause to be no X’. So to debug is to cause there to
be no bugs, to demilitarize is to cause there to be no military. The result is a reading which
implies removal. Clearly, with verbal bases derived with ‑ize, ‑ify, or ‑ate, the causative sense
may already be supplied by the suffix. With ‑less, a sense of absence or loss is implied; so
shoeless denotes the absence or lack of shoes.
We turn finally to pejoratives, which arguably lie somewhere between the morphology of
negation and evaluative morphology (see Chapter 18). The pejoratives we are considering
here are morphemes like mal‑ or mis‑, in which a negative evaluation is associated with the
base. For a base X, such forms can be paraphrased as ‘X wrongly or badly’. What is targeted
by the negative in such cases is the manner in which an action is done.
Let us summarize the above theoretical considerations by listing the different semantic
categories and their semantic interpretation. The brackets contain the argument on which
negation operates, ‘X’ stands for the denotation of the base and ‘x, y, z’ for specific properties
of X.

• Reversative: neg [result]

• Privative: neg [existence of X]
• Negative: neg [X]
Contrary: intermediate states possible between X and not X
Contradictory: no intermediate states between X and not X
Scale external: neg [scale defined by X]
Stereotype negation: X [x, neg-y, neg-z, . . .]
• Pejorative: neg [manner]

17.3.2 Ranges of meaning

What is most interesting to note with the cohort of negative affixes is that they do not carve
up the semantic nuances of negation in any neat one-to-one fashion. While a few of the
affixes appear to have more or less fixed semantic content, and a few have a tendency towards
one reading or another, the majority of them are polysemous, with specific readings depend‑
ing on the nature of the base, the degree of lexicalization, and in novel forms often on the
context in which a given form is used. In other words, for some affixes, multiple readings are
available and result from the interaction of a general affixal negative meaning ‘not’ with the
semantic composition of a base. We summarize the correlations of possible readings with
affixes in Table 17.2.
Table 17.2 Distribution of readings
Base Category Category change Negative Reversative Privative Pejorative
contrary contradictory stereotype negation scale extension
Adjective
in‑ ✓ ✓
un‑ ✓ ✓
non‑ ✓ ✓
dis‑ ✓ ✓
a(n)‑ ✓ ✓ ✓ ✓
mal‑ ✓
Noun
in‑ (✓)
un‑ sometimes A or V ✓ ✓
non‑ sometimes A ✓ ✓ ✓
dis‑ ✓
de‑ V ✓
mis‑ ✓
mal‑ ✓
‑less A ✓
‑free A ✓
Verb
un‑ ✓ ✓
dis‑ (✓) ✓ ✓
de‑ ✓ ✓
mis‑ ✓
mal‑ ✓
368 Negatives

We begin with the affixes that appear to be relatively fixed in meaning. Among these are
the pejoratives mis‑, and mal‑, which seem to be used almost exclusively in this reading. With
the exception of a couple of forms to be found in COCA (mislike, used in context to mean
‘dislike’; misfortunate, used in context to mean ‘unfortunate’), mis‑, and mal‑ consistently give
a negative evaluation of manner, that is, they add meanings like ‘wrongly, mistakenly, badly’
to the bases they attach to.
Also quite fixed in meaning is the suffix ‑less. It is consistently privative in interpretation,
but often with an evaluative component, especially if compared with corresponding com‑
pounds with free as the head (Górska 1994; Slotkin 1990). For example, although both salt-
less and salt-free denote the absence of salt, the former marks that the absence of salt is a
negative or at best neutral quality, whereas the latter generally suggests that the absence of
salt is something desirable. There are some instances, however, where the two rival forms
seem all but identical in force, as the examples in (14) suggest:

New York Times 2007: Kathleen Marshall directs and choreographs a charisma‑free ensemble,
(14) 
whose leads (Max Crumm and Laura Osnes) were cast via reality television (2:15).
Wolf at the Door 2006: “Drank three pints straight from the bags before I was fit company

again,” the charismaless wonder said, peppering his delivery with a few smug chuckles.
[…]

At best charisma-free conveys a slightly ironic edge that charismaless lacks, but the difference
is very small.
For the remaining affixes, it is usually the case that the wider the range of bases to which
an affix can attach, the wider the range of readings that it is compatible with. For example,
in‑ occurs almost exclusively with adjectival bases, there being only a few examples of in‑ on
nouns. We find that in‑ is consistently negative in force, either contrary or contradictory
depending on the gradability of its base. For example, inappropriate or incompatible are con‑
traries, it being possible for things to be more or less inappropriate or incompatible. But inani‑
mate and inexact are contradictory (for us, with the usual caveats). A very large proportion of
in‑ words are high-frequency and item-familiar, and with these characteristics, as might be
expected, we often find lexicalization. For example, indifferent does not mean ‘not different’
but rather ‘neutral’, and inhuman does not mean ‘not human’, but ‘cruel’.
The prefix a‑ is also confined to adjectives and bound bases, covering most of the range of
negative readings (contrary, contradictory, and scale external), with a few examples of priva‑
tives also attested:

(15) a. contrary
 Journal of Sports Behavior 2009: First, it may be that the same behavior is seen as
equally aggressive when exhibited by women or men but that the women are
punished more because, for women, the act is astereotypical.
 17.3 Semantic considerations 369

b. contradictory
 Symposium 1999: However, Todorov also demands a non-allegorical, apoetic
reading of the text, which means that fairy tales are necessarily and unfortunately
excluded from the genre, instead classified as the “merveilleux” (59).
c. scale external
 Raritan 1990: Viewed this way, the inversions of carnival where kings and paupers
exchange clothes, the subversions of jokes where, say, the well-dressed lady slips on
a banana peel, and the reversions from self-conscious and constrained adulthood to
playful or spontaneous childhood that are prominent in texts like Alice in
Wonderland let us abandon, imaginatively at any rate, conventional social
differences for participation in delightfully “real” human unions or communions
that are acultural and ahistorical in nature.
d. privative
 Fantasy & Science Fiction 2005: The Blemmye had no head; he was acephalous.

It would appear that the default affix for forming scale-external negatives from adjectives is
a‑, but a‑ is not restricted to this sense.
Also relatively limited in scope is de‑. Much of the semantic range of this prefix can be
attributed to its formal characteristics—namely that it attaches to nouns or verbs and uni‑
formly derives verbs. The vast majority of words derived with de‑ are either privative or
reversative in force. On nouns, we find almost exclusively privative forms, as these COCA
examples illustrate:

Prevention 2007: It significantly helps to degunk her pores, which, she says, “started
(16) 
looking bigger with age.”
US News and World Report 1993: It costs between $350 and $1,800—depending on

helicopter flight time—to dehorn a single rhino, and the procedure has to be repeated
every three years as the horn regrows.

This is especially consistent where the nouns are concrete, count nouns. Interestingly, differ‑
ent arguments can be referenced. While with most formations the entity denoted by the base
is the entity that is removed, there are cases where the source can be referenced, instead of
the object. Thus, dethrone means ‘remove from the throne’, not ‘remove the throne’ (cf. also
decart, debus).
On verbal bases, either reversatives or privatives may be found, although the distinction
between the two is sometimes hard to make. If the verbal base is itself derived from a noun
(as in decarbonize, dechlorinate), we often find the same sort of privative reading that we find
on purely nominal bases. Where the verbal base is derived from an adjective, the reversative
meaning is dominant:
370 Negatives

(17) a. privative
 ABC_Special 1994: On the other hand, if you take decaffeinated coffee, sometimes
the process to decaffeinate it has exposed the coffee to a chemical that isn’t good
for you.
 Popular Mechanics 2009: Mechanical, chemical and thermal energy act together in
your central air-conditioning system to cool and dehumidify indoor air.
b. reversative
 Washington Post 2008: Following the adoption of a new constitution in 2005 and the
first multiparty elections in decades, Congo is aiming to decentralize government
from the capital of Kinshasa to newly elected legislatures in 11 provinces.
 Review of Contemporary Fiction 2006: If the great sorcerer showed them under the
title, “How to de‑erotify the erotic object,” they would be the epitome of what not
to do.

For some verbs, either interpretation is plausible: for example, to dechlorinate can equally
well be construed as meaning ‘remove the chlorine from’ or ‘reverse the process of
chlorinating’.
The prefix non‑ attaches productively to both adjectives and nouns. On adjectives it shows
both contrary and contradictory negative readings:

(18) a. contrary
 Children’s Digest 2002: There are different types of fish that will be best suited for the
type of tank you choose. Choose from the four main categories of freshwater fish:
goldfish, non‑aggressive, semi-aggressive, or aggressive.
 CNN_LiveSun 2003: I considered it but because my parents are divorced I took it
very, very, very seriously and I’ve always said that if I was to be engaged that it would
be a very non‑traditional engagement and I would take a very long time.
b. contradictory
 African Arts 1996: The same associations underlie the geometric designs and pictorial
ideographs on both mnemonic devices like the lukasa and royal emblems in
non‑beaded media such as wood, metal, painting, pottery, and weaving.

Where non‑ occurs on nouns, we most frequently get a general negative reading, the distinc‑
tion between contrary and contradictory being neutralized, as nouns are not gradable or
scalar:

Denver Post 2005: Myers wears a green Pirates cap and is the only non‑coach allowed
(19) 
on the St. Mary’s sideline during games.
Houston Chronicle 1995: Composed of six dentists and six non‑dentists, the committee

figures to play an important role as the Health Department tries to improve its much-
criticized system of detecting fraud and abuse in the state’s Medicaid dental program.
 17.3 Semantic considerations 371

A non-coach is someone who is not a coach, a non-dentist someone who is not a dentist.
Contrasting with examples like these, we also sometimes find non‑ used as a stereotype
negator. A lexicalized example is the word non-person, which denotes someone who is indeed
a person but is not acknowledged as a person, or is not treated fully as a person. Algeo (1971:
92) offers the example of nonbook, which is intended to denote the sort of coffee-table book
that has lots of pictures, not much text, and is meant to be flipped through, rather than read
cover-to-cover. Non-lexicalized examples of stereotype negating non‑ are not too difficult to
find, as the examples from COCA in (20) suggest:

The Paris Review 1997: The man in the tweed suit wore his hair clipped short, in a crew
(20) 
cut. It was a flat metallic color, a non‑color, like his eyes.
NPR_Weekend 1992: Well, he’s not saying he has the answers. That’s—that’s what’s so

novel about this campaign. It’s almost like a non‑campaign. It—it’s like we’re running
to get him on the ballot.

In the given context, a non-color is a colour, but one that lacks the vibrancy of a true colour.
A non-campaign is a campaign, but one that is not characteristic in certain ways. As Algeo
(1971) points out, non‑ words of this sort are frequently used with a disparaging tone.
As mentioned in Section 17.2, nominal non‑ words often occur in prenominal position, and
in such cases we might argue that non‑ has had a category-changing function; where this
occurs, the non‑ word can have a privative reading. We have in mind here forms like non-fat,
as in non-fat milk, or non-fish, as in the example in (21):

San Francisco Chronicle 2000: My only regret with these non‑fish sushi was the absence
(21) 
of two wonderful flavor elements: shiso, a fragrant relative of mint; and ume or Japanese
plum, a tangy, pickled fruit paste that’s delicious in sushi.

In cases such as these, the interpretation of the non‑ words is close to that of adjectives derived
with ‑less, the difference being that whereas ‑less can evaluate the absence of the base nega‑
tively, non‑ is neutral in evaluation.
The affixes with the widest range of potential readings are un‑ and dis‑, both of which
attach to adjectival, nominal, and verbal bases. On adjectival bases, un‑ productively derives
both contrary and contradictory negatives, but apparently not scale external readings.
Although there are clearly many high frequency item-familiar forms among the un‑ words,
we in general find less lexicalization than we find with in‑, and the vast majority of forms
are completely transparent in meaning. Examples like unactorish, unbacked-up, unbeany,
un-chain-store-afflicted, uncherrylike, unduckable, uneventual, and so on, are easily found in
the corpora (these from COCA). They are, not surprisingly, not recorded in the OED, and
surely have no need to be, their meanings being so clearly compositional. Indeed it is hard
to conceive of any derived adjective that could not potentially be negated with un‑, even
372 Negatives

when lexicalized alternatives with in‑ are available; in COCA we find cases like unaccessible,
uncomplete, uncorrect, uneffective, among many others for which there are well-established
in‑ forms (see Section 17.3.4 for further discussion). Similarly, there are few underived
adjectives to which un‑ cannot be attached, even those adjectives that have clear mono‑
morphemic antonyms (ungood as opposed to bad), and those that have evaluatively nega‑
tive semantics (unugly as opposed to unattractive). Since the existence of forms like these
has been a matter of controversy in the literature, we will explore this issue in detail in
Section 17.3.5.
In contrast to un‑, the prefix dis‑ occurs only infrequently on adjectival bases, producing
either contrary or contradictory negatives, and as mentioned above, seems to be only slightly
productive with this category of bases. We find examples in COCA like displastic, dishistori‑
cal, disordinal, disconsonant, discontiguous, dishonest, disingenuous, disloyal, dissimilar, dissocial,
and distemperate. The first four of these are not recorded in the OED, and occur in medical or
scholarly contexts.
It has been pointed out several times in the literature ( Jespersen 1917; Zimmer 1964;
Marchand 1969; Algeo 1971; Horn 2002), that un‑ and dis‑ adjectives frequently carry a depre‑
ciatory nuance that is absent with non‑. Examples that are often given to support this claim
are pairs like unAmerican as opposed to non-American, or unprofessional as opposed to non-
professional (Horn 2002: 9). This appears to be the case from pairs like those above, but we
have pointed out that non‑ can be quite depreciatory, when used as a stereotype negator
(Algeo 1971). We will return to this topic in Section 17.3.5 and evaluate it in further depth
there.
Dis‑ and un‑ also both occur on nominal bases. The prefix dis‑ generally expresses priva‑
tion. For example, disanalogy is privative, according to the OED, which defines it as ‘want of
analogy’:

Monist 1998: Surprisingly perhaps, it does not follow from this concession that there is
(22) 
a radical disanalogy between divine creation and intrasubstantial causality.

We find a similar force in examples like disfunction, disbelief, disaccord, and distrust, all of
which express the lack of what is denoted by their base.
The prefix un‑ can form privatives, but unlike dis‑, it can also act as a stereotype nega‑
tor. Examples like unfaith, unfreedom, unhealth, and unsuccess seem to have the privative
reading:

Christian Century 1995: I have sought for some years to find a theological dialogue
(23) 
where a serious methodological discussion is taking place about how to draw some line
between faith and unfaith, between orthodoxy and heresy.
Anthropological Quarterly 2004: In a way, while the socalled real kin are re-assigning

the kinship a new morality of humanitarianism beyond consanguinity, as in the case
 17.3 Semantic considerations 373

of Mr. Yu’s daughters, the invocation of kinship relations is multiplying, reflecting

the strained conditions of unfreedom under the paternalistic totalitarian regime of
North Korea.
Southern Review 1999: The life had gone out of her red hair, and what should have been

pale Irish skin was pasty, the picture of unhealth.
Lens of the World 1990: I disapproved heartily and wished them every sort of unsuccess

if only they didn’t drag me down with them.

In examples like unchurch, undiet, and unsleepover party on the other hand, un‑ has the force
of a stereotype negator:

CNN_SunMorn 2006: It’s not the traditional place of worship [. . .] Deron Cloud,
(24) 
founder and pastor of the Soul Factory, called it the “unchurch”.
Shape 2000: She calls her new way of eating the “undiet” because it has no restrictions—

and it and her weight loss have been lasting.

Clearly in some ways unchurches and undiets really are churches and diets, but they lack cer‑
tain key characteristics of stereotypical ones.
Horn (2002: 30) rightly observes that in some cases, the un-X noun actually denotes
something that is in fact not an X, but which nevertheless can be compared to X or consid‑
ered in relation to X by virtue of sharing certain characteristics of X. Consider the example
in (25):

Parenting 1999: Host an “unsleepover” party: The guest comes dressed in pajamas and
(25) 
brings a sleeping bag to lounge around on, and the kids eat, play games, and watch a
video while they munch popcorn. At around 9 P.M., the parents pick up their child,
and everyone gets some sleep that night.

In the case of an unsleepover, we do not in fact have a sleepover party, as the children go home
to sleep. But we do have many of the qualities that can be found in a stereotypical sleepover:
wearing pajamas, bringing a sleeping bag, watching movies, and so on. Un-cola and un-turkey
are other examples of this sort. As Horn points out, not just anything can be an un-cola; in
order to be one, a product has to be at least functionally similar to a cola (e.g. something to
drink, preferably sweet and fizzy).
We turn now to un‑ and dis‑ on verbal bases. It appears that un‑ and dis‑ produce similar
readings where they attach to verbs that belong to the same semantic class; however, to some
extent the two prefixes select different kinds of verbal bases. Un‑ is quite consistent in select‑
ing for verbs that imply non-permanent results. The vast majority of these are straightfor‑
wardly causative in nature (26a), although a few can be found with both causative and
inchoative senses as well (26b):
374 Negatives

(26) a. 
unclip, uncompress, undelete, unentwine, unfix, ungarble, unhitch, unkennel,
unmold, unname, unpersuade, unrelease, unsell
b. untense, untwist, uncoil, uncurl, unfold

Whether the verbal base is purely causative or alternates between causative and inchoative
depends largely on whether the action denoted is one in which external causation is neces‑
sary. Pure causatives presuppose external causation; verbs that alternate between the senses
allow either internal or external causation. On bases such as those in (26), the un‑ verb is
reversative in force, that is, denotes the negation of the result.
There are, however, at least some instances where the verbal base lends itself to a privative
interpretation, or to an interpretation which could be either privative or reversative. Among
these, Marchand (1969) cites unsex and unvoice, and to these we can add unbridle, unblock,
uncap, unchain, unclog, uncover, undress, unkink, and a variety of others in which the base may
be analysed as demonimal.
Interestingly, although un‑ typically does not attach either to verbal bases that are non-
causative or to bases that denote results that are permanent, the prefixation of un‑ to such
bases is not impossible; where it occurs it has the effect of coercing the verbal base to be
semantically like the verbal bases that un‑ normally selects. That is, un‑ can take a stative, activ‑
ity, or other kind of verb and force it into a causative/inchoative verb that implies a reversible
result. Consider the examples in (27), where un‑ occurs on a stative verb (27a), a verb of per‑
ception (27b), and a verb whose result is normally taken to be non-reversible (27c):

(27) a. stative verb

 Iris 1997: From the broad window of the caf, the office buildings erupt. Bodies spill
out in their gray and black suits. In one of her classes, she shows an avant‑garde film.
A hand is opened. Insects swarm out. She thinks of this now as the buildings
uninhabit.
b. verb of perception
 Cross Currents 2001: When hunger is understood as an acceptable reality in the
politics of competing states, the conviction that hunger is intolerable is itself
revolutionary. The attempt to unimagine its necessity in our present actions
transcends and relativizes national commitments.
c. verb with non-reversible result
 Bioscience 2005: This result was most surprising and unexpected, yet consistent with
the hypothesis that ancestrally both sexes may have expressed thoracic horns in this
species, and that females lost their horns by evolving mechanisms to un‑grow them,
rather than not growing them in the first place.

In the context above, to uninhabit is ‘to become uninhabited’, to unimagine is ‘to remove from
the imagination’, and to un-grow is to reverse the process of growing. Horn (2002: 15) observes
 17.3 Semantic considerations 375

that such un‑ verbs always denote ‘telic achievements’, using the terminology of Vendlerian
verbal classes, regardless of the class to which the base would normally belong.
It is also relatively common to find verbs like those in (28) in what Horn (2002: 16) calls
counterfactual contexts, that is, contexts that cancel the seemingly anomalous interpretation
of the verb:

(28) Fantasy & Science Fiction 2003: And once you’ve seen it, you can never unsee it.
CNN_Sonya 1993: The other big difference is once you have AIDS, you can’t unhave it.

Southwest Review 2007: She saw that she had hit a nerve and that there would be no way

to unhit it.

The negative contexts in effect highlight the unreversibility of the implied result.
The prefix dis‑ frequently attaches to the same sort of verbs that un‑ favours: causative or
causative/inchoative verbs that imply reversible results. The resulting derived verbs are
reversative or sometimes indeterminate between reversative and privative readings. We find
examples like the following:

(29) disarm, disband, disbud, disconnect, discredit, disembed, disinfest, disinvite, disrobe,
disthrone

Many of these are transparent in meaning, but quite a few are lexicalized, among them dis‑
charge, disclose, discover, discredit, disgorge, disintegrate, dismantle, and so on. Indeed, lexicaliza‑
tion seems more common with dis‑ verbs than with un‑ verbs.
Where dis‑ differs significantly from un‑ is that when it attaches to non-causative verbs, it
generally results in either negative or pejorative verbs:

(30) a. negative
 disaffirm, disagree, disallow, disapprove, disbelieve, disconfirm, discontinue,
discount, disimprove, disincline, dislike, dislove, disobey, displease, dissatisfy,
disuse
b. pejorative
 discolor, disidentify, disinform, disremember, disvalue, disconstrue, dishear,
dispray

In other words, although dis‑ sometimes attaches to verbal types other than causative/inchoa‑
tive, affixation of this prefix does not have the effect of coercing these other verbal types to a
causative/inchoative interpretation. Rather, it delivers either a simple negative or a pejorative
reading, depending on the nature of the base verb.
376 Negatives

17.3.3 Redundancy
It has been noted in the literature that the negative verb-forming affixes are sometimes used
redundantly such that the sense of the bare base and the sense of the base with the negative
affix are identical. Both Thomas (1983) and Horn (2002: 19) observe that this is the case for
verbs like unravel, unloose, unrip, and unthaw. In addition to these examples, we find in COCA
examples like unpeel and unpry which in context mean precisely the same thing as their
bases:

American Studies International 1997: On a slanting stone lay a drowned man, naked,
(31) 
swollen, purple; clasping the fragment of a broken bush with a grip which death had so
petrified that human strength could not unloose it . . .
The Mercury Visions of Louis Daguerre 2006: Louis cracked an egg on the side of the

plate and began to unpeel it.
Sewanee Review 1998: It would take near an hour to unthaw the sand, which meant a

shorter day for the brick crew, which meant they’d be grumpy at Ray and old Henry,
the other hodcarrier.

Echoing Covington (1981: 34), Horn attributes this seemingly redundant interpretation to
the propensity of reversative un‑ to return the base verb to what he characterizes as a state of
entropy, that is, movement from a more marked state to a less marked one, for example, from
closed to open, from frozen to thawed, and so on. If the verbal stem intrinsically denotes
movement towards the state of entropy, the un‑ does not reverse the direction of movement,
but reinforces it.
As with un‑, the affixes de‑ and dis‑ are sometimes found in redundant forms. For de‑, Horn
(2002: 20) gives such examples as debone, dehusk, deworm, and for dis‑, disannul, and dissever.
To these we can add debar, degut, dehull, delimit, dedust, de-joint, dejuice, depulp, de-shell,
­de-skin, and de-stem, all attested in COCA with the redundant meaning. It is worth noting
that the bases to which de‑ attaches are sometimes ambiguous, having the capacity to mean
both the action of putting in (what Horn calls goal-orientation) and the action of removing
(source-orientation). It is likely in these cases that the affixation of de‑ is felt to disambiguate
the base. As Horn (2002: 22) says, ‘But when goal- and source-oriented readings are both
plausible for a given base verb in combination with a given patient, the un‑ or de‑ verb will
serve usefully and unambiguously to signal the source or entropic interpretation.’
An alternative analysis in all of these cases is that de‑ is a class-changing prefix and that the
bases are nouns. In some instances, conversion (to skin) and de-prefixation (to deskin) can
operate on the same bases and give rise to synonymous forms, without there being any real
redundancy.
While Horn generally seems correct about the propensity of negative verbal affixes to
signal a change of state moving towards entropy, we should probably see this as a tendency,
 17.3 Semantic considerations 377

and not a strict rule. Given the right context, it is quite possible for an un‑ verb to signal a
change of state away from entropy, as we see in this example from COCA:

Deep Rising 1998: FI[R]ST MATE We have a main frame meltdown!! CANTON Well
(32) 
unmelt it!!

Examples of either sort are relatively rare, however, and it does not seem possible to make a
conclusive determination on this point.

17.3.4 Rivals
As Section 17.3.2 has shown, there is significant overlap in the functions of several of the nega‑
tive prefixes. The prefixes in‑, un‑, non‑, and a‑ all attach to adjectives, and de‑, un‑, and dis‑ all
attach to verbs. Such clusters can be seen as cohorts of rival affixes, and the question therefore
arises whether the existence of a form with one affix blocks the derivation of forms with the
other affixes. Similarly, if we find doublets (or triplets or even multiples) with these affixes,
we can ask whether consistent semantic contrasts can be found, such that each affix can be
seen as specializing in a particular sort of negative semantics.
We can begin with blocking. It seems quite clear that there is little or no blocking to be
found in these cohorts of affixes. While there certainly are established forms in which a par‑
ticular prefix is associated with a particular base, and while native speakers often have a sense
of which forms are prescriptively ‘correct’, we nevertheless frequently find novel or low-­
frequency formations with two or more rival prefixes. Kjellmer (2005: 158), for example,
estimates that 17 per cent of the adjectives in the 57 million word Cobuild corpus he uses
occur with more than one negative prefix. Looking at the negative prefixes on adjectival
bases, we find sets like the following in COCA (not an exhaustive list):

(33) a‑ in‑ non‑ un‑

accessible inaccessible non‑accessible unaccessible
attentive inattentive non‑attentive unattentive
credible incredible non‑credible uncredible
obvious inobvious non‑obvious unobvious
plastic aplastic non‑plastic unplastic
poetic apoetic non‑poetic unpoetic
rational irrational non‑rational unrational
religious areligious irreligious non‑religious unreligious

As for our second issue, whether there is a consistent semantic value that can be associated
with each affix in sets like those above, we find ourselves at odds with the literature. Horn
378 Negatives

(1989: 282–3) argues, for example, that in‑ is ‘generally depreciatory’, evaluative, and lexical‑
ized, that non‑ tends not to be evaluative or depreciatory, and that un‑ stands somewhere in
between the two. Similarly, Bauer and Huddleston (2002: 1689), while acknowedging that
there is sometimes little difference between the affixes, nevertheless suggest that where pairs
of un‑ and non‑ words exist, ‘The forms with non‑ are emotively neutral and non-gradable,
while those with un‑ have a wider range of meaning, so that they may convey criticism and
gradability (allowing them to take such degree modifiers as very, extremely, etc.).’ Hamawand
(2009: 128) claims that synonymy of negatively prefixed forms is in fact impossible.
We, however, find the corpus data to be somewhat at odds with these characterizations.
First, it is clear, as we have suggested above, that non‑ words can be either gradable or not, and
either depreciatory or not, depending upon the base to which they are attached (see (18)).
Further, while it is true that there are sometimes clear distinctions among the words in
each set, often this is not the case. For example, it appears that inobvious, unobvious, and non-
obvious all mean more or less the same thing, namely ‘not obvious’.

Ploughshares 2002: Basha folded herself into me. Everything about her seemed perfect
(34) 
just then: her cheeks, the way her mouth smooshed vowels, her new decadence, her
pale body. She was emotionally inobvious.
Monist 1993: What these senses will tell us . . . is on occasion accurate enough as far as it

goes, but will never get us beyond descriptions of a world of variety and change and
obvious connections (ft. 54) to a description of that more significant, unified world of
unobvious connections that tends to lie concealed (ft. 123).
Harvard Journal of Law and Public Policy 2002: Thus, sexually explicit websites get traffic

through obvious and non‑obvious ways, enhancing their vast reach.

We find a similar triplet of apparently synonymous negative forms in COCA with the words
inattentive, unattentive, and non-attentive.
Where there are distinctions, these distinctions are not always to be attributed to the con‑
tribution of the affixes. For example, unplastic, non-plastic, and aplastic are all attested in
COCA and have different meanings, but their differences appear to have more to do with the
polysemy of the base plastic than with nuances of the prefixes:

(35) 
People 1992: More important, Ross captivated a whole generation with her lush-lipped
and decidedly unplastic beauty—here was someone worth snatching from the altar
(or spiriting down to Bolivia, as in the 1969 megahit Butch Cassidy and the Sundance
Kid).
USA Today 2005: He’s also added some products made from non‑plastic materials,

including wooden furniture and fabric inflatables.
Ear, Nose and Throat Journal 1996: The external canal is hypoplastic, aplastic, or ends

in a blind pouch with fistulous tracts leading toward a rudimentary TM.
 17.3 Semantic considerations 379

The un‑ and non‑ words both refer to the material ‘plastic’, although the un‑ word in (35) does
so in a metaphorical sense and the non‑ word in the literal sense. The un‑ form is clearly more
evaluative than the non‑ form, but we would observe that the metaphorical use of plastic is
itself evaluative; plastic is intended to mean ‘fake’ in this context. The a‑ form selects the
learned or medical sense of the base (something like ‘shapeable’ or ‘moldable’). The prefixes,
however, in each case add a simple negative force.
There are, of course, cases in which the base maintains more or less the same meaning, and
the affixes provide clearly different nuances. Consider, for example, the quadruplet irreligious,
unreligious, non-religious, areligious:

Ms 2003: “I also wanted to write about what a spiritual quest might look like at the end
(36) 
of the twentieth century—a spiritual quest by someone who, like so many of my
friends, was basically irreligious and resistant to spiritual thought.”
NBC_Matthews 2007: Can he be the unreligious candidate, the secular candidate who

wins the East Coast states of New York, Connecticut, New Jersey, Florida?
Anthropological Quarterly 2008: That is, the Turkish state promotes the notion that

Islam can be separated out from other types of social activities, including politics, to
create a neutral, non‑religious public space and institutions.
Church History 2002: Religious symbols and acts may have saturated life, but Guibert

had encountered someone who was (in our language) “areligious,” indifferent to such
tugs or claims.

What we find here is that the in‑ form does have a clear depreciatory flavour; irreligious in this
context implies hostility to religion. The un‑ form is more evaluatively neutral, though, being
more or less equivalent to ‘secular’. The non‑ and the a‑ forms are also evaluatively neutral, but
express scale-external negation, rather than contrary or contradictory negation. Given the pre‑
vious examples, the important observation is that the differences that we find appear to depend
more on context of use than on some inherent and consistent semantic property of the affixes.
We do, of course, find lexicalized meanings more frequently with in‑ forms than with any
of the other prefixes. For example, irrational means ‘unreasonable’, incredible can mean ‘excel‑
lent’, and inaccessible can mean ‘difficult to understand’. And such lexicalized forms are often
but not always depreciatory in sense (for example, incredible clearly is not). But there are also
many in‑ words that are quite transparent.
What we find, then, is that the data in the corpora do not clearly corroborate the claims
made by Horn, or by Bauer and Huddleston. At best we might say that in‑ is more likely to
have a depreciatory sense than un‑, and un‑ is more likely to have one than non‑, but that this
is no more than a tendency having to do with the degree of lexicalization found with each
prefix. All four rival affixes can be negative without being depreciatory. We must conclude
that lexicalization and depreciatory nuance are a matter of individual lexical items, and not of
the affixes themselves.
380 Negatives

A similar story can be told about the negative affixes that take verbal bases. With the rival
cohort un‑, de‑, and dis‑ on verbal bases, we only occasionally find triplets, but doublets are
not difficult to find, again suggesting that blocking is not a real phenomenon.

(37) a. triplets: unbar, disbar, debar; uninvest, disinvest, de‑invest

b. doublets (un‑, dis‑): unarm, disarm; uncharge, discharge; uncover, discover; uninvite,
disinvite; unmount, dismount; unelect, dis‑elect
doublets (un‑, de‑): uncenter, decenter; uncertify, decertify; uncode, decode;
c. 
uncompress, decompress; unconstruct, deconstruct; unlink, delink; unpretzel,
depretzel; unselect, deselect; untrain, detrain
d. doublets (de‑, dis‑): decolor, discolor; decompose, discompose; de‑identify,
disidentify

As with the adjectival cohort, with this set of affixes we find some pairs that appear to mean
precisely the same thing, and others in which one or more members are lexicalized or have
different meanings:

Military History 2006: British soldiers crossed the parade field to unbar the main gate,
(38) 
admitting more Redcoats.
ABC_GMA 2007: The question is, will this panel disbar him, take away his law license,

prevent him from practicing law at all, as a civil lawyer, as a corporate lawyer, as a
transactional lawyer?
Washington Post 1999: But some corruption experts said that if the accusations pan out,

the firms should be “debarred” by the World Bank—that is, prohibited from bidding
on projects funded by the bank, a major source of financing for dams in developing
countries.

Sewanee Review 1997: “People can invest other things besides money,” Ginger said,
(39) 
taking their dishes to the sink. “Maybe I already have.” # “Well, then you’d better
uninvest it,” Lucky said.
Journal of International Affairs 1992: Companies wishing to disinvest were forced to do

so in a buyers’ market; South Africans could therefore acquire former subsidiaries,
whose profits would no longer be taken out of South Africa in the form of dividends,
at cut-rate prices.
Boston College Environmental Affairs Law Review 1999: EPA has de‑invested in Part 503,

while diverting fiscal resources to other programs under the CWA.

So unbar appears to have a literal privative meaning ‘remove the bar’, disbar a lexicalized
meaning ‘to remove from the bar, i.e., to be disqualified from practicing law’, and to debar
an apparently redundant sense identical to ‘bar, i.e., to prevent from doing something’
 17.3 Semantic considerations 381

(see Section 17.3.3). On the other hand, there does not seem to be any difference in sense
between uninvest, disinvest, and de-invest; all three seem in context to mean ‘cease to invest’ or
‘remove investment’.
Nor is there any clear trend towards differentiation of meaning overall in sets of rivals. Of
the doublets listed above, nine appear from the contexts in which they are used in COCA to
have the same meaning (mostly reversative or privative), five pairs have different meanings,
with either one or both of the meanings being lexicalized, and two pairs display a complex
pattern of overlapping senses. What this suggests again is that blocking does not seem to
occur, and that to the extent that items with these prefixes differ in sense, the difference can
only be attributed to the effects of lexicalization on an item-to-item basis, and not to any
systematic difference in the meanings of the affixes.

17.3.5 Semantic restrictions on bases

The last issue we will examine in this section is one that has a venerable history in the study
of English morphology, namely whether negative affixes are prohibited from attaching to
certain semantic categories of base: bases which themselves are negative, bases that have
monomorphemic antonyms, or classes of bases like colours; the issue is discussed at least by
Jespersen (1917), Zimmer (1964), Marchand (1969), Funk (1986), Horn (2002), Kjellmer
(2005), and Dixon (2010).
We begin with the issue of negative bases. Jespersen (1917) frames the claim as follows:

Not all adjectives admit of having the negative prefix un‑ or in‑, and it is not always easy to
assign a reason why one adjective can take the prefix and another cannot. Still, the same
general rule obtains in English as in other languages, that most adjectives with un‑ or in‑ have
a depreciatory sense: we have unworthy, undue, imperfect, etc., but it is not possible to form
similar adjectives from wicked, foolish, or terrible. ( Jespersen 1917: 144)

Zimmer’s (1964) monograph is entirely devoted to the issue of whether negative prefixes are
restricted to non-negative (that is, evaluatively positive or neutral) bases. He states (1964: 15)
the hypothesis as follows:

Negative affixes are not used with adjectival stems that have a ‘negative’ value on evaluative
scales such as ‘good ‑ bad,’ ‘desirable ‑ undesirable.’ (Zimmer 1964: 15)

After a careful study of data from then-contemporary English dictionaries, Zimmer’s results
are equivocal (1964):

We can thus say that for the group of forms in question, and especially for the monomorphemic-
base component of it, our hypothesis that un‑ prefixation is not applied to ‘negative’
382 Negatives

bases seems to be substantially correct, although some exceptions do occur. An extension of

the corpus leads, however, to the addition of further rather surprising exceptions; thus
uncruel, unevil, unignorant, unignominious, unsick, unsilly, unstupid (and possibly other excep‑
tions that we have missed) are all listed in [Webster’s 1957]. (Zimmer 1964: 36)

He acknowledges that there are many complex bases that are negative in connotation to
which negative affixes can nevertheless attach (for example, unbigoted, undegenerate, unfaulty,
unguilty, uninjurious, unmalicious, unobnoxious, unselfish, untroublesome), but very few that are
simplex (unbitter, unhostile, unsordid, unvulgar, and possibly a few more, as the quote above
suggests).
What corpus data from COCA suggest, however, is that what Zimmer calls ‘surprising
exceptions’ are neither particularly exceptional, nor particularly surprising. For example, we
find the negative forms on negative bases in (40) in COCA.

(40) unafraid, unangry, unanxious, un‑bald, unbare, unbitter, unbogus, uncoy, uncrazy,
uncruel, undead, unevil, unfake, unfraught, unhostile, unhumid, unjealous, unlame,
unlazy, unmad, unsordid, unsurly, unvulgar, unweary, unpoor, unsick, untimid, unugly

While individuals might quibble about whether all of the bases above are indeed negatively
evaluative (maybe bald is beautiful?), it is clear that the vast majority of these are counterex‑
amples to Zimmer’s hypothesis.
Furthermore, there is a clear pragmatic reason why they occur: there is a need for both the
positive term and for the negative-prefixed term. What, for example, is the antonym of angry,
or of sordid? There are surely several possibilities for each, but the most straightforward is the
un‑ word. And even where there is an institutionalized antonym, it may be too strong for the
context in which the word is needed. Someone who is untimid is not necessarily bold. Horn
(2002: 9) says with regard to forms like unhappy or unintelligent, that ‘their meanings are
palpably different from (and in particular, weaker than) those of the corresponding under‑
ived sad and stupid’, and we might say precisely the same thing about what Zimmer calls un‑
adjectives on negative bases.
Kjellmer (2005: 160) goes a step further than Zimmer, in suggesting that any word which
has a morphologically simple antonym is excluded as a base in negative-formation. We also
disagree with this claim. COCA yields the following examples that Kjellmer would predict to
be unacceptable, with the obvious non-prefixal antonym in parentheses.

(41) unadult (child), un‑big (small), unclear (opaque), unclever (stupid), uncomplex
(simple), uncrazy (sane), undead (live), unfake (real, genuine), unfancy (simple),
unfresh (stale), ungentle (rough), ungenuine (fake), ungood (bad), ungreat (small),
unhappy (sad), unheavy (light), un‑hot (cold), unhumid (dry), unkosher (trayf),
unlarge (small), unlight (heavy, dark), unlive (dead), un‑meek (bold), unopaque
 17.3 Semantic considerations 383

(clear), unplain (fancy), unpoor (rich), unpresent (absent), unpretty (ugly), unprofane
(sacred), unpublic (private), unrare (common), unrich (poor), unsick (well), unsmart
(stupid), unstraight (crooked), unstrong (weak), untame (wild), untidy (messy),
untiny (large), untrue (false), unugly (pretty), unwarm (cool), unwell (ill, sick), unwet
(dry), unwide (narrow), unwild (tame)

As with the examples of negative prefixes on negative bases, these examples may be far less
frequent than the morphologically simple antonym in parentheses, but they clearly provide
an alternative, and sometimes serve a distinct purpose in identifying a region on a scale that
is more central, less close to the opposing pole than the direct antonym. So, ungood is not
exactly the same thing as bad:

Houston Chronicle 1992: Templeton, who isn’t very talkative these days, even with his
(42) 
wife and children, says he feels good. He’s seen numerous 10th-floor neighbors emerge
from the transplants feeling distinctly ungood.

Examples like those in (40) are also common in contrastive contexts, paired either explicitly
or implicitly with the corresponding positive:

Fantasy & Science Fiction 1995: And when I win, the Net tells me, “Congratulations,”
(43) 
with its changeless voice, nothing warm or unwarm about it, no trace of involvement,
and nothing behind the words but an unshakable politeness.
Esquire 2005: The island is a leveler. It takes the greatest players on earth and makes

them . . . ungreat.

In the first example in (43), the contrast is overt, in the second implied. In all three cases it
is clear that the un‑ form provides a nuance of meaning that the simple antonym would
not supply.
A final claim concerning potential semantic restrictions on the derivation of negatives that
we might evaluate is that of Dixon (2010: 53). He states that, ‘the prefix un‑ may be used with
many adjectives from the human propensity type, with some from the value and with a
few from physical property, but with none from dimension, color, or age.’ The list in
(41) provides a number of counterexamples (un-big, ungreat, unlarge, untiny, unwide, unadult),
to which we can add colour adjectives like unblue and ungrey from COCA, and unred and
unyellow from internet discussions of haircolouring products.
The conclusions that we must draw from the corpus data are that there is no real prohibi‑
tion against affixing negative prefixes to negative bases, that the existence of simple antonyms
does not block the attachment of negative prefixes to either the positive or the negative anto‑
nym, and that even adjectives denoting size, age, and colour may occur with a negative prefix
384 Negatives

in appropriate contexts. Indeed, there are contexts in which such words are indispensible,
and native speakers do not hesitate to produce them. Further, it seems important to observe,
given the extensive discussion that these issues have received, that derived forms that might
seem odd or marginal when encountered out of context, or when judged against our intui‑
tions, are both possible and natural when encountered in context.
 chapter 18

Size, quantity, and attitude

18.1 Prospectus
In this chapter we consider morphological ways of marking size, frequency, and quantity. The
pertinent expressions are also often used to show attitude or evaluation and this is the reason
why we treat all of them in one chapter. The categories in question often involve prefixes,
though some suffixes and also other types of process are considered. Finally we consider the
morphology of numbers in English.
In terms of the semantics of the processes considered here, there is also a certain amount
of leakage. Many of the processes that are used to mark quantity are also used to show loca-
tion (presumably via semantic extension of the notion of location to location on a scale of
quantity, as, for example, with the prefix super‑). We have divided the treatment of potentially
relevant morphological categories between this chapter and Chapter 16 according to what
seems to be the major focus of the respective category, but the reader may need to consult
both chapters for a full report on all potentially relevant categories.
We will begin with categories that primarily mark evaluation and size, then turn to those
expressing non-evaluative attitude, and then turn to quantification and numbers.

18.2 Evaluation and size

18.2.1 Preliminaries
In this section we will deal with those word-formation processes that are used to denote the
size of the referent associated with the base. The main products of such processes that come
to mind are diminutives and augmentatives. The notion of diminutive in particular is not
easy to define clearly. One problem with this notion is the semantics, the other the kind of
formal means employed to express diminutive meaning. Hence, before entering the discus-
sion of the pertinent morphological processes we will first lay out our understanding of this
domain at large.
Traditionally, the term diminutive refers to morphological categories that express small-
ness. However, as noted by many authors, diminutives also express familiarity with the refer-
ent, and positive or negative attitude (i.e. appreciation or depreciation, e.g. Jurafsky 1996;
386 Size, quantity, and attitude

Schneider 2003). In order to explain that the same form can have apparently opposite read-
ings, we must consider the specific context in which a certain diminutive form is used. Which
reading a given diminutive form receives in that context may depend on the nature of this
linguistic and situational context as well as on the meaning of the respective base. As nicely
pointed out by Schneider (2003), wifelet is probably always negative in evaluation, while
cubelet perhaps never is. Referring to a child acting as a king in a play, with the diminutive
kinglet may convey a positive attitude, while referring with the same word to an adult sover-
eign probably never does. The meaning ‘small’ is often extended to ‘young’ or to ‘dear’, but
these meanings are relevant only with nominal bases.
We will refer to the meaning complex of small size, familiarity, appreciation, or deprecia-
tion as the notion of ‘diminution’, and the forms conveying these meanings as diminutives.
English has a rich inventory of forms expressing diminution, most of them morphological in
nature. Schneider (2003) arrives at 15 suffixes (‑a, ‑een, ‑er, ‑ette, ‑ie, ‑ing, ‑kin, ‑le, ‑let, ‑ling, ‑o,
‑pegs, ‑poo(h), ‑pops, ‑s) plus truncation (referred to here as clipping), and reduplication. In
addition, there is also the possibility of expressing diminution analytically, with the adjective
little (see Schneider 2003 for details). We will only be concerned here with the morphological
expressions of diminution.
Precisely which word-formation processes should go under the label of diminutive seems
to vary a great deal in the literature, mostly depending on how much emphasis is laid on the
attitudinal component versus the size component of the meaning of the process. Thus one
could distinguish between terms of endearment, which lack the meaning component ‘small-
ness’ and other diminutives, which have both meaning components. In practice, however, it
is often very hard to distinguish properly between the two, as most diminutive morphologi-
cal categories can express both concepts, even if not always simultaneously in each of the
pertinent formations. Furthermore, clippings or reduplications often express attitudinal
meaning and are therefore also subsumed under the rubric diminutive by many authors (e.g.
Schneider 2003). We remain largely agnostic as to these taxonomic questions and have
grouped processes expressing evaluation and size using a mixture of formal and semantic
criteria. As in the other chapters, we focus on productive processes.

18.2.2 Diminution by affixation

18.2.2.1 Native affixes: ‑let, ‑ling, ‑ie, and the minor suffixes ‑o, ‑s, ‑er and ‑kin
The suffix ‑let is still slightly productive in a number of uses. In (1) we find examples from the
BNC illustrating ‑let with the meaning ‘small of a species’, occasionally ‘young of a species’.

(1) auklet, basslet, catlet, froglet, gooselet, murrelet, piglet, swiftlet, toadlet

The examples in (2) show ‑let with a simple meaning of small size used on inanimate
entities.
 18.2 Evaluation and size 387

(2) applet, bomblet, booklet, brooklet, budlet, cloudlet, droplet, eyelet, firmlet, flatlet,
leaflet, manorlet, moonlet, platelet, rootlet, spinelet, spirelet, statelet, streamlet, tartlet,
twiglet, wavelet

The examples in (3) show ‑let with a slightly disparaging meaning.

(3) godlet, playlet, starlet

The examples in (4) show ‑let with the meaning of ‘piece of jewellery or adornment’, where
the base denotes the place where it is worn (corse is an old form meaning ‘body’, from the
French).

(4) armlet, bracelet, corselet, necklet, wristlet

The examples in (5) illustrate the suffix with an obligatorily bound base, in the relevant meaning.

(5) chaplet, gauntlet, hamlet, pamphlet, rivulet

For bases ending in <l> it may be sometimes difficult to determine whether we are dealing
with the suffix ‑let or ‑et. For instance, it is not clear whether forms like tablet are best analysed
as containing the more common ‑let and some simplification of the resulting double <ll>, or
whether they contain the rarer suffix ‑et. Another problematic case would be hamlet, which,
according to the OED, has a historical root hamle, but neither ham nor hamle is available with
a suitable meaning in current English.
The suffix ‑ling is of extremely restricted productivity; even where a word containing this
suffix is used with a meaning not listed in the OED, which might indicate productive coinage,
the form itself is usually listed. The examples below are mainly from COCA. We can distin-
guish between words with adjectival, nominal, and verbal bases and there are very few exam-
ples, like underling, overling (OED) which have none of these. Within the denominal
formations, there seem to be various semantic classes (see below).
De-adjectival formations are given in (6). The base in darling is etymologically from the
root that gives dear, though this is no longer transparent.

(6) darling, firstling, foundling, grayling, greenling, lostling, scantling, weakling, wildling,
youngling

In the denominal series, we first have those which denote the young or a small exemplar or
type, as in (7). Codling has several meanings, here it means ‘small species of cod’.

(7) codling, duckling, gosling, kidling, mouseling, pigling, ratling, spiderling

388 Size, quantity, and attitude

In the next series, in (8a), the diminution may be read as implying insignificance, illustrated
with an example from COCA in (8b).

(8) a. godling, lordling, priestling, princeling

The Helper and His Hero 2007: Now it was Bandar’s turn to be surprised: even a
b. 
provincial lordling ought to have heard of Rovers.

In the series in (9), the words all seem to denote young creatures (people, animals, even plants),
but the relationship to the base is much less transparent. A fingerling (when it denotes a fish or a
potato) resembles a finger, a nestling is found in a nest, and a yearling is a year old, for example.

(9) fingerling, fosterling, nestling, nursling, seedling, stripling, yearling

In the examples in (10) the semantics is obscure, and the ‑ling appears to denote little more
than ‘thing or creature in connection with X’. Starling is included here following the etymol-
ogy suggested by the OED, from the noun stare.

(10) 
bratling, earthling, fleshling, groundling, halfling, heartling, potling, sideling,
starling, witling

The deverbal formations in (11) are object-referencing, and the suffix is similar to ‑ee in this
regard (see Chapter 11). The semantics fits with fatling ‘an animal which has been fatted’ (see
the OED). Note that weanling has a synonymous form weaner.

(11) changeling, fatling, hatchling, hireling, starveling, weanling

Finally, there is a small set of words from COCA in (12) which all denote some kind of crea-
ture. They are not listed in the OED, and all of them arise in the context of science fiction or
fantasy fiction. Such sources also account for the attestation of a number of other rare forms
in COCA (fleshling, halfling, kidling). This type of prose seems to provide the only reliable
domain of productivity for this suffix in contemporary English and can be seen as evidence
for the extended meaning ‘(small) creature’.

(12) sisterling, stepling, witchling, wyrmling

In spelling terms, note the lack of <e> in nursling and starling, but not in hireling, mouseling,
sideling (although this last has an alternative spelling with no <e>).
The suffix ‑ie (also frequently spelt as <y> as in <Tommy>, and much more rarely as <ey>,
as in <Barney>) has an obscure history. It is related to the Germanic suffix that gives us Dutch
 18.2 Evaluation and size 389

‑tje and German ‑chen. Zandvoort (1962: 302) claims that English is more restricted than those
languages in its use of diminutives (see also Strang 1968: 138), but we know of no empirical
study to show this with solid quantitative data. Quite to the contrary, Cannon (1987: 185)
presents evidence that ‑ie is among the most productive suffixes of English. Claims that perti-
nent forms are largely restricted to the language addressed to or coming from children (doggy,
hankie, girlie, grannie, mousie, nighty, piggy, sweetie, tummy) also seem empirically weak, as such
forms are also common outside this domain (e.g. bookie ◆ bookmaker, darkey ◆ dark (now
taboo sl.), goalie ◆ goalkeeper, roadie ◆ road manager, and the examples in (13)). Hypocoristic
names with this suffix are also very common (Annie, Betty, Bobby/Bobbie, Charlie, Davy,
Freddy, Georgie, Jackie, Kitty, Laurie, Lizzie, Mandy, Teddy, Tommy). These are dealt with in
more detail in Section 18.2.3, as they are subject to slightly different restrictions.
The use of ‑ie seems especially strong in ScE, providing forms like laddie, lassie, and
exporting the suffix to Australia and New Zealand, where it thrives. In South Africa it met a
homophonous Dutch diminutive, which must have supported this form although the Dutch
allomorphy is often retained (Silva 1996). The suffix is, however, also widely used in other
varieties of English. It is not entirely clear whether the rules for ‑ie suffixation are the same
in all these varieties, but in this chapter words from different varieties will not be treated
separately. Hypocoristics of place names will also be treated on a par with diminutives from
ordinary words, though personal names will be left to one side here, since they are often
formally more complex than other ‑ie/‑y forms (see Section 18.2.3 for more discussion).
In the examples of ‑ie formations that follow, many words are taken from the literature,
which has often concentrated on ‑ie formations in particular varieties. The words are marked
as being AusE, NZE, SAfrE, or ScE as appropriate, although there is no expectation that they
are different in kind, or that they might not at all occur anywhere else. The ScE forms are
from Bulloch (1924) unless otherwise stated. Many of the AusE and NZE forms are from
Ramson (1988), Sussex (2004), and Bardsley and Simpson (2009). South African forms are
from Silva (1996), other forms from Schneider (2003) or Lappe (2007).
Examples are given in (13). All or nearly all of these are established, although some forms
appear to be restricted in usage to certain groups of speakers and therefore will not necessar-
ily be known to every reader. The suffix is, however, highly productive. In (13), we use regional
designations if pertinent or if given as such in the sources. Such designations do not preclude
that the item in question may not also be found in other varieties.

(13) baddy ◆ bad

bankie ◆ bank ‘bench’ (SAfrE)
barbie ◆ barbecue (AusE/NZE)
boatie ◆ boat (ScE)
bootie ◆ boot
cardie ◆ cardigan (NZE)
390 Size, quantity, and attitude

ciggie ◆ cigarette
cocky ‘farmer’ ◆ cockatoo (AusE/NZE)
coldie ◆ cold beer (AusE)
combie ◆ combination
curlie ◆ curl (ScE)
drillie ◆ drill sergeant (AusE)
eggs benny ◆ eggs Benedict (NZE)
footie ◆ football
girlie ◆ girl
gummy ◆ gumdigger (NZE)
hammy ◆ hamster
hanky ◆ handkerchief
homie ◆ home boy ‘friend’ (NAmE)
indie ◆ independent
kindy ◆ kindergarten (AusE/NZE)
lairdie ◆ laird (ScE)
lassie ◆ lass (ScE)
Palmie ◆ Palmerston North (NZE) / Palm Bay (AusE)
Phillie ◆ Philadelphia (NAmE)
pollie ◆ politician (AusE)
pokey ◆ poker
pommy ◆ pomegranate ‘Englishman’ (AusE/NZE)
pozzy ◆ position (AusE)
pressy ~ prezzie ◆ present ‘gift’ (AusE/NZE)
prossie ◆ prostitute (BrE)
Rangy ◆ Range Rover (AusE)
rhodie ◆ rhododendron
roey ◆ Rohypnol (AusE)
scotchie ◆ scotch thistle (NZE)
steppie ◆ stepmother (AusE)
swaggie ◆ swagman (AusE)
Tassy /tæzi/ ◆ Tasmania (AusE)
titty ◆ tit

As expected from our discussion of the semantics of diminutives, derivatives with the suffix
‑ie do not necessarily imply a particularly small size, but are often about attitude, as, for
instance, illustrated by the examples steppie ◆ stepmother or homie ◆ homeboy listed in (13). The
connotations are perhaps mostly positive or mark familiarity, but may also be negative, as
 18.2 Evaluation and size 391

evidenced by the established forms blowie ◆ blowfly (AusE) and even with rellie ◆ relative,
relation (AusE/NZE).
The suffix attaches most frequently to nouns and forms count nouns, except in the case of
Aussie, which, like its base Australian, may be either an adjective or a noun. We also find for-
mations based on adjectives (e.g. baddie, blackie, goodie, greeny, softy, stiffy), in which case the
derivatives are nouns. Very occasionally it attaches to other parts of speech, such as preposi-
tions (again forming nouns, e.g. innie, outie ‘types of navel’). One might even be tempted to
analyse forms such as howdy, all righty, thanky as being derived from phrases, that is from how
do you do, all right, and thank you, respectively. At least with howdy and thanky such an analy-
sis is problematic, as these forms probably result from univerbation and phonological ero-
sion. Furthermore, there is the problem that all three forms are not nouns, as all other ‑ie
diminutives seem to be.
Our analysis implies that the adjective-forming ‑y suffix which gives cheery, crazy, dotty, shotty
(‘excellent’, NZE children’s sl) is a different suffix, despite its homophony with the diminutive
suffix ‑ie and—at least in the instances cited—similar connotations (see Chapter 14).
Let us now turn to the phonological aspects. Derivatives in ‑ie are almost all disyllabic
trochees, which necessarily leads to the loss of phonological material of the base with multi-
syllabic base words. A further peculiarity that arises through this prosodic requirement is the
fact that with many derivatives it is actually unclear whether they are suffixed. Derivatives
like indie on the basis of independent, or combie on the basis of combination, might be analysed
as not having a suffix at all and being simply truncated. However, from an output perspective,
such an analysis is unwarranted and cannot explain why the putatively unsuffixed forms
behave exactly like the clearly suffixed forms, both in terms of phonology and semantics. We
adopt an output-oriented view here and define ‑ie diminutives as diminutive forms that are
disyllabic trochees ending in /i/. For the sake of convenience we will nevertheless keep using
the traditional ‘suffix’-based terminology.
Monosyllabic bases take ‑ie and integrate the suffix phonologically (e.g. po.key ◆ poker).
The truncation patterns found with polysyllabic bases are quite intricate and are closely
linked, but not fully identical, to the ones found with clippings, which we discuss in Section
18.2.4. There are three main problems associated with base truncation in this morphological
category. The first is what material from the base survives. We will refer to this problem as
‘anchoring’. The second is the phonotactics and syllable structure of the derivatives, and the
third is the segmental stability, that is whether segments from the base have the same quality
as their corresponding segments in the derivative. Our treatment is largely based on the find-
ings in Lappe (2007).
The vast majority of ‑ie forms anchor to the first syllable of the base (e.g. hammie ◆ hamster,
pokey ◆ poker), irrespective of where the base is stressed. Thus we find forms anchoring to the
initial base syllable in spite of non-initial base stress (combie ◆ combinátion, indie ◆ indepéndent).
How much of the base is preserved is quite variable, but also clearly constrained. Diminutives
of this category preserve contiguous strings starting from the left edge of the first syllable.
392 Size, quantity, and attitude

There are no extra restrictions on the syllable structure of the derivative apart from the fact
that the second syllable needs to have an onset. As a consequence, deletion patterns vary
within the regular confines of English syllables and the prosodic requirement imposed by the
morphological category. Thus we find some variation in the preservation of consonant clus-
ters. Sonorant–obstruent clusters favour preservation (e.g. barbie ◆ barbecue, alkie ◆ alcohol),
while obstruent–obstruent clusters are more variable. For example, the diminutive of prosti‑
tute is attested as prossy, while hostess has the diminutive hosty. The only clusters that seem to
be always subject to deletion are those in which the second consonant is more sonorous than
the first consonant (seggy, *segry ◆ segregation; Chevvy, *Chevry ◆ Chevrolet)
The overwhelming majority of derivatives faithfully preserve the quality of the segments
of the base, but changes from schwa to full vowel can be observed with those forms where
the derivative anchors to a base-initial schwa syllable, as in s[ʌ]ssy ◆ s[ə]spicious. Consonantal
changes are even rarer and show only one recurrent, but not consistent, pattern, the voicing
of /s/ in intervocalic position, as in Au[z]ie <Aussie> ◆ Australian, or co[z]ie <cossie> ◆
costume.
Exceptions to the general patterns just described are occasionally found. They may retain
more than one syllable, may anchor to a non-initial syllable, or other phonological changes to
the root are made. It is unclear how much the pertinent forms may in fact reflect regular
patterns in regional dialects. Some such forms are listed in (14), with a few of them appar-
ently being neologisms.

(14) breakfastie (ScE)

bullocky ‘beef ’, ‘bullock driver’ (AusE)
cradlie (ScE)
gyppie ◆ Egyptian (AusE)
littlie ‘small child’ (NZE)
messagie ‘errand’ (ScE)
second handy (NZE)
Shorty/Landy/Streety ◆ Shortland Street (NZ TV soap)
smooy /smuːi/ ◆ smoodge ‘display of amorous affection’ (AusE)
steenbokkie (SAf E)
tammie ◆ Temazepam tablet (AusE)

Diminutive formation in ‑ie is certainly the most productive diminutive category in English,
but a couple of minor classes of words, those ending in ‑o, ‑a,‑er, ‑s, and ‑kin, should not go
unmentioned. Formations in ‑o (journo ◆ journalist) are the next most common, while forma-
tions in -er (also very rarely spelled <a>) are very rare (e.g. in Maccas ◆ Macdonalds).
Diminutives ending in ‑o seem more frequent in, but are certainly not restricted to,
Australasian varieties of English, and then in AusE rather than in NZE. Attested words across
 18.2 Evaluation and size 393

varieties are aggro ◆ aggravation, ammo ◆ ammunition, arvo ◆ afternoon, combo ◆ combination,
doggo ◆ dog (in the expression to lie doggo ‘keep a low profile’), journo ◆ journalist, kiddo ◆ kid,
muso ◆ musician, weirdo ◆ weird.
Some examples from Sussex (2004) and Bardsley and Simpson (2009) are listed in (15);
see Lappe (2007) for many more forms.

(15) ambo ‘ambulance officer’

Catho ‘Catholic’
hospo ‘worker in the hospitality industry’
lezzo ‘lesbian’
maco ‘immaculate’
metho ‘methylated spirits’
Missos ‘The Miscellaneous Workers’ Union’
obno ‘obnoxious’
Proddo ‘Protestant’
reffo ‘refugee’
rego ‘(car) registration’
Rotto ‘Rottnest Island’
Salvo ‘member of the Salvation Army’

As was the case with ‑ie, we find forms, such as demo ◆ demonstration, or condo ◆ condominium,
that appear to be simply truncated, and not necessarily to be analysed as suffixed with ‑o.
Following the same arguments as with ‑ie diminutives, ‑o diminutives are thus best defined in
an output-oriented fashion as disyllabic trochees ending in ‑o.
Derivatives in ‑o are mostly nouns that are based on nouns, but, as shown by some of the
above data, we also find de-adjectival and de-phrasal nouns (weirdo ◆ weird, metho ◆ methylated
spirit drinker, respectively). Monosyllabic bases take ‑o and integrate the suffix phonologically.
The derivatives anchor to the initial base syllable and preserve contiguous strings starting
from the left edge of the first syllable, and, unlike ‑ie derivatives, they almost categorically
preserve all clusters, even those in which the second consonant is more sonorous (obno ◆
obnoxious). Segmental alternations are very rare, one example is l[aɪ]no <lino> ◆ l[ɪ]noleum.
Diminutive formations in ‑o and ‑ie are in obvious competition and we find a number of
bases that are attested with both suffixes: prossie ◆ prosso ◆ prostitute, Proddie ◆ Proddo ◆
Protestant. Diminutives in ‑ie are however, much more common.
Another type of diminutive comprises disyllables ending in schwa, normally spelled <er>
in the non-rhotic varieties where this suffix seems to be mostly found. It has been noted to be
in-group marking, particularly in academic institutions and many words refer to concepts
pertinent in such environments (nipper ◆ nip, rugger ◆ rugby, and topper ◆ tophat, footer ◆ foot‑
ball, prepper ◆ prep(aratory) school, fresher ◆ freshman). The form triggers truncation in a
394 Size, quantity, and attitude

similar fashion to ‑ie and ‑o but there seem to be too few forms to establish more detailed
generalizations. Diminutives in ‑er derive nouns, mostly on nouns. Often, it co-occurs with
an additional ‑s as in champers ◆ champagne, starkers ◆ stark naked, butters ◆ butt ugly and preg‑
gers ◆ pregnant. Some forms are actually ambiguous between a person noun ‑er and a diminu-
tive ‑er interpretation, as in fresher ◆ freshman, tucker (‘food’) ◆ tuck in ‘eat heartily’ (see
Schneider 2003 for more discussion and further references).
The suffix ‑s frequently arises in instances where there is multiple marking of diminutive-
ness. Its function is often one of marking a hypocoristic (see Section 18.2.3). Various uses can
be distinguished. It is found by itself or with a clipped base on nouns which are either
hypocoristics or terms of address (the two may intersect, of course). Examples include Babs
◆ Barbara, Debs ◆ Deborah, ducks (BrE), Jules ◆ Julia/Julian, Mags ◆ Margaret, moms, pops.
As already mentioned above, it occurs after the diminutive ‑er in BrEsl, as in Honkers ◆
Hong Kong, Wimblers ◆ Wimbledon. This combination of forms is usually attributed specifi-
cally to British public schools (mainly Rugby) and the University of Oxford, though some of
the forms have become item-familiar for a wider portion of the British English community
and even the Australian community. Rudders ◆ Rudiments of Divinity (see the OED sv. ‑ers)
would not be easily interpreted outside of Oxford (possibly Oxford of the relevant period).
Bonkers and crackers (both meaning ‘mad’) may illustrate the same process, though the rela-
tionship with the base is less clear. Quirk et al. (1985: 1584) also put bananas in the same cat-
egory, but it is questionable whether this form expresses diminution.
The suffix ‑s also occurs along with the ‑ie suffix, for example in the language of the play-
ground. Bauer and Bauer (1996) cite examples such as ankles, kneesies, and under-bums as vari-
ants of a playground jumping game from New Zealand. Note the triple diminutive marking
in kneesies. The item-familiar child-language word tootsie (‘foot’) (and the more adult forma-
tion play footsie) seem to illustrate the same combination of affixes. It is less clear whether it
is found (or may be analysed) in words like the heebie jeebies, the (blue) meanies, walkies, and
the willies.
Let us turn to the suffix ‑kin. Most of the words where ‑kin can be analysed as a genuine
suffix are names, often surnames, where the ‑kin seems to function just like ‑son, so that
Dickson (or Dixon) and Dicken (or Dickens) are parallel forms. There are only a few examples
of ordinary words with the suffix ‑kin, and in most of these the base is no longer recuperable
from the word. Examples from the BNC include those in (16). All of these are established,
often old-fashioned. Many of them represent loans from Dutch rather than genuine English
derivations. Even where the elements are clear, the meaning relationship between the ele-
ments is often obscure, with only bunnikin, catkin, lambkin, and manikin being at all
transparent.

(16) bodkin (‘dagger, needle’), bumpkin, bunnikin (defined as a flower in the OED), catkin,
firkin (‘quarter barrel’), gherkin, lambkin, lumpkin, manikin, munchkin (Baum 1900),
napkin, pipkin (‘drinking vessel’), ramekin, siskin (‘small bird’)
 18.2 Evaluation and size 395

Mannequin is probably a parallel development of the same source as manikin, but one which
has come through French.

18.2.2.2 Non-native affixes: ‑ette, hypo‑, mini‑, micro‑, nano‑

The suffix ‑ette is French in origin, and there are still a large number of recognizably French words
which end in this sequence. The suffix occurs on bound bases, as well as on simple and compound
nouns, as the examples in (17) and (18) illustrate. It is no longer productive on bound bases.

(17) baguette, banquette, briquette, brunette, camionette, coquette, couchette, croquette,

florette, lazarette, lorgnette, marionette, marquisette, vinaigrette

The suffix ‑ette is one of the auto-stressed suffixes of English and as such carries the main
stress of the derivative (with the exception of the single example %cígarette). It is remarkably
productive in a number of distinguishable meanings. While there are many established items
in the lists from the BNC and COCA, there are many neologistic items which occur in only
one of the corpora.
A list of examples which appear simply to denote smallness is presented in (18). In some of
these cases there may be overtones of disparagement, but those do not appear to be dominant.

(18) cigarette, clubette, coffinette, crashette, crisisette, diskette, featurette, festival‑ette,

filmette, flingette, ghetto‑blasterette, iconette, jaunt‑ette, jobette, mealette, partyette,
recessionette, seatette, sermonette, snackette, spinnerette, statuette, suburbette,
novelette

There are rather more specific series of words where size appears to be more important than
attitude, distinguished here as businesses (19) and rooms/buildings (20), although that dis-
tinction may be slightly artificial at times. These seem to be semantic domains where ‑ette is
particularly favoured.

(19) laund(e)rette, luncheonette, ranchette, shoppette, washerette

(20) dinette, hangarette, hotelette, kitchenette, loungette, maisonette, pullmanette, roomette,

sleeperette

Note the example in (21), which specifically mocks the use of the suffix on the names of
rooms.

(21) I’ve been allocated a little house or ‘maisonette’, as it’s called (a twee fake-French word
I’ve always disliked) on the campus, at the end of a terrace of five. . . . An open-plan
living room with ‘kitchenette’ downstairs and a bedroomette and bathroomette
upstairs. (Lodge 2001: 10)
396 Size, quantity, and attitude

Another well-known meaning for this suffix is ‘fake, imitation’ when applied to fabrics. There
are few of these words in the corpora, and those that occur are all established, sometimes
having a non-transparent, lexicalized meaning (e.g. towelette ‘pre-moistened small towel
sealed in a foil packet’).

(22) flannelette, leatherette, towelette

The largest group of ‑ette words with an identifiable meaning are those where the suffix means
‘female’. Many of these are clearly disparaging, but not all are, though they are rarely com-
pletely serious words. Very few of these are item-familiar, which implies considerable pro-
ductivity in this meaning. The suffix ‑ette may, in fact, be the main gender-marking suffix
available in English, more used than ‑ess (see Chapter 11), but with far more strongly marked
connotations.

(23) 
bachelorette, baronette, bimbette, bumette, chaufferette, chefette, choir‑­
ette, columnistette, conductorette, daemonette, deanette, demonette, disciplette,
drumette, dudette, goblinette, gothette, grocerette, hustlerette, idolette, jockette, major­
ette, nerdette, nymphette, professorette, punkette, rabbinette, raiderette, redskin­
ette, reporterette, rock‑ette, santa‑ette, scionette, shrinkette, smurfette, suffragette,
trooperette, usherette, wenchette, wifette, yeomanette

The suffix ‑ette strongly disprefers bases ending in /t/, which can be interpreted as a haplo­
logy effect, in this case the avoidance of identical onset and coda in the final syllable of the
derivative (see Chapter 9 for discussion). Among the hundreds of derivatives in the BNC
and COCA we found only three forms of this kind, columnistette, Dustette (the name of a
vaccum cleaner model), and seatette.
The prefix mini‑ is extremely productive in modern English, a point which we illustrate
with just a handful of the examples we find in COCA which are not listed in the OED.

(24) 
mini‑DVD, mini‑illustration, mini‑mask, mini‑minivan, mini‑moon, mini‑novel,
mini‑researcher, mini‑supermarket, mini‑Supreme court, mini‑vineyard, minibottle,
minihomily, minipizza, minireferendum, minisat, miniski, minisnowbank, minispace,
minitelescope, minitremor, minizoo.

The examples in (24) show that hyphenation is inconsistent, that mini‑ can be prefixed to
clippings, initialisms, derivatives (including prefixed words), to native and neoclassical com-
pounds, and to phrases and that mini‑ can be recursive. Stress is variably placed on the prefix
or on its host, with longer base words tending to carry the main stress of the derivative.
Mini‑ appears to refer purely to size and not to be biased towards positive evaluation.
While nearly all of the bases are nouns, we occasionally find an adjective or a name used as a
base.
 18.2 Evaluation and size 397

(25) a. Popular Mechanics 1997: The undermounted handles and spout on this Monticello
mini‑wide faucet will fit any standard 4‑In. centerset basin
b. San Francisco Chronicle 2008: It was a mini‑Woodstock for the public transit set.

The explosion in the use of forms with mini‑ appears to have begun in the 1960s (despite a
few earlier examples—see the OED sv. Mini‑) and to continue to the present time. The prefix
mini‑ is probably now the preferred affix for marking small size (with no other implications)
in contemporary English.
The prefix micro‑ derives from the Greek word meaning ‘small’, but gains a number of dis-
tinguishable meanings in English, though these should almost certainly be attributed to poly-
semy of the prefix rather than to a series of homophonous affixes. The meanings include
‘small’, sometimes ‘abnormally small’ (26), ‘microscopically small’, ‘concerned with micro-
scopically small entities’, or perhaps ‘involving the use of a microscope’ (27), ‘smaller-scale
than with the contrasting macro-’ or ‘smaller than something denoted with the prefix mini-’
(28), ‘a one millionth part’ (29), ‘created by microphotography’ (30), and so on. While
derivatives in micro‑ are largely derived from nouns, there are occasional exceptions where
micro‑ is found in verbs, adjectives, and adverbs (e.g. microetch, micro-fresh, microneural,
microsocial, microthin, microurban). The examples below are all taken from COCA.

(26) micro‑adjustable, microbrewery, microcassette, microcellular, microclimate, micro­

expression, microhabitat, microhouse, microlivestock, micromastia, micropenis, microthin

(27) 
microbiology, microfauna, microfossil, micrometeoroid, micrometer, Micronesia,
microspectrography

(28) micro‑economic, microbus, microcredit, micromanagement, microskirt

(29) micro‑henry, micro‑kelvin, microliter, microsecond, microwatt

(30) microdot, microfiche, microfilm, microprint

The prefix micro‑ can carry primary or secondary stress on its first syllable (e.g. mìcrobiólogy,
mícrobrewery), with the distribution of the two patterns being essentially unclear. Primary
stress may also occur on the second syllable of micro‑ if the prefix is part of a form with a
stress-shifting element, as in micróscopy (see Chapter 10, ‑y).
The prefix nano‑ originated as a number prefix in the metric system, denoting 1/109 (see Section
18.5.4). It has made its way into the general language with the meaning ‘extremely small’, or ‘with
regard to very small entities’. COCA lists hundreds of derivatives, most of them nouns (e.g. nanoac­
celerator, nanobacteria, nanobacterium, nanobattery, nanobiology, nanoboat, nano-brain, nanobug).
Many adjectives can also be found, however (nanobased, nanobiological, nano-biomedical, nanocrys‑
talline, nano-enhanced, nano-generated, nano-inspired, nanoimpregnated, nanostructured). Some of
398 Size, quantity, and attitude

them appear to be back-formations from related nouns, others do not. Derived verbs are hard to
find and seem to be back-formations (e.g. nano-optimize ◆ nano-optimized, nano-processN ◆ nano-
processV, nano-fabricate ◆ nano-fabrication). Stress-wise, nano‑ behaves like micro‑, as shown by the
examples nánomachìne versus nànotéchnology versus nanógraphy.
The three prefixes micro‑, mini‑, and nano‑ appear to be interchangeable in some contexts,
with no precise difference in meaning in a non-contrasting environment.

Analog Science Fiction & Fact 2001: A stray reflection from the overhead lights, Pancho
(31) 
thought, an unavoidable momentary glitter off the array of nanocameras and
projectors.
Analog Science Fiction & Fact 2001: Arrays of the church’s microcameras scanned the
preacher, the audience, and the lifelike images of the elderly couple, transmitting
the memorial services around the world to other family members and friends who were
attending electronically, plus a few million interested or at least voyeuristic webbies.
USA Today 1997: Baseball has PC‑toting managers and minicameras attached to
catchers’ face-masks for TV viewers, basketball coaches churn statistics on laptops and
golfers boom big drives with titanium clubs.

Of the three, mini‑ seems most colloquial, micro‑ has a somewhat scientific nuance, and nano‑
is used in scientific or pseudo-scientific contexts. The quote from Tech Review illustrates this
excellently.

Technology Review 1999: It’s somewhere around here that the science starts getting
(32) 
mixed up with science fiction. If you can make a nanowheel, why not a nanogear? A
self-powered nanoboat? Why not build a nanorobot to move around the atoms for
you? # And while you’re at it, why not make nanorobots that can replicate themselves,
making it possible to staff nanofactories capable of piecing together almost anything
out of the basic building blocks of atoms? # Welcome to molecular manufacturing, as
preached by nanoevangelist Drexler.

Nano- is frequently used in brand names to make the product sound more scientific, even
when it’s not clear that the product involves anything small.

Sunset 2007: Sandwiched inside the glass panels is Nanogel aerogel (nanogel.com), a
(33) 
Cabot Corporation product that improves the insulating properties of glass.

Nano‑ is also the one that seems to be used disparagingly:

Southwest Review 2000: “He said, ‘Tell nano‑brain to write me when he learns how.’ ”
(34) 
Rufus gives his younger son a sort of shoulder punch of affection, in the oblique male
manner.
 18.2 Evaluation and size 399

The Albino Knife 1990: Finally, the little gadget’s microbrain could recognize a couple
dozen explosives by using no more than a few stray molecules.

18.2.2.3 Miscellaneous diminutive affixes and multiple affixation

There are a number of historical diminutives whose force has now vanished, minor diminu-
tive markers, and some isolated diminutive forms in English.
The OED recognizes a historical ‑en diminutive in words like chicken, kitten, and maiden.
This is not recognizable as such today, and is not productive. Marchand (1969: 281‑2) recog-
nizes a suffix ‑rel or ‑erel in words such as cockerel, gomerel (ScE ‘fool’), kestrel, mackerel, mon‑
grel, scoundrel, wastrel, but comments that this has not been productive since the seventeenth
century. The suffix ‑et may be difficult to separate from the suffix ‑let when the base ends in
<l>, and from the suffix ‑ette, with which it is segmentally homophonous. Nevertheless it may
be recognized in items like those in (35).

(35) bullet, freshet, hogget, midget, owlet, rivulet, snippet, spinneret

Marchand (1969: 288‑9) discusses this affix at greater length, but many of the words he dis-
cusses are no longer current. This affix is not productive.
The OED lists a diminutive suffix ‑ot, but the only current word it lists as an example is eyot
(‘small island’) perhaps from the base ait ‘small island’. Ballot, spigot might be seen as carrying
the same suffix. This affix has not been productive for many centuries.
The OED also recognizes a diminutive in ‑ola, though it is mostly present in borrowings
and not easily analysable in English. Some of the relevant words are in (36). This ‑ola should
probably be distinguished from the homophonous suffix which the OED describes as being
used for humorous effect, which occurs in payola (see also below Section 18.2.5.8).

(36) aureola, cupola, fasciola, foveola, pergola, roseola, rubeola

There are a few isolated forms which seem to function as diminutives, but these forms do not
seem to give rise to larger extendable series. A few pertinent forms are illustrated in (37a). In
(37b) we list all pertinent forms from COCA involving the suffix ‑poo. Such forms are typically
based on diminutives in -ie, which in turn are mostly, but not exclusively, based on nouns.

(37) a. beddy‑byes, daddypegs, girleen (IrE)

b. bonkers-poo, bunny-poo, chickie-poo, cutesy-poo, drinky-poo, Evey-poo, hanky-
poo, giggledypoo, grumpy-poo, icky-poo, kissy-poo, matchy-poo, Neddy-poo,
Putie-poo, sneaky-poo, stinky-poo

Markers of diminution are known cross-linguistically for being recursive, either in the sense
that the same affix may be repeated, or in the sense that multiple markers of the same category
400 Size, quantity, and attitude

may accrue. Although this is not automatically the case in English, there are many such exam-
ples, some of which we have already mentioned. Consider (38).

(38) Debs/Debsie/Debso, goofies (◆ goef ‘swim’ = ‘swimming baths’ SAf E), Honkers,
kneesies, micro‑miniskirt (COCA), micro‑minivan (COCA), microdroplet,
mini‑microwave (COCA), mini‑minivan (COCA), mumsy ‖ momsy, preggers, tootsie.

18.2.3 Proper noun diminution: Hypocoristics

Hypocoristics are pet names, usually for people but also occasionally for places. These are
marked in the same way as diminutives for the most part, but they are also marked in other
ways.
Various examples have been given in Section 18.2.2 of diminutive suffixes being used for
hypocoristics. Some of these are repeated below, according to the suffix used. Many of these
are based on a clipping.

(39) Babs, Debs, pops

(40) Honkers (◆ Hong Kong), Wimblers (◆ Wimbledon)

(41) Annie, Bobby, Charlie, daddy, Freddy, Georgie, Jamie, Jeannie, Lizzie, mummy ‖
mommy, Sally (◆ Salvation Army) (NZE); Brizzie (◆ Brisbane), Gissy (◆ Gisborne),
Palmie (◆ Palm Bay or Palmerston)

(42) Maccas (◆ Macdonald’s) (AusE)

(43) Salvo (◆ Salvation Army) (AusE); Rotto (◆ Rottnest Island) (AusE)

Hypocoristics can also be formed by clipping, that is by deleting phonological material of the
base word. Some clippings that show no modification beyond the shortening are exemplified
in (44). Many of these names have become names in their own right, but began as clippings
of a longer name.

(44) Alex, Chris, Dan, Dave, Deb, Di, Don, Ed, Geoff, Greg, Jake, Jan, Jess, Joe, Ken, Lena,
Liz, Lou, Manda, Mike, Nick, Pete, Ray, Reg, Rob, Sam, Steve, Sue, Tim, Tom, Trina,
Val, Vic, Will, Zach

The highly productive morphological categories of monosyllabic name clippings and suffixed
hypocoristics impose severe prosodic restrictions on their members. The restrictions are simi-
lar to those we find in the ‑ie diminutives discussed above and in the clippings discussed in
Section 18.2.4 below, but are not fully identical to them. The following treatment of the phono-
logical characteristics of clipped names is largely based on Lappe’s (2007) analysis of a large
 18.2 Evaluation and size 401

number of such forms. As will become clear, clipped names and suffixed hypocoristics are
highly regular, in spite of the existence of a significant amount of variation. As shown in Lappe
(2007), the regularity of the patterns is most insightfully recognized if one takes an output-ori-
ented approach that focuses on the relationship between what is preserved from the base word
and not on what is deleted (see, for example, Berg 2011 for the latter kind of approach). It is
useful to treat the different categories separately, as their restrictions differ slightly.

18.2.3.1. Clipped names

Clipped names are monosyllables that need to have a heavy rhyme and at least one consonant, so
that forms like *D[e] ◆ Dexter, or *A ◆ Abraham are ruled out. Clipped names anchor either to the
first syllable or to the main-stressed syllable (Pat ◆ Trish ◆ Patrícia). Very occasionally, the sec-
ondarily stressed syllable is chosen as anchor (as in Gail ◆ Ábigàil). Anchoring to an onsetless first
syllable is largely avoided if that first syllable is not stressed. In such cases the main stressed syl-
lable is the anchor (e.g. Tave ◆ NOc ◆ Octavia). Exceptions to these anchoring patterns are
extremely rare (e.g. Al ◆ Alónzo, Ron ◆ Áaron), but where found, the derivatives often have homo-
phone clippings derived in a regular fashion from other bases (cf. Al ◆ Álfred, Ron ◆ Rónald).
As was the case with ‑ie diminutives, the derived form preserves contiguous segments
starting from the left edge of the anchor syllable. To fulfil the constraint on heavy rhymes, the
first consonant of the following syllable often makes it into the clipping (as in Cliff ◆ Clifford).
Some forms show exceptional violation of the contiguity requirement Ike ◆ Isaac or Floss ◆
Florence. Such violations are systematic with derivatives that would otherwise end in /r/
(*Flor). Preservation of consonant clusters is generally variable (e.g. Clif ◆ Clift ◆ Clifton, Jasp
◆ Jas ◆ Jasper, Bren ◆ Brend ◆ Brendan, Mel ◆ Melv ◆ Melvin), but there are specific clusters that
are never preserved (e.g. stop-stop clusters, as in Hec ◆ *Hect ◆ Hector), while other cluster
types never seem to undergo deletion (e.g. /ŋk/, as in Frank ◆ *Fra[ŋ] ◆ Franklin).
The segments that make it into the name clipping are mostly unaltered, but some excep-
tions occur. A large group of rather systematic segmental alternations concern schwas in base-
initial syllables. In such cases the schwa is replaced by a full vowel. Which vowel is chosen is
mostly dependent on the spelling. For example, schwas spelled as <a> in the base may become
/æ/ in the clipping (Al ◆ Alyssa, Pat ◆ Patricia), schwas spelled as <e> typically becomes /e/
(Del ◆ Delila). Apparently more random vowel changes also occur, examples include Bab ◆
Barbara, Kate ◆ Kay ◆ Catherine, Meg ◆ Margaret, Ray ◆ Regina. Again it is noteworthy that these
exceptional forms may be homophonous to the completely regular derivatives of other bases.
The vast majority of name clippings do not alter the consonants taken from the base. If they
do, the change is systematic with regard to two phenomena. The dental fricative /θ/ is often
replaced by a /t/ (e.g. Bart ◆ Bartholomew, Ted ◆ Theodore), and post-vocalic /r/ is avoided as
a final segment and in that position is often replaced by a consonant occurring later in the base
(Chat ◆ Charity, Floss ◆ Florence, Jem ◆ Jeremy, Lon ◆ Lawrence), or it is replaced by /l/ (Hal ◆
Harold, Sal ◆ Sarah, but note the exception Lor ◆ Lorraine NAmE).
402 Size, quantity, and attitude

18.2.3.2. Suffixed hypocoristics: ‑ie, ‑o, ‑s, ‑zza

Of all suffixed hypocoristics, those in ‑ie are by far the most common. Suffixed hypocoristics ‑ie
and ‑o are phonologically restricted in much the same way as the ‑ie diminutives described in
Section 18.2.2.1, with two differences. The hypocoristics can freely anchor to the first and to
the stressed syllables, while the other diminutives have a tendency to be restricted to the first
syllable of the base. With regard to consonant clusters, hypocoristics are less tolerant of cer-
tain clusters. Thus, suffixed hypocoristics do not allow non-homorganic second consonants
in nasal-initial clusters, while clippings do. Derivatives exhibiting these two regular patterns
are, for example, Winny ◆ *Winfy ◆ Winfred versus confy ◆ confidential.
Hypocoristics in ‑s (e.g. Babs, Debs, Fats, Lyds) are monosyllabic and are perhaps best
described as being based on clippings, as all attested forms that we have come across have
bases that are also attested as clipped names. There seems to be a preference for bases ending
in a plosive, but the data sets are too small to make more reliable generalizations.
Another type of hypocoristic formation is comprised of trochaic disyllables ending in
‑zza, which are of relatively recent use, and especially prevalent in BrE and AusE. Some
examples are given in (45) from various sources including the BNC and COCA.

(45) Bazza ◆ Barry, Dazza ◆ Darren, Gazza ◆ Gary / Gascoigne, Hezza ◆ Heseltine, Lazza ◆
Larry, Mazza ◆ Amanda, Mozza ◆ Morrissey, Ozza ◆ Owen, Shazza / Shazzer / Shaz ◆
Sharon, Tezza ◆ Terry

The truncation patterns are very similar to the other suffixed hypocoristics (anchoring to first
or stressed syllable, contiguity, and relative stability of the segments taken from the base).
Base words with intervocalic /r/ or /s/ seem very much overrepresented, and in these deriva-
tives /r/ does not survive, which is reminiscent of clipped hypocoristics.

18.2.4 Clipping
Non-hypocoristic clippings express primarily the attitudinal component of diminution, mark-
ing familiarity with the denoted object or concept. Like abbreviations, they often convey the
in-group status of the speaker and are often restricted in usage to sub-groups of the speech
community at large. Like other diminutives, they may express equal status or reduced social
distance. Eventually, through the process of lexicalization, clippings may become the standard
words for the relevant denotatum. Clippings may be formed from nouns, adjectives, verbs,
and phrases. They do not change the syntactic category of their bases and in the case of phrasal
bases they have the category of the head of that phrase, even though the clipped word may not
be the head and have a different syntactic category (e.g. pubN ◆ [publicADJ houseN]NP).
Clippings based on polysyllabic bases involve loss of phonological material in rather com-
plex ways. In order to explore potential patterns, clippings have been classified according to
which part of the original base is removed, as reflected in the terminology ‘back-clipping’ and
‘fore-clipping’ (e.g. Marchand 1960; Bauer 1983; Berg 2011). However, it is more insightful to
 18.2 Evaluation and size 403

focus on what is preserved in the output and how the material preserved in the derivative relates
to the base (see Lappe 2007; see also Nelson 1998 for French, or Piñeros 1998 for Spanish).
The overwhelming tendency in clipping formation is to form monosyllables. The majority
of clippings anchor to the first syllable (e.g. 93 per cent in Lappe’s database, Lappe 2007:174).
This pattern is exemplified in (46).

(46) ad ◆ advertisement, app ◆ application, caff (BrE) ◆ café, dem ◆ democrat, doc ◆ doctor,
fax ◆ facsimile, gas (NAmE) ◆ gasoline, lab ◆ laboratory, mike ◆ microphone, pram
(BrE) ◆ perambulator, pub ◆ public house, sub ◆ submarine sandwich, submarine
vehicle, substitution, ute (AusE/NZE) ◆ utility truck

If the base starts in an unstressed syllable followed by a stressed syllable, we also find disyl-
labic clippings, as in (47). Example (47a) shows forms based on a main-stressed syllable,
(47b) forms based on a secondarily stressed syllable.

(47) a. artíc (BrE) ◆ artículated lorry, binócs ◆ binóculars, celéb ◆ celébrity, exéc ◆
exécutive, idént ◆ idéntity, posísh ◆ posítion, %postóp ◆ post-óperative, %prelím(s)
◆ prelíminaries
b. co-éd ◆ coèducátion, congráts ◆ congràtulátions, colláb ◆ collàborátion, exám ◆
exàminátion

Other clippings that are longer than one syllable involve truncation that is not purely phono-
logical, but targets sublexical morphemes. This can be found frequently with prefixed or
compound bases. In such cases the surviving material may anchor to and comprise initial,
medial or final (pseudo-)morphemic constituents. We can thus distinguish terminologically
between purely phonological clippings and morphemic clippings. Examples of morphemic
clipping are given in (48).

(48) blog ◆ weblog (with shifted morpheme boundary), bus ◆ omnibus, fiche ◆ microfiche,
fridge ◆ refrigerator, mini ◆ minicar / miniskirt, pike (NAmE) ◆ turnpike, phone ◆
telephone, plane ◆ aeroplane ‖ airplane, shrink ◆ head-shrinker ‘therapist’, stereo ◆
stereophonic record-player, uni ◆ university, video ◆ videorecorder / videotape

A closely related phenomenon that is usually not treated as a case of word-formation is ellip-
sis in lexicalized phrases, as in fries for French fries, the House for either the House of
Representatives or the State House, and the States for the United States of America.
Exceptions to the three major patterns of monosyllabic, disyllabic, and morphemic clip-
pings exist, but are rare. Some are collected in (49).

(49) flu ◆ influenza, gator (NAmE) ◆ alligator, jams ◆ pyjamas, tache (BrE) ◆ moustache, tec
(BrE now arch.) ◆ detective
404 Size, quantity, and attitude

Clippings are frequently used in compounds, either being compounded with a full word (50)
or with another clipping (51).

(50) behavior mod, con-man, interrobang, petrodollar, teletext

(51) biopic, edbiz, hi-fi, kidvid, sci-fi

As we have seen, morphemic clippings faithfully preserve the phonological structure of their
base morpheme. Phonological clippings are more intricate. In the following passages we
return to phonological clippings and their properties.
Phonological clippings preserve contiguous segments starting from the left edge of the first
syllable. In addition, they need to have a heavy rhyme. In order to satisfy both requirements,
they preserve as many segments as needed, sometimes even more (cf. cue ~ cuke ◆ cucumber).
Consonant clusters are variably preserved, as in amp ◆ amplifier versus sis ◆ sister, but there is a
preference for preservation. Clusters that begin with /l/‑ seem to be categorically preserved
(Alb ◆ Albania, alt ◆ alternative). Very rarely, contiguity is violated (e.g. crab ◆ karabiner). Such
violations are predictable, however, if the clipping is based on a form that has an ‑s suffix. In
such cases the suffix is often preserved (e.g. avs ◆ average results, binocs ◆ binoculars, comps
◆ comprehensive exams, maths (BrE) ◆ mathematics, stats ◆ statistics, specs ◆ spectacles).
The segments surviving in the clipped form are normally faithfully preserved except that
schwas in base-initial syllables are regularly turned into full vowels, as in pro ◆ professional, fan
◆ fanatic (with specialized meaning), based on the possible sound–spelling correspondences.
Other vowel changes occur very rarely, and seem to be triggered by the prosodic constraints
discussed above, for example the change from short vowel to long vowel (in an open syllable)
to fulfil the requirement of having a heavy rhyme, as in bro ◆ brother. Consonantal alternations
are rare (e.g. prink ◆ principle, spaz ◆ spastic), and do not seem to affect the consonants that are
prone to alternation in clipped names, that is /r/ and /θ/ (cf. cath ◆ cat ◆ catheter, or ter ◆ tel
◆ tess ◆ terrorist, as against Nat ◆ Nathaniel, Sal ◆ Sarah, Floss ◆ Florence).
Overall, it is clear that, in spite of the variation this morphological category exhibits, the
formation of clippings is heavily constrained and to a large extent regular and predictable.
Notably, the patterns displayed by hypocoristic clippings (as discussed in Section 18.2.3) are
similar, but not quite identical to the patterns observed by the clippings described in this sec-
tion, which substantiates the claim that we are faced with distinct morphological categories.

18.2.5 Augmentatives
18.2.5.1 The prefix hyper‑
The prefix hyper‑ is extremely productive in both technical and non-technical domains, attached
commonly to both nouns and adjectives and less commonly to verbs. Where it is relevant, the
use of the prefix hyper‑ often seems to imply some pathological case or, the non-technical
equivalent, something which is excessive. In most ways the semantics of hyper‑ is comparable to
 18.2 Evaluation and size 405

that of super‑ (with which it is cognate). While super‑ came into English as a prefix, hyper‑ was
originally a combining form (Bauer 1983: 215). The prefix hyper‑ continues to be able to attach
to other combining forms, but both prefixes now attach freely to words. A few of the many
examples from COCA are given in the following examples, divided according to the bases, with
nominal (52), adjectival (53), verbal (54), and neoclassical elements (55) treated separately.

nominal bases: hyperachiever, hyperagency, hyperarousal, hyper‑deflation, hyper‑elite,

(52) 
hyper‑investment, hyper‑pension, hyper‑steel, hypercar, hypercello, hyperclarity,
hypercoffee, hypercognition, hypercompetition, hypercomputer, hypercube, hyperdrive,
hypergravity, hyperhonesty, hyperindividualism, hyperinsecurity, hyperirritability,
hyperlaw, hypermarket, hypermedia, hypermiler, hypernova, hyperparasite, hyperplane,
hyperreggae, hyperspeed, hypersuccess, hypersurface

adjectival bases: hyper‑abrupt, hyper‑awake, hyper‑fertile, hyper‑grossing, hyper‑­

(53) 
health-conscious, hyper‑miniature, hyper‑optimistic, hyper‑rapid-fire, hyper‑svelte,
hyperacerbic, hyperactive, hyperalert, hyperambitious, hyperarid, hyperaware,
hyperblue, hypercarnivorous, hyperchic, hyperchronic, hypercool, hyperdense,
hyperelegant, hypermacho, hypermeticulous, hyperobsessed, hyperpermissive

verbal bases: hyperaccumulate, hyperanalyze, hyperbrowse, hypercharge, hypercorrect,

(54) 
hyperextend, hyperovulate, hypersoar

bound bases: hyperaemia, hyperalgesia, hyperaspis, hyperbole, hy­

(55)  per­
bo­
rea,
hyperdermic, hyperemia, hypergammaglobulinemia, hypergamy, hyper­golic, hyper­
graph, hyperhomocysteinemia, hyperlect, hypermnesis, hyperonymy, hyperostosis,
hypertrophia

In many of these cases the order of affixation is underdetermined. For example, hypervegitari‑
anism may be seen as a case of hyper‑ prefixation to a base vegetarianism, or as a case of ‑ism
suffixation to a base hypervegetarian. Some examples like this have been included to show
that potentially the bases are complex, but many more have been omitted for this reason.
The prefix is generally secondarily stressed (on the first syllable), but if the prefix combines
with a neoclassical form that attracts stress to the syllable immediately preceding it (see
Chapter 9.4.5 for discussion), the second syllable of the prefix receives the main stress of the
derivative, e.g. hypérgamy, hypérbola, hypértrophy.

18.2.5.2 The prefixes maxi‑ and midi‑

The prefix maxi‑ is attached mainly to nouns to show great size. It often contrasts implicitly or explic-
itly with the prefix mini‑ (see Section 18.2.2.2). Some examples from COCA are given in (56).

(56) maxiboat, maxicab, maxi(‑)coat, maxi‑manor, maxi(‑)pad, maxi(‑)skirt

The prefix is considerably less profitable than mini‑, but is nevertheless productive, as indi-
cated by the fact that most of the relevant words in COCA are hapaxes. The prefix is mostly
406 Size, quantity, and attitude

main-stressed (e.g. máxicircle, máxi-skirt), sometimes secondarily stressed (màxi per‑

fórmance). As with mini‑, the question can be raised as to whether this is a prefix or a com-
pound element, especially since maxi is frequently spelled as a separate word, which probably
reflects the writers’ intuition about the lexical tightness of the derivatives. Where maxi occurs
as an independently spelled word, it is frequently being used as an adjective, and it is not clear
that this usage is, in reality, anything other than a different representation of the prefix. Some
examples are given in (57).

(57) Bicycling 1998: Here’s a mini bike capable of maxi performance

Mechanical Engineering 2002: Select maxi systems can allow pump-down of a supply
cylinder as low as 150 psi

Just occasionally, maxi‑ is prefixed to an adjective, yielding an adjective that is intensified in

its meaning (paraphrasable as ‘extremely X’). Two examples from COCA are presented in (58).

(58) Gregory Macdonald 1994: You’re marrying into a maxi‑wealthy family

Town and Country 2007: The country may be minisized (999 square miles, slightly
smaller than Rhode Island), but it’s maxirich.

Closely related to maxi‑ is the prefix midi‑, which denotes medium size. Strictly speaking, it
should perhaps not be classified as an augmentative, but we treat it here nevertheless, as it
stands at least in a paradigmatic contrast with mini‑. It is not very productive and occurs
chiefly with reference to garments or audio equipment, as in midi(-)coat, midi-dress, midi-
length, midi-skirt, midi(-)cd‑player, midi-hi‑fi-system. Orthographically, it seems to behave like
maxi‑, with two-word spellings being preferred for most formations. Stress-wise it also shows
the kind of variability found for maxi‑.

18.2.5.3 The prefix mega‑

The prefix mega‑ has a number of distinguishable uses. First it can be used as a combining
form, that is, in combination with another neoclassical element, as in (59) (see Chapter 19).
Here it is sometimes in competition with megalo‑, which is used exclusively as a combining
form. A second use of mega‑ is in the names of units to multiply the unit by one million, as in
(60). In its third use it can be used as an ordinary English prefix to mean ‘extremely large’. In
this last kind, it is usually attached to nouns (61), but is also found on adjectival bases, mean-
ing ‘extremely’ (62). The examples below are all from COCA.

(59) megakaryocytes, megalith, megaphone, megapod, megapode, megapolis, megaptera,

megatherium
(60) mega‑decibel, mega‑dollar, megaflop, mega‑joule, mega‑kilowatt, mega‑pixel,
megabar, megabit, megabits/sec, megabyte, megacalorie, megadecibel, megahertz,
 18.2 Evaluation and size 407

megahorsepower, megajoule, megakilometer, megamile, meganewton, megaparsec,

megapascal, megapixel, megaton, megavolt, megawatt, megawatt‑hour

(61) mega‑acquisition, mega‑amount, mega‑artist, mega‑attraction, mega‑book, mega‑box,

mega‑budget, mega‑chain, mega‑crayfish, mega‑ego, mega‑event, mega‑market,
mega‑selling, megabrain, megabrand, megacorridor, megadealership, megafarm,
megamachine, megamansion, megamedia, megamind, megamouth, megamusical,
megaproject, megastore, megastructure, megatap, megatask, megateam, megatech,
megathrust, megatonnage, megazillionaire, megazit

(62) mega‑bad, mega‑best-selling, mega‑funky, mega‑personal, mega‑quick, mega‑scaled, mega(‑)

rich, mega(‑)sized, mega(‑)successful, megacool, megafamous, megapopular, megawide

In the first two types of formation mega‑ always carries main stress, while in the third type, mean-
ing ‘extremely (large)’, it seems variably main-stressed, or secondarily stressed. The spelling is vari-
able between one-word, hyphenated and two-word spellings, with established formations having
a tendency toward one-word spellings, and new formations occurring hyphenated or even as two
words (e.g. mega churches, mega famous, mega drool-inducing, mega success, all from COCA).

18.2.5.4 The prefix super‑

The semantics of super‑ is particularly difficult to specify. There are a number of sub-meanings
which can be distinguished, but they are often interchangeable for the same lexeme, and one
meaning often shades into another. Given that the original meaning of super‑ is ‘above’, there
are cases where super‑ means ‘in a class above’, and, indeed, this might be an over-arching
meaning for the prefix. The locative uses of the prefix are discussed in Chapter 16. Many of
the derivatives also carry an attitudinal shade of meaning, expressing either excessiveness or
appreciated large size in the eyes of the speaker. This attitudinal meaning can be found along-
side the other meanings in one and the same derivative, or may occur on its own.
The augmentative meaning can have various more specific readings that emerge in the interac-
tion of the prefix with its particular base. We find readings where the augmentative denotes quali-
tative superiority either with an adjectival base, as in (63) or with a nominal base, as in (64).

(63) 
super‑easy, super‑powerful, super‑quick, super‑sharp, super‑simple, super‑skinny,
super‑smart, super‑sophisticated, super‑thin, superabsorbent, superabundant,
superbright, supercheap, supercold, superfit, superpowerful, superskinny, superslim,
supersmooth, superthin

(64) supercar, superhero, superinsulation, superlawyer, supermodel, supermom, supernanny,

superstar, superwoman

The derived word can also denote something which is of a similar type to the thing denoted
by the unprefixed lexeme, but one step up some notional scale; again this can be with a nomi-
nal (65) or adjectival base (66).
408 Size, quantity, and attitude

(65) 
super‑ego, super‑glue, super‑heavyweight, super‑organism, superalloy, superbike,
supercollider, supercow, superdrive, superhumans, superintelligence, superspy,
supersymmetry, superweed

(66) supernormal, supersensory

The augmentative may also be interpreted as denoting something which is simply bigger than
the thing denoted by the unprefixed lexeme, as in (67).

(67) 
superbomb, supercluster, supercomputer, supercontinent, supercroc, supergiant,
supergun, superhighway, superjumbo, superliner, superspeedway, superstore,
supertanker

As already mentioned, the derived lexeme can, in the appropriate context, also carry the
meaning of ‘excessive(ly)’, based on a nominal, adjectival, or verbal base (68a). Example
(68b), from COCA, illustrates this reading.

(68) a. super‑expensive, superabundance, supercool, supercritical, superovulate, supersensitive

b. Redbook 2003: And I came to understand some of the theories experts had as to the
cause of IC-from the leaky-bladder hypothesis, in which microscopic tears allow
urine to irritate the bladder wall, to a theory that suggested supersensitive nerve
endings were to blame

Despite the classifications of the various shades of meanings provided in (63) to (68), the
readings mentioned so far are often not clearly distinguishable, as the examples in (69) illus-
trate. For example, a superministry may be one that is simply larger than a usual ministry, or it
may be one that is perceived as excessively large, or a step up on a scale of significance.

(69) super‑agent, superagency, superministry, superpredator, superregion, superscience,

supervillain

The prefix is either main-stressed or carries secondary stress. The distribution of the two
stress patterns is not quite clear, but more lexicalized words tend to have main stress on the
prefix (e.g. Súperbowl). Following the same patterns as the previously discussed prefixes,
super‑ is normally left-prominent, but may also be main-stressed on its second syllable with
some lexicalized forms, or if the following element attracts stress to that syllable (e.g. supér‑
fluous, supérlative, supérnal).
The potentially problematic form superette (‘small supermarket’) is not current in all parts
of the English-speaking world, but provides a superficial instance of a prefix and a suffix with-
out a base. However, it is probably best analysed as a morphemic clipping of supermarket-ette.
This analysis is supported by the fact that ‑ette avoids attachment to /t/‑final bases.
 18.2 Evaluation and size 409

18.2.5.5 The prefix ultra‑

Ultra‑, coming from the Latin word meaning ‘beyond’, has a scientific or social-scientific
usage (70) where it retains that Latin meaning and a more colloquial usage where it is pre-
fixed to adjectives in particular with a meaning of ‘extremely’ (71). There are instances where
it can be difficult to determine which use is which. The examples here are from COCA. They
show that ultra‑ attaches to a wide variety of bases, including adjectives, compounds, and
phrases. The prefix usually carries secondary stress, but a small minority of forms have initial
main stress (e.g. últrasound).

(70) 
ultra‑heat-resistant, ultra‑high-density, ultra‑high-efficiency, ultra‑Keynesian, ultra‑­
left-wing, ultra‑low-dose, ultra‑loyalist, ultra‑maoist, ultra‑uv, ultrabroadband,
ultracentrifuge, ultradense, ultralow-frequency, ultramarathon, ultramicrobacteria,
ultranationalism, ultraorthodox, ultrapacifist, ultrasaurus

(71) 
ultra‑absorbent, ultra‑accurate, ultra‑ambitious, ultra‑American, ultra‑aware, ultra‑­
exaggerated, ultra‑svelte, ultra‑swanky, ultraclassy, ultraclean, ultraexclusive, ultraexpensive,
ultralurid, ultralush, ultralusty, ultraminiature, ultramobile, ultrarapid, ultrarare,
ultrarational, ultrasexy, ultrasweet

The prefix ultra‑ is also found prefixed to nouns, as in (72). Here the meaning may not be
clear out of context and may vary from the scientific to the rather colloquial.

(72) ultra‑coach, ultra‑elf, ultra‑kiosk, ultra‑nerd, ultra‑poppy, ultra‑underdog, ultrabike,

ultracop, ultradog, ultrafluid, ultrahero, ultrahit, ultraluxury, ultramarina, ultrasun,
ultratorque, ultrawife

18.2.5.6 The prefix turbo‑

The element turbo‑ originally meant ‘driven by a turbine’ as in turbo-charge, turbo-fan, turbo-
jet, turbo-prop. In modern usage this has become a quantitative and attitudinal prefix indicat-
ing desirable high speed and power. COCA provides Turbolinux, Turbotax, TurboChip,
TurboChef (all brand names) and turbochat, turbonews, turbosled, turbospeed, and turbowash. It
carries the main stress of the derivative on the first syllable.

18.2.5.7 Prefix rivalry: hyper‑, maxi‑, mega‑, super‑, ultra‑

Given that the meanings of these prefixes are very similar to each other, and that they often attach
to the same kinds of bases, it is not surprising to find examples where these suffixes seem inter-
changeable, with no, or only very little, difference in meaning. Hyper‑ can be used neutrally, but
when it occurs in medical, technical, and science fiction contexts it is the one used to convey
‘large in a negative, pathological sense’. The prefixes maxi‑, mega‑, ultra‑, and super‑ seem most
often to be used in a positive sense. It is possible to find doublets for any pair of these prefixes. The
examples in (73) to (79) illustrate this, with (77d) being from the BNC, all others from COCA.
410 Size, quantity, and attitude

(73) a. Rolling Stone 1990: The Police were the first video-era rock & roll band of any
magnitude. They grew into that era, and it was sort of hyper‑accelerated through
video, through MTV.
b. Analog Science Fiction & Fact 2001: The engines don’t hum—no devices move within
them but the super‑accelerated particles pushing the Resurrection slowly to
greater and greater speed, and even when you stand next to the reactors, there’s no
sound.

(74) a. Ear, Nose, & Throat Journal 2004: Hyperacute forms progress in only a few hours,
but most cases progress over a period of days, probably because patients undergo
antibiotic treatment at the onset of symptoms.
b. USA Today 2003: Early in the film, blind attorney by day, vigilante by night Matt
Murdock uses his other ultra‑acute senses to trail the hard-to-impress Elektra, a
dazzling practitioner of martial arts.

(75) a. Total Health 2006: Yet if the immune system is continualIy on hyperalert,
inflammation can become chronic.
Outdoor Life 2007: If you really want to challenge an antelope with all your skills, try the
b. 
animal with a bow or muzzleloader. You’ll know you’ve been on a hunt once you do. The
need to get very close to this ultra‑alert speedster will test all your abilities as a hunter.
Patchwork Planet 1998: It wasn’t even ten o’clock, but she must have been in bed,
c. 
because she answered so immediately, in that superalert tone people use when they
don’t want to let on you’ve wakened them.

(76) a. Skiing 1998: I suggest that you spend at least a few hours on this hyperfriendly
terrain making very big, round long-radius turns.
b. Redbook 2010: Everyone was superfriendly—and very concerned about my
hydrating needs: Every single person who spoke to me asked if I wanted water.
c. Newsweek 1998: Acer preloads two especially useful programs on its machines: the
My Aspire Guide, an ultrafriendly software that will hold your hand should you
run into problems, and a !just for kids! browser for tapping into kid-safe Web sites.

(77) a. Smithsonian 1994: He is explaining why ostensibly mature men and the occasional
woman buy his hyperpowered, handmade ultracars like CheZoom.
b. Journal of International Affairs 1999: Ultimately, plug-in Hypercars could provide
five to 10 times as much generating capacity as all utilities now own—enough in
principle to displace essentially all central thermal power stations at a profit.
Popular Science 2008: Now that Nissan has proven that it can be done, expect more
c. 
accessible, technology-rich supercars in the years ahead.
 18.2 Evaluation and size 411

d. w_fict_prose 1991: And he had a status-symbol mega car which he drove very slowly
in the middle lane

(78) a. Consumers Research Magazine 2004: Independent Rear Suspension will, however, be
fitted to the more expensive, ultra‑performance SVT Cobra.
b. Bicycling 1998: Here’s a mini bike capable of maxi performance thanks to a
welded‑aluminum frame

(79) a. USA Today 1990: Reggie Fountain, in Fountain, averaged 107.94 mph to win the
Super Boat class at the Popeye’s Racefest in New Orleans
b. USA Today 1991: Goetz, who rolls up his sleeves and works alongside his employees,
has worked with Conner since 1979 and built Koch’s prize-winning maxiboat
Matador in 1990.

18.2.5.8 Non-prefixal augmentatives

The overall rule is that English does not have augmentative suffixes. However, there are per-
tinent forms that use ‑rama, what appears to be what we call a ‘splinter’ (see Chapters 2 and
23), and some with the putative suffix ‑ola. Examples are listed in (80).

(80) a. Hardboiled homicide cops dancing the fawnorama. (McClister 1999: 75)
b. Leaves us with a gigantic hole‑o‑rama in our December release schedule. (Cannell
2003: 252)
Christian_Science_Monitor 2007: I wrote about the sport in the 1960s and 1970s,
c. 
which was the medieval age compared with today’s big-bucks cashorama.
c. there’s a huge problemola with fetchmail (Newbie 2001)
d. the screen debut of Jimmy Durante, he of the large sch[n]ozzola (OED cit. 1930)

The suffix ‑ola shown here is presumably the non-diminutive suffix which is usually assumed
not to have a specific meaning in English. It is often assumed that the same suffix appears in
payola and granola (despite the different meanings in the two words, see Randle 1961; Glowka
1985).

18.2.6 Reduplication
There are a number of patterns of reduplication in English whose main function, apart from
the creation of new words, appears to be attitudinal. The examples can be classified in a num-
ber of different ways which interact. One classification is in terms of two words with some
phonological constituent in common that are both words, while in other instances one
412 Size, quantity, and attitude

or both of the elements may be items without independent occurrence. A formal classifica-
tion is in terms of the process of reduplication involved. The examples below are all estab-
lished, and are mainly taken from Matthewson (1991). The overwhelming majority of
examples have elements of matching length, and each element is usually no more than disyl-
labic in itself.
First we find examples where there is real reduplication, the two elements are homopho-
nous. The examples in (81) all illustrate nursery language.

(81) chuff‑chuff, choo‑choo, doo‑doo, quack‑quack, woof‑woof

Second we find instances where the two elements retain the consonantism from the stressed
syllable of the foot, but show ablaut between the two elements. There are two common pat-
terns of ablaut: items with /ɪ/ in the stressed syllable of the first element and /æ/ in the
stressed syllable of the second, see (82), and items with /ɪ/ in the first element and /ɒ/ ‖ /ɑː/
in the second, see (83).

(82) chitchat, dilly‑dally , fiddle‑faddle , flimflam, jimjam(s), kitcat, knick‑knack, mishmash,

pitter‑patter, rickety‑rackety, shilly‑shally, tittle‑tattle, whim‑wham

(83) 
clip‑clop, criss‑cross, dingdong, flipflop, pingpong, singsong, ticktock, tip‑top,
wishy‑washy

In a third case, the rhyme of the stressed syllable and any remaining material in the foot
remains the same, but the onset is changed. Examples are in (84).

(84) 
argy‑bargy, boogie‑woogie, easy peasy, fuddy‑duddy, fuzzy‑wuzzy, hab‑dabs,
hanky‑panky, heebie‑jeebies, helter‑skelter, higgledy‑piggledy, hocus‑pocus, hoity‑toity,
hokey‑pokey, holus‑bolus, humdrum, hurdy‑gurdy, hurly‑burly, incy‑wincy,
killer‑diller, lovey‑dovey, mumbo‑jumbo, pow‑wow, raggle‑taggle, roly‑poly, silly‑billy,
super‑duper, teensy‑weensy, teeny‑weeny, walkie‑talkie, wham‑bam, willy‑nilly

There is a slight preference for a bilabial consonant in the onset of the first syllable of the
second element where this is not predetermined by the original, and there is also a slight
preference for plosives in the same position. The examples illustrate, though, that neither of
these preferences is categorical.
There are also words which can be viewed as genuine compounds whose elements happen
to rhyme. The rhyme is certainly part of the motivation for the coining of the compound, and
the rhyme certainly makes the compound more striking. Examples are in (85). In these cases,
though, it is not clear that the alliteration is any different in principle from that which affects
expressions like those in (86).

(85) 
backpack, cop‑shop (BrEsl ‘police station’), culture vulture, hotpot, hot shot,
hurry‑scurry, peg leg, ragbag, sin bin
 18.2 Evaluation and size 413

(86) come hell or high water, cool as a cucumber, fit as a fiddle, kith and kin, might and
main, now or never, Peter Pan, ring‑a‑ring‑o’‑roses, vim and vigor

Bauer and Huddleston (2002) suggest compounds like those in (85) are trivializing, and that
is certainly true in cases like gang-bang and culture vulture, but not necessarily in the case of
hotpot. Perhaps the best we can say is that these words involve a ludic element.
A specific type of the rhyme-motivated construction arises in NAmE (based on Yiddish)
use of /ʃm/ to replace the onset. This is generally treated as syntax rather than word-forma-
tion, in the sense that the /ʃm/word is usually separated from its base by a comma (and
occasionally by more). Examples from COCA are given in (87). They are clearly dismissive
and disparaging.

(87) a. 
acting, shmacting; chemistry shmemistry; children, shmildren; cover, shmover;
eloquent, shmeloquent; fantasy shmantasy; innovative, shminnovative; millennium,
shmillennium; organic, shmorganic; parked, shmarked; peace, shmeace
b. Cobb 1994: I don’t know baseball from shmaseball

It is perhaps worth noting that many discussions of English word-formation include these
reduplicative constructions as compounds, seeing them as parallel to the examples in (85).
For reasons discussed in Chapter 19 we prefer not to count these as compounds at all.

18.2.7 Expletive insertion

Expletive insertion is a phenomenon in contemporary English whereby words like fucking,
bloody, and frigging can be inserted into the middle of a word, giving words like kanga-bloody-
roo, guaran-frigging-tee, Tala-fucking-hassee, abso-bloody-lutely. This process appears to be mar-
ginal in the sense that some speakers do not control it, though most speakers do. Some sources
(e.g. Aronoff 1976; Plag 2003) term this infixation, but it is not strictly affixation, since the ele-
ments that are inserted are words rather than affixes. Part of the puzzle of these formations is,
therefore, their precise nature. If it is a relevant process, it is one whose main function appears
to involve the attitude of the speaker: the formations created in this way are emphasized but in
a pejorative way, as well as having their style-level changed by the presence of the expletive.
Although we talk of expletive insertion here, in some instances, as is pointed out by McMillan
(1980: 163) non-expletives can be inserted: Jesus H. Christ, terra-extremely-firma, theo-jolly-logical.
Typically, such alternative inserts are perceived as euphemisms for an expletive.
The phonology of this process has been dealt with as an instance of prosodic phonology
(Chapter 9.6.3). The implication of the required prosodic structure of words into which
expletives can be inserted is that there are words which cannot be used as bases. In such cases
the expletive can still be used, but it is preposed to the noun, as in (88).
414 Size, quantity, and attitude

(88) bk:Senseless 2011: Nobody here ordered any fucking flowers.

HarpersBazaar 2011: We’ve spent the fucking lot.
ACAD_Commentary 2010: Finkler’s wife asks the increasingly obsessive Treslove if he
doesn’t wish that Jews would just “shut up about themselves . . . whether the world
hates them or it doesn’t, the fucking Holocaust, fucking Palestine?”

The implication of this might be that this is a case of prosodic syntax. The expletive is placed
in such a position as to strengthen a strong syllable in the original; where necessary, it
migrates across the word-boundary into the word in order to do this. The last example in (88)
suggests that this is not the whole story. Given that holofuckingcaust would fit the prosodic
pattern of typical outputs of this process, the writer must have chosen to use the preposed
expletive, which means that the two structures are alternatives, not variants of the same
underlying pattern.
There are other indications that the expletive insertion is a morphological process. The
first is that while the preferred structure is for the expletive to be placed relative to a stressed
syllable, in some instances it can be placed between morphs. Thus alongside prosodically
expected unbefuckinglievable, we find (89a), and there are even rare examples like (89b, c),
where morphology seems more important than prosodic structure.

(89) a. MAG_Esquire 2000: well . . . it was amazing, incredible, unfuckingbelievable.

b. Triquarterly 2005: He wanted a huge dog, a humongous dog, the world’s most enor-
fucking-mous dog in the universal pound.
c. In-goddamn-consistent (McMillan 1980)

Whatever the precise nature of this phenomenon, therefore, it has morphological implica-
tions, and in some cases the insertion point is determined by morphological structure. It
might be possible to view this phenomenon as a prosodically-driven version of compound-
ing with an expletive.

18.2.8 The prefixes pseudo‑ and quasi‑

Another affix which, at least in colloquial usage, denotes an attitude is pseudo‑, where it indi-
cates something sham or less than genuine. In scientific usage the negative implications are
absent. The colloquial usage is illustrated in (90), where the bases are nouns or adjectives.
The examples are from COCA.

(90) 
pseudo‑Arabic, pseudo‑celebrity, pseudo‑classical, pseudo‑conservative, pseudo‑diary,
pseudo‑environment, pseudo‑independent, pseudo‑issue, pseudo‑linguistics, pseudo‑news,
 18.2 Evaluation and size 415

pseudo‑philosopher, pseudo‑reality, pseudo‑religious, pseudo‑science, pseudo‑superhero,

pseudo‑Victorian, pseudocapitalism, pseudocoffeehouses, pseudofeminists, pseudoscience

The scientific usage is exemplified in (91a). There are also neoclassical formations with
pseudo‑ in the first position. A few examples are provided in (91b).

(91) a. 
pseudo‑cells, pseudoacacia, pseudoaneurysms, pseudoarthrosis, pseudobinary,
pseudocysts, pseudoephedrine, pseudopod, pseudorandom, pseudosarcoma,
pseudoscorpion, pseudotumor, pseudoxanthoma
b. pseudoceros, pseudodoxia, pseudomorphic, pseudonym

The prefix is either main-stressed or secondarily stressed. If main-stressed, it has three allo-
morphs. One with initial stress and a final schwa, as in pséudonym, one with initial stress and
a final diphthong, as in pséudopod, and one with final stress, which occurs if a stress-shifting
form follows it, as in pseudónymous. In scientific usage, the prefix may (sometimes variably)
lose its final vowel if followed by a vowel-initial base, as in pseudepisematic, pseudaposematic,
pseudarthrosis ~ pseudo-arthrosis (all from OED).
There is another prefix that expresses the speaker’s doubtfulness about the presence of all
the necessary properties in the referent denoted by the base, the prefix quasi‑. It is glossed by
the OED as ‘resembling or simulating, but not really the same as, that properly so termed;
having some but not all of the properties of a thing or substance’. The prefix is quite produc-
tive and attaches to nouns, adjectives, adverbs, compounds, and phrases as shown in (92),
with derivatives found in COCA. The prefix is either main-stressed or secondarily stressed.

(92) a. quasi‑editor, quasi‑homophony, quasi‑intuition, quasi‑radio, quasi‑shoe, quasi‑sleep,

quasi‑slavery, quasi‑zombie
b. 
quasi‑experimental, quasi‑hip, quasi‑hypnotic, quasi‑public, quasi‑stable,
quasi‑stellar, quasi‑utopian, quasi‑violent
c. quasi‑accidentally, quasi‑anonymously, quasi‑clockwise, ­quasi‑­elastic­ally, ­quasi‑­
faceti­ously, quasi‑simultaneously,
d. quasi‑backcountry, quasi‑flat‑rate, quasi‑newsgroup, quasi‑market-based
e. quasi‑American‑style, quasi‑free‑market, quasi‑global‑scale, quasi‑working-class

A comparison of quasi‑ and pseudo‑ reveals that the two prefixes share the meaning compo-
nent that the derivative does not refer to a genuine exemplar of its head’s category. However,
quasi‑ lacks the element of falseness that pseudo‑ generally carries. This can be illustrated with
the examples in (93) and (94), from COCA:

(93) a. Consumers Research Magazine 1999: One of the first countries to reform ATC in
this way was Britain, which created the National Air Traffic Services as a
416 Size, quantity, and attitude

quasi‑independent corporation in 1972. While still nominally part of the U.K.

Civil Aviation Authority (their equivalent of our FAA), NATS is funded mostly
by user fees and is regulated for safety by the safety branch of the CAA.
b. Africa Today 1998: Second, the plurality system provides greater opportunity to
manipulate the size of single-member constituencies through malapportionment.
An incumbent government (represented by a pseudo‑independent Electoral
Commission) can adjust the size and boundaries of constituencies according to its
needs.

The ‘not genuine’ meaning component leads to the negative evaluative meaning compo-
nent that can be found with many pseudo‑ derivatives as the falseness is often seen as an
intentional act of deception or obfuscation. Notably, and as already mentioned above, the
scientific usage of pseudo‑ lacks the evaluative attitude, which makes both prefixes eligible
for the coinage of a new scientific term, and we do find both, perhaps with a preference
for pseudo‑ (for example, astronomical quasi-stellar for quasi‑, and the examples in (91a)
for pseudo‑).
Other potential rivals of pseudo‑ and quasi‑ are the suffixes ‑ish, ‑esque, and ‑oid whose
derivatives also denote something that is similar, but not identical, in shape or quality to what
the base denotes (see Chapter 14). Like the prefix quasi‑, the suffixes lack the element of
falseness. The exact difference in meaning, if any, between quasi‑, ‑ish, ‑esque, and ‑oid is not
entirely clear and the distribution perhaps a matter of register or domain. The examples in
(94) illustrate the use of the five affixes with the same base, modern (examples from COCA).

Geographical Review 1998: One need not admire flimsy construction, the short-sighted
(94) 
planning, the overdramatized, over-colored, pseudo‑modern ranch houses which are
rising all over the country;
Foreign Affairs 2009: This is both more modest and more realistic than earlier notions
of building a modern democracy in a society in which only the urban sectors are more
or less quasi‑modern and the rural areas are in many respects still quite medieval.
Town and Country 2009: Since opening in December, Benjy’s has been strictly SRO.
The modernesque look sets the stage for boisterous fun in the upstairs lounge, where
you can enjoy the signature blood-orange margaritas, and downstairs, where the best
choices include tuna tartare on a tempura edamame roll, [ . . . ]
San Francisco Chronicle 1996: “I always like to do Shakespeare in modernish dress
because I want to put people into clothes rather than costumes,” he explains.
Public Interest 1992: [O]verall, these designs are refreshingly free from formula: no
modernoid lunar landscapes with lollipop lightbulbs, no ‘Olmstoid’ caricatures of
Frederick Law Olmsted’s pioneering work of a century ago.
 18.3 Non-evaluative attitude: anti‑ and pro‑ 417

18.3 Non-evaluative attitude: anti‑ and pro‑

Two affixes which show attitude in a rather more specific way are anti‑ and pro‑. These do not
show attitude in the sense that they have connotations attached to them, but in the sense that
they specify an attitude.
Anti‑ has at least three distinct uses. The first is the attitudinal meaning, where it attaches
to a noun, a name, or an adjective to show opposition to whatever is denoted by the base.
Examples are provided in (95a). The second is the medical or scientific usage, where the
prefix is usually attached to an adjective, though may also be attached to a noun, to show the
effect of something in attacking or counteracting something else. Some of the examples in
this set are less scientific, but retain that meaning. Examples are in (95b). In the third usage,
illustrated in (95c), the prefix seems to express a negative meaning with readings like
‘reversed’, ‘non‑’, or ‘opposite to’. Examples are from COCA.

(95) a. attitudinal
 antiabolitionist, antiabortionist, antiacademic, anti‑additive, anti‑appeasement,
anti‑Aquino, anti‑aristocratic, anti‑Aristotelian, anti‑Bergsonism, anti‑Blair,
anti‑British, anticensorship, antichoice, anti‑environmentalist, anti‑free‑speech,
anti‑hate‑speech, anti‑Kissinger
b. medical or scientific
antiacid, antiacne, antiaddiction, antiaircraft, antiamoebin, anti‑amyloid, antibody,
anticholera, anticholesterol, anticonvulsant, antidiuretic, antiflatulence, antifreeze
(note the verbal base), antifungal, antigenic, antineutrino, antioxidant, antitussive,
c. negative
anti‑alphabetical, anti‑biography, antichrist, anti‑election, antihero

I n a few instances, anti‑ combines with final combining forms (and thus behaves like a com-
bining form itself), creating neoclassical compounds, as in (96).

(96) antidote, antiphony, antipodes, antiptosis, antisoma

Anti‑ has the allomorph ant‑ before <h> or before a vowel in neoclassical formations as in
antagonist, anthelion, antonym. Occasionally this occurs where the element is added to a word
rather than to a bound morph, as in antacid. More often, though, as illustrated in (95) and
(96), the <i> is retained. The prefix can be recursive, as in anti-anti-muslims, anti-anti-porn,
antiantiquarks, antiantirelativism, anti-anti-semitism.
The prefix anti‑ is most often secondarily stressed (ànti-abórtion, ànti-magnétic), but can
also occur with main stress on the first syllable (ántibòdy, ántimàtter) and with main stress on
its second syllable, when stress‑shifted (antónymy, antípathy).
418 Size, quantity, and attitude

We now turn to the prefix pro‑. There are two types of instance of the string <pro> that
should not be confused with the prefix under discussion here. First there is the Latin prefix
pro‑, whose reflex can still be seen in words like produce, prosecute, provide, provoke, but which
is no longer analysable as a prefix in English. Second there is the clipped form of professional,
which can occur as a compound constituent (e.g. in pro-am, pro-football, pro-level). There are
two meanings of the actual prefix pro‑, only one of which is clearly productive. The productive
meaning is the attitudinal one (‘in favour of ’), illustrated by the words from COCA in (97).

(97) pro‑abortion, pro‑apartheid, pro‑Arab, pro‑Aristide, pro‑British, pro‑business, pro‑­

Clinton, pro‑communist, pro‑consumer, pro‑death, pro‑democracy, pro‑environment,
pro‑European, pro‑gay, pro‑German, pro‑industry, pro‑inflammatory, pro‑Iranian,
pro‑Islamic, pro‑Israel, pro‑life, pro‑NAFTA, pro‑Nazi, pro‑prosecution, pro‑war,
pro‑western, pro‑woman, pro‑Yeltsin, proactive, proslavery

The probably non-productive meaning is ‘in place of ’, which is found in established words
like those in (98).

(98) proconsul, pro‑form, pronominal, pronoun, pro‑Vice-Chancellor

Established uses of pro‑ in words like prolapse, prolong (‘movement forwards’) and prodrug
(‘before’) are typically hard to analyse in English words.

18.4 Quantification and measure

18.4.1 Nominal ‑ful
The suffix ‑ful to be considered here is the one which creates nouns (or, more specifically,
noun classifiers, see Dalton-Puffer and Plag 2000) from nominal bases which denote objects
viewed as containers. The derivative denotes a measure. This suffix is very productive in its
domain, as is illustrated by the sample of words from the BNC in (99). The words in (99)
show a preference for monomorphemic bases, but include derivatives and compounds.

(99) armful, barnful, barrelful, barrowful, basketful, bathful, beakful, bellyful, bottle­ful, bowlful,
boxful, caveful, cellarful, cupboardful, cupful, diskful, drawerful, earful, eggcupful, eyeful,
fistful, gutful, hatful, headful, houseful, jugful, kettleful, lungful, mouthful, mugful, neckful,
nestful, noseful, officeful, pipeful, planetful, pondful, potful, pramful, purseful, quiverful,
sackful, screenful, sinkful, skyful, sporranful, syringeful, tankful, teaspoonful, thimbleful,
townful, trainful, truckful, trunkful, tubful, tumblerful, wineglassful

There is a prescriptive position that these nouns pluralize by marking the plural on the base
(spoonsful, etc.), but it is more common to find the plural marked on the right-hand edge of
 18.4 Quantification and measure 419

the word (spoonfuls, etc.). The BNC has only one form with the internal ‑s, and many with
the final ‑s. While COCA has rather more forms with internal inflection, those with final
inflection still predominate.

18.4.2 Verbal re‑

The prefix re‑ is relevant to a chapter on quantification because it quantifies over events by
indicating iteration. It is primarily added to verbs (according to OED particularly transitive
verbs), though nouns and adjectives may subsequently be derived from these verbs. Examples
of relevant verbs from COCA are presented in (100). The examples illustrate the use of tran-
sitive and intransitive bases, simple, derived and compound bases, established (sometimes
idiomatized) and novel uses, and varying hyphenation.

(100) 
reabsorb, reaccess, reaccumulate, readapt, readdress, readjust, readmit, reaffirm,
re‑alphabetize, reapproach, reargue, rearm, re‑ask, reassemble, re‑attack, rebabyproof,
re‑bid, rebloom, reboot, rebottle, re‑button, recalculate, recapture, recategorize,
recertify, re‑choose, reclaim, recode, recolonize, reconfigure, recycle, redeciphering,
redevelop, redig, rediscover, redo, re‑download, re‑endear, re‑enlist, re‑erupt, re‑evolve,
re‑exit, re‑fall, refilm, refloat, refocus, reformulate, refuel, regraft, re‑grout, rehaunt,
rehistoricize, reimplant, reimprisoning, reinforce, re‑kill, relandscape, re‑listen, relook,
remortgage, repopulate, rerefine

As shown in Lieber (2004), arguing against Smith (1997), the suffix generally attaches to
those verbs that imply some sort of non-permanent result. Verbs denoting events with no
inherent result are impossible bases (e.g. *resleep, *repush), and so are verbs which imply a
result which cannot be reversed. For example, it would be impossible to *reeat the apple or
*reexplode the bomb, unless we imagine a world where such things can actually happen (cf. the
science fiction horror movie title Re-Kill).
Note the examples recork, refit, refresh, reinstate, renew, where the re‑ may indicate the return
to an original state rather than a repetition and where the bases are not all verbs. Another
example of this reading is intransitive re-eat as attested in COCA, shown in (101), signifying
the return to a state in which the speaker ate, and not the reversal of an eating event.

ABC_GMA 2009: No kidding, eat your breakfast. That’s what I didn’t do is I stopped
(101) 
eating, and I learned to re‑eat. So, that’s a very important thing.

What we might see here at work is the same kind of ambiguity we find with the free form
again, which can indicate repetition as well as ‘back’, that is the return to a former position or
state (OED, sv. again). It is not clear, however, that we have to recognize the latter reading as
a meaning of the prefix re‑ in contemporary English, where iteration seems the appropriate
interpretation in productive uses.
420 Size, quantity, and attitude

The prefix has a tense /iː/ and is secondarily stressed, but some lexicalized formations may
deviate from this pattern (e.g. recreation ‘refreshment’ versus re-creation ‘action of creating
again’). Hyphenation is usually used to distinguish between pairs which would otherwise be
homographs where one has and the other does not have the prefix (note also the respective
presence or absence of secondary stress on the first syllable). Examples include react versus
re-act, recover versus re-cover, reform versus re-form, represent versus re-present, review versus
re-view.

18.4.3 The suffix ‑fold

The suffix ‑fold is added to numbers and other semantically suitable quantifiers to create
adverbs meaning ‘times the value of the base’, so five-fold means ‘five times’. It is surprisingly
productive, as the short list of examples from the many in COCA provided in (102) indicates.
Note the variable hyphenation and variable writing of the number in the base.

(102) 1,700‑fold, 14‑fold, 162‑fold, 2.5‑fold, 37‑fold, 5‑fold, 800‑fold, billionfold, five-and-
a-half‑fold, manifold, many‑fold, one‑fold, sixty‑sevenfold, trillionfold, twelvefold,
zillion‑fold

18.4.4 The suffixes ‑some

There are two distinct suffixes of the form ‑some used with bases which are numbers, and in
usage they appear to be in complementary distribution.
The first of these is the nominalizing suffix used in twosome, threesome or with any other
number under ten, and also found in ownsome. It denotes a group (usually of people, but
occasionally of other entities) of a size denoted by the number in the base. Some textual
examples from COCA are provided in (103).

Harpers_Bazaar 2007: Going away can be an inflammatory enterprise for any twosome
(103) 
or moresome.
Golf_Magazine 2004: Balls must be at least 1.62 inches around and no heavier than 1.62

ounces. The reason for the 1.62 twosome, if any, is lost to history.
Parenting 2008: a toddler twosome eating ice cream.

The other suffix of form ‑some to be dealt with here is the one found, for example, in 20‑some,
which means the same as in the syntactic construction some twenty, that is ‘approximately’
(cf. ‑ish, Section 18.4.6). The majority of the bases are round numbers—numbers of twenty
or greater which end in zero. COCA does provide occasional exceptions, though, as shown
in (104).
 18.4 Quantification and measure 421

(104) Bicycling 1994: You’re talking 12‑some hours in the saddle.

NBC_Dateline 2008: the one act in combat that has been a burden for me for 33‑some

years.

The suffix ‑some is in obvious competition with its syntactic equivalent or so, which has a
wider distribution. For the ‑some which creates adjectives, see Chapter 14.

18.4.5 The suffix ‑ton

The nominalizing suffix ‑ton is established only in the word singleton ‘one that is isolated,
unique or alone’, but is occasionally found in some analogical forms, as illustrated in (105).
The OED suggests that this is the same suffix as in simpleton, but the meaning is so different
as to raise queries about this in the current state of the language. There is no evidence of
productivity for this suffix beyond what is listed in (105).

Journal_of_Sports_Behavior 2010: Beginning with a five-factor solution, a process of

(105) 
eliminating singleton and doubleton items was conducted.
New York Times 2007, cited in Wiktionary: It means that you lead low from a doubleton

and the middle card from a tripleton.

18.4.6 The suffix ‑ish

The suffix ‑ish creates adjectives of similarity or approximation, and occasionally adjectives of
nationality, as shown in Chapter 14, but is listed here because of its approximative use with
number bases, as in the examples in (106), all attested in COCA or the BNC.

(106) 10‑ish, 25‑ish, 500‑ish, fortyish, twelve‑ish, twentyish, two‑ish

18.4.7 The prefix hypo‑

The prefix hypo‑ is the counterpart for hyper‑ (see Section 18.3.4) in some of its uses. When
used in a locative sense, it can mean ‘under’ (sometimes a metaphorical locative, as with
‘lower’ orders of animals), in which case it may be in competition with sub‑ (see Chapter 16).
But it also has a quantitative sense in which it means ‘less than is normal or desirable’. It is
usually used with this quantitative sense in medical and scientific contexts, attached to a
noun or an adjective (107), or in a neoclassical formation (108).

(107) hypo‑allergenic, hypo‑production, hypo‑reactivity, hypo‑sensitization, hypo‑skeptical,

hypo‑ventilation, hypoactivity, hypocretin, hypocritical, hypodense, hypoechoic,
422 Size, quantity, and attitude

hypoelastic, hypofunction, hypoglycaemic, hypogonadism, hypointensity,

hypopharynx, hypophthalmus, hypopituitarism, hyporesponsive, hyposensitivity,
hypotension, hypotext, hypothermia, hypotragedy, hypotympanic, hypoventilation

(108) hypobaric, hypochloric, hypochondria, hypocrisy, hypoderm, hypodermic, hypogamous,

hypogram, hypokinesis, hypokinetic, hyponymy, hypoplasia, hypopnea, hyporheic,
hyposmia, hypostasis, hypotaxis, hypothalamus, hypothalmus, hypothesis, hypotrophy,
hypoxaemia, hypoxia

Unlike its congener, hyper‑, hypo‑ is rarely used in a colloquial context. Even hypo-skepticism
and hypotext, cited in (107), are technical terms of psychology and literary criticism
respectively.
The prefix hypo‑ carries secondary stress, unless it is followed by a neoclassical form that
attracts stress to the immediately preceding syllable (see Chapter 9.4.5 for discussion). In this
case, the second syllable of the prefix is main-stressed (hypónymy, hypóstatis, hypótenuse,
hypóthecate, hypóthesis). In one set of forms the first syllable of the prefix is reduced to a short
vowel, [ˈhɪ]pocrite, h[ɪ]pócrisy and h[ɪ]pocrítical.

18.5 Cardinal and ordinal numbers

The basic numbers of English are one, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve,
hundred, thousand, million and, since the 1940s, googol, and googolplex. Twelve can be considered
a suppletive form for an unattested *twoteen, and the tw‑ initial cluster is also found in two (in
the orthography only), twin, twain, and between, and possibly in twine. The orthographic <tw>
no longer has morphemic value in these words. It is not clear that there is a number word cor-
responding to the notation ‘0’, since nil, null, nought/naught, oh /əʊ ‖ oʊ/, zero, and, for some
speakers, nothing, divide this area between them in a not entirely regular way.

18.5.1 The suffixes ‑teen and ‑ty

The suffix ‑teen, meaning ‘plus ten’ can be added to the numbers from three to nine, and the
suffix ‑ty, meaning ‘times ten’ can be added to the numbers from two to nine as shown in
(109). The suffix ‑teen is auto-stressed, ‑ty is unstressed and base stress is preserved.

(109) base ‘plus ten’ ‘times ten’

two twelve twenty
three thirteen thirty
four fourteen forty
five fifteen fifty
six sixteen sixty
 18.5 Cardinal and ordinal numbers 423

seven seventeen seventy

eight eighteen eighty
nine nineteen ninety

Phonological alternations can be observed in four of the nine series, that is with derivatives
based on two, three, five, and eight. We find suppletion (two ◆ twelve ◆ twenty, three ◆ thirteen ◆
thirty), vowel alternation and consonant devoicing (five ◆ fifteen ◆ fifty), and degemination, or
loss of final /t/, depending on the analysis (eight ◆ eighteen ◆ eighty). The orthography reflects
these alternations but also introduces a further irregularity, namely a different spelling of the
base four in forty. The orthographic distinction between <our> and <or> in four and forty is
not reflected in the pronunciation in most varieties of English, but a phonological distinction
is made in ScE. Some Americans have a different vowel in fourteen from that in four and forty
(Wells 2008).
Because of the phonological similarity of ‑teen and ‑ty, the pronunciations are often exag-
gerated, and in speech the figures are often given alongside the figure: It cost $40, four oh.
There also seems to be an unspecified one-digit number ump‑, to which ‑ty can attach, as
in She’s just had her umpty umpth birthday, expressing the meaning ‘an unspecified number’.
The unspecified number ump‑ may also combine with ‑teen to form umpteen ‘many’, and an
ordinal number can be formed on the basis of umpty-ump and umpteen, as in umpty-umpth
and umpteenth.
There is also a free form teen, which can occur as a plural form in expressions like some‑
where in their teens ‘aged between 13 and 19’ and as a metonymic singular form teen ‘a teenager,
a person in their teens’. The formative ‑ty is also occasionally (and usually jocularly) added to
eleven (see, for example, Tolkien 1954: 43), is used on naught to give the punning naughties
(‘the years between 2000 and 2009’).

18.5.2 More complex numbers: beyond 20

More complex numbers are derived by the combination of the simplex and complex num-
bers described in the preceding section. The numbers between 20 and 99 are created by tak-
ing the number in ‑ty and then adding the relevant unit number: twenty-seven, thirty-two,
ninety-one, and so on. The hyphen is generally used, and the stress is on the second element.
These numbers might be argued to behave semantically like additive coordinative com-
pounds such as Alsace-Lorraine in that their meaning is the sum of the parts, and they are
exocentric, not being a hyponym of either of the parts.
Three digit numbers consist of one of the numbers from 1 to 9 followed variably by hundred
or hundred and (with and being often phonetically reduced to /ən/ or /n/), followed by one
of the numbers from 1 to 99: one hundred (and) thirty-nine (i.e. 139), six hundred (and) two (i.e.
602). One before hundred may be replaced by a, as in a hundred (and) sixty-three (i.e. 163).
424 Size, quantity, and attitude

An alternative way of reading or saying three-digit numbers is by having one of the num-
bers from 1 to 9 followed by one of the numbers from 10 to 99: two fifty-three (i.e. 253), eight
ninety-eight (i.e. 898). If the number following the first digit is below ten, the second digit is
pronounced as /əʊ ‖ oʊ/, as in two oh seven (i.e. 207).
Numbers for thousands and millions are added recursively before the numbers as
explained so far, except that the alternative variant just discussed is no longer possible
(10,898: *ten thousand eight ninety-eight versus ten thousand eight hundred (and) ninety-
eight). Typically there are intonational breaks possible after the word million and after
thousand when a number of hundreds follows. This is reflected by commas in the mathe-
matical notation of the numbers. Thus we get numbers like two thousand nine hundred
(and) thirty-three (i.e. 2,933), six hundred (and) twenty-nine million, nine thousand and five
(i.e. 629,009,005), seven million, nine hundred (and) twenty-one thousand, three hundred and
seventy-nine (i.e. 7,921,379). Notably, million and thousand must be followed by and (full
form or reduced form) if the overall number has more than four digits and there are only
zeros between the thousand, or million, and the last digit, see (110a). If the overall number
has only four digits and there are only zeros between the first and the last digit, the use of
and is variable (110b).

(110) a. 235,000,002:
two hundred (and) thirty-five million and two
*two hundred (and) thirty-five million two
345,006:
three hundred (and) forty-five thousand and six
*three hundred (and) forty-five thousand six
b. 8005:
eight thousand and five
eight thousand five

Although numbers like sixty-five may look like compounds, larger numbers look far more
syntactic in structure. Not only may they have the syntactic coordinator and in them, but
with numbers with millions and thousands, the intonation also suggests that they are phrasal
rather than lexical, especially if the number is said more slowly. Phonologically and semanti-
cally, therefore, a number like 7,302 may be parallel to a syntactic phrase like Peter, Paul (and)
Mary. The rather strong co-occurrence restrictions show, however, that we are dealing with a
tight-knit lexical system.
Although we have outlined a system for number names above, there is an alternative sys-
tem in operation for numbers between 1,100 and 9,999. This alternative system is not based
on a segmentation with three digits, but uses two digits as a boundary. This alternative system
 18.5 Cardinal and ordinal numbers 425

is more usual with the lower numbers in this scale. According to this system, the first two
digits are pronounced as numbers between 11 and 99 followed by hundred, followed by the
number given by the last two digits. Thus, instead of one thousand one hundred (and) twelve
the number 1,112 may be spoken as eleven hundred (and) twelve. This option is available even
when there are numbers in the tens and units: fifteen hundred (and) forty-nine. While ninety-
nine hundred (and) ninety-nine would be odd, such figures occur with lower numbers, such as
twenty-five hundred (and) six. This alternative system also has a variant without the word
hundred, used from 1,000 onwards, so that 1,549 could be fifteen forty-nine; this formulation is
frequently used for dates. Without the word hundred, 9,999 could well be ninety-nine ninety-
nine, and would then be homophonous with a price of $99.99 or £99.99.
In principle words containing billion, trillion, quadrillion, etc. are built up in the same way as
the numbers including millions; in practice, such numbers are rarely spelled out in full: they
are used in whole numbers (five trillion) or with decimal points (five point seven trillion).
Morphologically, however, these higher numbers are hard to deal with, since there are two
distinct systems in use. These are sometimes referred to as ‘British’ and ‘American’ systems,
but there is no such simple division. One system includes forms in ‑lliard, the other does not,
as shown in (110). Note that the use of these numbers is common only in specialist circles.

(111) 1024 septillion quadrillion

1021 sextillion trilliard
1018 quintillion trillion
1015 quadrillion billiard
1012 trillion billion
109 billion milliard

The system in the righthand column appears to be yielding to that in the lefthand column.
The formative ‑illion is also added to nonsense onsets or syllables to give bazillion, bajillion,
frillion, jillion, gajillion, gazillion, kajillion, kazillion, squillion, zillion, each of which simply
means ‘an indeterminately large number’.

18.5.3 Ordinals and fractions

The ordinal numbers are created from the cardinal numbers with the addition of the suffix ‑th,
except that *one + th has the suppletive form first, *two + th has the suppletive form second, and
*three + th has the suppletive form third. Following numerals in ‑ty, there is an extender /ə/
<e> before the ‑th: thirtieth, ninetieth and a spelling alternation of <y> ~ <i>. Five and eight
show the same allomorphy before ‑th as they do before ‑teen and ‑ty: fifth, eighth.
Fractions are formed using the same process, except that the word half is used instead of
the expected second: ½ is a half (or one half) and not *a second, and quarter is often used in
426 Size, quantity, and attitude

place of fourth. When fractions are made plural, the sequence of /θs/ is often simplified to
[sː] or [s], so that 5/6 may be pronounced [faɪv sɪksː], scarcely distinct from five six. On the
other hand, for some speakers, an epenthetic /t/ may be inserted before the /θ/ in particu-
larly clear speech, to give, for instance, /fɪftθs/, /sɪkstθs/, /naɪntθs/.

18.5.4 Number prefixes

A series of prefixes is used for fractions; these are presented in (112). The earliest such prefixes
come from French, Latin, and Greek, but invented elements based on other languages have
been added as smaller fractions have become required for scientific study.

(112) number prefix example

1/2 demi‑ demi‑cup
hemi hemisphere
semi‑ semicircle
1/10 deci‑ deciliter
1/100 centi‑ centigram
1/1000 milli‑ milliliter
1/106 micro‑ microsecond
1/109 nano‑ nanotube
1/1012 pico‑ piconewton
1/1015 femto‑ femtometer
1/1018 atto‑ attosecond

Demi‑, from French, can be used literally to mean ‘a half ’, but usually means ‘of lower quality’
as in demigod. Semi‑ and hemi‑ are etymologically Latin and Greek respectively, and this
sometimes affects their distribution, though note the BrE hemidemisemiquaver (‘1/32nd
note’), where the ordering seems random, and there is nothing in the environment to help
predict the distribution.
It is always possible to use the English numbers in combination with other elements to
create forms like six-ball, six-bedroom, six-sided, six-strong, six-wheeler. However, English has
specific prefixes from Latin and Greek to make single words that involve specific numbers
and quantifiers. Many of these are borrowed but analysable, and they tend to be stylistically
more formal than the English constructions.

(113) number/ quantifier Latin Greek

1 unicycle monologue
2 bicycle diphthong
3 tricycle tripod
 18.5 Cardinal and ordinal numbers 427

4 quadrangle, quadraphony, tetrahedron

quadrimanual, quadruped
5 quincentenary, quinquereme pentagram
6 sexfoil hexagon
7 septennial, septivalent heptathlon
8 octosyllable octahedron
9 nonagon ennead
10 decennial decalogue
20 vigesimal
100 centipede hectogram
1000 millipede kilogram
Many multi-national polyvalent
All omnivore pan-American
Both ambisyllabic

Some of these forms are subject to allomorphy whose analysis is unclear, possibly involving
extenders in the contemporary language.
For some larger multiples of 10 there are prefixes, as illustrated in (114), all carrying main
stress. These are used largely in scientific discourse, often within the fields of astronomy,
computer science, and economics.

(114) 106 megabyte

109 gigaflop
1012 terawatt
1015 petabyte
1018 exajoule
1021 zettabyte
1024 yottabyte
 Pa rt I V

Compounding
 chapter 19

Compounds: formal
considerations

19.1 Prospectus
This chapter and the following one are concerned with compounds. The present chapter will
deal with the formal properties of these formations while Chapter 20 discusses their seman-
tics. We will start out in Section 19.2 by clarifying one of the most controversial issues in the
study of compounds and their formation, that is the question of which kinds of constructs
should be classified as compounds and which ones should not. This is followed by a discus-
sion of the internal structure of compounds in Section 19.3, followed by an overview of the
formal properties of different types of compound (Section 19.4).

19.2 Boundary issues: What is a compound?

Derivation is conceived of in this book as the formation of words by combining affixes with
bases, and also the operation of some non-combinatorial processes on a base, as in conver-
sion or clipping. In contrast to this, compounding can be regarded as the formation of words
through the combination of bases (see also Bauer and Huddleston 2002: 1626). Since bases
are also concatenated to form syntactic phrases, the issue arises of how compounds can be
distinguished from syntactic phrases, and indeed the literature is full of discussions of this
question. Another issue is to determine whether a given base is really a base or perhaps better
analysed as an affix. A third issue concerns the problem that not every word that contains two
(or more) bases, and thus would fall under this rough definition of compound, arises as the
result of the same process of base-combination by which compounds are normally formed.
These three issues will be discussed in Sections 19.2.1 to 19.2.3.

19.2.1 Compounds: morphological or syntactic?

Discussions of the boundary between compounds and syntactic phrases have focused mainly
on noun–noun (NN) combinations, and especially on those NN combinations that are
432 Compounds: formal considerations

spelled as two separate orthographic words. In contrast, NN combinations spelled as one

word or hyphenated are generally taken to be words, that is, compounds, not phrases.
Examples of all three kinds are given in (1). Following the terminology introduced in Chapter
4.5.2, we will refer to words like those in (1b) as being written solid, those in (1c) as being
hyphenated, and we will say that those in (1a) are spaced. Compounds written solid or
hyphenated may be grouped as being ‘concatenated’. The examples are taken from the Boston
University Radio Speech Corpus (BURSC, Ostendorf et al. 1996), the transcriptions of
which were provided by native-speaking research assistants, and did not undergo a central-
ized editorial process.

(1) a. birth control, hospital staff, housing subsidies, state official, wind erosion
b. cornerstone, jailbird, lawmaker, workday, wristband
c. hoop-junkie, life-span, tissue-type

Instead of restricting our discussion to constructs spelled in a certain way, we will focus here
on general criteria that have been used to distinguish compounds from phrases. Furthermore,
we will look not just at NN combinations, but at combinations including verbs and adjec-
tives as well.

19.2.1.1 Criteria for NN compounds

There are a number of general criteria that are standardly adduced to test the status of NN
constructs as phrases or words. One problem with these criteria is that they sometimes rest
on questionable assumptions, which makes them potentially unreliable. Furthermore, as
shown, for example, by Bauer (1998a), it is frequently the case that the outcomes of the vari-
ous tests contradict each other, which further undermines the value of these diagnostics.
Huddleston and Payne (2002: 449) claim that it is enough to fulfil only one criterion of
phrasehood to decide the issue, but do not give reasons why such an approach would be
justified.
The first test criterion to mention is an aspect of syntactic atomicity, which says that words
are uninterruptable units, hence it should not be possible to insert an affix or another word
into a compound. The examples in (2) illustrate this. Examples like those in (3) are not coun-
terexamples, because they arise from the recursiveness of the compounding process, that is,
from the attachment of council to the left of health program and not from the insertion of
health into council program.

(2) council program versus *council recent program

state deficit versus *state big deficit

(3) council health program, state budget deficit

 19.2 Boundary issues: What is a compound? 433

The second criterion holds that compounds be listed in the lexicon, while phrases are not.
This criterion is unclear and thus questionable. It is unclear since the notion of listedness is
not well defined. Psycholinguistic studies (e.g. de Vaan et al. 2007) have shown that even a
single mention of a word leaves a trace in memory and that words may be present in the lexi-
con in varying degrees of strength. A way out of this problem may be to define ‘listed’ as
‘listed with some idiosyncratic property’. Such a move would, however, make listedness a
questionable, if not utterly useless, criterion, because it would lead us to classify all non-idio-
syncratic compounds as syntactic and therefore unlisted (apparently the view of Lyons 1977:
535). This is an undesired result, for at least two reasons. First, there is good psycholinguistic
evidence that non-idiosyncratic complex words are also stored in the mental lexicon (de
Vaan et al. 2007). Second, under this approach any regular complex word would have to be
regarded as a syntactic entity instead of a word.
A third criterion tests whether the second element can be replaced by anaphoric one,
which is taken as an indication that the construction is a phrase. The criterion itself is not
reliable, however, since it rests on the assumption that sub-lexemic elements are not accessi-
ble to pronominalization, which has been shown to be false (e.g. by Ward et al. 1991; Lieber
1992: 122–4). Instead, the possibility of anaphoric reference is restricted by pragmatic rather
than structural principles. The problem with this criterion is illustrated in (4), where all the
examples show that words are not anaphoric islands (all the extracts in (4) are attested, the
first from COCA, the second from a web-site, the others in conversation).

(4) a. ABC_Nightline 1991: as both a North supporter and a Reagan supporter, I still support
both without reaching the conclusions that Ollie reaches in his book [. . .]
b. . . . cash withdrawal fees generally average two per cent for credit cards and 1.5 per
cent for debit ones. (http://www.newsletter.co.uk/news/business/pack_the_right_
plastic_1_1839384 9 June 2011 cited by Bell 2011)
c. I’m trying to give myself a headache by banging it on the floor.
d. He opened the garage door and drove straight into it [viz. the garage]

A fourth criterion is modification. Payne and Huddleston (2002: 449) argue that the pos-
sibility of expanding NN constructs into [N[A N]] or [[A N]N] constructions can be
taken as evidence for their being phrases. According to Bell (2011) this argument is not
compelling. Based on the idea that a nominal head in syntax should be able to project full
phrases, the potential phrase headed by the left noun in the structure [[A N]N] would be
expected to allow complements, specifiers, and modifiers as dependents. Bell shows, how-
ever, that the noun is unusually restricted in its capacity to carry these, as is the adjective.
This would make this premodifying noun phrase very atypical, and the force of the crite-
rion dubious. She cites examples such as red tulip collection, which does not allow *very
bright red tulip collection.
434 Compounds: formal considerations

The fifth test to mention here is coordination, which assumes that sub-lexemic units may
not be syntactically coordinated, for example with and. This test is not very useful either,
because sub-lexemic units can easily be coordinated syntactically, provided that certain pro-
sodic (and semantic) requirements are met (see Plag 2003: 84; Bell 2011; unless otherwise
noted the examples in (5a, b) are from COCA).

(5) a. Education 2010: pre- and post-test methodology

 Ms. 2008: No alumnus, whether pro- or anti-feminist, should be able to determine
a college curriculum
 Mechanical Engineering 2001: the two or three large engines are replaced by many
mini or micro engines
b. child- and home-less (Plag 2003)
business and family-wise
c. speech-perception and production
word and sentence structure
sugar- and fat-free

The sixth criterion is stress, with the traditional distinction of right stress for phrases and left
stress for compounds. As will be discussed in detail in Section 19.3.3, this criterion only works
in one direction: left-stressed NNs are uncontroversially considered to be compounds, but
right-stressed ones are not necessarily phrases.
Spelling is the final criterion mentioned in the literature, but the observable variability in
spelling makes this criterion difficult to apply. If anything, compounds that frequently occur
with concatenative spellings are uncontroversially regarded as compounds.
In sum, there seems to be no established set of trustworthy procedures that could tell us
reliably and theory-neutrally for a given NN construction whether it is a noun or a phrase (see
also the survey in Lieber and Štekauer 2009). We will therefore be inclusive rather than exclu-
sive and consider NN constructs as compounds, unless there is clear evidence to the contrary.
However, even if we are as inclusive as possible, the question of determining what are
compounds and what other types of complex nominal remain, and we now turn to look at a
number of potential candidates for compound status.
For example, we do not consider names consisting of two proper nouns (e.g. Alice Miller)
or appositive NNs (such as Governor Dukakis) as compounds. We include in this set NN
constructions which are left-headed; the letter A, the number nine, a Model T, the year 2000, the
play ‘Macbeth’.

19.2.1.2 AN combinations
We now consider a number of types of AN construction. The first is made up of basic, usually
etymologically Germanic, adjectives which are usually written solid with the noun and
 19.2 Boundary issues: What is a compound? 435

stressed on the leftmost element. Examples are blackboard, greyhound, highlight, hotplate,
longboat, lowland, madman, shorthand, short wave, sweetcorn, whitehead. Typically the adjec-
tives are monosyllabic, but there is the occasional disyllabic adjective in this class, as in nar-
row-boat. Some of these words may be exocentric like redcap, redskin, whitehead, others are
endocentric. There is virtually no dispute in the literature that these are compounds (but see
Spencer 2003). Typically the adjective creates classes of objects denoted by the noun, and
cannot be sub-modified by very, rather, so, etc.: a very hot plate is not a hotplate, but a plate
that is very hot.
Next we have two classes made up of AN where the adjective is a relational adjective.
Relational adjectives are usually derived from nouns, cannot be graded or intensified, cannot
occur in predicative position and express the meaning ‘in relation to X’ with ‘X’ being the
denotation of the base: dental surgery, electrical engineer, financial advisor, polar climate, postal
service, vernal equinox. The first group of constructions with these adjectives are rather like
the AN compounds discussed above, except that they are not written solid with the noun.
Importantly, like the compounds discussed above, they take left stress. Examples in this
group include cúltural club, dramátic society, %flóral arrangement, prímary school, romántic
period, sócial worker, sólar system. The reason for the occurrence of left stress in these con-
structions is not entirely clear: to some extent, it is probably a function of the frequency of
the combinations, but the fact that we find sets like archeológical society, chóral society, dram-
átic society, horticúltural society, operátic society, zoológical society all with left stress suggest that
there is some influence of the paradigm, despite royal socíety and national socíety with right
stress (see Section 19.3.3 for a detailed discussion of stress). These constructions are not much
discussed in the literature, but would probably be generally accepted as compounds.
The second type with relational adjectives looks similar, but does not have left stress.
These are expressions such as atomic bomb, Chomskyan analyses, fungal spore, human genome,
linguistic behavior, moral philosophy, phocine distemper, prosodic analysis, and so on. Levi (1978)
classifies these constructions as ‘complex nominals’, along with constructions that would
generally be regarded as compounds. Some subsequent scholars (see Giegerich 2006; Bell
2011 for discussion) have classified them as compounds. The question remains controversial,
and may amount to no more than a matter of definition; because these constructions are not
clearly morphological, we do not treat them in any more detail here.
Next we have a set of constructions with usually monosyllabic and usually Germanic
adjectives that behave just like the first set except that they do not take left stress. Examples
are black íce, brown ále, green téa, high cóurt, light ráilway, red gíant, red squírrel, thin áir, white
wítch. The difference between these and the compound set is largely one of frequency (Bauer
2004a). Nevertheless, we know of no authorities who include these as compounds, and we
will treat them as lexicalized noun phrases. We note, however, that their semantics seems
indistinguishable from that of compound AN constructions, and that the adjective seems
similarly isolated, in that we cannot have a very red squirrel without losing the lexical meaning
of red squirrel, and that brown red squirrel is not necessarily a contradiction.
436 Compounds: formal considerations

Finally, there are free combinations of adjective and noun, typically adjectives not derived
from nouns, which can frequently be sub-modified, in which the adjective is descriptive of
the head (rather than relational), and which typically are not lexicalized, although there are
lexicalized examples: red light is lexicalized to some extent (especially when it occurs in a
larger collocation such as red-light district, but even in the sense of ‘traffic light’), but brown
trousers, green leaf, inveterate liar, short walk, sloppy kiss, tall tower, wet grass are all simple noun
phrases, and not part of word-formation at all.

19.2.1.3 Other cases

The first of the special constructions we need to consider involves the combination of two
words of the same category being used as modifiers of a nominal head, such as father–daugh-
ter (dance), mind–brain (connection), true–false (question), or push–pull (gizmo). Plag (2003:
221–2) advances three arguments for treating them as compounds, none of them being par-
ticularly forceful, however. First, insertion of adjectives seems to lead to odd results. Second,
the relation between the two constituents is normally expressed overtly by a conjunction, if
clearly syntactic means are chosen (e.g. the connection between mind and brain). Furthermore,
inflections seem to be obligatorily missing (fathers-daughters-dance). We will also treat such
structures as compounds; see Chapter 20.3.2 on coordinative compounds.
A second potentially problematic case is what has been labelled ‘descriptive genitives’, by
which we mean (following Rosenbach 2006) the whole N’sN construction. Some examples
are given in (6).

(6) Broca’s aphasia, driver’s license, fool’s errand, men’s room, men’s suit, mother’s milk,
smoker’s cough, writer’s block

Some descriptive genitives sometimes have rival NN constructions without ’s, some do not,
as the examples in (7) show. The existence of doublets is presumably partly dependent on
the degree of lexicalization.

(7) lawyer fees versus lawyer’s fees (both in COCA)

people power versus people’s power (both in COCA)
*mother milk versus mother’s milk
frog spawn versus ?frog’s spawn

Although descriptive genitives may look like straightforward phrases, their status as
phrases is arguable. Descriptive genitives differ systematically from more prototypical, so-
called ‘determiner genitives’ in that their pre-head structure is clearly a noun and not a
noun phrase, as is possible with determiner genitives (Huddleston and Pullum 2002: 354–5).
In addition, the pre-head structure in determiner genitives makes the whole possessive NP
definite, even if the possessor itself is indefinite, as in a smoker’s car, which corresponds to
 19.2 Boundary issues: What is a compound? 437

the car of a smoker (note also that a car of a smoker does not have a corresponding ’s genitive
construction).
Conversely, descriptive genitives are very similar to NN compounds in that their first ele-
ment has word status rather than phrasal status, has a classifying semantic function, and
tends to be non-referential. Furthermore, many descriptive genitives have left stress and have
lexicalized meanings. Applying the usual tests (such as coordination, modification, etc.)
leads, as with so many NN constructions, to contradictory results (see Rosenbach 2006 for
detailed discussion), and Rosenbach uses the mixed behaviour of genitives as an argument
for the view that categories are gradient.
We see here a number of constructions which are formally more or less syntactic and
semantically more or less compound-like, with some gradience even within the individual
subtypes. Arguments as to whether some of these ‘are’ or ‘are not’ compounds are likely to
become matters of definition rather than questions that can be answered by the production
of evidence. We will not treat descriptive genitives here, even though we acknowledge the
semantic similarities between the various constructions.
The next potentially problematic kinds of structure are so-called ‘phrasal compounds’.
This class is characterized by a phrase as first element of a compound. Examples are numer-
ous, some of them, as taken from COCA, are listed in (8a). They are often highly specific in
meaning and unlikely to become institutionalized, as shown in example (8b), which illus-
trates one in its original context.

(8) a. circle-a-word puzzle, soon-to-be-divorced wife, strawberries-in-July talk

b. Watchful parents need to know: Where will my precious child be safe from the
scourge of drugs? At the service academies, of course, which score four of the five
places in the “Don’t Inhale” rankings. (Boston Globe, 24 August 2010)

These structures have featured prominently in discussions of the syntax–morphology inter-

face, but it seems that they are less spectacular than assumed by many theorists. As shown in
the previous chapters and later in Chapter 22), phrases can function as bases for many deri-
vational processes, in other words phrases must be included anyway in the set of possible
bases in English word-formation. From this, and from our definition of compounding as the
combination of bases, it follows naturally that phrases can occur as compound elements, and
that phrasal compounds are compounds. It is less clear, however, why phrases may not
occupy the right position in compounds, unless it is axiomatically (and perhaps sensibly)
assumed that heads are lexical (see, for example, Zwicky 1985: 5).
Similar to phrasal compounds are those structures that contain phrase-like first elements,
given in bold in (9). The unproblematic insertion of an adjective between the hyphenated
modifier and the head noun (shown in (9a) ) suggests that the structure as a whole is not a
compound. Yet these first elements are not clearly phrases since they lack certain properties,
for example agreement marking (consider a three-syllable word). This lack of agreement
438 Compounds: formal considerations

marker in (9a) and the wider set of examples in (9b) suggest that these may all be compounds
(some of them exocentric ones) albeit non-canonical ones in some cases (see Section
19.2.3).

(9) a. a three-syllable English word, the two-year formative period, a four-term

congressman
b. a go-go dancer, a no-go area, a pass–fail test, a tow-away zone, before-tax profits

Rarely featuring in discussions of compounding, numerals (such as twenty-four, three hun-

dred (and) fifty-five) seem also to be borderline cases. They are discussed in detail in
Chapter 18.
Perhaps less controversial are combinations of adverbs with adjectives, whose spelling in
premodifying position with a hyphen may suggest compound status (e.g. ill-advised decision,
well-chosen example), but following the arguments developed in Chapter 4.5.3, we do not
consider these compounds.
Similarly, lexicalized phrases such as forget-me-not, jack-in-the-box, lady-in-waiting, matron-
of-honor, son-in-law or structures such as book-turned-movie, coach-turned-commentator (all
from COCA) are excluded from the set of compounds, though some authorities (e.g. Quirk
et al. 1985) list them as such.
Consider specifically constructions which consist of two or more words combined under
inclusion of a linking element cum or ‑cum‑, as shown in (10), with examples taken from
COCA. The construction is productive and works with nouns (10a) and (much less fre-
quently) with adjectives (10b).

(10) a. bookstore cum restaurant, café-cum-bookstore, cricketer-cum-politician, deli cum

takeout fish market, earth rock cum paperweight, intentionality cum responsibility,
soap opera cum sex farce, soccer player cum musician
b. cognitive-cum-affective, civic-cum-cultural, cosy-cum-corny

The element cum is listed as a preposition in the OED, but outside genuine Latin expres-
sions such as summa cum laude it occurs only as a conjunction-like coordinating formative
between words of the same syntactic category. The expressions are spelled either with or
without hyphens and may contain more than just two elements (e.g. wagon-cum-car-cum-
truck, from COCA). The resulting expressions refer to a single entity that combines char-
acteristics of the denotations of its constituents, which makes the constructions similar to
coordinate, more specifically appositional, compounds (see Chapter 20 for discussion).
The constructions have rightward stress, which is typical of both coordinate compounds
and phrases.
The phrasal analysis is, however, strongly supported by the placement of plural marking. In
the few attested constructions in COCA that have plural marking, the inflectional suffix
 19.2 Boundary issues: What is a compound? 439

always occurs on both constituents. In (11) we provide an exhaustive list of attestations of

pluralized constructions in COCA.

(11) a. boarding schools cum penal institutions, doctors-cum-soldiers, embassies-cum-

fortresses, guides-cum-chefs

This kind of plural marking is the same as we find with syntactically coordinated noun
phrases (e.g. boarding schools and penal institutions), and it is very rarely attested for coordina-
tive compounds (as in producers-directors, COCA). For example, COCA has 59 plural forms
of the rather high-frequency compound singer-songwriter with plural marking only on the
second noun, and no attestation of this compound with plural marking also on the first noun
(Nsingers-songwriters). Renner (2013) also finds for his collection of cum-constructions that
double plural-marking is the norm, with 19 out of 20 pluralized forms showing this pattern
(across different types).
The fact that cum-coordination is restricted to nouns and adjectives is interpreted by
Renner (2013) as evidence against the syntactic status of the construction. This interpreta-
tion is not compelling, however, as we know from other languages that coordinators may be
restricted to the coordination of certain types of constituent. For example, many languages
have different coordinating conjunctions for NPs as against clauses or verb phrases (see
Haspelmath 2007: 20–2).
Due to its rather controversial status as a morphological construction we do not treat cum-
constructions in any more detail in this volume. The reader is referred to Renner (2013) for a
recent and more detailed study of the properties of these expressions.
Some authors (Bauer 1983; Katamba and Stonham 2006: 305) take reduplicative com-
pounds such as easy-peasy, killer-diller, lovey-dovey to be compounds. Since they do not fit the
definition of being constructed of two bases, we reject this classification. We then extend this
to the few examples where the two elements do exist independently, such as silly-billy, wham-
bam. These constructions are dealt with in Chapter 18.2.6.
Finally, there are several lexical items with the head on the left that are originally borrowed
from French. These include items such as attorney general, court martial, governor general,
notary public, and so on (for further examples see Chapter 7). Some authorities (e.g. Quirk
et al. 1985: 313; Plag et al. 2008) treat these as compounds. We consider these to be pieces of
lexicalized syntax, just like the phrases such as forget-me-not discussed above, the only differ-
ence being that the original syntax here is French. The same may be true of expressions like
Chicken Kiev, Steak Diane. The only example of which we are aware that seems exceptional in
this regard is girl Friday (an analogy on man Friday, which is more clearly an instance of
apposition), which appears left-headed without any French syntax (although in COCA its
plural is always girl Fridays).
Having clarified the boundary between compounds and syntactic constructions, we will
now turn to the boundary between compounding and derivation.
440 Compounds: formal considerations

19.2.2 Compounding versus prefixation and suffixation

While for most compounds this boundary is not an issue, there is a set of complex words
where it is not so clear whether the constituent elements should be regarded as bases or
affixes. This set can be subdivided into one that is traditionally described as having affixoidal
elements and one that consists of neoclassical compounds. We will discuss each in turn.

19.2.2.1 Affixoids
The category of ‘affixoid’ or ‘semi-suffix’ has been invoked for those complex words where
the status of one of the elements as a base is felt to be ambiguous between an affix and a base
(e.g. Marchand 1969: 357). A standard example is ‑man in words like postman, chairman,
milkman.
Instances of this phenomenon are often brought about by historical processes of gram-
maticalization, accompanied by semantic bleaching, and it is uncontroversial that the devel-
opment of an affix out of a free morpheme is a gradual process. The question for the
synchronic linguist is, however, whether a better understanding of these phenomena can be
achieved by positing an intermediate category. It seems that this is questionable.
Again, we would need sensible criteria to classify formatives as either base or affix. Two
kinds of properties come to mind: morpho-phonological and semantic. The more integrated
morpho-phonologically a formative is, the more affix-like it is. We have to be careful, how-
ever, with this criterion. While it is true that phonologically highly fused elements tend to be
affixes, the reverse is not necessarily the case. Even among affixes, phonological integration
can vary a great deal (see Chapter 9 for discussion). To mention only one complication, most
vowel-initial suffixes in English are integrated into the host’s prosodic structure, most conso-
nant-initial suffixes are not, and many prefixes (e.g. pre‑, post‑, anti‑, de‑, dis‑, mis‑) behave
prosodically like independent words. The degree of phonological integration and morpho-
logical status thus do not always match and the criterion of morpho-phonological integration
does not tell us whether we are dealing with a compound element or an affix.
Another issue is the phonological weakening of formatives, as in postman, with a schwa
realization in ‑man. This indeed questions the degree of morphological complexity of this
particular form to some extent. But given that the reduction of the vowel seems to be an
epiphenomenon of lexicalization (appearing on very frequent forms like postman but not on
stunt man) linked with phonological effects (two unstressed syllables are not permitted
before an unstressed vowel in man, so that cameraman has /æ/), it is unlikely that we need to
posit two lexical entries, with one of them being an affix. This brings us to other criteria.
Semantic opacity and a high degree of lexicalization have been adduced in the literature as
typical properties distinguishing derivational morphology from inflectional morphology
(Plag 2003: 15–16, for example), with compounding being allegedly more transparent.
However, compounds can be non-transparent and productive derivation is by definition
transparent.
 19.2 Boundary issues: What is a compound? 441

The most important criterion in distinguishing a compound element from a suffix seems,
therefore, its relatedness to a free form. If the constituent in question occurs with the same
meaning as a free form, no additional affix should be assumed. If, however, the bound form
consistently differs in meaning from the free form, one should assume the existence of an
affix. If we apply this reasoning to formations with ‑man, there is clear evidence that ‑man and
man are semantically identical, as is clear from low-frequency formations such as balloonman,
barnman, gloveman, holyman, taxi-man (all from COCA or BNC). A second relevant case is
that of over‑ or under‑ which we have in earlier work individually analysed as compound ele-
ments (Plag 2003: 222–3) or as prefixes (Lieber 2004: 126–33) (see also Chapter 16), but
which by these criteria are prefixes. Finally, we mention ‑like which Marchand (1969: 356)
analyses as a ‘semi-suffix’ and Adams (2001: 36) as a suffix, but which we take to be a com-
pound element (see Chapter 14).
In any case, there seems no need for an in-between category of affixoid and we will not
make use of this term.

19.2.2.2 Neoclassical compounds

Another group of complex words that are made of formatives of potentially unclear status as
base or affix are so-called ‘neoclassical’ formations. These are generally characterized as forms
in which lexemes of Latin or Greek origin are combined to form new combinations that may
or may not be attested in the original languages; the classical elements may sometimes also
be used in English as free forms. Example (12) lists a selection of forms featuring the neoclas-
sical elements bio‑, photo‑, and ‑logy.

(12) biodegradable photometric geology

biorhythm photolyse narratology
biofuel photolabile eurology
bionomic photovoltaic analogy

It is not obvious whether these neoclassical elements should be regarded as affixes or as

bound roots. If such data are interpreted as evidence for the prefix status of bio‑ and for the
suffix status of ‑logy, we are forced to assume that biology would consist of a prefix and a suf-
fix, which in turn would go against our basic assumptions that complex words need to have
at least one base. The obvious alternative is that we are not dealing with affixes, but with
bound bases, hence with compounds. This would also nicely account for the interpretation
of these forms. As in other kinds of non-coordinative compounds, the first element modifies
the meaning of the head by specifying its type. Thus, biofuel refers to a kind of fuel, biorhythm
is a kind of rhythm, etc. The same argument would hold for photo‑ ‘light’ or geo‑ ‘earth’. The
only difference between the neoclassical forms and native compounds is that many of these
non-native elements are obligatorily bound. This is also the reason why the neoclassical ele-
ments are often called ‘combining forms’. We can thus state that neoclassical formations are
442 Compounds: formal considerations

best treated as compounds, and not as cases of affixation. For a detailed treatment of the
formal properties of neoclassical compounds, see Section 19.4.5.

19.2.3 Compounding and compounds: process versus product

There are two ways of determining what should be considered a ‘compound’. We can view a
compound as a synchronic product, anything that fits the definition of consisting of two
bases. If we take this view, we ignore the fact that such forms may have arisen historically by
any number of morphological processes (including, among others, back-formation and con-
version). The second way is to view compounds as entities which are formed by a process of
compounding (i.e. a process of putting two bases together to form a larger word, with certain
predictable qualities). In this section we will have a closer look at the consequences of these
two competing views on compounds. It is clear from the two definitions that the process-
oriented approach is more restrictive. In this section we will examine those cases where the
resulting structures were apparently not formed by a speaker combining two bases. We will
use the term canonical compound for compounds formed by productively combining
two or more bases, and non-canonical compound for forms that came about through
other processes. As will become clear, non-canonical compounds may differ from canonical
compounds not only in terms of their derivational history, but also in terms of some of their
properties, for example headedness, or internal structure (as discussed in the next
subsection).
One mechanism that can result in a non-canonical compound is univerbation, that is the
merging of two (or more) words due to their frequent adjacent co-occurrence in discourse.
Prepositions and other closed class items such as into, onto, throughout, whereafter, therefore,
notwithstanding, hereby are a case in point.
Another group of non-canonical compounds is best analysed as being historically derived
through a process of inversion (see Berg 1998 for details). This group involves forms with a
preposition in first position and a verb or adjective in second position, as in download, out-
source, input and inbuilt, upcoming, outgoing.
Third, there are forms derived by conversion into nouns from phrasal verbs, accompanied
by a stress shift (as in, for example, break dówn ◆ bréakdown, push úp ◆ púsh-up, rip óff ◆ ríp-off).
Sometimes, the suffixation of ‑er to phrasal verbs may also lead to non-canonical compounds,
as in hanger-on, passer-by (see Chapter 11 for details).
Fourth, there is a large group of non-canonical compounds that are created through back-
formation (see Chapter 23 for detailed discussion). This process is especially prevalent with
compound verbs that are back-derived from synthetic compounds, as evidenced in babysit ◆
babysitter, chainsmoke ◆ chain-smoker, brainwash ◆ brainwashing.
Finally, consider the rather transient formation pattern of the form verb plus ‑in, as in sit-in,
love-in, sleep-in, teach-in. These non-canonical compounds systematically violate the right-
headedness that is characteristic of canonical compounds, as discussed in the next section.
 19.3 Internal structure 443

19.3 Internal structure

In this section we will look in more detail into certain properties of canonical compounds,
that is headedness, constituency, stress, and spelling.

19.3.1 Headedness
It can be observed that canonical compounds have the peculiar property that, in general, the
righthand member determines the grammatical properties of the whole compound, and is
therefore called the head of the compound. For example, if the righthand member is a count
noun, the compound will also be a count noun, irrespective of the grammatical properties of
the left member (consider blackboard, rice cooker, milk bottle). This generalization is known as
the righthand head rule (Williams 1981). The head also carries the inflection, as in pluralized
compounds (blackboards, milk bottles). Despite some claims in the literature (see, for exam-
ple, Pinker 1999: 178–87) that NN constructions with a plural in the lefthand element are not
canonical compounds, we treat such forms as canonical. Crucially, and in accordance with
the righthand head rule, these compounds do not receive a plural interpretation, unless the
head is also plural, as can be seen in examples like clothes basket, admissions officer, parks com-
missioner, assists record, mammals CD-ROM, spoons player versus whales deaths (COCA).
The concept of headedness is quite problematic in its application to coordinative com-
pounds. Knowing the head of a non-coordinative compound allows us to predict certain
properties of that compound. For example, if the head is a count noun, the compound will be
a count noun, if the head is a gradable adjective, the compound will be a gradable adjective.
With coordinatives, matters are different. As the constituents of a coordinative compound are
always very similar in terms of their morphosyntactic (and semantic) properties, either con-
stituent might be considered the source of the compound’s properties. The fact that the inflec-
tion is attached to the rightmost constituent might be taken as an argument for this constituent
being the head, but this argument is theory-dependent. Under a theory in which the phono-
logical host (in this case the rightmost constituent) need not be identical to the morphosyn-
tactic host (in this case the compound), it loses much of its force. In view of this situation
headedness seems not to be a useful concept in the analysis of coordinative compounds.

19.3.2 Constituency
In the above definition we used the formulation ‘two or more bases’. In fact, we find many
compounds that consist of more than two constituents. However, all canonical compounds
(except coordinatives, to be dealt with shortly) can be analysed as having an essentially
binary structure. In the case of more than two bases involved, one can find binary structures
embedded in binary structures, as shown in the examples in (13).
444 Compounds: formal considerations

(13) [Alabama [ [health department] official] ]

[ [Magnolia Terrace] [Guest House] ]

The interpretation of these compounds follows straightforwardly from the assumption of

binarity and the righthand head rule. An [Alabama [ [health department] official] ] is an offi-
cial of a health department and is located in, or employed by, the state of Alabama. Note that
the meaning of such long compounds may be ambiguous as a result of different possible
bracketings. Hence, Alabama health department official could also be analysed as denoting an
official employed by the Alabama health department, in which case the constituency would
differ slightly from that shown in (13).
Coordinative compounds are not restricted to binary structures (see further Chapter 20).
Examples like those in (14) (from COCA, see also Olsen 2001) illustrate structures with
more than two linguistically equal elements (in terms of history, it may be possible to deter-
mine what the internal structure of a name like AOL-Time-Warner-EMI is, but that is not
reflected in the linguistic structure).

(14) Bk:LiarsGame 2001: She took my digits, gave me her red-white-blue business card.
News_CSMonitor 2000: Chuck D has a message for the AOL-Time-Warner-EMI

conglomerate.
Mag_Time 2007: Knocked
 Up certified leading-man status for this
writer-producer-actor.
Acad_Education 1992: The study attempted to identify the personal-social-career

development needs of the school.

19.3.3 Stress
Compound stress is an extremely intricate phenomenon. First, we have to distinguish
between compounds that have only two bases and those that have more than two bases. The
standard view of compound stress with two-member compounds is that they are stressed on
the left constituent (e.g. Bloomfield 1935: 180, 228; Chomsky and Halle 1968: 1–4). However,
it has also been observed that many compounds do not seem to observe that rule. In order to
solve this problem some researchers, like Bloomfield, claim that right-stressed compounds
cannot be compounds, but this kind of reasoning has been proven to be circular and
unconvincing.
Another complication arises from the acoustic manifestation of the two stress patterns. As
shown in Farnetani et al. (1988), Kunter and Plag (2007), Nguyên et al. (2008), and Kunter
(2011), right-stressed compounds are characterized by more or less level pitch, while left-
stressed compounds have a clear drop in pitch from the first member to the second. It is these
acoustic properties that have apparently led some researchers to posit three stress levels: left
 19.3 Internal structure 445

stress, right stress, and an additional type labelled ‘level stress’ or ‘double stress’ (e.g.
Marchand, 1969: 22; Faiss, 1981: 132). However, the phonetic studies mentioned above sug-
gest that so-called level stress is simply a variant of right stress, not a separate category at all.
There is also some debate on the phonological status of compound prominence as either
a kind of lexical stress or a kind of accentuation. Thus an accent-based account of English
compound stress would hold that left-prominent compounds have only one pitch accent,
namely on the left member, while right-prominent compounds have two pitch accents, one
on each of the two members (e.g. Gussenhoven 2004: 19, 276–7; Kunter 2011). We remain
agnostic here and use the terms ‘prominent’ and ‘stressed’ interchangeably, without any theo-
retical implication.
The important question remains: what determines the distribution of leftward and rightward
stresses? One factor that plays a role is the type of compound in question. Thus, noun–noun
compounds seem to behave slightly differently from adjective–noun compounds, and other
combinations of word classes may show yet other characteristics. In the following we will mainly
deal with stress assignment to noun–noun compounds for the simple reason that all empirical
studies available deal exclusively with this type of compound. We will complement our discus-
sion of noun–noun compounds only with a few tentative remarks on other compound types.
In the literature on noun–noun stress, a number of influential factors for the distribution
of the two stress patterns have been proposed, such as argument structure, semantics, lexi-
calization, spelling, analogy, and informativity. Recently, some empirical studies have become
available that have tested various hypotheses using large amounts of experimental and corpus
data, and we will summarize their results. The theoretically most important result is perhaps
that none of the determining factors works in a categorical fashion. Rather, different studies
find different effects at work, all of them having a probabilistic effect. We will discuss each
factor in turn.
Before discussing the above-mentioned individual factors that determine the distribution
of left and right stress, it should be pointed out that compound stress may also vary along
other dimensions. First of all, contrastive stress may generally override the stress pattern
normally assigned to a compound (as in She meant Park Stréet, not Park Róad). In the follow-
ing, we will ignore this source of variation and only deal with compound stress in non-­
contrastive environments. Second, we find regional differences, that is dialects may differ in
the stressing of particular forms, or even whole sets of forms (e.g. NAmE drý-clean versus BrE
dry-cléan). Third, a given compound may show variation even within and across speakers of
the same dialect. Kunter’s (2011) corpus study indicates that this type of variation seems to be
restricted to certain compounds and is absent from other compounds. It remains unclear,
however, what makes a given compound variable (e.g. ice-cream) and another one non-variable
(e.g. ice-cap). Speaker-variation will inevitably lead to disagreement about the stress of some
forms exemplified below. All forms we cite with a particular stress pattern have been taken
from speech corpora, from the literature on compound stress (including dictionaries) or
have been elicited from or judged by phonetically aware native speakers.
446 Compounds: formal considerations

Let us now turn to factors that have been more thoroughly investigated. Giegerich (2004)
claims that what he calls argument-head compounds such as ópera singer (a subset of our
argumental compounds, see Chapter 20) have left stress, while modifier-head structures like
steel brídge (our attributive compounds) are right-stressed. He concedes that there are many
left-stressed modifier-head compounds (such as ópera glass) and explains this phenomenon
as a lexicalization effect. There is also potentially an important distinction to make between
subject and object arguments. Liberman and Sproat (1992) claim that subject-referencing
compounds (such as government claim) are right-stressed, while object-referencing com-
pounds (such as bookseller, squadron leader) are left-stressed, and Kunter (2011: 134–5) finds
empirical evidence for a tendency in that direction. In view of the many counterexamples
(e.g. Chapter 20, example (5)) further study is called for.
Empirical evidence for the general argument structure effect as proposed by Giegerich
was found, for example, by Plag (2006) and Plag et al. (2007, 2008), but the latter two
studies (using different corpora) both found that the effect was restricted to compounds
whose head ended in agentive -er (as in opera singer), which corresponds largely to the
group of object-referencing compounds mentioned above. Furthermore, the lexicaliza-
tion effect as gauged by frequency (with higher frequency indicating a higher degree of
lexicalization) was found to hold equally for argumental compounds and attributive
compounds. Overall, the structural distinction does not turn out to be a good predictor
of compound stress.
On the other hand, a very robust general effect of lexicalization has been empirically sub-
stantiated in a number of studies. While Plag (2006) finds no correlation between frequency
and stress for his small set of experimental items, studies using larger numbers of compounds
(e.g. Plag et al. 2007, 2008; Bell and Plag 2012) have supported the claim that with rising fre-
quency the chances of left stress increase. Another way to assess lexicalization is spelling.
Sepp (2006) and Plag et al. (2008) showed that spelling and frequency are related. The higher
the frequency of a compound, the more likely a concatenated spelling becomes. Since con-
catenated spellings are most prevalent among lexicalized compounds, it can be predicted that
the proportion of left stresses is highest with compounds that occur with a concatenated
spelling. The studies by Plag and colleagues (2007, 2008; Bell and Plag 2012) show that this
prediction is correct.
Semantics has also played a role in discussions of compound stress. For example, it has
been claimed that certain semantic relations between the two members go together with
particular stress patterns (see e.g. Kingdon 1958; Fudge 1984; Olsen 2001; Plag et al. 2008; and
Lieber and Štekauer 2009: 8–11 for a review). A very robust generalization in this domain is
that coordinative compounds (such as singer-sóngwriter, nerd-génius) are stressed on the right
constituent. Other relations taken to trigger rightward stress are, for example, ‘N2 is tempo-
rally or spatially located at N1’ (summer dréss, Boston hárbor), ‘N2 is made of N1’ (aluminum
lég), ‘N2 is created by N1’ (Mahler sýmphony). However, although they have been shown to
be influential, the effects of these semantic classes do not turn out to be categorical in any of
 19.3 Internal structure 447

the empirical investigations that tested them (e.g. Plag et al. 2007, 2008; Bell and Plag 2012).
In other words, the semantic approach works for many, but by no means all compounds for
which it makes predictions. Other semantic approaches have taken recourse to the properties
of individual constituents or of the compound as a whole. Often, the respective claims coin-
cide with claims based on semantic relations and have the same drawbacks. More exotic cat-
egories held to trigger rightward stress are also mentioned in the literature, for example, first
nouns denoting food items (e.g. Gussenhoven and Broeders 1981), but empirical investiga-
tion has not found any independent effect (Bell and Plag 2012).
From a theoretical perspective the question arises as to what unites the mixed bag of
semantic categories and relations that are held responsible for rightward stress assignment. In
other words, why would a compound showing a locative relation behave stress-wise in the
same way as a coordinative compound? Giegerich (2009b) suggests that the semantic factors
can be subsumed under the notions of semantic transparency and ascriptive versus associa-
tive semantics. According to Giegerich, in ascriptive attribution, the modifier denotes a
property of the head, as in metal bridge which can be paraphrased as ‘the bridge is metal’. In
contrast, in associative attribution, the modifier denotes an entity associated with the head
(e.g. metal fatigue), where a paraphrase ‘the fatigue is metal’ is nonsensical, but the fatigue ‘is
associated with metal’. Giegerich (2009b: 5–6) claims that only modifier-head compounds
(‘attribute-head’ in his terminology) can be right-stressed, and that right stress typically
occurs with semantically transparent compounds and with compounds that express an
ascriptive relationship.
There are at least two problems with this attempt to unify the semantic effects. First, as
shown in the empirical studies mentioned above, the assumption that only modifier-head
compounds can carry rightward stress is empirically wrong (examples like party léader prove
the hypothesis wrong). Second, the distinction between an ascriptive and an associative
relationship is often very hard to apply. For example, it seems that linguists may disagree as to
whether door in doorknob is associative (the knob is associated with a door) or ascriptive (the
knob has the property of being on a door) in nature.
The most successful approaches to the compound stress problem to date are those that do
not relate to categorical rule application but to effects of lexical relatedness and lexical repre-
sentation. One set of these approaches focuses on analogical effects, another on informativ-
ity. We will discuss each in turn.
It has long been observed that compounds with the same rightward or leftward constitu-
ent behave rather uniformly in terms of stress assignment. The sets of compounds that share
a constituent are called ‘constituent families’, and a prime example of such constituent family
effects are compounds with the right constituent street as against compounds with the right
constituent road, the former of which are all left-stressed (e.g. Óxford Street), the latter of
which are all right-stressed (Oxford Róad). Corpus-based studies (e.g. Plag 2010; Arndt-
Lappe 2011) have shown that the constituent family effect is an extremely powerful mecha-
nism across the board in predicting the stress pattern.
448 Compounds: formal considerations

Another promising take on compound stress is inspired by the idea that generally,
those expressions tend to receive prosodic prominence that are most interesting or
informative for the speakers and hearers. What does it mean for an expression, or a lin-
guistic sign, to be informative? Shannon (1948) operationalized the informativity of a
sign as the inverse of the probability of occurrence (i.e. the frequency) of that sign.
Applying the idea of the connection between informativity and stress placement we
arrive at the prediction that the right member of a compound receives stress if it is espe-
cially informative. This idea has been around for a long time (see, for example, Sweet
1892: 288), but has been tested only recently. Sweet himself suggests using frequency and
semantic specificity as correlates of informativity, with high frequency and semantic non-
specificity being indicators of low informativity, and Bell and Plag (2012) find the pre-
dicted effects in a large corpus.
To summarize, stress assignment to noun–noun constructs is an intricate phenomenon
where many different, sometimes even conflicting, influences play a role. It is presently
unclear how the different factors can be unified or related to each other. What is clear is that
deterministic rules do not lead to adequate results.
Let us turn to other kinds of compound, for which, however, no systematic research has
been carried out. Adjective–noun compounds (such as blúeprint, hóthouse) normally have
initial stress. This is scarcely surprising, since stress was taken as one of the crucial factors in
identifying adjective–noun compounds (see Section 19.2.1.2). However, adjective–noun con-
structions without left stress are generally accepted as being normal syntactic structures, so
the weight of having left stress is greater here than in the noun–noun cases where the stress
is not necessarily decisive. Note that as well as the stress pattern, which distinguishes these
adjective–noun compounds from the corresponding phrases, there is often reduced transpar-
ency. Right stress can also be found with adjective–noun compounds which act as adjectives
(e.g. small-scále). The leftward-stress pattern can also be commonly found with verb–noun
compounds (crýbaby, spóilsport).
Adjectival compounds show the additional complication that with many compounds
stress may vary according to sentential position. In premodifying position, there is a ten-
dency towards left stress (e.g. a ský-blue dress), while in predicative position we tend to find
right stress (e.g. it is sky-blúe). Spoken in isolation, coordinative adjectival compounds (e.g.
sweet-sóur, grey-gréen) as well as most modifier-head adjectival compounds (e.g. knee-déep,
bone-drý, dog-tíred, London-básed) are right-stressed. Argument-head adjectival compounds
may take either pattern. While some are more prone to right stress (university-contrólled),
others are left stressed (áwe-inspiring, háir-raising).
Verbal compounds seem to generally prefer left-stress in NAmE, although right-stressed
items are also found (cf. swéet-talk versus soft-sóap). In BrE, coordinative verbal compounds
show a certain amount of variation (e.g. dry-cléan, freeze-drý ~ fréeze-dry, stir-frý ~ stír-fry),
while all other verbal compounds have a tendency towards left stress (déep-fry, shórtcut,
blíndfold, próof-read, tálent-spot, cháin-smoke, but see also spring-cléan).
 19.3 Internal structure 449

Let us turn to stress assignment on compounds with more than two members. Again we
concentrate on compounds with only nouns as constituents, since there are no other combi-
nations of bases that have been systematically investigated. The standard assumption for tri-
constituent nominal compounds is that the branching-direction determines stress assignment.
In left-branching compounds, that is those of the structure [ [NN] N ], the leftmost noun
receives main stress, whereas in right-branching compounds, that is [ N [NN] ], the second
noun is the most prominent one (so-called ‘Lexical Category Prominence Rule’, e.g. Liberman
and Prince 1977). This assumption is built in turn on the ill-conceived idea that all compounds
are left-stressed. And indeed it has been shown that branching direction does not work prop-
erly as a predictor of triconstituent compound stress (e.g. Giegerich 2009a; Kösling and Plag
2009; Kösling 2011). Giegerich (2009a: 10) gives the following examples, which show that all
constituents may in principle be stressed, no matter what the bracketing is.

(15) a. right branching

steel [wáre-house]
university [spring térm]
ówl [nest-box]
b. left-branching
[sand-stone] wáll
[óil-tanker] driver
[garden shéd] exhibition

The crucial point is that the stress of triconstituent compounds seems to be subject to the
same mechanisms as the stress of two-member compounds (Kösling 2011). The only differ-
ence between the two kinds of compound is that with triconstituent compounds these
mechanisms work at two levels, namely at the level of the immediate constituents and at the
level of the embedded compound. Still, the Lexical Category Prominence Rule makes cor-
rect predictions for the majority of compounds. As shown by Kösling and Plag (2009), the
many cases of correctly predicted stress patterns result from the fact that there is a skewed
distribution of different kinds of bracketings and stress patterns. Thus most triconstituent
compounds are left-branching, and most two-member compounds are left-stressed, which
gives the Lexical Category Prominence Rule a quantitative advantage which belies the rule’s
theoretical and empirical inadequacy.

19.3.4 Orthography
As already mentioned, the spelling of compounds is variable, with three possibilities: solid
(butterfly, wristband), hyphenated (child-care, loop-hole) and spaced (state official, seat belt).
Some types have three variants (e.g. cell phone, cell-phone, cellphone), but most compounds
450 Compounds: formal considerations

prefer one spelling. Thus, in the corpus of noun–noun compounds used in Sepp (2006),
there are 91,868 compound types, but including all spelling variants increases the numbers by
only an additional 3 per cent (N = 3,009). Depending on the corpus one chooses the differ-
ent spellings occur in variable proportions. Table 19.1 summarizes the results for noun–noun
compounds from three corpora. Sepp (2006) is a balanced corpus with texts from different
domains, Plag et al. (2008) extracted their compounds from the Boston University Speech
Corpus (BURSC, Ostendorf et al. 1996), which contains news speech, and CELEX (Baayen
et al. 1995) is a lexical database with compounds taken from two dictionaries.
From such results we can conclude that in normal usage there is a strong tendency for
noun–noun compounds to be spelled as two words. The discrepancy between the text cor-
pus-based figures and the dictionary-based figures indicates one important factor determin-
ing the choice of the spelling, that is lexicalization. The more lexicalized the compound the
more likely it becomes that we find a solid spelling. Thus, as already mentioned, Sepp (2006),
Plag et al. (2007, 2008) and Bell and Plag (2012) find a correlation of frequency and spelling
(and stress), with solid spellings being more probable among more frequent compounds.
Other factors influencing the spelling choice (empirically verified in Sepp 2006: 88–116) are
length (longer compounds tend to avoid solid spellings) and the segmental make-up across
the internal boundary. Identical consonants across the word-boundary and vowel hiatus dis-
favour a solid spelling. Finally, we find analogical effects based on constituent families. For
example, a solid spelling for a given compound is more likely if its constituents also occur in
many other solid compounds. It also seems that there may be in-house spelling rules for
compound orthography; for example, The Chambers Dictionary (Marr 2008) is more con-
servative in using large numbers of hyphens than Collins English Dictionary (2006).
The above findings all refer to noun–noun compounds, and it is not clear whether they
easily transfer to other types of compound. On the contrary, it seems that the spelling of
compounds with an adjectival or verbal head show a different kind of variability. Unfortunately,
there are no systematic empirical studies available on the spelling of adjectival and verbal
compounds, but adjectival compounds seem to generally prefer hyphenation (e.g. dog-lean,
girl-crazy, sugar-free, university-controlled, Cambridge-based), with solid or spaced spellings
being much rarer (e.g. footloose, threadbare, toll free). For illustration, consider the

Table 19.1 Proportion of spellings in three corpora in percent

Spelling Sepp (2006) types Sepp (2006) tokens BURSC tokens Plag et al. CELEX
(N = 94,877) (N = 265,991) (2008) (N = 4,353)
Solid 5.4 27.5 8.4 42.1
Hyphenated 10.3 7.5 0.1 29.6
Spaced 84.2 65.1 90.9 28.2
 19.4 Types of compound 451

distribution of spelling variants for the noun–adjective combination N-prone (as in accident-
prone). In COCA the vast majority of relevant forms are hyphenated. Solid spellings (e.g.
repairprone, floodprone) are extremely restricted and can be found for only 12 types (as against
227 hyphenated types), with only 18 tokens (as against 935 hyphenated tokens).
A similar picture holds for the verbal compounds, which also show a clear preponderance
of hyphenated spellings (stir-fry, finger-catch), with only the occasional solid spelling or
spaced spelling (e.g. smellsip, flash photograph, from Erdmann 1999).
The above discussion tacitly related to contexts in which the different spelling variants
are functionally equivalent. This is, however, not always the case. Hyphenation, for example,
is frequently used to indicate correct parsing with premodificational adjectival compounds
(a disease causing headache versus a disease-causing headache) or with triconstituent com-
pounds (e.g. company whistle-blowers, Kösling and Plag 2009: 210). In larger compounds,
especially phrasal compounds, the use of hyphenation also reflects structural relations as
represented by syntactic bracketing (cf. [noun-plus-adjective] compounds, or [late-twentieth-
century] Britain. In such cases, the hyphenated portions represent the first immediate con-
stituent of the compound. Especially with adjective–noun compounds, solid or hyphenated
spellings may be used to indicate the status of the construct as a (lexicalized) word instead
of a phrase (whiteboard versus white board). However, this is also linked with frequency so
that relatively less frequent red squirrel is lexicalized but spelled spaced.

19.4 Types of compound

In this section we will look in more detail at the formal properties of specific types of com-
pound. These types are grouped according to the formal (syntactic, morphological, phonologi-
cal) properties of the kinds of elements that they consist of. Table 19.2 gives an overview of the
compound types by syntactic category, using a product-oriented definition of compounding,
including canonical and non-canonical compounds. The examples in Table 19.2 simply illustrate
in a pretheoretical manner forms with words of the stated category in the stated position.
Quite strikingly, all combinations are attested. As we will see, however, the different
combinations differ widely in their derivational history and productivity. In the next four
subsections we will discuss each of the cells in Table 19.2. We will do so column-wise, that
is according to heads. The semantic properties of these compounds will be discussed in
detail in the next chapter; for questions of spelling and stress assignment see the preceding
subsections.

19.4.1 Nominal compounds

Noun–noun compounds are the most productive kind of compound in English, and they are
semantically extremely diverse (as will be illustrated in Chapter 20). The productivity of
452 Compounds: formal considerations

Table 19.2 Compound types by syntactic category

Noun Verb Adjective Preposition
Noun film society brainwash knee-deep year-in
Verb pickpocket stir-fry go-slow breakdown
Adjective hothouse hot glue light-green tuned-in

adjective–noun compounds is hard to determine, since there may be some variation in stress
in individual collocations and since the compounds seem to arise from stress shift in phrases
(Bauer 1983: 205–6). An example like slow food may be indeterminate, even though its conge-
ner fast-food has left-stress in attributive positions as in fast-food outlets. Spelling is not neces-
sarily helpful, since the majority of adjective–noun constructions linked by a hyphen are in
attributive position, and the hyphen is used according to standard orthographic norms. It is
sometimes implied that exocentric adjective–noun compounds cannot be productive
( Jackendoff 1975), although this seems unlikely since various meanings for a form like redskin
(North American native, potato, apple, tomato) can readily be found.
Next, there is a class of compounds in which the first element is ambiguously analysed as
a noun or a verb, for example, boarding pass, livingroom, surfing lesson, killing field. These some-
times alternate with forms that lack the ‑ing, as in board pass, surf lesson, in which case they are
unambiguously verb–noun compounds. New endocentric verb–noun compounds in which
the noun plays various roles in relation to the verb are regularly found; examples include dive
suit, drown-proofing, dump truck, write-performance, think-tank, wait-time, as well as those
immediately above (see Chapter 20.3.1.1 for discussion). The pickpocket type illustrated in
Table 19.2 is exocentric, and not really productive in current English.
Finally, we find numerous constructions with prepositions as first elements, as in after-
birth, afterthought, backseat, back-office, downtown, downside, upland, uplight. However, there
are good reasons to analyse these as derived words rather than compounds (see Chapter 16).

19.4.2 Verbal compounds

Most compounds with verbal right elements seem to be non-canonical, that is derived by
processes other than the combination of two bases. The effect is so strong that Marchand
(1969) claims that there are no verbal compounds in English. Erdmann (1999) finds the fol-
lowing sources for compound verbs.
They can be converted from nominal or adjectival compounds or even phrases (bear-
proof, breath test, carbon-copy, cold shoulder, head-shake, white-lie), derived by inversion from
phrasal constructions (e.g. upgrade ◆ grade up ‘to improve the breed of (stock) by grading’
OED), back-derived from nominal or adjectival compounds (de-nominal: ghetto-blast
 19.4 Types of compound 453

◆ ghetto-blaster, crash-land ◆ crash-landing, de-adjectival: horror-strike ◆ horror-struck, tailor-

make ◆ tailor-made). In many cases, however, the evidence (e.g. from earliest attestations) is
not very clear concerning the question of whether the compound verb was first or some
nominal or adjectival compound from which it was potentially derived. Erdmann (1999)
gives many examples of compound pairs where the dates of the earliest attestations are very
close to each other, or where there is even an earlier attestation for the verbal compound.
Especially among the latter group, one often finds whole sets of verbal compounds with the
same right element, which suggests that analogy is also a very important factor at work. For
example, the twentieth-century neologisms drip-dry, spin-dry, blow-dry are all attested
before their adjectival counterparts came into existence (drip-dried, spin-dried, blow-dried).
Crucially, at the time there was already a large right constituent family of verbal compounds
in ‑dry that served as an analogical model for these new canonical formations (kiln-dry,
smoke-dry, sun-dry).
There are, however, also verbal compounds that apparently have been coined without
recourse to any of the above mechanisms and should be considered canonical compounds.
Such compounds can be either attributive, as in tip-touch, window-shop, sugarcoat, or possibly
coordinative, as in blow-dry, stir-fry, trickle-irrigate.
Given the diverse derivational histories of compound verbs, it comes as no surprise that
we find nouns, adjectives, verbs, and prepositions as first elements.

19.4.3 Adjectival compounds

Canonical adjectival compounds can be productively formed with nouns or adjectives as first
elements. With nouns in initial position the compound is generally argumental (e.g. ankle-
deep, butter-yellow, crime-prone, resource-expensive, word-final) with adjectival non-head it is
coordinative (blue-green, spicy-mild, icy-hot). Non-canonical adjectival compounds arise
through the use of phrases in premodifying position (as in they want a go-slow approach,
COCA). Whether the slow in go-slow should be analysed as an adjective or as an adverb may
be controversial, and it is a problem which faces many such forms. Adjectival compounds
with prepositions in non-head position seem to be restricted to participles as heads (down-
sized, in-turned, out-sourced, up-graded), and are derived by inversion from phrasal verbs, or
are participles of compound verbs.

19.4.4 Prepositional compounds

None of the four types of prepositional compounds are canonical. The combination of two
prepositions as in into, upon has arisen via univerbation. Noun–preposition combinations are
not restricted to fixed expressions such as year-in, year-out, but appear to be moderately pro-
ductive, as shown in (16).
454 Compounds: formal considerations

(16) 2010 FIC Bk:AmorFugit: I hold the book face-out so he can read the lettering
2004 MAG TodaysParent: Can be worn in five positions (front face-in and face-out,
back face-out and face-in and a fifth “nursing” position)
1999 MAG PopScience: Each has 16MB of memory, a color LCD, and a video-out port.
2005 FIC SCarolinaRev: they drank until sun-up

The syntactic category of these formations is not always clear, but it seems that they cannot
be straightforwardly analysed as prepositions (sun-up is a noun, face-out an adverb, video-out
is even less clear), in other words, these forms are syntactically exocentric. Yet another case
of noun–preposition compounds are forms arising from the suffixation of ‑er to phrasal verbs
(e.g. runner-up, passer-by, picker-up).
Verb–preposition compounds are either nouns converted from phrasal verbal construc-
tions (e.g. breakdown, hold-up, sleep-over), or analogical formations on the basis of converted
phrasal constructions (e.g. sleep-in on the basis of sit-in—there are also other meanings of
sleep-in). Adjective–preposition compounds are adjectives and seem to be restricted to ver-
bal participles as heads (e.g. tuned-in, tied-up).

19.4.5 Neo-classical compounds

Neoclassical compounds have a number of peculiarities that deserve special attention. First,
the position and combinatorial properties of the elements involved; second, the phonologi-
cal properties of the resulting compounds; third, the status and behaviour of the vowel that
often appears in the middle of such forms; and finally the syntactic category of the elements
in question.
The list of forms that can be argued to belong to the class of neoclassical forms is long, and
it is not entirely clear which elements should belong to this class (see, for example, Bauer
1998b; Baeskow 2004 for discussion). (17) shows some of them.

(17) form meaning example

a. astro‑ ‘space’ astro-physics, astrology
bio‑ ‘life’ biodegradable, biocracy
biblio‑ ‘book’ bibliography, bibliotherapy
electro‑ ‘electricity’ electro-cardiograph, electrography
geo‑ ‘earth’ geographic, geology
hydro‑ ‘water’ hydro-electric, hydrology
morpho‑ ‘figure’ morphology, morpho-genesis
philo‑ ‘love’ philotheist, philo-gastric
retro‑ ‘backwards’ retroflex, retro-design
 19.4 Types of compound 455

tele‑ ‘distant’ television, telepathy

theo‑ ‘god’ theocratic, theology
b. ‑cide ‘murder’ genocide, insecticide
‑cracy ‘rule’ bureaucracy, democracy
‑graphy ‘write’ iconography, bibliography
‑itis ‘disease’ laryngitis, lazyitis
‑logy ‘science of ’ astrology, lichenology
‑morph ‘figure’ anthropomorph, pythonomorph
‑phile ‘love’ bibliophile, jazzophile
‑phobe ‘fear’ bibliophobe, commitmentphobe
‑scope ‘look at’ iconoscope, telescope

Most of the forms in (17) do not occur as free forms. There are elements like morph‑/‑morph
and phil‑/‑phile, which can occur in both initial and final position, while most forms occur in
only one position. Hence a distinction is often made between initial combining forms and
final combining forms. Combining forms can either combine with bound roots (e.g. glaciol-
ogy, scientology), with words (lazyitis, hydro-electric, morpho-syntax), or with another combin-
ing form (hydrology, morphology).
With regard to the phonological properties of neoclassical elements, we see that they vary
in their segmental structure and in their stress contour, depending on whether initial com-
bining forms combine with free forms or with certain other combining forms.

(18) a. astro-phýsics b. astrólogy

biodegrádable biócracy
biblio-thérapy bibliógraphy

Initial combining forms that take a word as second element, as in (18a), regularly have their
main stress on the righthand member of the compound. Many final combining forms, such
as ‑graphy, ‑cracy, and ‑logy, as in (18b) impose antepenultimate stress on the compound,
accompanied by a different vowel quality (/əʊ/‖/oʊ/ versus /ɒ/‖/ɑ/). Combining forms
such as ‑graphy, ‑cracy, and ‑logy thus behave phonologically like certain stress-influencing
suffixes (such as ‑ity).
Note that we have accepted here a general analysis whereby words such as geology are
usually seen as being made up of geo‑ and ‑logy (see below for the status of the ‑o‑). In
many instances we might consider the final ‑y here an affix (as in allomorphy, for example),
parallel to ‑ist in geologist (see Chapter 12.2.5 on ‑y in particular). If we accept items like
‑logy as units, rather than as sequences of ‑log‑y, it is because they are treated as independ-
ent items by speakers (e.g. Urdang et al. 1986: 7, where several such items are treated as
unanalysable).
456 Compounds: formal considerations

The next characteristic of neoclassical compounds that needs discussion is the status of the
vowel that appears at the boundary of the two elements. This vowel, orthographically repre-
sented mostly as either <o> (mainly with Greek-derived bases) or <i> (mainly with Latin-
derived bases), could be analysed as the base-final vowel of the initial element (e.g. Bauer
1998b) or as an independent element intervening between the two bases, similar to linking
elements in native compounds in other Germanic languages (e.g. Baeskow 2004). In English,
combining forms are used with the attached vowel when added to lexemic bases: insect-icide,
morpho-syntax, geo-thermal, lichen-ology. This might lead to an analysis whereby the vowel
belongs with each combining form and is part of the lexical specification of the element. This
would account for the differing vowels in tele-printer, insect-icide and socio-linguistic. It would
also imply some kind of haplology to simplify a sequence of vowels in cases like *geo-ology.
There is an alternative analysis of what is going on here. We could see the vowel as an ele-
ment placed between the two roots to ameliorate the phonotactics so that morphology comes
from *morph-logy with an inserted ‑o‑ (this is historically the case in Greek, but, of course, it
does not follow that this is the best analysis for modern English). This fails to account for the
variation in the vowel, but does account for the fact that the vowel may be omitted where the
morphology does not bring two consonants into contact across the boundary: homo‑taxis,
laryng‑itis, mono‑semy, poly‑gon, It also accounts for the vowel in formations such as ster-
oid‑o‑genesis (COCA) which appear to be made up on the basis of two English lexemes.
Ancient Greek formations and modern English formations differ slightly in their patterns of
constructions where there is an initial <h> in the second combining form. Such differences are
also seen with Greek prefixes. For instance, consider the difference between anthelion and anti-
hydrophobic, both of which contain the prefix anti‑. In anthelion, the final <i> of anti‑ is deleted to
avoid the vowel–vowel sequence because the <h> is treated as a vowel. In anti-hydrophobic, the
<h> is treated as a consonant, and there is no deletion. Similar instances are found with neoclas-
sical formations. Glycemia and glycohemia are synonymous words made from the same elements,
the first with vowel (and <h>) deletion, the latter with no deletion, in the modern manner.
Finally, we turn to the problem of syntactic category. In general, we find that neoclassical
formations are nouns. This may be surprising since bound roots are normally assumed to
have no syntactic category specification (e.g. Giegerich 1999: 72–6). Neoclassical compounds
may undergo further morphological processes, in which case the new formation largely
inherits the formal properties of its base. For this reason such derivations are sometimes
called ‘neoclassical derivations’ (Baeskow 2004).

19.4.6 Phrasal compounds

Phrasal compounds are right-headed structures and seem to be largely restricted to nominal
heads, although, for example, adjectival heads do allow phrases as left sisters. This class of
compounds is very productive, some examples from COCA are given in (19).
 19.4 Types of compound 457

(19) burned-rope-and-sugar taste, soon-to-be-divorced wife, less-than-successful husband,

empty-nest syndrome, one-on-one conversation, work-at-home husband, on-air
puzzle, this-person-is-a-jerk attitude

There seem to be very few restrictions on the phrases that can be used in first position: we
find noun phrases, adjective phrases, verb phrases, prepositional phrases, and clauses. It is
even possible to use strings that are not syntactic constituents as initial elements (e.g. thumbs-
up sign). There is, however, a restriction on the kinds of noun phrases that may occur in
phrasal compounds. Thus, it has been pointed out in the literature (e.g. Lieber 1992: 12) that
determiner phrases seem not to be allowed (*a the-burned-rope-and-sugar taste). Why it is
exactly that these phrases are excluded is still under debate.
Although many phrasal compounds feature phrases that seem lexicalized or rather fre-
quent, it is by no means impossible to use completely new phrases (e.g. this-person-is-a-jerk-
attitude, COCA).

19.4.7 Reduplicative compounds

There are various types of reduplication whose main function appears to be to indicate atti-
tude, and these are dealt with in Chapter 18.2.6. Here we deal with a pattern of full reduplica-
tion also known as contrastive reduplication (Ghomeshi et al. 2004) or identical constituent
compounds (Hohenhaus 2004) which involves the full phonological copying of bases, that is
of words or phrases. Ghomeshi et al.’s (2004) examples reproduced in (20) illustrate the
phenomenon.

(20) a. Are you leaving-leaving? [i.e., are you “really” leaving (for good), or are you just
stepping out for a minute?]
b. A: Are you nervous?
B: Yeah, but, you know, not nervous-nervous. [i.e. not “really” nervous]
c. Lily: You have to get up.
Rick: I am up.
Lily: I mean up-up.

Semantically, contrastive reduplication restricts the meaning of the head to some type of
prototypical meaning. As a consequence, Ghomeshi et al. (2004) suggest that function words
lacking the necessary semantic variation cannot be contrastively reduplicated.

(21) A: I will visit my mother next week.

B: *You will, or you will-will?
458 Compounds: formal considerations

Note, however, the title The Foods you Must Must Eat Before You Die (Men’s Health 2011)
which suggests that such constraints may be overstated.
Contrastive reduplicative compounds are categorically stressed on the first immediate
constituent, which is, however, a fact that automatically follows from the contrastive prag-
matics and semantics.

19.4.8 Blends
Blends are compounds where at least one constituent lacks some of its phonological mate-
rial. Semantically, they behave like other compounds, as shown in more detail in Chapter 20,
but stress-wise they behave like a single word, normally adopting the stress pattern of one of
the two source words.
There is no agreed definition of the phenomenon and the boundaries of what is called
‘blend’ in the literature are fuzzy. For example, a distinction has been proposed (e.g. Bauer
2006b) between ‘blends’ (which involve the loss of medial segmental material, as in brunch)
and ‘clipped compounds’ or ‘complex clippings’ (which involve the loss of final material of
both bases, as in modem < modulator-demodulator). The two patterns can be formalized as in
(22), where AB stands for the left base (with its two parts A and B) and CD stands for the
right base, with its two parts C and D.

(22) a. AB + CD → AD
b. AB + CD → AC

In many blends, the surviving material may be shared by the two bases, as in boatel or
scrapnel, where the rhyme of the first syllable of the blend corresponds to the same
sequence of sounds in both bases (boat and hotel, scrap and shrapnel, respectively). It is
also possible to find blends in which the whole left base word is preserved (e.g. painstation
◆ pain + playstation). Either of these patterns may be analysed as an AD blend, in accord-
ance with (22a).
It is not immediately obvious whether the distinction between blends and clipped com-
pounds is theoretically or empirically informative. Gries (2006) shows, however, that the
two kinds of formation do indeed differ in important respects, so that it may be justified to
posit two distinct processes with their own properties. Thus, in Gries’s data set, clipped com-
pounds systematically preserve much less material than blends normally do. Furthermore,
overlaps of segmental material are underrepresented in clipped compounds. Finally, blends
and clipped compounds differ with regard to their base words. While bases of blends tend to
be orthographically and phonologically highly similar to each other, clipped compound
bases are significantly less similar to each other (only half-way in similarity between blend
bases and arbitrarily chosen words in the lexicon).
 19.4 Types of compound 459

There is also a clear difference in productivity, with clipped compounds being in the clear
minority, with a proportion ranging from 3 to 6 per cent in different empirical studies (cf.
Kubozono 1990; Gries 2006; Arndt-Lappe and Plag 2013).
Given that the literature on clipped compounds is scarce, we will say very little about this
category and concentrate on blends of the type given in (22a). The reader should note, how-
ever, that some of the empirical results we summarize here are taken from studies that may
not have neatly distinguished between the two types of blend.
Another problem in delimiting the category of blend is the occurrence of splinters. We
defined splinters as originally (mostly) non-morphemic portions of a word that have been
split off and used in the formation of new words with a specific new meaning. The line
between a derivative with a splinter and a blend may sometimes be hard to draw (see also
Chapter 23 for discussion). For example, freeware might be treated as a blend of free and soft-
ware, or as a case of suffixation of the splinter ‑ware to the base free (as in crimeware, eduware,
malware, netware, spyware, trialware).
In a large collection of more than 1,018 formations, Reischer (2008) finds that nouns are
the most frequent output category of blends in English (almost 80 per cent, e.g. brunch, buf-
feteria, diagnonsense), followed by adjectives (11 per cent, e.g. bleen ◆ blue + green, dramastic ◆
dramatic + drastic, and verbs (4 per cent, guesstimate). Reischer’s figures, however, include
instances of syntactic wordplay that we do not consider to be part of word-formation such as
I drink, therefore I am or A car is born. Such examples amount to only 1 per cent of all forms in
Reischer’s collection.
The four most frequent input combinations (in descending order) are noun + noun, adjec-
tive + noun, adjective + adjective, and verb + verb. Example (23) illustrates the different
patterns.

(23) noun + noun

brainiac, infotainment, netizen, painstation (◆ pain + playstation), sitcom (◆ situation
+ comedy), smog, spork (◆ spoon + fork), suicycle, skort (◆ skirt + shorts), textpert,
wreader (◆ writer-reader)
adjective + noun
britcom (◆ British + sitcom), prosumer (◆ professional + consumer), quicktionary
adjective + adjective
abnormous, delishful, fabtastic, obsolute, prezactly, rurban, solunar
verb + verb
correctify (◆ correct + rectify), gesplain (◆ gesture + explain), scarify (◆ scare + terrify),
suspose (◆ suspect + suppose)

Such examples raise a number of questions, the most important of which is, given two input
forms, how one can determine the phonological or orthographic structure of the output?
460 Compounds: formal considerations

Which parts survive, which parts are deleted? Where does the stress pattern come from?
What kinds of entities play a role in the process (e.g. graphemes, segments, onsets, codas,
rhymes, syllables, or feet)?
As will become clear, the diversity of the observable patterns is considerable and the
apparent variability in the outcomes has led researchers to claim that the phonological-
orthographic output properties of blends are largely unpredictable (e.g. Marchand 1969;
Bauer 1983: 225; Cannon 1986: 744). Notably, such pessimistic statements stem from an era
in which deterministic rules predominated the conceptualization of structural linguistic phe-
nomena. More recent developments in probabilistic grammar, prosodic morphology, and
statistical linguistic modelling have contributed to the emergence of many studies that have
shown that blends are not to be conceived of as ‘extragrammatical’ (Dressler 2000). Instead,
the formation of blends is prosodically highly constrained and essentially predictable within
a well-defined space of variation.
Constraint-based approaches in the theoretical framework of Prosodic Morphology
such as Lappe (2007) have shown that subtractive morphological processes such as the
formation of clippings are largely regular, despite the conclusions in much of the earlier
literature (see Chapter 18). These processes rely heavily on the manipulation of prosodic
constituents instead of on the agglutination of morphemes. Blending is another such pro­
cess in which prosodic categories play a prominent role in shaping the phonological form
of the output. Approaches using this perspective have achieved considerable success in
describing the phonological properties of blends and in suggesting mechanisms through
which these properties emerge (e.g. Kubozono 1990; Bat-El 2006; Bat-El and Cohen 2012;
Arndt-Lappe and Plag 2013). Complementing constraint-based approaches, statistical, and
computational investigations such as Gries (2004a, 2004b, 2004c, 2006, 2011), Reischer
(2005, 2008), or Cook and Stevenson (2010) have revealed that many aspects of blends
lend themselves to successful quantitative and computational modelling. In the following
we summarize some of the most important findings with regard to the above-mentioned
questions.
In general, there are two competing forces at work that generally constrain the form of phono-
logically reduced compounds (and other prosodic-morphological formations). One is the short-
ening of the combination into a single word, the other the need to preserve as much material as
possible in order to establish a formal relationship between the output and the input forms.
The two counteracting principles just mentioned conspire to influence not only the size,
but also the order of the two elements, such that in coordinative blends, where order is not
determined by semantic considerations, there is a strong tendency to prefer an order that
preserves as many segments as possible. Thus, we get, for example, spork instead of *foon from
the combination of spoon and fork. However, this tendency does not seem to hold for forma-
tions that involve source words with more than one syllable, a point which illustrates the
complexity and interaction of the restrictions involved.
 19.4 Types of compound 461

One general restriction on blends is that the output must obey the normal constraints on
the phonological structure of words in English. For example, Davis (1988) suggests a form
like *smang ◆ smash + bang) preserves more material than the competitor bash but is ruled
out for phonotactic reasons, since English words of the structure sCVC may not contain two
nasals in the two C positions (Davis 1988).
Another combinatorial mechanism that reconciles the two basic competing principles
with regard to the size of the output form is the tendency of blends to have as many syllables
as one of the bases, usually the second base (as in boatel, brunch, guesstimate). Counterexamples
preserving more syllables can be found (e.g. correctify), but they often involve a considerable
overlap of phonological (or sometimes purely orthographic) material, which facilitates the
recoverability of the relationship between base words and blend. Example (24) provides an
illustration of overlaps, with the overlapping part in bold print. The form brainiac shows a
phonological overlap, netizen an overlap at the phonological and orthographic level, smog an
orthographic overlap, and suicycle an intricate mixture of both, with the first <i> overlapping
orthographically, but not phonologically, and the onset and nuclei of the respective second
syllables (<ci> and <cy>) overlapping phonologically but not orthographically. Finally,
especially with longer words, there can be no overlap (as in infotainment).

(24) brain + maniac ◆ brainiac

net + citizen ◆ netizen
smoke + fog ◆ smog
suicide + bicycle ◆ suicycle
info/information + entertainment ◆ infotainment

The examples in (24) may suggest that overlaps are always and only medial. This is, however,
not true. Overlaps may also occur in other positions, and given the fact that the two bases are
often phonologically very similar to each other, this is even to be expected, consider Chunnel
◆ Channel + tunnel, hesiflation ◆ hesitation + inflation, Turlish ◆ Turkish+ Polish. There is no
empirical study available that specifically addresses the question of where the cut-off points
are located in blends that have a medial overlap as against blends that do not have a medial
overlap. It seems, however, that blends with medial overlaps also conform to the restrictions
that hold for blends with a different kind of overlap or without any overlap. In general, in an
output-oriented analysis, forms that show overlaps can be more easily accommodated to
structural restrictions than blends with no overlap, as correspondence relations between
bases and blend can be established more flexibly.
Some of the prosodic restrictions on blend formation concern syllable structure and stress.
For the description of the pertinent phenomena it is useful to distinguish between monosyl-
labic and polysyllabic blends. With monosyllabic blends there is a strong tendency to combine
the onset of the first word with rhyme of the second, as we see it in br#unch, sp#ork,
462 Compounds: formal considerations

Sw#atch. However, certain initial clusters of the left base may be broken up if the initial part
of the left word onset can form a new onset with the right word onset (irrespective of the
number of syllables). For instance, Arndt-Lappe and Plag (2013) find systematic variation in
their experimental data with forms such as bleen ~ breen (◆ blue + green) and scinter ~ scrinter
~ sprinter (◆ scanner + printer). They also show that base stress has an influence of the cut-off-
points. Thus, the cut-off point of the second word has a tendency to be chosen in such a way
that (parts of) the stressed syllable makes it into the blend (e.g. ginórmous ◆ gigántic + enór-
mous), a tendency that we do not find for the left base.
How much of the base survives is also dependent on the overall length constraint men-
tioned above, namely that blends are usually as long as the longer base word, which, if the
two words differ in length, is mostly the second word. In general, we find that the second
word also provides more material than the first and that the stress pattern of the blend tends
to be the same as the stress pattern of the second word.
Interestingly, these structural restrictions or tendencies have psycholinguistic correlates.
Gries (2006) shows that the cut-off points are correlated with what he calls ‘selection point’,
that is the position in the word where the selection of the base word among its lexical com-
petitors becomes highly likely (according to the particular metric he develops).
To summarize, blends are a productive word-formation process in English which, in spite
of the considerable variability, conforms to a number of general principles and tendencies
that highly restrict the structure of possible formations. Given the intricacies involved and
the comparative scarcity of pertinent studies, blends certainly merit much further study.
 chapter 20

Compounds: semantic
considerations

20.1 Prospectus
In this chapter we move from a formal consideration of compounds in English to a considera-
tion of semantics, a subject that has drawn an enormous amount of attention in the linguistic
literature. Previous accounts of the semantics of English compounding can be found in
Jespersen (1942), Lees (1960), Hatcher (1960), Marchand (1969), Brekle (1970), Adams
(1973, 2001), Levi (1978), Bauer (1978, 2008, 2010c), Lieber (1983, 2004, 2009, 2010), Ryder
(1994), Plag (2003), Jackendoff (2009), among others. Blends and neoclassical compounds
have received less attention, but we will see that their semantic properties are similar to other
compounds.
We begin in this section with a consideration of some general semantic characteristics of
English compounds and with a justification of the basic terminology we will need in order to
organize further discussion, especially the distinction between argumental and non-
argumental compounds that will be the basis for our classification. We will treat argumen-
tal compounds in Section 20.2, and move on to non-argumental compounds in Section 20.3.
Section 20.4 will discuss cases in which the boundary between argumental and non-argu-
mental compounds is blurred. In Section 20.5 we will look at the semantics of both blends
and neoclassical compounds and assess the extent to which their semantics is similar to that
of other compounds. Finally, in Section 20.6 we will look at several marginal or disputed
compounding patterns including compounds with verbs as their second elements (henpeck),
phrasal compounds (over-the-fence gossip), and reduplicative compounds (friend friend).

20.1.1 General remarks

We intend here to explore the semantics of those compounds whose meanings are composi-
tional in nature, but we should make clear at the outset what we mean by compositionality.
Compounds, like any other complex words, can be semantically opaque and be lexicalized
with non-transparent meaning. Lexicalization is not a homogeneous phenomenon, however.
464 Compounds: semantic considerations

Here we will not be concerned with the semantics of item-familiar compounds whose mean-
ings are in no way predictable from their parts—for example, compounds like butterfly or
honeymoon. Unlike derived forms, however, newly coined compounds frequently have the
potential to be multiply polysemous, at least when considered out of context. In context and
especially if they become established and item familiar, many of the potential meanings of
course fall by the wayside; at this point, the meaning of the compound becomes lexicalized,
although still compositional.
We will be concerned here with the semantics of newly coined compounds and those
lexicalized/institutionalized compounds whose meanings continue to be transparent, and
we will leave aside the semantics of compounds that have ceased to be compositional in
nature.
There are two general characteristics that form the basis of most accounts of the semantics
of English compounds: first, that the compound as a whole is a hyponym of the second ele-
ment of the compound (or put otherwise, that English compounds are right-headed seman-
tically as well as syntactically); and second, that the first element of the compound is
non-referential in nature. While both of these generalizations are true of the majority of
compounds in English, they are of course not exceptionless. We will have occasion in the
sections below to explore in depth the extent to which hyponymy obtains in English com-
pounds, but we should first say a few words about the issue of non-referentiality.
The problem of referentiality in compounds is necessarily restricted to compounds in
which the first element is a common noun. It is certainly the case that in the vast majority of
such cases, that noun is intended non-referentially; clearly, cat in cat litter does not and can-
not refer to any specific cat, nor can truck in truck driver refer to any specific truck. It should
be kept in mind, however, that compounds can be formed in English with proper nouns as
the first element, as we find in examples like Amadinejad supporter or Beatles fan. In these, the
first element is clearly meant to refer specifically to the former President of Iran or the musi-
cal group comprised of Lennon, McCartney, Harrison, and Starr. Similarly, in compounded
names of businesses, countries, or individuals (e.g. Hewlett-Packard, Bosnia-Herzegovina,
etc.), both elements of the compound are clearly referential. Similarly, compounds with
common noun modifiers that have unique reference, such as moonlight, earth sciences, sunset
have referential first elements. Furthermore, it seems that the discourse context may make
the first element of a given compound referential. Thus, in a parliamentary debate about the
budget for the nation’s army, the word army in army budget will receive a referential interpre-
tation, namely as referring to the army of that country. These complications aside, it seems
safe to say that the first element of English compounds is usually non-referential.

20.1.2 Classification
Any discussion of the semantics of compounding must inevitably be organized on the basis
of some underlying scheme of classification. The classification of compounds, however, has
 20.1 Prospectus 465

been a matter of much controversy among morphologists; see for example Scalise and Bisetto
(2009), and references cited therein. One significant problem has been the failure of many
researchers to look at the full scope of compounding in English or to look at English in the
context of types of compounds found in other languages. Another has been the welter of
overlapping and conflicting terminology. It behooves us, therefore, to begin by setting out
clearly the terminology we will use and briefly justifying the way we propose to divide up the
landscape of compounding.
In Chapter 19, we looked at formal properties of compounds, including their categorial
composition, (morphosyntactic) headedness, and phonological properties. Here, we con-
centrate on surveying the semantic types of compounds that can be found in English and the
extent to which particular semantic types and subtypes can be deemed productive. We will
therefore not divide compounds on the basis of their syntactic categories; for example, where
nominal and adjectival compounds share semantic characteristics, they will be treated
together. There have been a number of recent surveys of the semantics of English compounds
that will be useful to us, including Adams (2001), Plag (2003), Bisetto and Scalise (2005),
Bauer (2008, 2010c), Scalise and Bisetto (2009), and Lieber (2009, 2010). The system that we
will introduce here is a synthesis of various proposals introduced in those works. We will
begin with an overview of the basic terminological distinctions of which we will make use.
A distinction is commonly made between endocentric and exocentric compounds.
The most comprehensive discussion of exocentricity can be found in Bauer (2010c), which
we will largely follow here. Briefly, endocentric compounds are ones in which the compound
as a whole denotes a subset of the head element of the compound. In semantic terms, in
endocentric compounds the compound as a whole is a hyponym of the head. Exocentric
compounds are ones in which the compound as a whole does not denote a kind or subset of
the head.
However, as Bauer (2010c) points out, exocentricity is not a homogeneous phenomenon;
there are a number of ways in which compounds can count as exocentric. One type of exo-
centricity might be seen as purely semantic, that is, cases in which the compound as a whole
is not (literally or figuratively) a hyponym of its head. Items like pickpocket or cutpurse are
exocentric in the sense that they denote types of people, and not objects like pockets or
purses; these are frequently literal in nature: a pickpocket literally is ‘someone who picks
pockets’. Items like blockhead or air head are also exocentric, and might be termed posses-
sive compounds (similar to the Sanskrit designation bahuvrihi). These compounds fre-
quently denote types of people, and are often meant metaphorically or metonymically: a
blockhead, for example, is a person whose head is like a block, hence a stupid person, a red-
shank a bird with red shanks.
The next basic division we make is between argumental and non-argumental com-
pounds. The former are compounds, one element of which is interpreted as an argument of
the other element. Argument-taking elements can be relational nouns (e.g member, as in club
member), or verbs, or nouns derived from verbs via affixation or conversion (e.g. soccer
466 Compounds: semantic considerations

player). In the case of an argumental compound with a verbal or deverbal head the non-head
constituent is interpreted as a subject, object, or prepositional object. The argumental com-
pounds that have overt affixes (for example, truck driver, cost containment) have frequently
been referred to as synthetic compounds in the English tradition. Those that involve unde-
rived verbs or nouns formed from verbs by conversion (e.g. attack dog, dog attack) have fre-
quently not been identified as a coherent type, and indeed have sometimes only been
discussed in passing (but see Bauer and Huddleston 2002, and especially Lieber 2010, who
calls them ‘non-affixal (de)verbal compounds’).
Non-argumental compounds are those in which neither member bears an argumental
relation to the other. Among these, we can distinguish two major types, attributive and
coordinative compounds. In attributive compounds the relation between the first element
and the second is one of loose modification. Such compounds have often been called root
or primary compounds. In coordinative compounds we find a variety of relationships, but
always ones in which both elements of the compound have equal semantic weight; in other
words, where neither element can be said to modify the other. Note that the distinction
between attributive and coordinative compounds is not always clear. For example, it seems
that the putative coordinative compound zebra-dog could also be analysed as an attributive
one, as can be seen by the potential semantic difference between zebra-dog ‘a dog with black
stripes’ and dog-zebra ‘a zebra with the appearance of a dog’.
We should note here that the endocentric/exocentric distinction and the argumental/
non-argumental distinction are orthogonal; argumental compounds can be endocentric or
exocentric, as can non-argumental compounds. It is also worth pointing out that the dividing
line between argumental and non-argumental compounds is not always crystal clear as
deverbal nouns and adjectives can serve as the second element in non-argumental as well as
in argumental compounds. We will return to this subject in Section 20.4.

20.2 Argumental compounds

We will first discuss argumental compounds whose head is argument-taking and deverbal
(Section 20.2.1), then turn to argumental verbal compounds (Section 20.2.2), and to those
with relational nouns and non-deverbal adjectives as heads (Section 20.2.3) and then treat
argumental compounds whose non-head is argument-taking (Section 20.2.4). In Section
20.2.5 we will briefly discuss the extent to which we find exocentric examples among argu-
mental compounds in English.

20.2.1 Argumental compounds with deverbal heads

This group of compounds comprises those with a deverbal element derived by suffixation
with, for example, ‑er, ‑ation, ‑ment, ‑ure, ‑al, ‑ing, ‑ent, ‑ive, or a past participle form
 20.2 Argumental compounds 467

(henceforth affixal argumental compounds), and those in which there is a noun formed
from a verb via conversion. The semantics of argumental compounds with deverbal heads,
otherwise known as synthetic, verbal, or deverbal compounds, has been well-explored
in the literature on English word-formation, both descriptively ( Jespersen 1942; Marchand
1969; Adams 2001; Bauer and Huddleston 2002) and theoretically (Lees 1960; Levi 1978;
Roeper and Siegel 1978; Selkirk 1982; Lieber 1983, 1992, 2004; Roeper 1988; Ryder 1994, among
many others). Nevertheless, we find that scrutiny of the corpus data adds depth and nuance to
our understanding of compound use and interpretation in contemporary English.
Most commonly discussed in the literature on affixal argumental compounds are com-
pounds whose second elements are nominalizations in ‑er (truckdriver), ‑ation (air circulation),
and ‑ing (ball-playing), but such compounds can in fact occur with any deverbal nominaliza-
tion as head; we find compounds with ‑ment nouns (cost containment), ‑al nouns (garbage
disposal), ‑ure nouns (base closure), and ‑ee nouns (city employee). Also possible, although
somewhat harder to find, are cases in which the second element is a deverbal adjective, as we
find in compounds like AIDS preventive or drug dependent. The interpretation of the first ele-
ment of such compounds to a large extent depends on the argument structure of the base of
the second element, as well as on the semantic profile of the nominalizing or adjectivalizing
affix (see Chapters 10, 11, 12, 14). Nevertheless there is a surprising range of interpretation
available depending on context.
The syntactic terminology (‘subject, object, prepositional object’) we are employing to
refer to the different arguments may appear somewhat problematic, as it invites invoking
sentential paraphrases for the compounds in question. The methodological or theoretical
status of such paraphrases is, however, unclear, and there are certainly additional complexi-
ties involved in the mapping of semantic argument structure onto phrasal syntactic represen-
tations. These complexities are outside the scope of our treatment since we are dealing with
compounds and do not claim that there is any direct relationship between a given compound
and a putatively corresponding sentence. For example, heart in heart attack can be conceived
as an object argument of attack in spite of the fact that one might not be able to find an
acceptable or convincing sentential paraphrase (‘something attacks the heart’).
As we saw in Chapter 11, the affix ‑er most frequently forms agent/instrument nouns.
When ‑er derivatives with this interpretation constitute the second element of an argumental
compound, the first element of the compound is typically object or prepositional object ref-
erencing, depending on the diathesis of the base verb.

(1) a. object referencing: air heater, alcohol abuser, antelope hunter, asphalt spreader
prepositional object referencing: adoption worker, asthma sufferer, army deserter,
b. 
apartment dweller

So with a verb like heat, which can take a subject and object, air as the first element of the
compound air heater receives the object interpretation. With a verb like suffer that takes a
468 Compounds: semantic considerations

subject and a prepositional object (someone suffers from something), the ‑er form is agentive
and the first element of the compound is interpreted as the prepositional object.
The affix ‑ee typically forms patient nouns, as we saw in Chapter 11, and when used as
the second element of the compound, the first element is frequently subject referencing, as
the examples in (2a) illustrate. But with ‑ee forms that are more agentive in flavour, like the
‑er forms discussed immediately above, the first element of the compound can receive an
object or prepositional object interpretation, as the examples in (2b, c) show.

(2) a. subject referencing: Bush appointee, Delta acquiree, army detainee, Ford nominee
b. object referencing: conference attendee
c. prepositional object referencing: army retiree, catalogue devotee, band escapee

So a Bush appointee is someone who Bush has appointed (subject referencing), a conference
attendee someone who attends a conference (object referencing), and an army retiree some-
one who retires from the army (prepositional object referencing).
With nominalizing affixes like ‑ing, ‑ation, ‑ment, ‑al, and ‑ure we find argumental com-
pounds where the interpretation of the first compound element expresses one of the argu-
ments of the second (deverbal) element. As with the readings exhibited by nominalizations
themselves (see Chapter 10), the readings exhibited by compounds of which they form the
second element are highly context-dependent. We illustrate this with compounds whose
second elements are nominalizations in ‑ation, ‑al, or ‑ing, but comparable examples can be
found for items formed with the other nominalizing and adjectivalizing affixes as well (data
from COCA).

(3) a. subject referencing: army investigation, party nomination, airline hiring, administration
refusal
New York Times 1996: An army investigation later found that most of the Israelis

killed were hit by gunfire directed at the armored convoys on their way to and from
the tomb.
New York Times 2006: And it was unclear how much political support Mr. McGreevey

had among Democrats: even before Mr. Cipel first threatened to sue, Mr. McGreevey’s
spotty performance and the lingering scandals around his associates had led some
party leaders to lobby for Mr. Corzine to get the party nomination for governor
next year.
Atlanta Journal Constitution 2003: Then terrorism and a recession all but halted major

airline hiring and put thousands of pilots on furlough.
PBS Newshour 1991: Mr. Veliotes, isn’t that one of the other arguments that is behind

administration refusal to get involved, that success by either the Kurds or the Shiites
would mean either an administration by one of them or the dismemberment of
Iraq?
 20.2 Argumental compounds 469

object referencing: abuse investigation, candidate nomination, baby swapping, anger

b. 
arousal
Atlanta Journal Constitution 2001: After a four-month impasse over an abuse

investigation at the northwest Atlanta church, state child welfare officials say they’re
tired of such intransigence.
Arab Studies Quarterly 2006: “Electoral law” stands for the family of rules governing

the process of elections from the calling of the election, through the stages of
candidate nomination, party campaigning and voting, right up to the stage of
counting votes and determining the actual election result.
Dateline NBC 2003: The tests are supposed to help circumvent kidnappings and baby

swapping, but the private investigator says there are ways to get around DNA tests.
Journal of Social Psychology 1997: When experiencing anger arousal, isolated people

without a support system may suppress their anger simply because no one is present
to receive their cathartic expressions.
prepositional object referencing: Emmy nomination, aid application, activity
c. 
participation, abortion ruling, ash disposal
Parenting 2007: Mel Brooks copped an Emmy nomination for his meaty, bleaty

performance in the title segment.
USA Today 2007: Your tax return is not required with an aid application, but you do

have to provide income and other numbers from your return.
Health & Social Work 2006: Research participants reported their sociodemographic

information, physical health status, vision status, subjective experience of agerelated
vision loss, functional ability, social support from family and friends, activity
participation, rehabilitation service use, and coping strategies.
PBS Newshour 1992: The Supreme Court ended its term today with its long-awaited

abortion ruling.
Boston College Environmental Law Affairs Review 1993: Public opposition usually

involves the following: concern with health and environmental risks from air
emissions and ash disposal; . . .

As the examples in (3) illustrate, the interpretation of compounds whose second elements
are nominalizations in ‑ation, ‑al, or ‑ing can vary widely. For example, for an ‑ation noun
formed on a transitive verb like investigate either subject or object may be referenced in the
first element of a compound. For a ditransitive verb like nominate, it is apparently possible for
any of the three arguments—subject (party nomination), object (candidate nomination), or
prepositional object (Emmy nomination)—to be referenced by the first element of the com-
pound. This is an important point to note, as there have been claims in the theoretical litera-
ture that the subject interpretation is not permitted (see Selkirk 1982: 34; Adams 2001: 78–9,
for example). We find to the contrary that the subject interpretation is often available, and
470 Compounds: semantic considerations

indeed occurs with some frequency in journalistic writing. This said, however, we should
acknowledge that forms in which the first element is interpreted as object or prepositional
object are clearly more frequent in the corpora than forms in which the first element receives
a subject interpretation. We note as well that the first element of an affixal compound need
not bear an argumental relationship to the second element, a point that we will return to in
Section 20.4.
Affixal argumental compounds in which the second element is a past participle are also
frequent. The first element may be interpreted as a subject or prepositional object of the
second, but cannot receive an object interpretation.

(4) a. subject referencing: teacher written, sun baked

Journal of Instructional Psychology 2004: Teacher written tests can be quite valid in

testing which cover what was taught.
Saturday Evening Post 2001: their verdure now beautifully shaded the quaint, narrow

lanes, and transformed into cool wooded roads what once had been only barren sun
baked wastes.
prepositional object referencing: achievement related, arsenic exposed
b. 
Journal of Sports Behavior 1994: Subsequent research investigated causal attributions

predominantly within achievement related contexts
Journal of Environmental Health 2008: Results We found nine relevant studies within

three general domains: 1) four studies begin with arsenic exposed populations and
seek excess childhood cancer,

As with other argumental compounds, the first element can also be interpreted as something
other than an argument, a point to which we will return in Section 20.4.
Let us now turn to non-affixal deverbal compounds, that is to those where the argument-
taking head is a noun that can be analysed as being derived from a verb by conversion (e.g.
government claim). As with the affixal argumental compounds we discussed above, with non-
affixal argumental compounds a wide range of interpretations may be found, interpretations
again being highly dependent on context.
We give examples of compounds with subject, object, and prepositional object orientation in
(5). The examples in (5) and (6) are taken from Lieber (2010: 129–30), which is itself a compila-
tion of examples from Jespersen (1942), Marchand (1969), Bauer and Huddleston (2002),
Jackendoff (2009), Plag et al. (2008), Bauer and Renouf (2001), the Morbocomp corpus (http://
morbocomp.sslmit.unibo.it), and Lieber’s own collection from American newspapers. Given the
coding available in corpora like COCA and the BNC, it is unfortunately not possible to amass
examples of these compounds systematically, as it is with items containing overt affixes.

(5) a. subject referencing: bee sting, bellyache, brain bleed, bus stop, cloud burst, daybreak,
dog attack, dogfight, earthquake, eyewink, flea bite, foot step, footfall, frostbite,
 20.2 Argumental compounds 471

government claim, government collapse, ground-swell, headache, heartbeat,

heartbreak, heartburn, land slide, moonrise, mouse squeak, nightfall, nosebleed,
plane crash, rainfall, sunburn, sunrise, sunset, sunshine, thunderclap, troop advance,
waterfall
object referencing: age limit, air traffic control, ball kick, birth control, blood test,
b. 
bloodshed, bodyguard, car park, clambake, cost control, court reform, dress design,
energy audit, fare increase, fee hike, funding increase, gun control, haircut, handshake,
heart attack, manslaughter, rate hike, robot repair, spending cut, sun worship, tax
hike, wind break
c. prepositional object referencing: baby care, linguistics lecture, peace talks, tax vote

Lieber (2010) claims that it is far easier to find the subject interpretation among non-affixal
argumental compounds than it is with affixal argumental compounds. We tested this claim
with two different data sets. The first set is 324 argumental compounds taken from CELEX as
used in Plag et al. (2007). These compounds are largely taken from dictionaries. The second
data set consists of the 641 argumental compounds found in the Boston University Radio
Speech Corpus as used in Plag et al. (2008) and Kunter (2011). The most striking finding is
that, with ‑er compounds, subject interpretations are completely absent from either data set
(376 ‑er compounds all together). Subject interpretations are, however, not infrequently
found with all other types of compounds, with varying proportions across subtypes and data
sets. For example, among non-affixal argumental compounds we find 43 per cent subject
interpretations in the CELEX data and 14 per cent in the Boston corpus data. Among com-
pounds ending in ‑ation or ‑ing there are 30 per cent and 20 per cent, respectively, with a
subject interpretation in the Boston corpus (CELEX has too few of these types of compound
to allow for any meaningful quantitative analysis). In sum, the quantitative analysis has shown
that subject interpretations can and do arise with all morphological categories whose seman-
tics allows for such interpretations.
It should be noted that for many argumental compounds non-argumental interpretations
are also readily available. For example, government claim could in principle also refer to a
claim sent to the government (non-argumental interpretation) instead of to a claim made by
the government (subject argument interpretation). Any of the forms cited in (5) could thus
also be cited as an example of non-argumental readings. We cite them as argumental because
the argumental interpretation is at least available.

20.2.2 Argumental verbal compounds

We have seen in the previous chapter that in many cases we can analyse verbal compounds as
back-formations from compounds whose second elements are either deverbal nouns in ‑er or
‑ing, or participles in ‑ed. In such cases the first element in the verbal compound is interpreted
in whatever way the first element would be in the corresponding deverbal compound. So if to
472 Compounds: semantic considerations

birdwatch is a back-formation from birdwatcher or birdwatching, the element bird receives the
same object interpretation in all cases. If carbon-date is a back-formation from carbon dating,
then the carbon receives the same interpretation in both cases. That is, the verbal compounds
can be classed as argumental or non-argumental, just as their sources can. Further, to the
extent that we allow for the direct coining of verbs where there does not seem to be an obvi-
ous pre-existing ‑ing or ‑er form (for example, blind bake, sugarcoat, blow-dry) we can still
classify the resulting compounds as argumental or non-argumental on the basis of the rela-
tionship displayed between the verbal second element and the first element.
One interesting point might be noted, however. Whereas we have seen that the first ele-
ments in typical argumental compounds can express either subject, object, or prepositional
object interpretations, we seem not to find subject interpretations in verbal back-formations.
That is, although we have found that compounds like administration refusal or army investiga-
tion in which the first element is interpreted as subject do exist, to our knowledge we do not
find back-formations like to administration refuse or to army investigate. It is unclear whether
this state of affairs is simply the consequence of the low proportion of subject interpretations
among affixal argumental compounds, or whether we are dealing with a true fact that needs
explanation.

20.2.3 Argumental compounds with relational nouns and

argument-taking adjectives as heads
There is yet another class of argumental compounds that is rarely discussed in the literature
but deserves mentioning here. Apart from verbs, there are also nouns and adjectives that can
take arguments. Such nouns are also known under the label ‘relational’ (as against ‘sortal’
nouns). According to Löbner (1985: 292), sortal nouns classify objects while relational nouns
describe objects in their relation to other objects. Examples of relational nouns are birth,
bride, distance, head, status, surface, victim. Similarly, adjectives can take arguments as exem-
plified by free, prone, or proud. Such argument-taking nouns or adjectives freely occur as heads
in argumental compounds. Consider, for example, the adjectival compounds fat-free, freckle-
free, toll-free, divorce-prone, drug-prone, storm-prone, America-proud, fraternity-proud, neighbor-
hood-proud. Analogous formations with relational nouns as heads are child birth, company
birth, galaxy birth, club member, university member, village member, glacier surface, moon surface,
pond surface, cancer victim, crime victim, fire victim.

20.2.4 Non-head is argument-taking

In the preceding sections we dealt with argumental compounds where the first element—the
non-head—functions as an argument to the second element—the head. There is also, how-
ever, a type of non-affixal argumental compound in which it is the second element
 20.2 Argumental compounds 473

that functions as an argument to the first element. As we do not find compound-internal

inflection in English, it is usually not possible to determine the categorial status of the first
element in this type of non-affixal argumental compounds. In most cases we have first ele-
ments that could be analysed equally well as verbs or as (relational) nouns derived from
those verbs. In a few cases, we do find first elements for which a nominal conversion form is
unattested in the relevant sense (e.g. go in go cart), but in most cases either analysis seems
consistent with the data. We will therefore refrain here from deciding on this matter. What
argument is expressed in the compound again depends on context and on the verbal diathe-
sis of the first element. We did not find any prepositional verbs that participate in this type of
formation.

(6) a. subject referencing: attack dog, blowtorch, call bird, clamp screw, cover letter, cry baby,
drag man, driftwood, finish coat, flashlight, go cart, jump jet, punch press, rattlesnake,
rip saw, screech owl, scrub woman, slide rule, spark plug, stop watch, tow truck,
tugboat, watch dog
object referencing: bore hole, call girl, draw string, drawbridge, dropcloth, jump rope
b. 
(NAmE), kick ball, kickstand, mincemeat, punch card, push boat, push cart, rip cord,
row boat, show bread, showplace, skim milk, throw stick, tow net

Note that compounds of this type are easily interpreted as non-argumental, especially when
the head is a sortal noun, and as such all the usual interpetations of NN compounds become
available. For example, a rattlesnake may be so called because it rattles (subject-referencing
argument interpretation), or because it has a rattle (non-argumental interpretation). The
same problem holds for some compounds with a base ending in the suffix ‑ing (e.g. manufac-
turing plant, ramming device, heating system, all from the Boston University Radio Speech
Corpus).
None of these interpretations is necessarily ‘correct’, and any of these could also be cited
as examples of non-argumental compounds. We put them here because the argumental inter-
pretation is at least available, while for root compounds that have sortal nouns as both heads
and non-heads an argumental interpretation is not available. See Section 20.3.1 for a discus-
sion of verb–noun non-argumental compounds.

20.2.5 Exocentric examples

English has only a small cohort of argumental compounds that are exocentric, apparently all
of them of the non-affixal variety. Among these are examples like pickpocket, cutpurse, scare-
crow, tattletale, killjoy, scofflaw, know-nothing, turncoat, spoilsport, breakwater, rotgut, turnkey
(with the now archaic meaning of ‘jailer’), shearwater, stopgap. The majority of them denote
types of people, animal, or object. According to Marchand (1969: 380–1), which contains the
most exhaustive listing of such compounds in the literature, these were formed on a pattern
474 Compounds: semantic considerations

of French agentive and instrumental compounds that remains productive today in many of
the Romance languages. Although Marchand (1969: 382) claims that this pattern ‘has proved
exceedingly productive in English’, this is certainly not true in the contemporary language. It
seems doubtful that any new examples are being coined, and indeed, many of the examples
that Marchand cites have become archaic or obsolete.

20.3 Non-argumental compounds

Non-argumental compounds are those in which neither element of the compound denotes
an argument of the other. Among the non-argumental compounds we can make a basic divi-
sion between attributive and coordinative compounds. The former are compounds in
which the relationship between the first and second elements is one of loose modification;
this applies both to endocentric attributives, in which the second element is clearly the head
and the compound as a whole is a hyponym of the second element, and to exocentrics which
may be syntactically headed by their second element, but which are not hyponyms of those
elements. In coordinative compounds a more equal relationship between first and second
elements obtains, such that one could argue that this type of compound has more than one
semantic head (e.g. Haspelmath 2002: 89) or has no semantic head (e.g. Booij 2007: 80).
We will examine attributive compounds in Section 20.3.1 and coordinative compounds in
Section 20.3.2.

20.3.1 Attributive compounds

20.3.1.1 Endocentrics
Quite a few theorists have attempted to catalogue a finite set of relationships that can be
expressed in attributive compounds: see Lees (1960), Hatcher (1960), Marchand (1969),
Brekle (1970), Levi (1978), Ryder (1994), Jackendoff (2009) for just some of the attempts
since the mid-twentieth century. As is widely acknowledged, however, nearly any conceivable
relationship can obtain between the first and second elements of attributive compounds. The
relationships between the elements are fluid and frequently determined by context, except in
compounds that have lexicalized meanings (see for example Downing 1977). Consequently,
the fit between any given compound and any gloss from a fixed list is, in context, likely to be
approximate; indeed, the same compound in the same context can frequently be read in dif-
ferent ways, so that police dog could be interpreted as a dog used by the police or a dog work-
ing for the police or a dog trained for police work.
That said, we should acknowledge that the nature of the compounded elements and espe-
cially of the second element does have some effect in circumscribing the range of meanings
conceivable for any given attributive compound. If the second element is a relational noun,
 20.3 Non-argumental compounds 475

for example, a body part noun like leg or arm, or a kinship noun like mommy or father, the first
element of the compound is likely to express the expected argumental relationship ‘leg of X’,
‘mommy of Y’ as we see in compounds like table leg or rabbit mommy, and in the vast majority
of other compounds with leg or mommy that can be found in COCA. But it does not have to
express this relationship, as the following leg compounds and their meanings in context from
COCA illustrate.

(7) aluminum leg ‘made of aluminum’

back leg ‘located at the back’
bench leg ‘human leg that is located on bench’
bow leg ‘shaped like bow’
cigarette leg ‘shaped like cigarette’
fishnet leg ‘covered in fishnet’
polio leg ‘affected by polio’
summer leg ‘cosmetically suitable for summer’
support leg ‘intended for support’

Similarly, if the second element is an adjective like crazy or friendly that can express a relation-
ship (‘X is crazy about Y’, ‘X is friendly to Y’), the first element is likely to express the expected
relation, as we find in trout-crazy or beginner-friendly, although others are conceivable
(for instance, compounds like angry-crazy or goofy-crazy that can be found in COCA).
The less relational the second element, the wider the range of meanings conceivable
between the first and second elements of the compound. For example, with a sortal count
noun like bed in its central meaning as a piece of furniture in which one sleeps, we find the
following relationships exhibited by compounds attested in COCA. Alternative paraphrases
are of course possible and nothing hinges on any particular paraphrase.

(8) adult bed ‘for an adult’

air bed ‘filled with air’
alcove bed ‘located in alcove’
army bed ‘used by the army’
bamboo bed ‘made of bamboo’
birth bed ‘place where birth takes place’
bunk bed ‘with bunks’
custom bed ‘custom made’
day bed ‘for use during the day’
designer bed ‘made by a designer’
dream bed ‘that you dream about’
476 Compounds: semantic considerations

emergency bed ‘for use in an emergency’

fairy-tale bed ‘like something in a fairy-tale’
foldout bed ‘that folds out’
heirloom bed ‘that is an heirloom’
marshmallow bed ‘that is soft like a marshmallow’
observation bed ‘where one undergoes observation’
sex bed ‘in which one has sex’
Shaker bed ‘from a Shaker design, in the Shaker style’
sofa bed ‘that is also a sofa’
tanning bed ‘designed to allow the tanning of skin’
travel bed ‘intended to be used while traveling’

It seems safe to say that given an appropriate context just about any relationship is conceiva-
ble. However, some relationships between the first and second elements are more predictable
or pragmatically more likely (see Štekauer 2005). In particular, it has been shown in psy-
cholinguistic experiments that new compounds tend to be interpreted with the same mean-
ing relationship as other compounds with the same head or non-head (Gagné and Shoben
1997 and Gagné 2001, see also Chapter 23).
Basically the same general principles of interpretation apply to endocentric adjective–
noun compounds. Thus we find very straightforward cases of adjectival modification, as
in sweetcorn, which is a sweet-flavoured kind of corn, or brownstone, which denotes a dark-
brown sandstone. But other adjective–noun compounds show that less straightforward
relationships are also readily available, as in silly-season, which is the name of a period of
a time when newspapers publish articles on trivial, silly matters due to the scarcity of
real news.
A special case of adjective–noun compounds are those in which the adjective is rela-
tional in nature, as in medical school, social worker, solar system. Here the head of the com-
pound is semantically associated with the base noun of the adjectival modifier, and many
readings become available in principle. The correct interpretation requires world knowl-
edge, which again stresses the similarity of adjective–noun compounds and noun–noun
compounds.
A final group of non-argumental nominal compounds are those that can be analysed as
having a verb in the first position. As already discussed in Section 20.2.4 above the categorial
status of the first element as either verb or noun is often impossible to decide, but at least in
some cases the pertinent noun is unattested in the relevant sense (e.g. surf in surf lesson, board
in board pass), which rules out a consistent treatment of this pattern as noun–noun. The
pattern is illustrated in (9), with data from Dollinger (2008) and from COCA (not all of
these are usual in all varieties of English).
 20.3 Non-argumental compounds 477

(9) call card, carry bag, disappear act, dive suit, drain board, draw power, dress gown, drill
platform, drill rig, drink box, dump truck, finish line, frypan, patch kit, sail boat, ship
town, surf lesson, swim meet, wait time

The interpretation of these compounds works along the same lines as that of other nominal
non-argumental compounds. Any type of semantic relationship seems possible, based on the
meaning of the constituents and pragmatic considerations. A very common interpretation is
instrumental, with the head denoting an entity with which the action denoted by the verb is
performed (e.g. carry bag, dive suit, drill rig), but many other types of interpretation can be
found. A disappear act is an act in which someone or something disappears, a finish line is a
line where something finishes, and so on. Note that non-argumental compounds with a ver-
bal non-head may sometimes have rival forms with a verb ending in ‑ing as non-head (e.g.
dressing gown, drilling platform, see Chapter 19). The latter type of compound behaves gener-
ally like other noun–noun compounds, as illustrated in (8) above (cf. tanning bed), and
appears from examples in the BNC to give rise preferentially to a purpose meaning.
With compounds consisting of two verbs, we can get an interpretation in which the first
verb denotes the manner in which the second verb is performed. So to blow-dry is to dry by
blowing, and to trickle-irrigate is to irrigate by trickling. Also in this category we might put
dive-bomb, freeze-dry, skim-read, and slam-dunk. As we saw in Chapter 13, we also find verbal
compounds in English that are created by conversion of nominal compounds, for example,
blacklist, break dance, carbon-copy, cold shoulder, end-run, lipstick, litmus-test, machine-gun,
pitchfork. If we analyse these as typical cases of noun to verb conversion, we would expect
them to display the same wide range of interpretations we find with noun to verb conver-
sion with non-compound nouns (see Chapter 13). This is indeed what we find: so blacklist is
locative (‘to put on a blacklist’), carbon-copy is resultative (‘to make a carbon copy’), break
dance or end-run performative (‘to do a breakdance or endrun’), lipstick ornative (‘to put
lipstick on’), machine gun or pitchfork instrumental (‘to use a machine gun or pitchfork’),
and so on.
Let us turn to attributive compounds whose second element is adjectival. When the sec-
ond element of the compound is a qualitative adjective the relationship between the first and
second elements is frequently one of similitude (for example, butter yellow, blizzard cold), but
it need not be. We find other relationships expressed as well: ‘compared to’ (ankle high),
‘composed of ’ (cadmium yellow), ‘in relation to’ (color blind). It appears that the relationship
between first and second elements is no more fixed in adjectival compounds than in nominal
ones.
It is of course possible in English to find compounds composed of more than two ele-
ments, and in such the opportunity arises for multiple interpretations. The semantic inter-
pretation of multi-element compounds depends on the bracketing of the elements. In a
compound like toy poodle groomer manual, at least four interpretations are possible, corre-
sponding to four different syntactic bracketings.
478 Compounds: semantic considerations

(10) a. ‘a manual for groomers of toy poodles’

[[[toy poodle] groomer] manual]
b. ‘a groomer manual for toy poodles’
[[toy poodle][groomer manual]]
c. ‘a poodle groomer manual that is a toy’
[toy [[poodle groomer] manual]]
d. ‘a manual for toy groomers of poodles’
[[toy [poodle groomer]] manual]

Granted that some interpretations are more plausible pragmatically than others, none of
these can be ruled out a priori.

20.3.1.2 Exocentrics
Exocentric attributives are highly productive in English. The bahuvrihi or ‘possessive’
type are all metonymic or metaphorical in nature. Typical of such compounds are the
examples in (11) with the second element head, the majority of them taken from
COCA.

(11) a. stupid, annoying, or malevolent people: air head, bastard head, blockhead, bone head,
bubble head, dick head, egg head, fuck head, hot head, knuckle head, meat head,
penis head, pin head, poop head, shit head
b. people addicted to various substances: acid head, crack head, dope head, hash head,
helium head, java head, pot head, smack head
people with obsessive interest in something: football head, baseball head, cheesehead,
c. 
gear head, granola head, muscle head, petrol head (BrE), radio head
d. offensive terms for minorities: handkerchief head, towel head
e. animal or plant name: steel head (trout), fiddle head (fern), butter head (lettuce)
f. inanimate object: bobble head (figurine)
g. miscellaneous others: redhead, bedhead, blackhead

So, a meat head is someone ‘having a head like meat’, a gear head someone ‘whose head is full
of gear’ (that is, in colloquial NAmE, someone with obsessive interest in equipment or gear
connected with a particular sport or hobby), a fiddle head a kind of fern whose emerging
shoot looks like the top of a fiddle, and so on. The main difference between these and the
endocentric attributive compounds is simply that the characterization of their referents is
figurative, often being metaphorical or metonymic. Thus, an analysis is possible that treats
them as regular endocentric compounds with a metonymic or metaphorical reading of the
 20.3 Non-argumental compounds 479

head noun. Many compounds are not to be taken literally, and appear to fit the definition of
an exocentric as a result: elbow grease, mind fuck, rug rat.
A different kind of exocentricity can be found in cases of ellipsis. There are, for example,
binary compounds that have originated through ellipsis of a final third base, as in roll-neck for
roll-neck sweater/shirt/jersey/pullover, as shown in (12), from the BNC.

CDX W_biography 1991: I knew she didn’t like my roll-neck pullover.

(12) 
BMW W_fict_prose 1991: she recognised the slight figure in black roll-neck sweater

and skin tight pants.
AMU W_fict_prose 1992: spare combat jacket and two pairs of jungle trousers; for night

work, black cotton roll-neck, black tracksuit bottoms and black cotton gloves; slacks
and a light jumper for Mariana.
BMW W_fict_prose 1991: Never a big man, overnight his frame seemed to have become

almost frail and the sinews in his neck were raised and stringy above the cotton roll-
neck.

Other examples are hardtop (a car), house-warming (a party), turtleneck (a garment, NAmE),
or underground (a railway). It seems that this kind of exocentricity is a special case of a much
wider pattern of metonymy that affects both complex and simplex lexical items. Many two-
member compounds also lose their final element and thus become metonymic expressions,
as in business for business class, chair for chairperson, Northwestern for Northwestern University,
the Tate for the Tate Gallery,Tasman for Tasman Sea, vacuum for vacuum cleaner. The semantic
consequences of these kinds of ellipsis in compounds seem no different than the metonymy
that arises from syntactic ellipsis (as in the House for the House of Representatives) or even
without ellipsis, as in Washington for the government residing in Washington.

20.3.2 Coordinative compounds

Coordinative compounds are ones in which the relationship between the first and second
elements is not one of hyponymy but rather one in which the two elements bear equal
semantic weight. As with the other major categories of compounds coordinative compounds
may be either endocentric or exocentric. As Bauer (2008) argues, within each of these sub-
classes a number of different equative relationships can obtain. Coordinative compounds can
be appositional, additive, or compromise, and we will discuss each type in turn. Examples in
this section are taken from Olsen (2001) and Bauer (2008), and from the internet.
Appositional compounds, as given in (13), are compounds which refer to a single indi-
vidual or object that represents the intersection of two sets: a singer-songwriter is a person
who is both a singer and a songwriter, a scholar-athlete someone who is at the same time a
scholar and an athlete. Appositional compounds are not limited to two nouns, as compounds
480 Compounds: semantic considerations

like actor-producer-director are entirely conceivable.

(13) singer-songwriter, scholar-athlete, advertiser publisher, tent-office, comedy-drama,

nerd-genius

The second type of coordinative compound that is highly productive in contemporary

English is what we might call additive, again following Bauer (2008). Among these are
many that consist of proper nouns, including coordinated place and company names.

(14) Austro-Hungary, Alsace-Lorraine, Hewlett-Packard

Such names denote a merger of territory or business, and to them we might add the
hyphenated surnames, as for example, the name Koslosky-Pappafilovich that the couple
might adopt when Ms Koslosky marries Mr Pappafilovich. Additive compounds of this
sort are of course not limited to two proper nouns. Especially with company names we find
coordinations of more than two names, as for example in the corporate name AOL Time
Warner.
Additive compounds are also possible with common nouns or adjectives, although our
impression is that such compounds are less frequent in English than appositional compounds
or additive compounds formed on the basis of proper nouns.

(15) angry-crazy, cough-laugh, deaf-mute, historical-philosophical, goofy-crazy, murder-

suicide

The distinction between these and the appositional compounds is delicate, and sometimes
not entirely clear. Our instinct is that in the additive type we have two events, objects, or
qualities bundled or conceptualized together whereas in the appositional type we are think-
ing of a single item with two distinct labels. So a murder-suicide is a murder closely followed
by a suicide, a cough-laugh could probably be interpreted as a noise that muffles a laugh with
a cough, deaf-mute a state of being both unable to hear and unable to speak.
To the extent that we find verbal compounds in the native speaker’s mental lexicon, we can
offer a semantic analysis. In the few cases where the compound verb is made up of two verbs
(for example, blow-dry, dive-bomb, shrink-wrap), we find at least two sorts of interpretation,
one of which is a coordinative interpretation in which two actions are performed simultane-
ously or sequentially (cf. drink-drive, stir-fry, sleep-walk), the other being resultative, as in
blow-dry. The borderline between the two types is often unclear: is freeze-dry ‘to dry by freez-
ing’ or ‘simultaneously to freeze and to dry’? Pragmatically there is little difference, and both
classifications are found (see Renner 2008; Bauer 2010b).
A final type of coordinative compound is what Bauer (2008) terms compromise coordi-
natives. These are the type represented by adjectival compounds like blue-green, and nominal
 20.3 Non-argumental compounds 481

compounds like northeast, which designate an intermediate property, for example, between
blue and green or between north and east. This type of compound is limited pragmatically
by the nature of objects, qualities, and events: there are ontological limits to what can be
blended or seen as a compromise between two entities. But where pragmatic considerations
allow, we do find such compounds formed rather productively, as might be evidenced by
names for mixed breed (or ‘designer’) dogs such as Afghan spaniel, Akita shepherd, Cocker
pug, and the like (although the vast majority of these seem to be blends, see Section 20.5.1
below).
There are also exocentric coordinative compounds, which denote a relationship between
the elements. This can be seen most clearly in compounds like Arab-Israeli (conflict) where
the compound as a whole does not denote a merger or aggregate of Arabs and Israelis, or
some sort of mixture of the two, but rather a relationship between two separate entitities.
Such compounds can consist of two adjectives, as with Arab–Israeli (conflict) or Dutch–English
(dictionary), or of two nouns (father–daughter [dance], Boston–Philadelphia [flight]) or even
two verbs (pass–fail [test]). In the latter cases the compounds are functionally as well as
semantically exocentric: the majority of exocentric coordinative compounds, regardless of
categorial composition, function as premodifiers of nouns.
The precise relationship expressed between the two compounded elements depends to
some extent on the noun that they are predicated of. We might distinguish three types of
relationship. In the conjunctive type, the relationship between the first and second ele-
ments of the compounds is one of aggregation, for example, both father and daughter partici-
pate in a father–daughter dance; this type includes what Bauer (2008) calls ‘co-participant’
coordinative compounds. In the translative type, the relationship between the first and
second elements is one of ‘between-ness’: a Boston–Philadelphia flight is one between Boston
and Philadelphia, a mother–daughter conversation is one between mother and daughter. The
final type is one that we might call disjunctive (this is one not mentioned by Bauer 2008)
in which the relationship between the first and second elements of the compounds is one of
mutual exclusivity; we have in mind here compounds like pass–fail (test) or true–false
(question).

(16) a. conjunctive: father–daughter (dance), brother–sister (duet), fall–winter (collection),

Pitt–Aniston (wedding), love–hate (feelings)
b. translative: Arab–Israeli (conflict), Dutch–English (dictionary), Boston–Philadelphia
(flight), mind–brain (connection), mother–daughter (conversation), Utah–Arizona
(border), subject–verb (agreement), hand–eye (coordination)
c. disjunctive: pass–fail (test), true–false (question), push–pull (gizmo)

Above we said that the majority of exocentric coordinative compounds function as premodi-
fiers, but not all do. Consider, for example, the compound push–pull as it is used in (17).
482 Compounds: semantic considerations

People 2007: What Wife gets right is the push–pull of marriage, the way the comfort of
(17) 
the familiar can butt up against the siren call of the new.

Here, the coordinative compound is both semantically and categorially exocentric, but does
not function as a premodifier. It is reminiscent of the sort of dvandva compounds more fre-
quently found in other languages (see for example, Wälchli 2005).

20.4 The border between argumental

and non-argumental compounds
The classificatory schema that we have followed in the previous two sections covers the
majority of compound types to be found in contemporary English. As with most classifica-
tory systems, however, there are potential problems.
One problem may be the boundary between arguments and adjuncts. In our discussion we
have treated compounds as argumental if the first element is interpreted as a core argument of
the verb, that is, as subject, object, or prepositional object. In those cases where the first ele-
ment would be considered an adjunct, we treat the compounds as non-argumental. We list
some of those cases in (18). In all of these examples the first element receives an interpretation
that would be expressed syntactically by an adjunct phrase rather than a core argument. Note
that we intend the labels ‘location’, ‘time’ etc. to refer to possible but not necessary readings.

(18) a. location: aerial observer, airport parking, alley fighter, backyard gardener, bakery
worker, vine-ripened
b. time: advance preparation, after-hours trading, afternoon trainer
c. status: absentee landowner
d. concerned with: account officer, AIDS meeting
e. purpose: acne medication, asthma inhaler, backup singer
f. instrument: air popper, arsenic poisoning, hand-washed
g. containing: action thriller, adventure education
h. made of: adobe construction, almond filling, bamboo steamer
i. caused by: acid indigestion
j. comprised of: activist organization

Just about any relationship between first and second elements is conceivable when the first
element is not interpreted as a core argument, and the relationship is largely determined by
context. Indeed, the range of interpretations available for such compounds is unlimited, just
 20.5 The semantics of blends and neoclassical compounds 483

as we find with attributive non-argumental compounds. We take the position here that a
compound should be considered argumental only if one element may be interpreted as a
core argument of the other. Certain borderline cases may exist, but this is unavoidable in any
classificatory system.
Another problem arises with compounds that have relational nouns as heads. Although
we have used a seemingly clear definition of how argument-taking nouns, that is relational
nouns, can be distinguished from non-argument-taking nouns, that is sortal nouns, we
acknowledge the existence of cases where the boundary is not so clear.

20.5 The semantics of blends

and neoclassical compounds
In Chapter 19 we treated the formal characteristics of blends and neoclassical compounds.
Here, we look at their semantic characteristics, which are to a large extent similar to those of
compounds, especially of non-argumental compounds. We begin with blends and proceed to
neoclassical compounds, keeping in mind that the dividing line between the two types is
sometimes not entirely clear.

20.5.1 The semantics of blends

The vast majority of blends in English display the semantics of non-argumental compounds,
with both attributive and coordinative types amply attested. Attributives are illustrated
in (19). All examples of blends in this section are taken from Wordspy (www.wordspy.com).

(19) attributive blends

agritourist ◆ (agriculture + tourist)
beersicle ◆ (beer + popsicle)
carbage ◆ (car + garbage)
daycation ◆ (day + vacation)
dormcest ◆ (dorm + incest)
foodoir ◆ (food + memoir)
menoporsche ◆ (menopause + Porsche)

In semantically attributive blends, as in attributive compounds, the blend as a whole is a

hyponym of the second blended element, and the first element bears some contextually
plausible relationship to the second. So a beersicle is a popsicle made of beer, carbage is ­garbage
located in one’s car, and a daycation is a one-day holiday.
For blends that have coordinative interpretations, we find both appositive and compro-
mise types. Note that for some of these an endocentric reading is also possible.
484 Compounds: semantic considerations

(20) a. appositive coordinative blends

actorvist ◆ (actor + activist)
aireoke ◆ (air guitar + karaoke)
celesbian ◆ (celebrity + lesbian)
fictomercial ◆ (fiction + commercial)
b. compromise coordinative blends
avoision ◆ (avoidance + evasion)
broccoflower ◆ (broccoli + cauliflower)
chofa ◆ (chair + sofa)
freshmore ◆ (freshman + sophomore)
puggle ◆ (pug + beagle)

As with compounds of this sort, the appositives denote the intersection of two types of entity
or action. So an actorvist is both an actor and an activist, and aireoke is the simultaneous
action of playing air guitar and singing karaoke. The compromise coordinatives denote hybrid
entities or concepts. A chofa is a piece of furniture somewhere between a chair and a sofa, a
puggle a cross between a pug and a beagle. The creation of names for new mixed breed dogs
through blending is in fact highly productive (see, for example, http://www.dogbreedinfo.
com/hybriddogs.htm), but blending can also be found with other kinds of hybrids in the
animal kingdom (e.g. jagupard, liguar, tigon, zebrule, zorse). There seems to exist also a con-
vention to place the dam’s name in final position in such blends (see, for example, the
Wikipedia entries Panthera hybrid and Zebra). If people do this, or to the extent they do this,
such formations may be considered no longer strictly coordinative.
Only rarely do we find argumental compounds among blends, although there does not
seem to be any principled reason why this should be the case. In the Wordspy corpus, we find
a few examples, among them the ones in (21).

(21) argumental blends

agrimation ◆ (agriculture + automation) ‘automation of agriculture’
bitlegging ◆ (bit + bootlegging) ‘bootlegging of software’
buttlegging ◆ (butt + bootlegging) ‘bootlegging of cigarettes’
Coca Colonization ◆ (Coca Cola + colonization) ‘colonization by Coca Cola
and similar companies’
kidfluence ◆ (kid + influence) ‘influence by kids’
spamdexing ◆ (spam + indexing) ‘indexing spam’

As these examples illustrate, argumental blends are frequently affixal (agrimation, Coca
Colonization, spamdexing), and can be either object-referencing (agrimation, bitlegging, butt­
legging, spamdexing) or subject-referencing (Coca Colonization, kidfluence).
 20.5 The semantics of blends and neoclassical compounds 485

There is a small residue of blends that is difficult to characterize. Some blends can be rather
opaque, as we find with the blend Boyzilian (◆ boy + Brazilian), which is the name for a
bikini-wax for men. A few blends seem to be examples of idiosyncratic word play as much as
they are blends, for example Internot (◆ internet + not) which denotes a person who refuses
to use the internet or transwestite (◆ west + transvestite), which denotes a person who likes to
dress up as a cowboy. In such cases, some semantic element needs to be added to arrive at
the appropriate interpretation, in other cases, the intended meaning is clear. Nevertheless,
it seems safe to say that blends generally are interpreted in the same way that compounds are,
though not necessarily in the same proportions.

20.5.2 The semantics of neoclassical compounds

The interpretation of neoclassical compounds raises special issues. For one thing, in the case
of conventional compounds and the sorts of blends illustrated above, there is no particular
question as to the semantic status of the individual elements that make up the complex word:
they mean what they mean. For neoclassical combining forms, however, the question may
always be raised to what extent the individual elements are meaningful to native speakers.
Second, it is sometimes difficult to distinguish the elements that make up neoclassical com-
pounds from derivational affixes. Finally, as neoclassical combining forms have no clear cat-
egorial status, they present problems for our classificatory scheme, at least for the distinction
between argumental and non-argumental compounds. We will take up each of these issues
in turn.
As to the first, we might say that the forms that make up neoclassical compounds are
meaningful to at least some speakers of English. Indeed, some of them are quite common and
probably known to most speakers of English. Many speakers of English know that ‑osis has
something to do with disease or illness and neuro‑ to do with the nervous system or the brain,
for example. Other combining forms constitute part of the learned vocabulary of the lan-
guage and are often specific to particular scholarly disciplines. For those conversant in the
relevant disciplines such elements have as much meaning as free forms. For example, while
many English speakers know the meaning of the combining form neuro‑, far fewer know that
sarco‑ means ‘flesh’ or that ‑odynia means ‘pain’, but this is knowledge that we might expect
from doctors or biologists.
Nevertheless, neoclassical combining forms do sometimes present special semantic chal-
lenges, as McCray et al. (1988: 166) note. They cite the example of the initial combining form
leuko‑, which literally means ‘white’, and occurs in such neoclassical compounds as leukocyte,
leukotomy, and leukopathy. The issue here is that in each case what is modified by leuko‑ is left
implicit: in leukocyte, leuko‑ implicitly modifies ‘blood’ (leukocyte = white [blood] + cell), in
leukotomy it modifies part of the brain (leukotomy = white [part of brain] + cutting), and in
leukopathy, it implicitly modifies skin pigmentation (leukotomy = white [skin pigmentation]
486 Compounds: semantic considerations

+ disease). What we have here appears to be a sort of semantic ellipsis that is not predictable.
This kind of ellipsis can be found even with combining forms that are quite well-known, as
with the compound cardiophobia, which, as McCray et al. point out, means not ‘fear of hearts’
but ‘fear of heart disease’.
As for the dividing line between neoclassical combining forms and derivational affixes, it
must be acknowledged that no firm distinction may be drawn. As Bauer (1998b) has pointed
out, there is a cline from most semantically contentful to least semantically contentful bound
morphemes (see also Prćić 2005, 2008; Amiot and Dal 2007). At the more contentful end we
have what we would call neoclassical combining elements; these are the forms that are most
like independent lexical morphemes in meaning, having what Bauer (1983: 215) refers to as ‘a
higher density of lexical information’. At the less contentful end we have what we would tend
to call affixes. But there are many forms that fall somewhere between the two extremes.
Probably the thorniest issue for our purposes is whether it makes sense to fit neoclassical
compounds into the classificatory system we have used here, given that neoclassical combin-
ing forms have no syntactic category and therefore cannot be said to have arguments (or to
not have them) in the same way that free forms do. This is an open issue. If we argue purely
on the basis of semantics, as opposed to categorial status, it is possible to classify neoclassical
compounds according to the same schema we used for other compounds. But this analysis
comes with some cost.
Neoclassical compounds are typically right-headed, as other compounds and blends are;
that is, the compound as a whole is a hyponym of the second element. For the most part they
are endocentric. So far, our comparison is unproblematic. If we wish to further classify neo-
classical compounds, however, we must proceed strictly on the basis of ‘translation’. This
translation sometimes ends up with argument-taking deverbal nouns, as in (22a), and some-
times with sortal nouns, as in (22b).

(22) a. gastrostomy ◆ (gastro/stomach + stomy/opening)

neurogenic ◆ (neuro/brain + gen‑ic/producing)
neurology ◆ (neuro/brain + ology/study of)
bibliocide ◆ (biblio/book + cide/killing)
omnicide ◆ (omni/all + cide/killing)
b. gastrodynia ◆ (gastro/stomach + dynia/pain)
keratosis ◆ (kerato/cornea + osis/disease)
mycosis ◆ (myco/fungus + osis/disease)

The compounds in (22a) might be considered argumental insofar as we consider the transla-
tion of their second elements to be verbal or deverbal in nature, and the first element seems to
bear a relation like that of object to the second. So in omnicide, the second element translates
as a deverbal noun ‘killing’, and the first element as the erstwhile object of the verb. The neo-
classical compounds in (22b), on the other hand, have a relational noun as the second element
 20.6 Miscellaneous compounds 487

and can thus also be classified as argumental. Still, as McCray et al. (1988) point out, with
neoclassical compounds, as with other attributive compounds, it is frequently not p­ redictable
how the relationship between the first and second elements should be construed.
Following our classificatory schema, we find that some neoclassical compounds can have
coordinative interpretations, as we find, for example, in a compound like alveo-palatal, which
designates a point mid-way between the alveolar ridge and the palate, and therefore consti-
tutes a compromise coordinative. Or we can have an additive interpretation, as we find in
the compound laryngeotracheobronchitis, cited by McCray et al. (1988: 167) as meaning ‘an
inflamation of the larynx, trachea, and bronchi’. Indeed, as McCray et al. point out, parts of
neoclassical compounds can be coordinated in quite complicated ways. In acrocephalosyndac-
tyly, acrocephaly ‘high skull’ is coordinated with syndactyly ‘together (i.e. fused), digit’ which
denotes a condition characterized by both a high skull and fused digits, whereas in myoendo-
carditis, the elements myo and endo are coordinated with each other and modify carditis, so
the whole denotes ‘inflamation of the muscle and membrane of the heart’.
So far, we have looked at neoclassical compounds that are composed entirely of neoclassi-
cal combining forms, but of course, combining forms can be compounded with free mor-
phemes, as we find in the examples in (23).

(23) a. gastropub (gastro/stomach + pub)

gastro-porn (gastro/stomach + porn)
neuro-chip (neuro/brain + chip)
neuro-jargon (neuro/brain + jargon)
assholiosis (asshole + osis/disease)
b. bitchicide (bitch + cide/killing)
roboticide (robot + cide/killing)
oceanaut (ocean + naut/traveler)
neuro-numbing (neuro/brain + numbing)
neuroadaptations (neuro/brain + adaptations)

Semantically, these behave much the same as both pure neoclassical compounds and com-
pounds composed entirely of free morphemes. They are largely right-headed and endocen-
tric, and can be classified as argumental or non-argumental in the same way that we classified
the purely neoclassical compounds above.

20.6 Miscellaneous compounds

There are a number of types of compound or compound-like phenomena that we have not
yet discussed, either because they are of only marginal productivity, or because their status
488 Compounds: semantic considerations

as compounds has been disputed. We take up those types here (see again Chapter 19 for
discussion of their morphological status).

20.6.1 Phrasal compounds

Phrasal compounds are highly productive in contemporary English, as the examples in (24)
from COCA suggest.

(24) anybody-but-Bush syndrome

black-hole-in-a-bag syndrome
boy-on-a-bike syndrome
breaking-out-in-song syndrome
catch-as-catch-can syndrome
dad-needs-a-sports-car syndrome
failed-back-surgery syndrome
fear-of-failure syndrome
first-to-know syndrome
get-rich-quick syndrome
grass-is-greener syndrome
guilt-gets-grants syndrome
guitar-as-phallus syndrome
holier-than-thou syndrome
I’m-OK-you’re-not syndrome
it-wasn’t-me syndrome
look-at-me syndrome
not-in-my-backyard syndrome
out-of-the-nest syndrome
sheriff-of-Nottingham syndrome
staying-young syndrome
too-much-of-a-good-thing syndrome
tortoise-and-hare syndrome
use-it-or-lose-it syndrome
woman-behind-the-man syndrome

Although rarely, if ever, recorded in dictionaries (because their token-frequency is generally

low), such compounds can be frequently found in informal writing, as in the following exam-
ples from a novel.
 20.6 Miscellaneous compounds 489

(25) He was the groundsman, handyman, if-there’s-any-sort-of-difficulty-ask-William-

and-he’ll-fix-it-for-you person about the place. (Meynell 1978, p. 10)
We’ve got a what-the-unions-will-allow-us-to-print press. (p. 80)
The old manage-somehow-on-a-shoestring days were definitely gone (p. 125)

The examples in (24) and (25) illustrate two important semantic properties of phrasal
compounds. First, the phrasal part of the compound is not in any way restricted to lexical-
ized, institutionalized, or idiomatic phrases (although such phrases are of course not ruled
out), contrary to claims in the literature (see, for example, Bresnan and Mchombo 1995).
Second, although in some examples, the phrasal element has a clear quotative flavour
(anybody-but-Bush syndrome, it wasn’t me syndrome), more often than not there is no real
justification for calling this element quotative, as both Bresnan and Mchombo (1995) and
Wiese (1996a) do.
The interpretation of phrasal compounds is in most cases that of endocentric attributive
compounds. The initial phrasal element bears some plausible relationship to the second ele-
ment, with the relationship often being obvious only in context.

(26) hour-by-hour gossip (time)

quick-to-spread gossip (manner)
below-the-stairs gossip (location)
hide-and-seek house (function)
stone-and-redwood home (material)

In other words we expect to find the same range of interpretations in phrasal compounds that
we find in non-phrasal endocentric attributive compounds.
Phrasal compounds with argumental interpretations are harder to find, but they do seem
to exist. Consider the examples in (27), again from COCA.

(27) limits-and-consequences conversation

burned-rope-and-sugar taste
letters-to-the-editor writer

In the first example, we might say that the phrase limits-and-consequences bears the preposi-
tional object relation to the deverbal noun conversation; similarly for the second example. The
phrase in the third example seems clearly to be interpreted as object of the verb write.
We do not, however, seem to find phrasal compounds with a coordinative interpretation,
but that should not be surprising. Coordinative compounds are characterized by the juxtapo-
sition of the same kinds of formal entity, that is nouns, or verbs, or adjectives. For phrasal
490 Compounds: semantic considerations

compounds this would mean that both compound constituents be phrases. However, as
discussed in Chapter 19, phrases are generally ruled out in morphosyntactic head position.

20.6.2 Reduplicative compounds

Reduplicative compounds are frequently found in colloquial spoken English, and in more
informal genres of writing, and only infrequently in formal writing. Examples like the follow-
ing from COCA are typical.

Sports Illustrated 1992: John then fires another of his surprise questions: “When was the
(28) 
last time you read a book?” She pauses in mid-grip to reply, “You mean a book book?
Not having anything to do with sports? In college, I guess. That was three years ago.”
Triquarterly 1990: “I have to go home,” I said when she woke up. She thought I meant

home to her house in the Canyon, and I had to say No, home home.
New England Review 2002: I asked her if she wanted to go have a drink afterwards and

she said Drink! and I said it didn’t have to be a drink drink (I swear people who
overheard us sometimes must have thought we were retarded) and she said Sorry?
Lawrence Block 2000: “Mine is green.” “Green green? Or more like an olive green?”

Tamara Leigh 2008: “Yes, hot!” The “ruched” young woman jabs the air again, looks

around, and startles. “Er, not hot hot. Hot, as in under the collar. “Ticked off.” // That’s
my cue to appear relieved that she didn’t mean hot as in carnal, as she’s obviously
connected to this company – at least to the receptionist.

To the extent that we consider reduplicative compounds to be compounds, we must acknowl-

edge that their semantic properties are a bit different than the compounds we have looked at
thus far. They appear to be endocentric, with the compound as a whole being a hyponym of
the head. However, as pointed out by Hohenhaus (2004) and Ghomeshi et al. (2004), for the
most part, such compounds have a standard interpretation in that they serve to pick out or
identify a prototypical exemplar of the second element of the compound. So a book book
would be something with more pages than a pamphlet or a magazine, and probably with
paper and hard or soft cover, as opposed to, say, an e-book. Green green would be a color like
kelly green or apple green, as opposed to olive or chartreuse.
 Pa rt V

Interaction
 chapter 21

Combination of affixes

21.1 Prospectus
In this chapter we will discuss complex words that have more than one affix. This topic has
been of considerable theoretical interest and there are different approaches that try to account
for the existence or non-existence of given affix combinations. One of these approaches
focuses on lexical strata, with the native versus non-native distinction at its centre. This dis-
tinction played a role in our treatments of various morphological categories throughout this
volume and is investigated at a more general level and in considerable detail in Chapter 27.
The present chapter is concerned with all issues of affix combinations beyond the native/
non-native issue, focusing on derivational affixes, but also taking a look at inflectional
suffixes.
In the next section we will look at combinations of suffixes and combinations of prefixes,
the areas which research on affix combination has traditionally focused on. Section 21.3
will then be devoted to types of derivative that have received hardly any attention so far,
those in which a prefix and a suffix are found in a single word. For the sake of simplicity, in
Sections 21.2 and 21.3 we will restrict discussion to formations that feature only two affixes
at a time. We will also take up the intersection of conversion with affixation (Section 21.4).
Section 21.5 will deal with inflection, and Section 21.6 with affixation involving more than
two affixes.

21.2 Suffix combinations and prefix

combinations in derivation
2.1 Suffix combinations
With the rise of lexical phonology and morphology the question of affix combinations started
receiving a lot of attention (see Chapter 27). Beginning with Fabb’s (1988) influential paper,
a number of publications (e.g. Plag 1996; Hay and Plag 2004) showed that the stratal approach
is not well-suited to account for the empirically attested patterns of prefix combinations and
suffix combinations. The rather comprehensive new investigation of the assumed stratal
494 Combination of affixes

division in English morphology presented in Chapter 27 further supports this. Given the
empirical and theoretical inadequacy of the stratal approach (see also Plag and Baayen 2009
for a summary), the intriguing question remains what is responsible for the occurrence or
non-occurrence of a particular combination of affixes.
The most successful approach to this question is to predict combinations of prefixes and
of suffixes on the basis of the selectional restrictions of the affixes (and bases) involved (e.g.
Fabb 1988; Plag 1996, 1999, 2002; Hay and Plag 2004). These restrictions are structural in
nature and may refer to various properties of the elements involved: phonological (segmental
or prosodic), semantic, syntactic, or morphological. For example, the suffix ‑ify is restricted
to bases ending in a stressed syllable, the suffix ‑ess is restricted to nominal bases that denote
higher sentient beings, and the suffix ‑ation attaches productively only to the verbal suffixes
‑ize, ‑ify, and ‑ate. Apart from the morphological restrictions, selectional restrictions are not
specifically geared towards other affixes but are general selectional preferences that can be
conceptualized as being part of the lexical entry of the affixes in question. In the preceding
chapters such preferences or restrictions have been described for each of the morphological
processes under investigation.
Crucially and inevitably, such preferences have important repercussions for the combi-
nation of affixes, namely in those cases where the base is already affixed. To give a simple
example, if a suffix attaches only to adjectives and nouns, all combinations of that suffix
with bases that contain an outermost verbal suffix are ruled out. Or consider again the
suffix ‑ify, which strongly tends to attach only to bases ending in a stressed syllable. This
restriction has the consequence that ‑ify may not attach to very many already suffixed bases
since English suffixes that create final syllables in their derivatives are mostly unstressed
(e.g. ‑able, ‑age, ‑al, ‑ant, ‑ary, ‑er, ‑ery, ‑ful, ‑ic, ‑ish, ‑ism, ‑ist, ‑ive, ‑less, ‑ly, ‑ness, ‑ous, ‑ure, ‑y;
these are suffixes which are neither consonantal nor auto-stressed). The selectional restric-
tions still allow a great number of combinations, long lists of which can be found in Chapter
27, or, for example, in the appendices of Plag and Baayen (2009, for suffixes), and Zirkel
(2010, for prefixes).
On the basis of the affix-specific restrictions it becomes possible to predict possible and
impossible combinations of affixes quite successfully. For example, Hay and Plag (2004)
investigate 15 suffixes with 210 theoretically possible combinations (excluding combina-
tions of two identical suffixes). Of these, only 36 (i.e. 17 per cent) are actually attested in
their large database. It turns out that all attested combinations are also structurally possi-
ble, that is allowed by the selectional restrictions of the suffixes involved and the vast
majority of unattested combinations are actually ruled out by the selectional restrictions.
There are only six additional combinations that are structurally clearly possible but that are
not attested in Hay and Plag’s database. Apparently, their large database was not large
enough, since attestations of all six expected but missing combinations can be found in
COCA, on the internet, or in other linguistic works. Example (1) lists some of the perti-
nent forms.
 21.2 Suffix combinations and prefix combinations in derivation 495

(1) a. ‑ling-ly
 USA Today 1993: Elegant women wearing evening gowns—a few in darlingly short
chemises layered with fringe—lace shawls draped over their shoulders, jewels
flashing, sat back in the velvet-covered easy chairs, sipping cocktails. (COCA)
b. ‑ling-dom
Anyways, screw this weak attempt at weaklingdom, and check out the real deal at
www.theweaklings.com. (http://www.amazon.co.uk/review/R16DF1MLUJAH7W)
c. ‑ling-ish
Philip Leverington took this of the Lilac Muscovy ducklings at the weekend (yes,
I know, there’s nothing very ducklingish about them any more . . . . !) (http://www.
facebook.com/pages/Gillamoorganics/146169232118586)
d. ‑ness-less (from Plag and Baayen 2009: 115, who give more examples)
Eight Deadly Sins Of Web 2.0 Start-Ups. [. . .] Happinessless: Your start-up has no
future if you are not happy. (http://www.slideshare.net/imootee/eight-deadly-sins-
of-web-20-startups/)
e. ‑ee-less
 employeeless (not comparable) Having no employees. (http://en.wiktionary.org/
wiki/employeeless)
f. ‑ee-ness
 Theological Studies 2010: If the incarnation reveals that the Divine is not averse to the
conditions of human existence (sickness, refugeeness, and impoverishment), it also
sets a standard for the community that claims to be the concrete manifestation of the
‘glory, grace and truth’ of God in human history. (COCA)

Two important points emerge here. First, the suffix-particular restrictions that are needed
independently to account for the behaviour of a suffix straightforwardly account also for the
existing and the non-existing combinations of that suffix with suffixed bases, that is with
other suffixes. Second, very large databases are needed to find certain affix combinations due
to the fact that the co-occurrence of two affixes in one word is generally not particularly
common. This has been noted before by Hay and Plag (2004), who calculate, for example,
that the probability of finding a combination of ‑ee and ‑hood among any set of words (i.e.
types), based on CELEX type frequencies, is 0.0000006.
Hay and Plag (2004) further observe that suffixes form a hierarchy, such that a suffix
occurring inside another suffix cannot occur outside that suffix. Given a suffix sequence A–B
and another B–C, we can expect to find A–C, but not C–A. We will see in Section 21.5 that
this is overly simple for English, where there are some strings that operate against such pre-
suppositions. Nevertheless, it is largely true, especially with two-suffix sequences: that is the
presence of A–B and B–C largely excludes B–A or C–A. Crucially, the position of a given
suffix in the hierarchy is largely determined by the separability of that suffix. What exactly
does ‘separability’ refer to?
496 Combination of affixes

The underlying idea is that the morphological segmentability or separability of affixes is a

gradient phenomenon. Some words are clearly composed of more than one morpheme (e.g.
clueless, concreteness, unhelpful, dethrone), other words are clearly monomorphemic (e.g. fish),
and there are words whose status as complex words may be not so clear (e.g. analyze, business,
interview, listless, perceive, rehearse). In most current models of morphological processing,
access to morphologically complex words in the mental lexicon works by direct access to the
whole word representation (‘whole word route’) or by access to the segmented elements
(‘decomposed route’). The separability of an affix therefore depends on which route is more
prominent for derivatives with this affix. This in turn depends on various properties, for
example the frequency, phonotactics, and semantics of derivatives with the given affix (see,
for example, Plag 2003: 175–8 for an introductory treatment).
From what we have said so far, interesting insights follow. The same suffix will be differ-
ently separable in different words depending on the respective properties of the words in
question. For example, discernment is more decomposable than government. The derived
word government is much more frequent than its base govern, which induces as a whole-word
bias for government. This whole-word bias is reflected in the non-transparent phonology and
semantics of government. The form is pronounced with either assimilation of base-final /n/ to
the suffix-initial /m/, or even with the complete loss of the last syllable of the base, and
semantically government does not show the expected event meaning ‘action of governing’, but
the extended meaning ‘people who govern’. In contrast, discernment is rather infrequent and
much less frequent than its base discern, which leads to a strong bias towards decomposed
processing for discernment.
The more low frequency forms we find with a given affix, the more separable that affix will
appear. For example, ‑ish has many derivatives with low frequencies (such as housewifish, out-
of-the-way-ish, or soupish whereas ‑ic has few low-frequency words but many derivatives with
higher frequencies (e.g. democratic, fantastic, terrific) so that ‑ish tends to be more separable
than ‑ic.
Now, returning to the problem of affix ordering, we can use this notion of separability to
predict that more separable affixes will occur outside less separable affixes (cf. also Burzio
1994: 354). This is because an easily decomposable suffix inside a non-decomposable suffix
would lead to difficulties in processing, whereas a less easily decomposable inside a more
easily decomposable suffix is easy to process. Based in this reasoning, Hay (2000: 23, 240)
sets up the hypothesis that ‘an affix which can be easily parsed out should not occur inside an
affix which can not’. This mechanism leads to the kind of hierarchy observed in Hay and
Plag’s study, which has earned it the name of ‘complexity-based ordering’ (Plag 2002).
This implies a measure of separability. One simple measure of separability is productivity.
Productive processes are characterized by a large proportion of low frequency, that is easily
separable, forms, while unproductive morphological categories show a low proportion of
rare words. Thus, following Hay’s hypothesis, and confirmed by the empirical findings in
Hay and Plag (2004), suffixes that are closer to the base in the hierarchy (e.g. nominal ‑th,
 21.2 Suffix combinations and prefix combinations in derivation 497

verbal ‑en in Hay and Plag’s suffix set) are less productive, that is less easily segmented, and
suffixes that are further away from the base in the hierarchy (e.g. ‑ness, ‑ful in Hay and Plag’s
suffix set) tend to be more productive, that is more easily segmented.
In a follow-up study, Plag and Baayen (2009) tested a larger set of suffixes (N=31) and
their potential 930 combinations against behavioural data (lexical decision and naming laten-
cies) to get a better idea of the processing issues involved. Again a similar hierarchy was found
and a similar (though weaker) correlation between a suffix’s productivity and its position in
the hierarchy. But the behavioural data do not fully support the idea of complexity-based
ordering. The processing costs of a given word with a particular suffix are not related to the
rank of that suffix in the hierarchy. However, the average processing costs of a given suffix (i.e.
across all words with that suffix) and its rank in the hierarchy are correlated. Suffixes at both
ends of the hierarchy enjoy a processing advantage, either through easy memory retrieval
(the innermost suffixes), or through easy parsing (the outermost suffixes). Overall, it seems,
however, that the individual selectional restrictions are the mechanism that decides about
the possibility of coining a particular word with two given suffixes (see Plag and Baayen
2009: 142–4 for more detailed discussion).
We have given a lot of space to this processing account of suffix-ordering because it seems
to us to be a very explanatory approach. Others have considered, in general terms, rather
than related specifically to English, semantic explanations for ordering, either in terms of
relevance (Bybee 1985) or in terms of semantic scope (Rice 2000). While these factors appear
relevant in the ordering of English derivational suffixes, they cannot explain which of several
synonymous adjectival suffixes will be used in an individual position after a given nominal
suffix, for instance. That is, purely semantic explanations could not rule out a sequence of
‑ess‑ic (not attested in COCA) while allowing ‑ess‑y (attested in COCA in actressy, mistressy,
waitressy).

21.2.2 Prefix combinations

A similar picture emerges with prefixes. In Zirkel’s (2010) study of 15 prefixes and their com-
binations, of the 225 potential prefix combinations (including repeated prefixes) 35 per cent
were actually attested. A closer analysis reveals the existence of a hierarchy, similar to that of
suffixes, such that more productive prefixes attach quite freely to less productive prefixes and
more productive prefixes tend not to occur inside less productive prefixes.
With regard to their combinatorial properties, prefixes are generally less restricted than
suffixes. They normally do not impose phonological restrictions on potential bases, and they
are on average more flexible with regard to the syntactic categories they may attach to (Plag
2004: 201). Given this higher flexibility of prefixes in general, the larger proportion of 35 per
cent attested combinations as against 17 per cent for suffixes in Hay and Plag (2004), is not
surprising. In fact, 81 per cent of all 225 potential prefix combinations are permissible in terms
498 Combination of affixes

of the selectional restrictions of the prefixes involved, according to Zirkel (2010: 253). So for
prefixes, selectional restrictions do not nearly constrain their combinations as severely as
they do for suffixes. Why less than half of the structurally possible prefix combinations are
attested is unclear, however.
One possibility is that they are ruled out by processing considerations, since many of them
would violate Zirkel’s (2010) prefix hierarchy. Another possibility is that the lack of pertinent
forms is simply a matter of restricted corpus size. To test these two possibilities in a prelimi-
nary fashion, we checked 18 randomly picked combinations that violate the hierarchy, and
that are nevertheless structurally possible, but are not attested in Zirkel’s (2010) version of
COCA (February 2008 version, 360 million words). These 18 combinations (involving the
prefixes inter‑, sub‑, and pre‑ in first position) were checked against the COCA version of
April 2012, which, containing 425 million words, is almost 20 per cent larger than Zirkel’s
COCA version. Of the 18 formerly unattested combinations we actually found four newly
attested ones using this larger database.
The results for prefixes and suffixes thus point in the same direction, with complexity-based
orderings the norm and violations of the prefix and suffix hierarchies being dispreferred. For
sequences of native and non-native in regard to prefixes, see Chapter 27.
It appears to be relatively easy to add prefixes to etymological prefixes (that is, of Latin or
Greek origin) whose semantic value has become lost. Examples such as disconnect, ex-
researcher, for instance show bases (connect, researcher) where no prefix can be analysed in
English, even though there was one in the donor languages. Even instances like disavow,
unrefined, where we can analyse a prefix in English, have bases which are item-familiar and
well-established. This is, of course, in line with the generalizations made above about ease of
processing. Also in line with these generalizations are cases of prefixes whose selectional
restrictions would allow for either ordering and where in fact both orderings can be found. In
such cases, we often find that the inner prefix is less separable than the outer prefix, as we see
in examples like underrepresented versus re-understand, inter-react versus re-interview, unpoly‑
chrome versus polyunsaturated.
Non-native prefixes, both learned and very productive recent forms like mini‑, co-occur,
with the main factor determining order apparently being semantic scope: the outermost
prefix has scope over not only the base, but the combination of base and innermost prefix.
Examples from the corpora are given in (2).

(2) anti-postmodern, ex-nonentity, ex-superstar, hypermultilateralist, mega-supermarket,

micro-cogeneration, micronanosecond, mini-pseudo-Medici, mini-submarine,
multimegabyte, multitransnational, nanotransformer, neo-antibiotics, neo-metaphysical,
non-hyperactive, nonmicroporous, nonperinatal, nonpoststructural

Note that alongside anti-postmodern (cited above) we find postantibiotic, illustrating that the
order is not fixed by grammar or phonology. From a descriptive point of view, however, it is
 21.2 Suffix combinations and prefix combinations in derivation 499

difficult to say whether this is true of all such pairs, since combinations are rare in texts.
While COCA provides ex-co-presidency, even Google fails to find a hit for co-ex-wives,
which one might expect to be possible.

21.2.3 Recursion
Although it is possible to find sequences of two homophonous affixes in English (as in
friendlily), and sequences where an affix is found on a base which ends with a string
homophonous with the suffix (as with adulterer), it does not seem to be the case that
words in standard English end freely with a repetition of the same suffix. The nearest we
get to this is those instances where the affix is redundantly repeated to guarantee that it
occurs on the right edge of the word: the BNC provides asphalterer-over, but more gener-
ally words like fixer-upperer occur (see Chapter 11). An example such as that in (3) (from
COCA) may or may not count as real recursion, since ‑ation is added to a verb made of a
converted noun.

MAG_ScienceNews 2011: Elements undergo what’s called physical and chemical

(3) 
fractionation during processes such as going from liquid to solid.

Prefixes can be used recursively, though such uses are not particularly common. The prefix
re- usually alternates with again to avoid meaningful sequences of re-re- (see (4) below from
COCA), though re- on a word beginning with non-morphemic <re> is relatively common
(re-record, re-review) and COCA provides the recursive example re-re-revised. Recursive
examples with other prefixes are provided in (5).

(4) ACAD_Africa_Today 1993: separatist tendencies re-emerged once again

SPOK_NPR-Talk_Nation 2002: the people of Newark, who are ready to re-elect him

again

(5) NEWS_SanFrancisco 1997: We’re in the pre-prelaunch stage

NEWS_Atlanta 1991: They’ve got all these post-post-post-feminist things at their
fingertips.
NEWS_Denver 2009: with mandatory prison sentences on any individual, company,
subcontractor, sub-subcontractor or sub-sub-subcontractor or corporation that
receives taxpayer money and hires illegal aliens
MAG_NatlReview 1994: why should the existing information superhighways be
upgraded to super-superhighways?
FIC_Bk;HabeasCorpses 2005: she seemed content: being “un-undead” suited her just
fine.
500 Combination of affixes

MAG_RollingStone 2000: “It was a coming-of-age film, shot entirely in India—very,

very ambitious.— # Very, very is an under-understatement.
FIC_IowaRev 2004: there’s nothing in the world I want more than to have her as my
ex-ex-wife, to try again

Prefixes showing size seem to be particularly prone to recursion, with some examples from
COCA in (6), but even then, the additional use of quasi-synonymous prefixes may be pre-
ferred (see (7)).

MAG_ScienceNews 1991: A “hyper-hyperparasite,” known more simply as the “cheater,”

(6) 
sneaks into the placid community
SPOK_MSNBC_Cosby 2006: They are becoming mega-mega-stars.
MAG_Bicycling 2004: What with CO2 cartridges and those nifty mini-minipumps,
the traditional, super-effective frame pump has become a sign of retrohood.
NEWS_CSMonitor 1994: One self-described “ultra-ultraconservative Democrat”
defected

(7) MAG_PopMech 2007: It’s like a hip micro-minivan.

CNN_Showbiz 2009: And its also scary to see how isolated these mega-superstars are.
FantasySciFic 1994: he was making time payments on one colossal, outsize, super-
mega-omniprayer of his own asking being answered some day.

21.3 Combinations of prefixes with

suffixes in derivation
In this section we will look at a phenomenon that has received very little attention in the lit-
erature, namely the combination of a prefix and a suffix in the same word. There are four
types of formation that one could distinguish based on which affix is the outermost one. In
(8), the four possibilities are illustrated with data from COCA. In (8a) the prefix is attached
to an existing suffixed derivative, in (8b) a suffix is attached to an existing prefixed word, and
in (8c) either of the previous analyses is synchronically possible. Finally, (8d) gives examples
of parasynthetic formations, which are defined as derivatives that come about through the
simultaneous prefixation and suffixation to a base. The attachment of prefix and suffix is
‘simultaneous’ in the sense that neither a suffixed base nor a prefixed base is attested before
the emergence of the prefixed and suffixed form, which differentiates parasynthetic forma-
tions from the cases of multiple hierarchical formations as shown in (8a) to (8c). We will deal
with clearly hierarchical multiple pre- and suffixation in the next subsection and with para-
synthetic formations in Section 21.3.2.
 21.3 Combinations of prefixes with suffixes in derivation 501

(8) a. Prefix as outermost affix (right-branching)

after-retirement, dedogmatize, meta-learning, non-additive, semi-attractive,
vice-president
b. Suffix as outermost affix (left-branching)
embodiment, interviewer, midcontinental
c. Ambiguous cases (right-branching or left-branching)
semi-blindness, mistreatment, unhumanize
d. Parasynthetic formations
deasbestocize, decaffeinate, embolden, enliven

Combinations of these patterns are also possible, as in [[micro[dis[continu-iti]]]es],

[non[anti[bacteri-al]]], [[non[intervent-ion]]ist] (where discontinuity provides a possible
instance of ambiguity).

21.3.1 Multiple hierarchical affixation

The cases (8a) to (8c) are usually considered unproblematic as they follow straightforwardly
from the selectional restrictions of the affixes involved, and such words are also quite com-
mon in English. Exactly how common has rarely been investigated (but see Berg 2003), and
it is also unclear whether the combination of a prefix and a suffix in a single word is princi-
pally constrained beyond what we would expect on the basis of the selectional restrictions
involved.
In order to take a closer look at this, we investigated left- and right-branching structures
in more detail. For the study of right-branching structures we randomly picked five pre-
fixes with different degrees of productivity (de-, mid-, non-, semi-, and vice-) and checked
for each prefix a random sample of 50 derivatives from the unlemmatized COCA word
lists we had generated as part of the research for this volume. These word lists contain
derivatives in which the affix in question can be analysed as the outermost suffix, thus
including also the ambiguous cases shown in (8c). We ended up with an average propor-
tion of 47 per cent of the derivatives in the samples from the five lists having at least one
derivational suffix in addition to the prefix, with the individual percentages ranging
between 34 per cent for mid-­and 70 per cent for semi‑. This means that the prefixes in
question are quite likely to attach to already suffixed words, forming right-branching
structures as in (8a).
We then applied an analogous procedure to left-branching structures. For this we took
our word lists with the randomly picked suffixes ‑ize, ‑ness, ‑age, ‑able, and ‑some as the
outermost affix to see how many of their bases are prefixed. In this sample, only 5 per cent
of the pertinent derivatives were found to be based on prefixed words (on average, the
suffixes ranged between zero per cent prefixed bases for ‑some and 12 per cent prefixed
502 Combination of affixes

bases for ‑able). In other words, these suffixes are rather unlikely to attach to prefixed
words.
Assuming that these exploratory results are representative, how can this asymmetry
between prefixes and suffixes be explained? Why would prefixes attach more happily to
already suffixed words while suffixes are much less prone to attach to already prefixed
words? Interestingly, this asymmetry in our two samples is predicted by complexity-based
ordering. In general, prefixes tend to be more easily separable than suffixes due to the left-
to-right nature of the speech signal. Prefixes enter the processor before the base, whereas
the base comes first with suffixed words. In either case, the leftmost morpheme has an
advantage in lexical access and representation (see, for example, Plag 2002, for more
detailed discussion). With regard to multiple affixation this means that under the assump-
tion of complexity-based ordering the average prefix should be outside the average suffix,
which is equivalent to saying that, given a word with a prefix and a suffix, we expect the
structure of this word to be right-branching rather than left-branching. This expectation
fits nicely with our finding that among the words with an outermost prefix, we find very
many bases that are suffixed.
And complexity-based ordering would also predict that in those cases where we deal
with a left-branching structure (as in embodiment) we expect the prefix to be less easily
segmentable than in the average right-branching case (e.g. semi-attractive). This predic-
tion seems to be borne out by the facts: Berg (2003) finds that in his data set left-­
branching is generally uncommon among productive prefixes, but is highly frequent
with unproductive prefixes. Recall that unproductive prefixes are less easily segmentable
and thus have a greater chance of occurring inside suffixation, as part of a left-branching
structure.
Overall, these exploratory investigations into the combination of prefixes and suffixes in
the same words present a picture similar to that of prefix combinations and suffix combina-
tions. The prefix hierarchy and the suffix hierarchy interact in ways predicted by complexity-
based ordering, with unproductive prefixes being located at one end of the hierarchy (perhaps
together with certain unproductive suffixes), and productive prefixes being located at the
other end of the affixal hierarchy. More research is obviously called for to further study
the interaction of prefixes and suffixes and to test the tentative claims put forward here on the
basis of an investigation of a small sample of affixes.

21.3.2 Parasynthetic affixation

Parasynthetic formations as defined above are rare in English, but they raise a number of
theoretical problems, the most prominent of which is branching. For illustration, we show in
(9) three potential analyses of decaffeinate, with three kinds of constituent structure and
branching.
 21.4 Conversion and affixation 503

(9) a. [[de-caffein]-ate] b. [de-[caffein-ate]] c. [de-[caffeine]-ate]

decaffein caffeinate

de- caffein -ate de- caffein -ate de- caffein -ate

de- ... -ate

In some theories, morphological branching is strictly binary (e.g. Aronoff 1976), which would
rule out a representation like (9c). We do not want to discuss here all the theoretical and
empirical problems related to the binary branching hypothesis, but it seems that with regard
to parasynthetic formations, there is a problem for this approach. With our example decaffein‑
ate, there was no verb caffeinate at the time of creation to which the reversative prefix de‑
might have attached. There was also no potential base decaffein for the attachment of the
suffix -ate (Plag 1999: 101). This means that the meaning ‘remove X’ is expressed through both
the prefix and the suffix, with the suffix contributing the verbal semantics and the prefix the
privative meaning. It might seem that we would be forced to posit a circumfix de‑ate with that
meaning, as shown in the lower part of the tree in (9c), but both de‑ and ‑ate occur independ-
ently elsewhere with the relevant meanings, which is not typical of a circumfix. The alterna-
tive is to assume, as in (9b), the existence of a putative verb caffeinate ‘provide (with) caffeine’
(such a solution is sometimes termed an ‘overgenerating morphology’—Allen 1978). This
putative verb is then prefixed with the reversative prefix de‑ (see e.g. Guevara 2007 for a gen-
eral discussion of this kind of solution). The same kind of analyses and problems apply to the
other forms cited in (8d).
None of these analytical possibilities seems particularly attractive from a rule-based or
generative perspective, but there is yet another approach possible, one that is based on anal-
ogy or constructional schemata. There are quite a number of derivatives that feature both the
prefix de‑ and the suffix ‑ate on a nominal base with the meaning ‘remove X’ (where X is the
denotation of the base), such as deacylate, decapacitate, dechlorinate, dehyphenate. The para-
synthetic form may be analysed as simply following this pattern (or schema, in construction-
ist terms).

21.4 Conversion and affixation

In principle it might be expected that conversion would interact with suffixation in the same
kind of way that affixation processes interact with each other. The reality is rather more dif-
ficult to establish than this would suggest.
504 Combination of affixes

There seems to be no particular reason why a converted word could not act as a base for
subsequent affixation. Once the verb empty has been created by conversion from the adjec-
tive empty, it can inflect as a verb (empties, emptied, emptying), and we might expect it to be
used as a derivational base as well, as it is in (11). However, there do not appear to be any
established derivatives from the verb empty, and the question is whether this is typical or just
happenstance.
There are so many cases of conversion in English that there are times when it is difficult to
decide which member of the pair of converted forms is involved in a derivation. Our data sets
are full of such instances. For example, the derivatives designer, flapper, harpooner, peeper,
plasterer, scorer, toner, trooper, whisperer might all be derived from either a noun or a verb
base. We would probably decide that all of them are derived from verbs since ‑er appears to
attach more frequently to verbs than to nouns and since the semantics is consistent with the
base being a verb. If that is the case, though, it seems that at least harpooner and plasterer must
be formed on the basis of nouns used as the base for conversion. Thus there is no barrier in
principle to derivational affixation applying to instances of conversion. In the very nature of
things, however, it is difficult to quantify how often this occurs.
Conversion of affixed forms is slightly easier, at least where suffixes are concerned.
Marchand (1969: 372–3) claims that derived nouns rarely undergo conversion. He argues
that this is because of blocking (although he does not use the term): the converted verb
would have the same meaning as that of the base verb. However, it is shown in many chapters
in this book that blocking is never a strong force, and an argument based on it is correspond-
ingly weak. In (10) we provide some examples of words whose derivational morphology
suggests that they are not verbs and yet which can be found inflected as verbs.

(10) adventure, bandage, blinker, cashier, champion, commission, condition, disadvantage,

function, package, pilgrimage, pressure, provision, sanction, scooter, section, signal,
streamer, stopper, tenant, tension, waitress

We have drawn attention elsewhere to forms such as intellectual and disposable (Chapter 25)
which are marked as adjectives and yet which can be marked for plural and head noun
phrases. Similarly we have drawn attention to forms in ‑an and ‑ist (Chapter 14) which regu-
larly provide both adjectives and nouns. All of this suggests that conversion of affixally
derived bases is quite normal. The examples in (11) all illustrate established nouns or adjec-
tives being used as bases.

ACAD_Symposium 1993: In this dynamic, space is always convertible, emptiable or

(11) 
fillable.
FIC_IowaRev 2012: she can go back to doing her job, which is princessing, which
delights Audrey Hepburn
 21.5 Inflection 505

Psychology Today 2011: Researchers also identified two other categories of nonerotic
thoughts—“spectatoring,” or stepping outside your body to judge your own
performance, and body image concerns

21.5 Inflection
Although English inflection is exclusively suffixal and also rather impoverished, two ques-
tions still deserve answers: What is the ordering of inflectional versus derivational suffixes,
and what is the ordering of inflectional suffixes if more than one suffix ends up on the same
base? We will deal with each question separately.

21.5.1 Inflection in interaction with derivation

It is generally agreed that in English (as in many other languages), inflection is outside deriva-
tion. This is illustrated in (12), with adverb formation given in parentheses due to its debated
status as an inflectional process (see Chapter 15 for discussion).

(12) Nouns: plural and genitive

curiosities, princesses; discussant’s, manager’s
Verbs: tense, person, and aspect
hyphenated, personifies, verbalizing
Adjectives: gradation (and adverb formation)
unofficialer, winningest (adaptively, aimlessly)

Apparent exceptions to this general rule can be found when the inflection is irregular and not
clearly affixal, but even then only a few examples of a few types are found, as shown in (13).

(13) a. better]a]v ment, lessen, moreish, mostly, worsen

b. *feetage, *geesish, *menly, *micelings, *womenize; *ateing, *caughtment, *dranker,
*grewage

Another set of apparent exceptions occurs when things which might be analysed as inflec-
tional affixes (and in particular, ‑s, ‑ed, and ‑ing) are found inside one of a very few extremely
productive derivational affixes. Examples are given in (14). Note that in (14) adverbial ‑ly is
treated as derivational; if it is inflectional, these examples belong elsewhere.

(14) folk-s-y, interest-ed-ly, interest-ed-ness, interest-ing-ly, interest-ing-ness, sud-s-y, wedd-

ing-ish
506 Combination of affixes

There are also a few examples of other irregular plurals occurring inside derivational affixes,
but the clearest example we have found of this, mediawise, has the base media, which may no
longer be perceived as a plural noun.
Although it is strictly irrelevant, in that there is no interaction between inflection and deri-
vation, it is perhaps worth noting here that plurals are often found internal to compounds, as
illustrated with examples from COCA in (15) (see also Chapter 19.3.1).

(15) alumni relations, arms embargo, books editor, criteria levels, feet movement, grounds
maintenance, mice feces, universities consortium

Finally, there are also cases where it is not so clear that the inflectional suffix is the outermost
suffix; these cases concern parasynthetic formations, in which a derivational prefix and an
inflectional suffix come together. This is quite common with verbal participles. For example,
we have noted in Chapter 13 that ornative verbs with the suffix be‑ are often coined as past
participles, that is with the simultaneous attachment of the prefix be‑ and the suffix ‑ed to a
nominal base (e.g. bedenimed, bejowled, besmogged, bestuccoed, all from COCA, see also
Marchand 1969: 148, and the OED s.v. be‑7 for discussion). Note that such formations do not
pose a problem for a binary-branching approach to word-formation since these forms, in
contrast to the parasynthetic formations discussed in Section 21.3.2, can be analysed straight-
forwardly as left-branching, that is as a prefixed form with a suffix, as in, for example, [[be-
denim]‑ed]. In other words, with regard to these formations, the term ‘simultaneous’ refers to
the speaker’s act of coining the word, and not to a theoretical notion of derivationally simul-
taneous attachment (e.g. via ternary structures). However, because the ‑ed in these forma-
tions could be interpreted as an ornative ‑ed (see Chapter 14), and thus derivational, it is not
clear how far these examples are relevant here.

21.5.2 Inflection interacting with inflection

We now turn to the ordering of multiple inflectional endings. Such multiple inflectional end-
ings cannot be found with verbs due to the non-agglutinative nature of the English verbal
inflectional system, but they may appear, at least theoretically, with nouns and adjectives.
Nouns may be inflected for genitive and plural at the same time, in which case the genitive
marker follows the plural marker (as in brethren’s, oxen’s). The picture is of course more com-
plicated than that since the combination of the regular plural marker ‑s with the genitive
marker ‑’s triggers haplology in rather intricate ways (e.g. genitive plural boys’). The details of
this conflation of plural and genitive marking are discussed in Chapter 7 and need not be
repeated here.
With adjectival inflection, we can have the stacking of gradation and adverb formation,
but of course only under the premise that we consider adverb formation as inflection. If we
 21.6 Derivatives with more than two affixes 507

do so, it seems that the two kinds of suffix may simply not combine. With ‑ly adverbs, the
periphrastic comparative (e.g. more boldly) is obligatory, with only very few variant excep-
tions (e.g. archaic quicklier). If these exceptions are taken as sufficient evidence, it seems
that gradation is outside adverb formation.
Finally, we note in Chapter 24 that there are some processes which might be considered
inflectional, but are not usually taken to be inflectional, particularly formations involving
numbers. Examples like fif‑th‑s and fif‑teen‑th‑s show that these processes are less peripheral
than the clearly inflectional plural. Examples in (14) are relevant here if adverbial ‑ly is
assumed to be inflectional.
Overall, English is very poor when it comes to the stacking of inflectional suffixes. Where
it would be possible at all, it is highly restricted and occurs only with morphologically irregu-
lar or otherwise exceptional forms (oxen’s and the examples in (13a)).

21.6 Derivatives with more than two affixes

So far, we have limited the discussion to words with no more than two derivational affixes.
Needless to say, these words may take inflectional endings, to the effect that the combination
of one or more derivational affixes with an inflectional suffix is unproblematic (e.g. decontex‑
tualizations, disauthenticated, semi-customized). The number of derivational affixes that can be
found in a single word seems in principle unlimited, but processing becomes difficult with
longer forms. This is probably the reason why finding derivatives with four affixes is still quite
possible, but finding words with five or more derivational affixes becomes increasingly hard,
even though the attested forms could even be further expanded by additional suffixation or
prefixation. In (16) we list some pertinent examples from COCA. For the reader’s conven-
ience, we provide hyphens to separate the affixes. A particularly instructive example is the
recursive prefix great‑, which is attested in COCA a number of times with four instantiations
in one word, but never with more (see (16b) for one example). Note that while we list forms
in (16) in terms of the number of affixes overall, if we look at the number of prefixes or the
number of suffixes it is, in most cases, even more constrained in actual use.

(16) a. Four affixes:

 neo-en-light-en-ment, counter-re-form-ation-al, anti-americ-an-iz-ation, anti-
constitut-ion-al-ist, epi-phenomen-al-iz-ation, institut-ion-al-iz-ation
b. Five affixes:
 great-great-great-great-grand-parents, anti-inter-nat-ion-al-ism, (cf. anti-inter-
nat-ion-al-ist-ic, which would have one more affix and is not attested), neo-retro-maxi-
minim-al-ist (cf. neo-retro-maxi-minim-al-ist-ic)
508 Combination of affixes

One of the differences between looking at sequences of just two affixes and looking at longer
sequences is that we can see recursion over longer sequences that are not easily observable
over just two. English does allow sequences such as ‑ation-al-ize and al-ist-ic (Bauer 1983: 69).
We find words like sens-ation-al-ize, compartment-al-iz-ation, organ-iz-ation-al which show
that the order between the three affixes, while fixed, is recursive. Examples with the same
affix twice are rare, though COCA provides sens-ation-al-iz-ation, and nat-ion-al-iz-ation (and
derivatives of that). Equally we find individ-ual-ist-ic, ego-ist-ic-al, theatr-ic-al-ist which indicate
that recursion with certain constellations of ‑ic, ‑ist, and ‑al should be possible, and COCA
provides us with phys-ic-al-ist-ic and form-al-ist-ic-al-ly which illustrate actual recursion. That
recursion of this type appears to be restricted to the sets ‑ation-al-ize and ‑al-ist-ic is of some
interest; on the other hand, the presence of recursion at all is a point of theoretical
importance.
 chapter 22

Affixation on compounds
and phrases

22.1 Prospectus
In this chapter we address the issue of derivation on bases that are compounds or phrases. It
is normally assumed that derivation applies either to simplex bases (bound or free), or to
already affixed forms (see Chapter 21). Compounds can themselves be formed from derived
bases (knowledge economy), and phrases can certainly contain derived bases (knowledge of
linguistics), but general wisdom suggests that the opposite is only rarely the case. Theoretical
constructs such as Level-Ordering (regarding compounds) (e.g. Siegel 1978; Kiparsky 1982a),
and the Lexical Integrity Hypothesis (regarding phrasal bases) (e.g. Lapointe 1981) are
grounded in these assumptions. We will show below, however, that it is not at all rare to find
derivational affixes in English appearing on compound bases, and that phrasal bases are per-
haps not as rare as our intuitions might suggest. We will first review the data here, much of
which can be found in previous chapters treating specific types of derivation, and then con-
sider the factors that might contribute to the ability of an affix to take a compound or phrasal
base. Finally, we will raise the issue of the nature of the compounds and phrases to which
affixes attach, that is, whether affixation is restricted to compounds and phrases which are
highly lexicalized.

22.2 Data
22.2.1 Affixes on compound bases
We divide affixes into three categories according to the frequency with which they attach to
compounds: those that frequently allow compound bases (more than ten types found in
COCA), those that occasionally allow compound bases (one to nine types), and those that
never allow compounds as bases. In (1) we list those affixes which are not found on com-
pound bases.
510 Affixation on compounds and phrases

(1) a. Suffixes: ‑acy, ‑age, ‑alA, ‑alN, ‑ian, ‑ance, ‑ant, ‑ary, ‑ate, ‑eer, ‑en, ‑ese, ‑ess, ‑ette, ‑fulA,
‑ic, ‑ical, ‑ify, ‑ine, ‑ation, ‑ite, ‑ity, ‑ive, ‑ize, ‑ling, ‑ly, ‑ment, ‑ory, ‑ous, ‑some, ‑ure
Prefixes: a‑, ante‑, back‑, be‑, by‑, circum‑, cis‑, contra‑, counter‑, demi‑, dis‑, down‑,
b. 
en‑, hemi‑, hypo‑, LOCin‑, NEGin‑, mal‑, meta‑, mis‑, mono‑, nano‑, off‑, on‑, para‑, peri‑,
pico‑, poly‑, supra‑, uni‑

There may be instances where it is not clear whether the base should be interpreted as a
compound or a phrase. Although we have tried to be accurate and consistent here, the point
of the chapter is not affected if individual examples are mis-classified.
It is perhaps somewhat surprising that there are so many prefixes and suffixes that do take
compound bases. Table 22.1 lists examples.

Table 22.1 Suffixes that frequently take compounds as bases

Suffix Examples
‑dom automobiledom, bestsellerdom, blockbusterdom, breakfastdom, bubblegumdom,
couch-potato-dom, crackpotdom, highwaydom, newspaperdom, stakeholderdom,
statepriesterdom, teenagerdom
‑er babysitter, backloader, bagpiper, basketballer, beachcomber, bedsitter, bellyacher,
curbstoner, footballer, gen-exer, grade-schooler, grandstander, jitterbugger, keelhauler,
keynoter, kneecapper, landscaper, mainlander, motocrosser, motowner
‑ery ass-kickery, blacksmithery, crackpottery, dogoodery, dumbassery, folk-freudery,
gasbaggery, head-squishery, henpeckery, hornswogglery, housewifery, ironmongery,
jackassery, knickknackery, peacockery, pudfootery, ragtaggery, slapdashery, sleazebaggery,
tightwaddery, tomcattery, tomfoolery, wedding-cakery, werewolfery, whizbangery
‑esque all-terrain-esque, appleseed-esque, broadway-esque, disneylandesque, horror-movie-
esque, hummingbirdesque, merchant-ivoryesque, prom-queenesque, snowbirdesque,
snowboard-esque, strobe-lightesque, thanksgivingesque
‑ish art-nouveauish, backcountryish, bargain-basement-ish, bedfellowish, bedroomslipperish,
block-busterish, blue-greenish, boy-scoutish, bulldoggish, cheerleaderish, comicbookish,
cookie-cutterish, cowboyish, curtain-roddish, daredevilish, doomsdayish, eggplantish,
everyman-ish, fairy-taleish, farmhand-ish, film-scorish, fire-escape-ish, french-fryish,
hay-feverish, hefnerpaddish, hip-hoppish, homeboyish, and many others
‑ism bandwagonism, big-shotism, blue-bloodism, copycatism, deaf-mutism, foot-ballism,
front-runnerism, fuddy-duddyism, he-manism, highbrowism, hotshitism, lame-duckism,
meatballism, policemanism, red-tapism, short-termism, stakeholderism, thirdworldism,
trickle-downism, turntablism, watchdogism
‑ist automobilist, avantgardist, bagpipist, dotcomist, fluegelhornist, folklorist, keyboardist,
landscapist, storyboardist, third-worldist, turntablist, watercolorist
‑less air-conditionless, airportless, backboneless, breakfastless, box-scoreless, cocktailless,
earbobless, eyebrowless, footnoteless, girlfriendless, grandchildless, hairdresserless,
keyboardless, lifestyleless, lipstickless, makeupless, mass-transitless, newspaperless,
nicknameless, octoberfestless, roommateless, shoelaceless, sidewalkless, suitcaseless,
sunspotless, swordfishless
 22.2 Data 511

Suffix Examples
‑ness absent-mindedness, accident-proneness, air-headedness, airportness, airworthiness,
babyfacedness, baseballness, bigheartedness, bond-indebtedness, bone-tiredness, bushman-
ness, butterflyness, checkmateness, childfriendliness, church-relatedness, class-consciousness,
cocksureness, colorblindness, commonplaceness, costeffectiveness, couch-potato-ness,
crazy-busyness, cross-borderness, down-homeness, earth-embeddedness, everlastingness,
hanky-pankiness, headstrongness, hip-hopness and many more
‑ship caretakership, headmastership, role-modelship, roommateship, uptownship; lots of
examples in x-manship, x-personship
‑yA asskissy, backwoodsy, cardboardy, headachy, hot-tubby, hound-doggy, householdy,
kiss-assy, knick-knacky, name-droppy, networky, new-agey, old-maidy, oldtimey,
pepperminty, pin-cushiony, pine-woodsy, pipsqueaky, piss-stinky, red-necky, red-peppery,
sandpapery, secret-agenty, singsongy, stand-offy, sunshiny, woodworky

Table 22.2 Prefixes that frequently take compounds as bases

Prefix Examples
anti‑ anti-age-discrimination, antibackflow, anti-blacklist, anti-breast-cancer, anti-car-theft,
anti-comic-book, anti-cookie-baking, anti-direct-lending, anti-drug-abuse, anti-hair-mussing,
anti-home-birth, anti-kidney-stone, anti-land-reform, anti-meat-eating, and many more
co‑ co-anchorman, co-babymaker, co-boat-owner, co-chairman, co-concertmaster, co-decision-
making, co-filmmakers, co-headlines, co-leaseholders, co-scriptwriter, co-songwriters
ex‑ ex-alcohol-addict, ex-ballplayer, ex-bandmates, ex-boomtown, ex-bus-driver, ex-case-
worker, ex-cheerleader, ex-churchgoers, ex-councilwoman, ex-farmhand, ex-fisherman,
ex-grad-student, ex-hothead, ex-jailbird, ex-love-parade, ex-newscaster, ex-redneck,
ex-schoolteacher, ex-sidekick, ex-skinhead, and many more
mega‑ mega-bailouts, mega-best-seller, megablockbuster, mega-bookstore, mega-boxoffice,
mega-crayfish, megadatabase, mega-earthquakes, mega-feedlot, mega-girl-group,
mega-hardware, megahorsepower, megajackpots, megalandfills, mega-landlord,
megalawsuit, megamainframe, mega-man-hours, mega-mastermind, mega-meltdowns,
mega-network, megatracksites, mega-wildfire
micro‑ micro-bloodvessels, microbroadcasters, micro-buzzbaits, microcornrows, micro-cowlicked,
micro-earthquake, micro-hula-hoops, microkeyboard, microlandscape, microlivestock,
microtiddlywinks, microwarehouse, microwatershed
mid‑ midafternoon, mid-backstroke, midbroadcast, mid-cockpit, mid-dogfight, mid-driveway,
mid-photo-spread, mid-price-point, mid-running-step, midsidewalk, mid-single-digit,
mid-treetop, midwatercourse
mini‑ mini-aircraft, mini-backpack, mini-battlewagon, mini-bloodbath, minibugsucker,
mini-cheeseburgers, mini-comeback, mini-crossbow, minidesktop, mini-firestorm,
mini-food-processor, minigreenhouse, mini-guitar-case, mini-jumpsuit, mini-landslide,
mini-meatballs, mini-nutcracker, mini-playgrounds, mini-spreadsheet
multi‑ multibedroom, multi-cactus-plant, multi-care-level, multi-decisionmaker, multi-earth-
mass, multi-fall-lined, multihousehold, multinetwork, multiseafood, multi-snow-sport,
multispacecraft, multistakeholder, multi-touchdown, multiwarhead, multiwavelength
neo‑ neococktail, neo-gentleman, neo-glam-rock, neo-hunter-gatherer, neo-icehouse,
neo-loungeheads, neo-main-street, neo-norsemen, neo-slapstick, neo-zenbohemian
512 Affixation on compounds and phrases

Prefix Examples
non‑ nonabstinence-based, nonadhesive-backed, non-African-born, non-airmen, non-art-world,
non-barrier-island, nonbedridden, non-bullet-proof, noncheesy-looking, non-college-track,
non-control-top, non-credit-card, nondatabase, non-drug-using, non-empty-net,
non-elf-sized, non-fairtrade, non-full-face, non-grade-level, and many more
post‑ post-affirmative-action, post-babyboom, postbaseball, post-blacklist, post-bluegrass,
post-childbirth, post-classroom, post-closing-bell, post-cough-syrup, post-digital-camera,
post-foster-care, post-goal-scoring, posthomework, post-knee-injury, post-lovemaking,
post-phone-number, and many more
pre‑ pre-air-conditioned, prebaseball, pre-birthday, pre-bodice-ripping, pre-book-tour,
prebullfight, pre-ceasefire, pre-computer-literate, precowboy, pre-cram-school, pre-
dogfighting, pre-drywall, pre-fieldwork, pre-freeze-frame, pre-heart-attack, pre-
homeroom, pre-job-offer, and many others
semi‑ semibackbreaking, semi-bandwidth, semi-baseball, semi-beavertail, semihardshell,
semi-homemade, semi-landlocked, semilowland, semi-open-stance, semisidelined,
semi-sideways, semi-star-crossed, semi-supply-sider, semi-tractor-trailer, semi-
warehouse, semi-widetake, semi-work-related
sub‑ sub-cabinet-level, subfootnotes, sub-freshman, sub-golfsize, sub-hide-outs, sub-
lightspeed, sub-lightyear, submachinegun, submainstream, sub-medium-grade,
subnetwork, sub-notebook, sub-passageway, sub-record-breaking, sub-room-
temperature, sub-watershed, sub-wave-length
super‑ super-airliner, super-bookstores, superboomtown, super-bottom-heavy, super-broadband,
superchecklist, super-featherweight, super-first-class, superflywheel, supergirlfriend,
superguardsmen, superheavyweight, super-high-end, super-landfills, super-leftwing,
super-mudroom, super-muscle-car, superpipelining, supersalesman, supershareholder,
super-stonewashed, super-sunflowers, super-side-angle
ultra‑ ultrababyfat, ultrabroadband, ultra-cold-hardy, ultra-deadpan, ultra-deep-field, ultra-
deepwater, ultra-fast-acting, ultrafine-grained, ultra-free-market, ultra-fuel-efficient,
ultra-hands-on, ultra-heat-resistant, ultra-heavy-duty, ultrahigh-capacity, ultra-left-
wing, ultralightweight, ultra-thinwall, ultra-underdog, ultrawideband
un‑ unairbrushed, unairconditioned, unbackcountry, un-blindfold, unbulldozed, un-chain-
store-afflicted, uncomputer-savvy, undinosaur-looking, undoublespace, un-dye-packed,
unflightworthy, unfloorplanned, unfootnoted, ungridlocked, unhandcuff, unhousebroken,
unloveworthy, unmakeshift, un-model-perfect, un-nerve-wracking, un-Paris-Hilton-
looking, unsolaraided, unstreamlined, un-Texas-size, un-user-friendly

Table 22.3 Suffixes that only occasionally attach to compounds

Suffix Examples
‑able backpackable, bushwhackable, copyrightable
‑ee blackmailee
‑fuln classroomful, hatboxful, mailboxful, papercupful, shoeboxful, tablespoonsful, trash-canful,
windshieldful
‑hood beach-bunnyhood, jackasshood, statesmanhood, wallflowerhood
‑let bookmarklet
‑ward northeastward, northwestward, southeastward, southwestward
 22.2 Data 513

Table 22.4 Prefixes that only occasionally attach to compounds

Prefix Examples
after‑ after-bedtime, after-breakfast, after-federal-tax, after-lovemaking, after-math-time,
after-thanksgiving, after-workout
arch‑ arch-dipshit
auto‑ auto-treadmill
crypto‑ crypto-deadhead, crypto-right-winger
de‑ debottlenecking, de-mothballed
extra‑ extra-bubble-butt, extra-classroom, extrahousehold, extra-lightweight, extra-sure-footed
fore‑ fore-topmast
hyper‑ hyper-health-conscious, hyper-rapid-fire, hyper-sports-utilities, hyper-weight-conscious
inter‑ intergenebank, interhousehold, interjumppoint, interkeystroke, inter-landau-level,
internecktie
intra‑ intra-beltway, intrahousehold, intra-landau-level, intra-network, intra-working-class
over‑ over-air-conditioned, overautomobiled, over-lifesize, over-lip-synced, overpipelined,
overstreamlined, overwallpapered
out‑ out-back-stabbed, out-cheap-shot, out-comeback, out-french-horn, out-horsepower,
out-redneck, out-swashbuckle, out-tailgate, out-tax-cut
proto‑ proto-bluegrass, proto-daily-paper, proto-dipstick, protodownscale, proto-highlands,
proto-hip-hop, proto-jumpsuit, proto-performance-art, proto-punk-rock
re‑ rebabyproof, rebroadcast, relandscaped, rewaterproofing
retro‑ retroopensource, retro-roadside
tera‑ teraelectronvolt
trans‑ transboundary-scale, trans-lightspeed, trans-windshield
under‑ under-fingernail, under-gunwale, under-tablecloth
up‑ up-full-court

22.2.2 Affixes on phrasal bases

It is much rarer to find affixes on phrasal bases. The examples in Table 22.5 can be found in
COCA.

Table 22.5 Affixes on phrasal bases

Affix Examples
‑dom low-carbdom, middleclassdom, wife-and-motherdom
‑er back-bencher, bottom liner, do-gooder, do-it-yourself-er, down-easter, fifth-grader,
first-rounder, free-trader, hall-of-famer, in-and-outer, in-towner, left-hander, no-brainer,
one-liner, out-of-towner, threewheeler, and others
‑ery old-fartery, white-trashery
‑hood above-averagehood, one-step-behindhood, senior-skier-hood
‑ish dark-reddish, dog-in-the-mangerish, feelgoodish, fox-in-the-henhouseish, girl-next-
doorish, last-mannish, low-keyish, metal-lumpish, up-your-buttish
514 Affixation on compounds and phrases

Affix Examples
‑ism can-doism, do-goodism, donothingism, don’t-give-a-damnism, don’t-rock-the-boatism,
feel-goodism, get-even-with-themism, holier-than-thouism, know-nothingism,
me-firstism, me-tooism, not-in-my-backyardism, old-fartism, old-fogeyism, one-
worldism, over-the-topism, stand-patism
‑ness all-over-ness, at-homeness, at-largeness, at-oneness, day-to-dayness, does-nothingness,
do-it-yourselfness, down-and-outness, down-to-earthness, every-girlness, feel-
goodness, fifth-wheelness, getting-on-ness, here-and-now-ness, high-and-mightiness,
hot-button-ness, I-can-do-it-too-ness, ill-at-easeness, in-loveness, in-placeness,
in-shapeness, just-so-ness, larger-than-lifeness, know-nothingness, letting-in-ness
‑ward down-and-southward
after- after-initial-sale, after-real-life
anti‑ anti-big-business, anti-big-government, anti-business-as-usual, anti-free-speech,
anti-global-capitalism, anti-poor-women, anti-reproductive-rights
co‑ co-first-place
ex‑ ex-friends-with-benefits, ex-man-of-steel, ex-olympian-turned-pro
extra‑ extra-longrange
out‑ out-crash’n’burn
post‑ postbaby-on-board, post-connery-as-bond, post-death-of-god, post-fall-of-Baghdad,
post-hunter-and-gatherer, post-Pearl-and-Jacob, post-state-of-nature, post-two-year-
college
pre‑ pre-cold-war, pre-end-of-world, pre-fifth-century, pre-Redford-and-Hoffman,
pre-shock-and-awe
semi‑ semi-fast-pitch
ultra‑ ultra-low-cal, ultra-right-wing

22.3 Factors influencing affixation

on compounds and phrases
It is clear from the data presented in Section 22.2 that it is not at all unusual for affixes to
attach to compounds, and that even on phrasal bases, affixation is less unusual than intuition
might suggest. The obvious question to ask is what the factors are that influence the ability of
affixes to take compounds and phrases, and whether there is anything systematic to be said
about the phenomenon. In what follows, we will look at possible factors influencing the abil-
ity of affixes to attach to compounds and phrases.
Etymological origin seems only to be of minor significance in determining the propensity of an
affix to attach to compounds and phrases. Suffixes of native origin (‑dom, ‑ish, ‑less, ‑ness, ‑ship) do
seem to be more likely than non-native suffixes to attach to compounds and phrases, although
some non-native suffixes are quite productive on compounds and phrases (‑ery, ‑esque, ‑ism, ‑ist).
Among prefixes, however, the opposite seems to be the case: it is the non-native ones (anti‑,
mega‑, mid‑, non‑, post‑, sub‑) that show somewhat more of a tendency to accept compounds and
phrases. But again, we can find productive use of native prefixes (after‑, over‑, out‑, under‑) as well.
 22.3 Factors influencing affixation on compounds and phrases 515

Etymological origin is sometimes seen as a proxy for phonological structure in some way.
By denying an absolute correlation between etymological origin and affixation on com-
pounds and phrases, we are also implicitly denying a clear correlation between phonological
behaviour and such affixation. This is illustrated in (2) and (3), with all sorts of affixes (native
or non-native, auto-stressed or not, stress-neutral or stress-shifting) showing either
behaviour.

(2) affixes that do not appear on compounds and phrases

‑al, ‑ance, ‑ant, ‑ation, ‑eer, ‑fulA, ‑ic, ‑ical, ‑ify, ‑ity, ‑ive, ‑ize, ‑ment, ‑ory, a‑, counter‑,
dis‑, en‑, in‑, mal‑, meta‑, supra‑

(3) affixes that do appear on compounds and phrases

‑able, ‑dom, ‑er, ‑ery, ‑esque, ‑fulN, ‑hood, ‑ish, ‑ism, ‑ist, ‑ness, ‑ship, ‑y, co‑, de‑, mega‑,
multi‑, non‑, post‑, proto‑, re‑, sub‑, un‑, up‑

The phonotactics of the base–affix boundary is equally not an influencing factor. Among the
affixes that do not attach to compounds and phrases, the preponderance have a vocalic
boundary segment, but there are also some where the boundary segment is consonantal.
And there is no particular bias towards consonantal boundary segments in the affixes that do
attach to compounds and phrases; among the suffixes, for example, ‑less, ‑ness, and ‑ship have
consonantal boundary segments, but ‑ery, ‑ish, ‑esque, ‑ism, ‑ist, and ‑y do not. We must there-
fore look elsewhere for factors influencing the combination of affixes with compounds and
phrases.
Another factor that might be thought to influence the ability of affixes to attach to com-
pounds and phrases is categorial selection. Since the majority of compounds formed in
English are nominal or adjectival and verbal compounding is significantly less productive, we
might expect that affixes that attach exclusively or primarily to verbal bases would be the least
inclined to accept compounds and phrases as bases. This is true to some extent, as we find
that affixes like ‑aln, ‑fula, ‑ation, ‑ive, ‑ment, ‑ory, ‑ance, ‑ant, and ‑ure do not appear at all on
such bases. But some affixes (‑able, ‑ee, ‑er, reversative un‑, and re‑) do appear on compound
verbs. So it is not clear that categorial selection figures in a major way in the ability of affixes
to appear on compounds and phrases. While the category of the base is only one of a host of
potential restrictions on the base that might be relevant, and it may in principle be impossible
to check them all, we can find no such constraint which appears relevant in our data.
Perhaps the clearest and most significant factor in determining whether affixes will accept
compounds and phrases is the extent to which those affixes are otherwise productive. We
stress otherwise here to eliminate the risk of circularity. What we mean to say, then, is that
the closer an affix is to being fully productive on non-compounds and phrases, the more
likely it is to accept compounds and phrases as well. For example, the suffix ‑ness comes as
close as any derivational affix in English to being fully productive, and it is also perhaps the
516 Affixation on compounds and phrases

one most frequently found on compounds and phrases. Similarly with ‑dom, ‑er, ‑esque, ‑ish,
and the other suffixes shown in Table 22.1. This is not to say, of course, that all productive
affixes take such bases; the affixes ‑age, ‑an, ‑ify, ‑ation, ‑ize, a‑, counter‑, dis‑, and mis‑, for
example, do not. Whether it is a matter of their lesser degree of productivity, or other restric-
tions that prevent them from attaching to compounds and phrases is unclear at this point.
But what is clear is that high productivity in other respects is correlated with high productiv-
ity on compounds and phrases.

22.4 The role of lexicalization

So far we have concentrated on trying to determine the kinds of affixes that can attach to
compounds and phrases. In this section we will look at the question from the opposite direc-
tion: that is, it is also worth considering whether there are any restrictions on the types of
compounds and phrases to which affixes can attach. Specifically, we should consider whether
affixes can attach only to compounds or phrases which display some degree of lexicalization
(or at least item-familiarity), or whether the affixes that attach to compounds and/or phrases
attach freely to any sort of compound and/or phrase.
The data given in the tables above does suggest prima facie that item-familiar or lexicalized
compounds and phrases more frequently serve as bases for derivational affixation than item-
unfamiliar or non-lexicalized ones. However, we should not be too quick to jump to the con-
clusion that derivation is restricted to item-familiar or lexicalized compounds and phrases. For
one thing, we do find a significant number of derived forms on compounds or phrases that are
not item-familiar or lexicalized, among them: ass-kickery, folk-freudery, head-squishery, pud­
footery, merchant-ivoryesque, crazy-busyness, earth-embeddedness, pine-woodsy, piss-stinky, anti-
cookie-baking, ex-love-parade, micro-buzzbaits, mid-running-step, mini-bugsucker, multi-fall-lined,
neo-loungeheads, neo-zenbohemian, non-empty-net, non-elf-sized, semi-open-stance, sub-golfsize,
super-side-angle, ultre-deep-field, ultra-thinwall, un-chain-store-afflicted, un-dinosaur-looking, un-
dye-packed, unloveworthy, un-model-perfect, un-Paris-Hilton-looking, un-Texas-size, senior-skier‑
hood, dark-reddish, metal-lumpish, ex-olympian-turned-pro, and so on.
For another, as pointed out in Chapter 19, the more item-familiar or lexicalized a com-
pound is, the likelier it is to be written as one word or with a hyphen. This leaves open the
possibility that the high ratio of item-familiar to item-unfamiliar compounds might be due to
the way we have extracted data from COCA and other corpora. Orthographically spaced
compounds would not show up in the sort of string search that we are limited to. String
searches for affixes only work on strings that are uninterrupted by spaces. For example, with
the search string ‘ex*’ we might extract ex-baseball-player (on the item-familiar hyphenated
base baseball-player), but not ex-kazoo player (on the non-item-familiar kazoo player spelled
with a space between the two words). Both examples are compound bases, but the unlexical-
ized one would simply not show up in our searches. This methodological difficulty leads to
 22.5 Conclusion 517

an over-representation of item-familiar and lexicalized forms and an under-representation of

item-unfamiliar forms.

22.5 Conclusion
We conclude that derivation is not infrequent in English on compounds and phrases and that
few generalizations can be drawn about the conditions that favour such derivation. The
strongest generalization that emerges from our scrutiny of corpus data is that affixes that are
highly productive in other ways are also likely to be highly productive in this way, a conclu-
sion which should not come as a surprise to us. Productive affixes are more easily segmented,
that is have stronger morphological boundaries (see Chapter 21.2.1) and thus seem to be able
to also accommodate bases with stronger morphological boundaries.
 Chapter 23

Paradigmatic processes

23.1 Prospectus
In general, most phenomena of English morphology can be described in either a syntagmatic
or paradigmatic way, that is, either in terms of concatenation or in terms of a set or grid of
related words. Which of the two analyses is to be preferred for the majority of inflectional,
derivational, or compounding processes may depend on the researcher’s theoretical prefer-
ence. However, there are also a number of phenomena described in the previous chapters for
which a syntagmatic approach seems very hard to implement and which strongly suggest an
analysis in terms of paradigmatic structure or analogy.
Recent psycholinguistic research has also substantiated the important role of paradigmatic
relationships for lexical access and processing, but it is far from clear how linguistic model-
ling and theorizing can deal with the wealth of paradigmatic relationships that seem to play a
role in the processing of complex words and in the emergence of the diverse properties such
words can have in a language. In this chapter we will take a closer and more systematic look
at the pertinent phenomena and describe them in more detail. The more general implica-
tions of the data for morphological theory will then be discussed in Chapter 29.

23.2 Terminological preliminaries

In the chapters on inflection, derivation, and compounding we mention a number of cases
where the data do not lend themselves to a straightforward analysis in terms of the concatena-
tion of established morphemes into a larger string. Instead, it seems that in such cases the
resulting complex word with its specific properties can only have arisen through some
mechanism (or mechanisms) that involves other, often morphologically complex, words.
Before discussing these cases in more detail in Section 23.3, some terminological clarifica-
tions seem necessary.
In this chapter we use the term ‘paradigm’ in a wide sense to refer not only to sets of
inflectionally related word-forms of the same lexeme, but also to the morphological related-
ness of derived words and compounds. Thus, any complex word stands in a paradigmatic
morphological relationship to those words that contain the same base or the same suffix.
 23.2 Terminological preliminaries 519

Throughout this volume we have referred to sets of words with the same affix, such as leader,
reader, singer, walker, etc. as ‘morphological categories’. Sets of words with the same base, for
example impress, impression, impressive, are known as ‘morphological families’. The psycholin-
guistic reality of such paradigmatic relationships has been shown in many studies. For exam-
ple, reaction times in lexical decision tasks vary significantly across morphological categories
(Plag and Baayen 2009) and are dependent on the size of the morphological family of a given
complex word (Schreuder and Baayen 1997).
Morphological relatedness is part of an even larger web of lexical relationships. Lexical
relatedness may be based on a wide range of lexical properties, including orthographic, pho-
nological, syntactic, semantic, or pragmatic ones. These kinds of relatedness may interact
with morphological relatedness in intricate ways. As an illustration, consider rival morpho-
logical processes. Different suffixes such as ‑er, ‑ant, and ‑ist may all be used to derive agent
nouns (as in writer, discussant, artist). As these derivatives belong to the same semantic class
(that of agent nouns), they are all lexically related even though they might not share either
the same base or the same affix. Nevertheless, this relatedness may have repercussions for the
morphology since the existence of sets of lexically related forms in the lexicons of individual
speakers may influence the choice of a particular affix to form a new word in a given situation.
On a larger scale, the result of many such choices is known as morphological productivity.
Another instance of a relationship between semantic lexical relatedness and morphologi-
cal relatedness can be found with verbal inflection. As discussed in Chapter 5.3.2, irregular
verbs differ from regular verbs not only in how they inflect, but also in their degree of seman-
tic relatedness to each other and to other words in the lexicon. Irregular verbs tend to have
more semantic relations to other words, and are semantically more similar to each other than
regulars are.
In the discussion that follows we include cases where the evidence strongly suggests an
analysis that makes reference to morphological or lexical relatedness, and we will refer to
such cases loosely as ‘paradigmatic morphology’. In general, paradigmatic morphology is
based on some sort of resonance or similarity between words in the lexicon (e.g. semantic,
phonological, orthographic, etc.).
To account for such similarity-based paradigmatic effects, the notion of analogy is often
invoked, although it is rarely spelled out in detail how exactly this mechanism would work,
either in general, or in a given case. In those cases where it is, proportional relationships are
sometimes appealed to, in which the comparison of three forms leads to the computation of
the fourth form; see, for example, Becker (1993), Arndt-Lappe (2013) for discussion. Becker
(1993: 3) gives the example repeated in (1).

(1) deceive : deceivable

= syntagmatize : X (= syntagmatizable)
520 Paradigmatic processes

Apart from ‘analogy’ at least two other concepts are currently used in referring to phenom-
ena that imply the existence of paradigmatic effects in the make-up of complex words. The
first of these concepts is back-formation, or ‘back-derivation’ as Marchand (1969) also terms
it. Back-formation is traditionally defined as the deletion of an affix or affix-like sequence of
segments on analogy to pairs of base and derivative that feature the affix in question (see also
Chapters 2 and 13). A standard example is edit ◆ editor on the basis of pairs such as credit ◆
creditor or exhibit ◆ exhibitor. In spite of the common conception of back-formation as the
deletion of an affix (or something that appears to be an affix), however, back-formation is not
actually a strictly syntagmatic process. Rather, back-formation critically depends on an ana-
logical pairing of forms with and without the suffix:

(2) exhibitor : exhibit

= editor : X (= edit)

Therefore, back-formation is at least partially paradigmatic in nature and can even be viewed
as a special kind of analogy (see again Becker 1993 for discussion).
The second concept is variably known under the names of ‘paradigm uniformity’
(Steriade 2000), ‘multiple correspondence’ (e.g. Burzio 1998), ‘stem selection’ (Raffelsiefen
2004), or the ‘split-base’ effect (Steriade 1999) (see also Chapter 9.4.2 for discussion). The
variation in the terminology reflects the range of theories in which the pertinent phenom-
ena are treated, but these theoretical differences need not concern us here. All four terms
refer to the fact that complex words often show phonological properties that strongly sug-
gest influences not only from a single base, but from other words, or even a whole network
of related words, that is from a word’s paradigm. For example, humídify has the stress pattern
of humídity, not of húmid, and predáte (that is, ‘act as a predator’) has inherited the stress pat-
tern of predátion.
The notion of paradigm here again refers to sets of words that may share either the same
base or the same affix, as described above, so that there are two kinds of paradigm. What is
important in this particular context is that these two kinds of paradigm may impose conflict-
ing requirements on a given derivative. Thus, all derivatives with a given base ideally have the
same formally and semantically recognizable base, as in, for example, crystal, crystalize, crys‑
talline. If they do not, we refer to the relationship between the bases as allomorphy if the
difference concerns their phonological shape, as in, produce, production, productive, or as
semantic opacity if apparent bases do not share the same meaning, as in, for example list and
listless. At the same time, each of the different derivatives with that base has to conform to the
formal and semantic requirements of its particular morphological category. This may result in
conflicts between the two paradigmatic dimensions, for example, if a particular morphologi-
cal category imposes a prosodic restriction on its derivatives that would involve some adjust-
ment of the phonological shape of a given base, thus leading to base allomorphy. The base
explain is a case in point, as it appears as [ɪksʹpleɪn] in explain, explains, explaining, explained,
 23.3 Inflection 521

explainer, explainable, but it has different allomorphs in explanatory or explanation due to the
phonological requirements imposed by the morphological categories of ‑ory and ‑ation. Such
conflicts often lead to non-uniform paradigms (from the perspective of the morphological
family of the base, as with explain), or to the exceptional behaviour of individual derivatives
(from the perspective of a particular morphological category, as with, for example, predáte,
which does not conform to the regular stress pattern of verbs in ‑ate).
It will become clear along the way that some pertinent phenomena can be subjected to
equally convincing analyses in terms of either multiple correspondence or back-formation,
and, furthermore, that analogy is such a wide notion that it can practically be applied to any
case of back-formation and multiple correspondence. In the next sections we give an over-
view of paradigmatic effects in English morphology, with many of them having already been
mentioned in previous chapters. In this overview we will use the terms ‘analogy’, ‘back-for-
mation’, and ‘multiple correspondence’ to informally label and classify different kinds of
phenomena in a preliminary fashion. In the following subsections we will remain largely
agnostic as to what kind of mechanism might be best suited to formalize and explain the
pertinent cases. This problem will be taken up in Chapter 29.

23.3 Inflection
We saw in Chapter 5 that the patterning of regular and irregular verb-forms in the different
ablaut classes and the behaviour of nonce verbs in experiments suggests that irregular verbal
inflection is best accounted for by some kind of analogical mechanism, according to which
verbs of a particular phonological make-up tend to show similar inflectional behaviour. For
example, the variation patterns found with verbal lexemes that have variable preterite forms and
past participles suggests that the preferential a–b–b variant ablaut pattern (e.g. keep, kept, kept)
attracts verbs on the basis of phonological similarities. In the same vein, ablaut forms in general
are most likely with monosyllabic bases that also share some segmental characteristics (such as
/s/-initial onset, /ɪ/ as their base vowel, or a final velar nasal, in various combinations).
In noun inflection, the variation in the pluralization of foreign loan words, but also occa-
sionally of native words, suggests the work of analogy, too, as shown in the forms in (3). The
variant ‑i plural of octopus is non-etymological and apparently coined on the basis of similar-
sounding pluralized words of Latin origin. The variant plural sistren is found as a correspond-
ent to the male brethren.

(3) alumni ◆ octopi ◆ octopuses

brethren ◆ sistren ◆ sisters

With regard to genitive marking, there is a priming effect in the choice of synthetic marking
with ‑’s and periphrastic marking with of. Given a particular noun in a genitive choice
522 Paradigmatic processes

context, the presence of a particular genitive variant in the preceding discourse influences the
choice of the variant with the given noun (see Chapter 7). This means that the exponent of
the genitive depends not only on the properties of the noun phrase in question but also on
other exponents of the same morphosyntactic category that happen to be co-activated at the
moment of production. A similar online priming effect can be found in adjectival inflection,
that is with the synthetic versus periphrastic encoding of the comparative and superlative (‑er
versus more, ‑est versus most, see Chapter 6).

23.4 Derivation
In derivation, we find many cases of back-formation into verbs. A large number of verbs in
‑ate are back-formed from diverse kinds of complex noun. The evidence for this kind of analy-
sis comes mainly from their semantic or phonological properties, which clearly relate them
to their complex bases (see Chapter 13 for detailed discussion). Where available, the earliest
dates of attestation often support a back-formation analysis. The forms in (4) illustrate some
uncontroversial cases of back-formation. Those in (5) are cases of multiple correspondence
with derivatives from various categories.

(4) advection ◆ advect, escalator ◆ escalate, escalation ◆ escalate, fabrication ◆ nano-

fabrication ◆ nano-fabricate, optimized ◆ nano-optimized ◆ nano-optimize

(5) húmid ◆ humídity ◆ humídify, cátholic ◆ cathólicism ◆ cathólicize, confér ◆ cónference ◆

cònferée, remedyn/v ◆ remédial ◆ remédiable

The forms in (6) seem especially interesting as, depending on the theoretical background,
they can be analysed with equal justification as being back-formations, cases of multiple cor-
respondence, and analogical formations.

(6) formátion ◆ formáte ‘fly in formation’, gelátion ◆ geláte, lenátion ◆ lènáte, notátion ◆ notáte
(~ nótate), olátion ◆ oláte, predátion ◆ predáte, solátion ◆ soláte, solvátion ◆ solváte

They may be seen as back-formations as they are based on corresponding complex

nouns, they may be analysed as cases of multiple correspondence as their morphosyntax
is that of other ‑ate formations, but their stress pattern is not in accordance with the regu-
lar pattern of ‑ate verbs. Instead, it is faithful to the complex base from which the verb is
derived. And, since the creation of the forms in (5) involves patterns of similarity that
could be expressed in terms of proportional analogy, all these forms can be classified as
analogical formations, too.
The items in (7) are ‑ate formations that are also based on individual (sets of) model words,
but do not involve back-formation (all twentieth-century neologisms from the OED).
 23.4 Derivation 523

(7) active ◆ activate ◆ inactive ◆ inactivate

active ◆ radioactive ◆ activate ◆ radioactivate
regular ◆ regulate ◆ stereoregular ◆ stereoregulate
plasticate ◆ masticate

Finally, (8) presents a similar case of analogy based on a single model word, but outside the ver-
bal domain, and also without the paradigmatic support of a morphological category. Thus, before
the coinage of the new words, singleton was the only established word with the suffix ‑ton.

(8) singleton ◆ doubleton ◆ tripleton

We also count as paradigmatic morphology sets of words with related suffixes in which every
word of one morphological category has a potential correspondent in another category.
Sometimes all derivatives with the same base are attested, sometimes not. Consider, for
example, the suffixes ‑ism, ‑ist, and the combination ‑ist‑ic. For each abstract noun in ‑ism
there is a potential personal noun in ‑ist denoting the follower of the theory or framework
denoted by the corresponding form in ‑ism, and a potential adjective in ‑ist‑ic (for example,
communist ◆ communism ◆ communistic; anarchist ◆ anarchism ◆ anarchistic). We are faced in
such cases with the curious situation that the meaning of the personal noun in ‑ist and the
adjective in ‑ist‑ic incorporate the meaning of the abstract noun in ‑ism, although the abstract-
noun-forming suffix is no longer present in the personal noun or adjective.
This is illustrated with two pertinent forms in (9), from COCA. The derivative boycottist
in (9a) is not synonymous with the agent noun boycotter ‘someone who boycotts’ but means
something like ‘someone supporting boycottism’, with boycottism referring to the ‘use of boy-
cotts as a form of protest or punishment’ (OED). Similarly, as shown in (9b), an abstractionist
is someone who is part of the artistic school of abstractionism, as implicitly indicated by the
adjective major preceding the derivative in question.

(9) a. Africa Today 1990: The ANC itself, as well as the CP and the COD, supported
candidates for parliamentary elections in the 1950s and only moved gradually to a
totally boycottist position (and had cordial relations with Buthelezi until 1979).
American Craft 2003: Marvin Lipofsky explored many forms in his artistic
b. 
development, achieving a style that defines him as a major abstractionist.

There are other examples of morphological categories that appear to participate in paradig-
matic relationships. One of these is the set ‑ate, ‑ant, and ‑ance (e.g. hesitate ◆ hesitant ◆ hesi‑
tancy, alternate ◆ alternant ◆ alternance, participate ◆ participant ◆ participance) discussed in
Chapter 10. In a case like this, it is impossible to decide whether the base of the ‑ance nominal
is a verb in ‑ate or an adjective in ‑ant. Other cases of paradigmatic morphology that have
been mentioned in earlier chapters are the corresponding forms in ‑ive and ‑ify (subjective ◆
subjectify) and those in ‑ize and ‑ism (baptize ◆ baptism) (see Chapter 13).
524 Paradigmatic processes

From the point of view of paradigmatic relatedness, it is quite striking that new words of a
particular morphological category are often coined in the textual neighbourhood of morpho-
logically related forms with the same base, but a different suffix. This is illustrated in (10) (see
also (21)). Examples (10b) to (10e) are from COCA.

(10) a. The fact is, saying Yes hadn’t been a pointless exercise at all. It had been pointful
(Wallace 2008:394, italics in the original)
b. Parenting 2001: Both the haircutter and the haircuttee will need a little psychological
preparation.
Today’s Parent 2000: Whether your child is the snubber or the “snubbee,” your
c. 
discreet guidance now is going to go a long way toward laying the groundwork for
the turbulent teens.
d. Outdoor Life 2007: For the observer, it may seem as if the falling person arrives at his
landing spot in a mere fraction of a second. For the fallee, however, there is a great
deal of time to think.
Atlanta Journal Constitution 2005: Compared to my colleagues who struggle with
e. 
more challenging commutes, I’m a lucky man. . . . Trim 8-hour commute, then let’s
talk # You can learn a lot being a “commutant.”

23.5 Compounding
In compounding, there are many forms that are modelled on particular existing compounds,
with the new compound inheriting crucial components of the institutionalized meaning of
the model compound. A selection of such compounds is given in (11).

(11) aftermath ◆ beforemath (Wordspy)

carpooling ◆ cowpooling ‘Purchasing a whole cow or side of beef from a local farmer
and sharing the cost among multiple families’ (Wordspy)
earthquake ◆ seaquake ‘a submarine eruption or earthquake’ (OED)
hangover ◆ hangunder ‘the funny feeling you get when you wake up after a night of not
drinking and you’re not hungover like usual’ (Urban Dictionary)
honeymoon ◆ babymoon ‘romantic vacation during a pregnancy’ (COCA)
horse whisperer ◆ dog whisperer ‘A person who has, or claims to have, a natural ability
to relate to or connect with dogs’ (Wordspy)
housewife ◆ househusband ‘male partner who carries out the household role and
duties traditionally associated with a housewife’ (OED)
rat race ◆ mouse race ‘lower-stress life-style’ (Wordspy)
strawman ◆ straw poll ‘an apparent but not real poll, a test poll’ (COCA)
 23.6 Splinters 525

Back-formation also frequently leads to non-canonical compounds, as is frequently the case

with verbal compounds, illustrated in (12).

(12) a. denominal: babysitter ◆ babysit, brainwashing ◆ brainwash, chainsmoker ◆ chain-

smoke, crash-landing ◆ crash-land, ghetto-blaster ◆ ghetto-blast
b. de-adjectival: horror-struck ◆ horror-strike, tailor-made ◆ tailor-make

There are also some general effects of morphological families with compounds. As discussed
in more detail in Chapter 19, there is very good evidence that stress-assignment to NN com-
pounds is largely predictable from the stress-behaviour of related compounds that have the
same left or right constituent (e.g. Plag et al. 2007; Plag 2010; Arndt-Lappe 2011). Such con-
stituent families are also predictive of compound stress in a different way. The smaller the
constituent family of the second constituent of a given compound, the more likely it is that
the NN compound will be right-stressed (Bell and Plag 2012).
Similar paradigmatic effects can be found with compound interpretation. Psycholinguistic
experiments have shown that the semantic interpretation of new compounds heavily relies
on the distribution of pertinent meanings in the pertinent constituent families (Gagné and
Shoben 1997; Gagné 2001). Consider, for example, the right constituent family of mountain
magazine, which in the majority contains compounds for which the semantic interpretation
is one in which the left constituent denotes something that is the main topic of the press
organ denoted by the second constituent.

23.6 Splinters
Another pertinent phenomenon, which shows a mixture of syntagmatic and paradigmatic
traits, is the emergence of splinters, which were defined in Chapter 2 as originally (mostly)
non-morphemic portions of a word that have been split off and used in the formation of new
words with a specific new meaning. The use, recognition, and interpretation of a splinter as
found in a new word thus necessarily requires a paradigmatic relationship. Interestingly, the
new morpheme, that is the splinter, receives its new meaning only through that paradigmatic
relationship. An oft-cited example of a splinter which still yields the occasional neologism (e.g.
Breastgate in 2010, COCA) is ‑gate ‘scandal, disaster’, based on Nixon’s Watergate scandal. The
meaning assigned to the string ‑gate only emerged when the first new word containing it was
coined on the basis of a semantic and formal similarity to the well-known existing word. The
birth of splinters is thus a paradigmatic process itself, and so is the coinage of further similar
words, at least until the new formative is established with that meaning in the lexicon. Given
that splinters seem to be only moderately productive, it may even be the case that most of them
are never firmly established as bound morphemes. But when they do become more productive,
they may even start a life as a free form, as in the case of burger ‘patty served on a bun’.
526 Paradigmatic processes

In (13) to (30) we present some more splinters with their meaning paraphrased, their pre-
sumed model word(s), some of their derivatives, and one example in context. Examples are
from COCA, unless otherwise specified; the orthography found in COCA is preserved:

‑ati ‘elite or would-be elite group related to base word’: literati ◆ arterati ◆ chatterati ◆
(13) 
designerati ◆ glitterati ◆ hip-erati ◆ hipsterati ◆ jazzerati ◆ Niggerati ◆ Twitterati
U.S. News & World Report 2000: Both Burns and Ward see their work as an attempt to

reach beyond the factional discord of the “jazzerati” to introduce nonaficionados, and
particularly younger Americans, to the story of one of the nation’s most distinctive
musical traditions.

(14) ‑bot ‘robot’: Cajunbot ◆ Hotbot ◆ Frankenbot ◆ Spybot ◆ Stickybot

National Geographic 2009: Engineers studied these terrific toes to make a robot called

Stickybot. Stickybot uses a material like the hairs on the gecko’s toes.

(15) 
‑burger ‘patty served on a bun’: hamburger ◆ bambiburger ◆ beanburger ◆ beefburger ◆
bisonburger ◆ buffaloburger ◆ cheeseburger ◆ fishburger ◆ nutburger ◆ pauaburger
(found on the menu of a New Zealand restaurant) ◆ peaburger ◆ quailburger ◆
shrimpburger ◆ slugburger ◆ tofuburger ◆ steakburger ◆ kosherburger ◆ kiwiburger
(Note that in the last two examples, the first element is not an ingredient. A kosherburger is
just a burger that is kosher, and a kiwiburger a kind of burger advertised in New Zealand,
rather than a patty made of kiwi.)
Southern Living 2003: There are as many stories in Corinth surrounding the history of

the slugburger as there are places serving them (and that’s a lot).

(16) 
‑delic ‘wild, mind-altering’: echodelic ◆ folkadelic ◆ funkadelic ◆ plaque-adelic ◆
psychedelic ◆ sample-delic ◆ scallydelic ◆ shagadelic ◆ shankadelic ◆ slumadelic
Rolling Stone 2003: “We call it slumadelic,” said Big Boi of OutKast’s far-reaching blend

of hip-hop, funk, rock and otherworldly sounds.

(17) 
‑holic ‘person addicted to’: bagoholic ◆ beanoholic ◆ chocoholic ◆ dealaholic ◆ eventaholic
◆ fishaholic ◆ hoardaholic ◆ playaholic ◆ sexoholic ◆ workaholic ◆ yogaholic
Prevention 1990: If today Bob is a “yogaholic,” it has a lot to do with the fact that up

until four years ago he was a workaholic.

(18) ‑illion ‘a very large number’: million ◆ billion ◆ trillion ◆ zillion ◆ bazillion ◆ bajillion ◆
frillion ◆ jillion ◆ gajillion ◆ gazillion ◆ kajillion ◆ kazillion
Scholastic Scope 2007: I mean, like I was going to give myself a facial, touch up my

highlights, pick out an outfit, and about a gajillion other things.

(19) ‑tarian ‘someone with a diet restriction’: vegetarian ◆ eggitarian ◆ flexitarian ◆ fruitarian
◆ pescetarian ◆ pollotarian ◆ value-tarian
 23.6 Splinters 527

USA Today 1997: Parnevik first tried volcanic sand after meeting a Swedish “fruitarian”

who sold him on the benefits of an all-fruit diet, which purports to clean the toxic
waste from your system.

(20) 
‑licious ‘appealing’: delicious ◆ bagelicious ◆ barfalicious ◆ bootielicious ◆ diva-licious ◆
goodielicious ◆ lowcarbolicious ◆ pig-a-licious ◆ snugglelicious ◆ thug-a-licious ◆
weavealicious
So not the drama 2007: Cool, laid-back Sarah was the cream, light cocoa-complexioned,

thick curly hair to her ears, to Jessica’s coffee-bean skin, broad nose, thick lips, and
extremely straight weavealicious hair down past her shoulders.

(21) 
‑matic ‘automatic (device)’: automatic ◆ airmatic ◆ bowlamatic ◆ chop-o-matic ◆
dripomatic ◆ futurmatic ◆ inflatomatic ◆ mince-o-matic ◆ randomatic ◆ sculpt-o-matic ◆
towelmatic ◆ vote-o-matic
ABC_20/20 1993: How many of those did you sell? Mr. POPEIL: Oh, hundreds of

thousands. Hundreds of thousands of the Chop-O-Matic, Dial-A-Matic, Veg-A-
Matic, Mince-O-Matic, the smokeless ashtray, Miracle Broom, the Ronco Clean Air
Machine.

(22) ‑o ‘language production error’: typo ◆ speako (Wordspy)

Speech recognition systems are seeing increased use in warehouses, although users
must avoid substituting ‘speakos’ for ‘typos’ when designing the system. (2001 cited in
Wordspy)

(23) ‑orama ‘event or display of considerable size’: panorama ◆ Adorama ◆ Audiorama ◆

Astrorama ◆ cashorama ◆ Censorama ◆ Clitorama ◆ comporama ◆ fawnorama ◆ Junk-o-
rama ◆ Motorama ◆ Pianorama ◆ Pornorama ◆ Sensorama ◆ spin-orama ◆ Sponge-o-
rama ◆ Teaserama ◆ Striporama ◆ tack-orama ◆ voodoorama
Christian Science Monitor 2007: I wrote about the sport in the 1960s and 1970s, which

was the medieval age compared with today’s big-bucks cashorama. The players then
uniformly were underpaid.

(24) 
‑rific ‘extremely (good), characterized by’: terrific ◆ ballerific ◆ calorific ◆ cheeserific ◆
moisturific ◆ slow-jammerific ◆ splatterific ◆ stuporific ◆ terrorific ◆ Tiggerific ◆ Twitterific
Entertainment Weekly 2007: The “Run It” singer has love—or at least lust—on his mind

with songs like the T-Pain-featuring “Kiss Kiss,” and the slow-jammerific “Take You
Down.”

(25) 
‑scape ‘view or picture of a scene or scenery’: landscape ◆ Aquascape ◆ cityscape ◆
dreamscape ◆ hardscape ◆ Netscape ◆ memoryscape ◆ moonscape ◆ manscape ◆
nightmarescape ◆ streetscape ◆ soundscape ◆ townscape ◆ tablescape ◆ Xeriscape
Studies in the Novel 2008: These representations associate images of the slave trade with

typical post-traumatic responses in characters who live hundreds of years after that era
of violence and loss that marred the African memoryscape.
528 Paradigmatic processes

(26) 
‑stan ‘country’: Afghanistan ◆ Pakistan ◆ Bin Ladenstan ◆ divorcistan ◆ Londonistan ◆
Islamistan ◆ New Yorkistan ◆ Refugeestan ◆ Richistan ◆ skate-istan ◆ Youngistan
The song is you 2009: Julian Donahue married in optimistic confusion, separated in

pessimistic confusion, and now was wandering toward a mistrustful divorcistan, a
coolly celibate land.

(27) 
‑tainment ‘entertainment’: entertainment ◆ advertainment ◆ agritainment ◆
asthetainment ◆ autotainment ◆ charitainment ◆ communitainment ◆ dinotainment ◆
edutainment ◆ eatertainment ◆ exertainment ◆ infotainment ◆ infomerciatainment ◆
intertainment ◆ kid-ertainment ◆ newsy-tainment ◆ psychotainment ◆ shoppertainment
◆ shop-o-tainment ◆ skankertainment ◆ retailtainment ◆ technotainment
Denver Post 2002: The Mills Corp. and developer Greg Stevinson are spending $ 300

million to build the ‘shoppertainment’ mecca, anchored by discount outlets of
Neiman Marcus and Saks Fifth Avenue, SuperTarget and Gart Sports.

(28) 
‑tronic ‘electronic, futuristic’: electronic ◆ animatronic ◆ conceptronic ◆ daytronic ◆
easytronic ◆ ethnotronic ◆ geartronic ◆ mediatronic ◆ parktronic ◆ psychotronic
Futurist 2000: Using credit-card databases, the companies develop profiles of road users

and adjust their mediatronic and other outdoor displays to offer products and services
of interest to drivers with these profiles.

(29) 
‑ware ‘software’: software ◆ crimeware ◆ eduware ◆ freeware ◆ malware ◆ netware ◆
spyware ◆ trialware ◆ CaptureWare
PC World 2007: Then there’s trialware, or preloaded software that functions only for a

set period, generally 30 to 90 days.

(30) 
‑zilla ‘overbearing exemplar’: Godzilla ◆ bridezilla ◆ Babyzilla ◆ Clonezilla ◆ Cordzilla ◆
cowzilla ◆ Dogzilla ◆ ErrorZilla ◆ groomzilla ◆ Hogzilla ◆ Nunzilla ◆ Ringzilla ◆ Rockzilla
◆ Shopzilla ◆ Stopzilla ◆ teenzilla
Money 2006: I’m fairly certain that I’m not the only parent who doesn’t want a teenzilla

living in the upstairs bedroom

There are a number of interesting observations to be made. First, some of the splinters have
the same meaning as the full forms on which they are originally based. This is, for example,
the case with ‑bot, ‑tainment, and ‑ware (in computer terminology), perhaps even with ‑stan.
This being the case, the pertinent new derivatives could also be analysed as blends, as we
will discuss below. What may speak against such an analysis, however, is the general struc-
ture of blends. Thus, in their vast majority, the first elements of blends lose some of their
phonological material (see Chapter 19), whereas the forms in (14), (26), (27), and (29) do
not show such a clear patterning. Nevertheless, those forms that do feature phonological
loss in the first part, such as intertainment ‘internet entertainment’ or exertainment ‘exercise
 23.6 Splinters 529

entertainment’, seem to lend themselves to a straightforward analysis in terms of blending.

In general, a blend analysis is impossible in those cases where the splinter carries a meaning
that is different from that of the original word. This is actually the case for the majority of
splinters shown in (13) to (30), that is for ‑ati, ‑illion, ‑tarian, ‑o, ‑orama, ‑rific, ‑scape, ‑zilla. In
sum, only a small minority of potential splinter derivatives lend themselves to a convincing
analysis as blends.
Second, some cases in (13) to (30) lend themselves more than others to an analysis as
compounds instead of derivatives involving splinters. Formations with ‑burger, for example,
are clearly more on the side of compounding than ‑tarian, as burger is also used as a free form
by now, but with bot it is perhaps unclear whether it is a free form for speakers outside the
robotics scene, in spite of its being attested as a free form in COCA. With most other splin-
ters a compound analysis is certainly not convincing. For instance, although ware is a free
form and can be found in many compounds (such as glassware, hardware, silverware, stone‑
ware, or warehouse) the splinter ‑ware has a distinct meaning ‘software’ and seems not to be
usable as free form with that meaning. Again, all other splinters are clearly bound morphs,
which speaks against a compound analysis.
Third, splinters may easily become polysemous, as for example, with value-tarian, in which
the object of the restriction is extended to non-food consumables, as shown in (31). Another
case in point may be ‑gate, whose meaning oscillates between ‘scandal’ and ‘disaster’, illus-
trated in (32).

Money 1995: Bob Rodriguez, 46, who has led $181 million FPA Capital since 1984, is a
(31) 
devout value-tarian. He buys only small-cap stocks that look cheap on the basis of
their earnings or book value (the per-share worth of the company’s assets).

San Francisco Chronicle 1992: Perhaps eventually we will look back on our government’s
(32) 
performance at the Earth Summit as an aberration, an environmental Riogate or Bay
of Pigs, a lapse that woke us up to the need for our country once again to become the
leader of the Environmental Revolution that is sweeping the globe.

Fourth, as indicated by the many forms in the corpus that are found with capitalization of the
first letter, a large number of words with splinters are proper nouns, often referring to brand
names or names of specific products. This suggests that coinages using splinters are perceived
as rather unusual and are seen as linguistic forms that attract more attention than many other
words would. Such a view seems justified on psycholinguistic grounds since the online rec-
ognition and interpretation of pertinent forms seems to require more processing effort than
simple lexical access to a memorized form would.
Fifth, as splinters heavily rely on analogy it is not surprising that we find variability in the
possible phonological shape of formations. In other words, the degree of phonological simi-
larity to the model word may not be uniform across formations, as can be seen with retailtain‑
ment versus shop-o-tainment versus kid-ertainment, or eggitarian versus fruitarian.
530 Paradigmatic processes

It is also worth mentioning that splinters seem to have a strong tendency to emerge from
nouns and to derive nouns. Although there are several splinters that derive from adjectives
(‑delic, ‑holic, ‑licious, ‑matic, ‑rific, ‑tronic), even among these the new formations are often
nouns (especially proper nouns) rather than the expected adjectives.
Finally, a closer look at the various sets of forms shows that the initial elements often share
certain semantic features. This may be due to the meaning of splinter, as in the case of ‑tarian,
which, due to its own semantics, primarily takes bases that denote food items. However, even
when the meaning of the splinter is not so restrictive, we still find rather coherent sets of
bases. For example, ‑zilla derivatives seem to be primarily based on animate bases, and ‑ati
derivatives have bases denoting concepts related to art, music, and culture. Note also the
choice of speak as a base for ‑o, which is semantically closely related to type, which in turn
leads the highly specific meaning of the splinter ‑o assumed here. All of this suggests that not
just any form is selected by speakers to participate in analogical processes, but that these
forms are selected on the basis of already existing, more or less salient similarities.

23.7 Conclusion
In this chapter we have looked at morphological processes that seem to necessitate an
analysis in terms of paradigmatic structure and analogy. We have seen that such processes
are too numerous to be dismissed as isolated exceptions in an otherwise well-behaved
system. Paradigmatic morphology can be found in any subdomain, that is in inflection,
derivation, and compounding, and paradigmatic processes are based on orthographic,
phonological, morphological, semantic, and syntactic similarities between related forms.
Furthermore it can be observed that the computations leading to new word-forms or new
lexemes seem generally to suggest an analysis that makes use of proportional analogy.
Notably, we often find a web of words contributing to the emergence of a new word,
instead of only a single model word. Any theory of morphology, or of analogy, would need
to be able to account for these phenomena. We will return to this subject in Chapter 29,
where we will consider how paradigmatic facts can best be accounted for, how general a
mechanism of analogy might be needed, and how the difference between regular and idi-
osyncratic cases might be dealt with.
Pa rt VI

Themes
 chapter 24

Inflection versus derivation

24.1 Prospectus
In this chapter we take up the vexed question of the difference between inflection and deriva-
tion as it applies to English. We consider the kinds of criteria that might distinguish between
the two, the value of the distinction where English is concerned, and features of English that
make applying the distinction awkward.

24.2 The basis of inflection and derivation

The distinction between inflectional morphology and derivational morphology is an ancient
one. Fundamentally, it is a matter of the means used to create new lexemes (derivational affixes
among other processes) and those used to mark the role of the lexeme in a particular sentence
(accidence, inflectional morphology). In the description of many languages, including most
obviously the so-called inflected (flectional, fusional) languages, the distinction is clear and
uncontroversial in the vast majority of instances. The problem with the distinction has come,
historically, in giving a definition to the two types of morphology, particularly a definition
based on necessary and sufficient criteria. Part of the difficulty here arises because the distinc-
tion seems to depend on a recognition of the distinction between lexemes and word-forms of
lexemes, yet that distinction is defined in terms of the distinction between inflection and deri-
vation, so that there is a constant danger of having a circularity built into the definitions. In
Chapter 2 we attempted to avoid this particular problem by providing an ostensive definition
of inflection for English, and allowing everything else to be derivation by default.
The literature provides us with at least two ways of dealing with this awkwardness. The first
is to define inflectional and derivational morphology as canonical categories rather than as
classical categories. This approach is taken by a number of scholars, including Scalise (1988),
Dressler (1989), Plank (1994), Stump (1998), Bauer (2003b). These scholars prioritize differ-
ent features of inflection or derivation in their discussions, and Plank (1994) gives a lengthy
list of possible features of one or the other prototypical category. This includes features such
as the following.
534 Inflection versus derivation

• Inflection is typically formally regular; derivation may not be.

• Inflection is typically semantically regular; derivation may not be.
• Inflection is typically fully productive; derivation typically shows (unmotivated) gaps
in productivity.
• Inflection does not add significantly to the meaning of the base, but contextualizes that
meaning; derivation adds to the meaning of the base.
• Inflection allows the prediction that a form will exist to cover some notion; derivation
does not allow the same security of prediction.
• Inflection does not change the major category (noun, verb, etc.), but derivation may.
• Inflection is what is relevant to the syntax; derivation is not syntactically relevant.
• You cannot replace an inflected form with an uninflected form in a sentence, but you can
replace an inflectional base (which may have been created by derivational morphology)
with a morphologically simple form (that is, one belonging to a different lexeme).
• Where both occur in the same word-form, derivational affixes are typically closer to the
root than inflectional ones.

These features are an attempt to get at the canonical characteristics of inflection and deriva-
tion, but none of them quite gets to the heart of the distinction. Inflection must be available
to be added to words which the speaker has never encountered before, its meaning and form
must thus be regular, and the precise form is determined by the syntactic structure in which
it occurs. Derivation changes lexemes, which may mean changing word class, but provides
the input to inflection, on a par with morphologically simple bases. Some fusional languages
provide challenges for the individual criteria or the distinction as a whole, but do not invali-
date the basic division. They may, for instance, have lexemes which do not inflect, they
may have defective paradigms, in which certain expected forms do not occur, they may order
some affixes in unexpected ways, they may have very regular derivation, particular semantic
categories may belong to an unexpected morphological category. The existence of such
instances does not upset the canonical distinction. Neither does the fact that the application
of individual features may be difficult to determine in particular instances.
Particularly important among these criteria has been the notion that inflection is what
is relevant to the syntax (Anderson 1992). But this criterion presupposes that we are clear
on what we mean by ‘relevant to the syntax’. The problem is that the decision as to what
is relevant to syntax is at least in part theory bound, so that adherents of one theory might
be inclined to see as inflectional what in another theory might be deemed derivational.
A second (and not necessarily mutually exclusive) way to deal with the awkwardness of the
inflection–derivation distinction might be to break inflection and derivation down further into
subcategories. However, the only further division which has become generally accepted is that
between inherent and contextual inflection (Booij 1996a). Inherent inflection is that inflection
which is meaningful for the sentence, representing distinctions that are semantically intrinsic to
 24.3 Some problem cases 535

specific syntactic categories—the expression of tense on the verb, for example, the expression
of person and number on nouns, or the expression of degree (comparative and superlative) on
adjectives. Inherent inflection is thus conceptually closer to derivation, in the sense that choice
of a particular form (say a singular noun) affects the semantic interpretation of the sentence.
Contextual inflection, on the other hand, is entirely driven by syntactic environment: for exam-
ple, the expression of subject person and number on the verb, or the expression of case on the
noun are all instances of contextual inflection.
Subdivisions of the category of derivation are not widely accepted, and yet it is well-known
that in some languages evaluative affixation (diminutive and augmentative affixes) often behaves
as a rather unusual instance of derivation: it may re-apply to its own output; the order of diminu-
tive affix and plural affix may not be as predicted by general rules (see Bauer 1997a); and evalua-
tive affixes may occur in paradigmatic relationships with more canonically inflectional categories
like singular and plural. Accordingly, it seems reasonable to assume that the category of deriva-
tion is not monolithic either, even if the divisions within it are not so well established.
In principle, at least, it might be open to an investigator to claim that there is, in a given language,
no distinction to be drawn between inflection and derivation. To our knowledge no such claim
has ever been made about English, but various points have been made about English which indi-
cate that it is not necessarily clear where to draw the line between the two categories in English.

24.3 Some problem cases

In this section we examine some of the problem cases that have been raised in the literature
or where we are aware that evidence for a particular affix’s being either inflectional or deriva-
tional is compromised.

24.3.1 The nominal plural

In Chapter 7 we defined the plural as part of inflection in English. On the other hand, Beard
(1982), within a framework in which derivation (in his terms, the process of changing the
semantic or morphosyntactic content of a lexeme) is distinguished from affixation (the actual
process of adding phonological material to a base), argues that the plural, across Indo-
European languages, is a derivational category.
First, he argues that plural marking in English, like derivational marking, is formally irregu-
lar. The existence of plurals such as alumni, children, deer, feet, matrices, mice, oxen, wives and
so on illustrates this. The problem with this claim is that it is formulated as a claim about
classical categories: if something does not fit into category C, it must fit into category not-C.
If, on the other hand, we look at our categories as being prototypes, then a few examples that
fail to show the feature of formal regularity do not discredit the notion that the prototype is
536 Inflection versus derivation

still relevant. Of course, we may then get into an argument as to how many irregular forms
there are and how many there need to be for the feature not to hold, an argument which, it
seems to us, would be unfruitful. But we note that in Chapter 7 we pointed out that Latin
plurals are usually not obligatory, and that Italian plurals are really cases of code-switching
and not cases of English inflection. In other words, the number of irregular forms that are
obligatorily part of the English system is considerably smaller than Beard would imply.
The next argument for considering plural-marking in English as derivational is that there
are a number of nouns marked with the plural morph which fail to behave like plurals: blues,
greens, measles, oats, scissors, shorts and so on. However the semantics of such forms is to be
represented, grammatically most of them are treated as plurals by the grammar. Your greens/
oats/scissors/shorts are/*is on the table. Measles is treated variably by speakers of English (pos-
sibly on a geographical basis). Some words like draughts, fives, ninepins are either singular by
elision—(a game of) draughts—or by metonymy.
A further argument is that many nouns do not have plural forms, so that the plural is not
fully productive. It is certainly true that some nouns such as knowledge, mahogany, peace, pork,
waving are unusual in the plural form, but we argue elsewhere (Chapter 25) that nouns that
normally lack plural forms are almost always subject to coercion ( Jackendoff 1991; Pustejovsky
1995), such that given the right context they are quite naturally inflected for the plural. The
example in (1) from COCA is illustrative.

(1) Chicago Sun-Times 1996: Armour now has a line of seasoned and marinated porks.

To summarize, although an argument can be made that the plural sporadically fails to exhibit
a number of the features of inflection, for the most part plural marking is consistent with a
canonical notion of what inflection is.

24.3.2 Adverbial marking with ‑ly

Traditionally, because ‑ly produces adverbs from adjectives and so brings about a change in
word class it has been treated as derivational. However, several theorists, among them
Haspelmath (1996) and Giegerich (2012) have argued that it is inflectional. We have consid-
ered the status of this suffix at length in Chapter 15, concluding that the evidence is inconclusive,
and we have been very tentative in our classification elsewhere as a result. We will not reiterate
those arguments here, but refer the reader to Chapter 15, where they may be found.

24.3.3 Ordinal ‑th

Ordinal ‑th is rarely discussed in the literature on inflection and derivation, perhaps because its
status is not clear. First we have to recognize that there is considerable suppletion in the
 24.3 Some problem cases 537

derivation of ordinals, with first, second, and third replacing *oneth, *twoth, *threeth. Suppletion,
except in the case of proper names, is probably more usual in inflection than in derivation in
English (see Chapter 2). However, it is not clear that this argument is helpful: unless we can
prove in some independent and non-circular way that suppletion is unique to inflection, the
fact that there is suppletion in these forms cannot be taken conclusively to prove their status
as inflectional. The suffix ‑th is formally regular (apart from the suppletion), semantically regu-
lar, productive (see below), and obligatory. Whether or not the change from cardinal to ordinal
number involves a word class change is perhaps controversial. Yet the change from five to fifth
is probably felt as being lexical rather than grammatical: standard dictionaries tend to treat
these words as separate lemmas and as independent lexical items linked by their meanings.
The criterion of productivity is also problematic since the productivity of ‑th can be viewed
two ways (Bauer 2001: 148). Either there is no productivity, because ‑th is already listed in
virtually every possible environment, and 32,300,426th simply illustrates a new occurrence of
sixth, or 32,300,426th is a new application of ‑th. The application of ‑th to words like zillion and
squillion and its application to non-numerals like n and umpteen argue for a limited degree of
productivity, even if the first of these two solutions is adopted. However, the first solution
suggests that we are dealing with derivation rather than with inflection.
As with other suffixes dealt with here, the features of ‑th split between inflection and
derivation.

24.3.4 Other numerical formatives: ‑teen and ‑ty

Both ‑teen and ‑ty have things in common with ordinal ‑th. The question of their productivity
is dubious, for the same reasons as were presented above (although, again there are occa-
sional productive uses as in umpteen, umpty). They can be followed by ‑th, which gives
ambiguous information as to whether they are inflectional or derivational, as was seen in
Section 24.3.3. They can be followed by ‑fold, whose semantics suggest that it is a derivational
affix rather than a compound element. Both ‑teen and ‑ty are formally and semantically regu-
lar, though with suppletion for eleven and twelve. They both seem to be treated by standard
dictionaries as lexical affixes. Overall, they both look rather more derivational than ordinal
‑th, but they do not precisely match the canonical pattern.

24.3.5 The participles

When forms ending in ‑ing and ‑ed (or one of its irregular congeners) are used in the creation
of complex verb groups there is no doubt as to their status as inflectional affixes. So the
instances in (2) illustrate inflectional ‑ing and ‑ed. However, when the same forms are used to
create words used as premodifiers (3) or when ‑ing is used to create more nominal forms (4),
there is doubt as to their status in the inflection/derivation categorization.
538 Inflection versus derivation

(2) He is driving to Cologne today.

She has implicated one of her colleagues.

(3) It’s the driving dynamics that make the Toyota appealing. (adapted from a COCA example,
Popular Mechanics 2009)
The implicated students deny any involvement.

(4) The driving of speedboats requires immense concentration.

His driving is atrocious.

We have the option, of course, of viewing the ‑ing in (2) as being a different affix from the ‑ing
in (3) and (4), or of seeing the examples in (3) and (4) as some form of conversion or
­zero-derivation (though see Chapter 25). It is also possible to see the examples in (3) and (4)
as instances of class-changing inflection as suggested by Haspelmath (1996). These examples
will be taken up in more detail in Chapter 25, where we suggest that any decision as to
whether the participles should be treated as instances of derivation or inflection is an intri-
cate matter that, like so many of the examples discussed in this chapter, is intimately depend-
ent on one’s theoretical framework.

24.3.6 Summary
In short, English has several borderline examples which seem to throw the distinction
between inflection and derivation into doubt. However, where we have borderline cases for
canonical categories, they do not invalidate the basic distinction, as long as there are clear-cut
examples at both ends, and as long as the distinction is useful where it can be applied. It is to
those examples that we now turn.

24.4 Clear-cut distinctions

In this section we will consider the difference between the nominal plural (regularly of the
form <(e)s>) compared with the nominal collective (the affix ‑ery) (following Plank 1994),
and the present participle ‑ing compared with the deverbal nominalization ‑ation. The point
is to show that these relatively similar processes can be appropriately divided into inflectional
and derivational along the expected lines, and that the division can be drawn in English.
Formally, most of these affixes seem to be regular. The <(e)s> marking the regular plural
covers, as is well-known, three distinct pronunciations (/z/, /ɪz/ and /s/), each of which
occurs in a well-defined environment. The ‑ery affix usually has only the single form ‑ery (and
where that occurs following a base spelled with a final <e>, there is elision of one of the <e>s,
 24.4 Clear-cut distinctions 539

as in tracery). The ‑ing affix also has a single form. Only ‑ation shows a range of forms whose
distribution might be unpredictable: competition, compulsion, definition, hospitalization, inhi-
bition, resolution, starvation (see Chapter 10). The suffix ‑ation is also attached to unpredictable
base allomorphs in compulsion, consumption, perception, redemption, reduction, and so on. So
‑ation is clearly less regular than the other affixes.
Although there are instances where we might have the <(e)s> form with no clear plural
meaning (see Chapter 7), in general the meaning of this affix is regular. While it arguably has
a different meaning when added to non-count nouns as in breads, toothpastes from what it has
when added to count nouns such as heads, tastes, the meaning is predictable from the count-
able or uncountable nature of the noun phrase as a whole.
The ‑ing suffix is again entirely semantically regular, both when used for strictly inflec-
tional purposes as part of a verb group, and also when used as a nominalization (see
Chapter 10).
The suffixes ‑ery and ‑ation are not semantically regular. In cookery the suffix can be
interpreted as an event or product, in fernery as a location or a collective, in baffoonery as a
noun denoting a type of behaviour, in snuggery as a location on an adjectival base. We
suggested in Chapter 12 that this range of polysemy is not unexpected, although the read-
ing we get for any given form derived in ‑ery is highly dependent on the nature of the base
and the context in which the form is found (see also Lieber 2004). Like ‑ery, the suffix
‑ation has a range of readings depending on the base verb and the syntactic context in
which the derived form is found, including event, state, location, product, and means (see
the discussions in Chapter 12).
Since it is quite possible to have plural locations, forms like rookeries are totally legitimate
and show inflectional ‑s further from the root than derivational ‑ery. Similarly with plurals
like vacations or transmissions, the plural occurs outside of ‑ation. COCA gives the example of
transformationing from the spoken part of the corpus. So these suffixes seem to show canoni-
cal behaviour of derivational and inflectional affixes. However, in order to make the observa-
tion, we have to know that they are, indeed, derivational and inflectional, so this observation
risks being circular.
On the whole, plural marking is not obligatory, since the choice between singular and
plural marking on nouns is meaningful: this is inherent, not contextual inflection. However,
we can construct environments in which a plural noun is required: a pair of ~, between the ~,
both ~ (e.g. a pair of feet, between the feet, both feet). We cannot construct an environment
where we must have an ‑ery derivative: amongst the jewellery, amongst the detritus; we will meet
in the fernery, we will meet in the house. Similarly, ‑ing is obligatory in They are ~ the food (eat-
ing, over-salting, preparing—at least, the only things that might fit in there with no ‑ing are
prepositions like on), but there is no position in which an ‑ation word is required: Their ~ of
the food (preparation, characterization; but also cooking, love).
While ‑ery or ‑ation always add lexical meaning, ‑s might be argued to do so in a minority of
cases (see Section 24.3.1 above) and ‑ing may do so on some occasions, all of which are quite
540 Inflection versus derivation

lexicalized (e.g. boring, building, interesting, railing, smashing (‘excellent’ BrEsl.), warning). So
while this criterion is not necessarily clear-cut, at least the tendency is in the expected
direction.
Number and tense/aspect are among the semantic categories which tend to be inflectional,
while collectives and nominalizations are among the semantic categories which tend to be
represented by derivational morphology. Since we recognize that these are no more than
cross-linguistic tendencies (although fairly robust ones), this cannot be strong confirmation,
but fits with the expected findings.
Plurality can affect most nouns in English (see Section 24.3.1 above) and ‑ing can be applied
to any non-modal verb in English. They are as productive as anything gets in English. The
suffix ‑ation has a rather restricted productivity; ‑ation is productively added mainly to verbs
ending in ‑ize or ‑ify (see Chapter 10). As we have shown in Chapter 12, ‑ery is surprisingly
productive in contemporary English, but still not comparable to the plural or ‑ing in
productivity.
Finally, as it happens both ‑ery and ‑ation are borrowed affixes, while both ‑s and ‑ing are
native. This is relevant insofar as inflectional affixes are only rarely borrowed, but derivational
affixes often are.
So although we might not want to say that any of these four affixes is totally canonical in
its behaviour, it is clear that we can draw a firm line between them in terms of the kinds of
distinction that were made earlier in this chapter. The implication of Plank (1994) is that this
is an expected and generalizable result. But we do have to stop and ask ourselves how much
this helps.

24.5 Why might we need to know?

It is one thing to say that, in general terms, we can distinguish between instances of inflec-
tional morphology and derivational morphology in English, but a more important question
is why this distinction is likely to be useful. There seem to be at least three points for discus-
sion here: the question of productivity, the question of affix ordering and the question of
headedness, and these will be treated in turn below.

24.5.1 Productivity
One of the big differences between inflection and derivation is, as we have seen, the degree
of productivity of the affixes. It is certainly true that in English we see a huge range of levels
of productivity, from third-person singular ‑s or ‑ing on all non-modal verbs to the suffix ‑red
which appears only in the word hatred with no apparent tendency to move beyond that word.
But to what extent full productivity correlates with inflectionality is a different question.
 24.5 Why might we need to know? 541

If we move from third-singular ‑s and ‑ing, and look at past tense and past participle inflec-
tion in English, we start to get a slightly different picture of what is happening. There are
many verbs in English which do not have a past-tense form in ‑ed. These verbs have alterna-
tive past tense forms which, it is frequently claimed, block the regular form. So the particular
affix is not fully productive, but the morphological category of past tense probably is.
At this point we start to get into impossible questions such as whether the pattern of ablaut
in give/gave, the irregular suffixation in spent and the regular suffixation in wandered are all
different renditions of ‘the same morpheme’ {past tense}. But we have to take care that the
argument does not become circular. If we define all these things as allomorphs rather than,
say, as synonymous morphemes (Bloch 1947; Hockett 1947), we cannot then claim that {past
tense} is fully productive, because it has become so totally circularly by way of our definition.
In any case, we need to take a wider look. While, as we have shown in Chapter 10, not all
English verbs have a nominalization other than with ‑ing, the productivity of a putative cate-
gory {nominalization}, which includes the affixes -ation, ‑ment, ‑al, and so on, as well as con-
version, is far higher than the productivity of any one of the individual processes involved. If
we are going to consider the productivity of {past tense} rather than the productivity of
{‑ed}, why would we not also consider the productivity of {nominalization}? There are
answers to this question: the obvious one is that the various ways of creating nominalizations
are not strictly in complementary distribution, and may occasionally contrast on a given
base, while the various markers of the past tense, while they are not strictly in complementary
distribution in the wider speech community, do not seem to provide any semantic contrast.
Nevertheless, the point we wish to make here is that the difference in productivity between a
so-called inflectional category and a so-called derivational one may be less clear-cut than
appears on a superficial consideration of the data.
If we look at past participles, the situation is slightly worse. Although English does not
have as many defective verbs as we might find for more highly inflected languages, it does
have verbs whose past participle is either missing or at least subject to great variation in the
community, with a consequential lack of confidence about the form of the participle and a
tendency to avoid it. Perhaps the best-known such example is the case of stride, although
the participle(s) of the verbs cleave and tread are also instructive. With stride the past
participle is used occasionally, but varies between strided, strid, strode, and stridden. With
cleave, which of course historically is two separate verbs, the past participle not only varies
between cleaved, clove, cloven, and cleft, but some of these forms are required in specific uses:
cleft palate but cloven hoof, for example. Such specialization suggests that there is some level
of contrast between the various patterns. In NAmE, tread comes close to being an example
of defective verb: except in fixed phrases (a well-trodden path), speakers are often reluctant to
use any past participle.
Plurals have already been discussed in this chapter, and they are less obviously fully pro-
ductive than past participles, but the arguments are very similar. Note, though, that there
may be semantic distinctions between appendixes and appendices, for instance.
542 Inflection versus derivation

At the other end of the scale, we need to discuss a suffix like ‑able. This suffix is often believed
to be restricted to transitive verbs, though as we saw in Chapter 14, its use is wider than that.
However, there is nothing to compete with ‑able, and on transitive verbs it seems to be abso-
lutely productive. Even though hospitalizable was not attested in COCA or the BNC at the time
this chapter was written, we cannot say that it would be an impossible formation. So we have a
derivational affix like ‑able which may be more productive within its domain than inflectional
affixes like the markers of the past participle. Perhaps the domain is what is crucial here: but
even with third-person singular ‑s we had to restrict its domain to non-modal verbs rather than
all verbs, so it is not clear that reduced domains should make a difference to our argument.
In other words, productivity is a feature of the individual affix, varies according to the
individual affix, but does not necessarily correlate in any straightforward way with the catego-
ries of inflection and derivation.

24.5.2 Affix ordering

A distinction between inflectional and derivational affixes allows a simple statement of some
instances of affixal ordering. Other things being equal, inflectional affixes will occur further
from the root than derivational affixes.
One of the difficulties with such a statement is that it requires a certain amount of ad hoc
modification because it is not absolutely true. Some counterexamples (many of which stand
for classes of words rather than being isolated instances) are given in (5).

(5) 
gutsy, interestedly, interestingly, looksism (COCA), lovingness, moreish, mostly,
outdoorsy, prohibitedism (COCA)

Of course, if ‑ly is inflectional, examples of this sort would not be problematic, but as we have
argued in Chapter 15, we do not find the arguments for the inflectional status of this affix
completely convincing.
The examples in (5) fall into a few recurrent classes and there are no doubt good reasons
for the exceptions. Nevertheless, the overall generalization does not hold.
An alternative approach would be to consider the ordering of affixes as subject to semantic
scope (see, for example, Rice 2000). This generalization appears to hold over the examples in
(5), as well as over more canonical examples such as entrained, fantasizing, hospitalizes, king-
doms, and so on. That being the case, the generalization about inflection and derivation is not
only partly misleading, but unnecessary.

24.5.3 Headedness
Theoretically, the inflection/derivation distinction might be of use in clarifying the notion of
‘headedness’. Using the most simplistic notion of ‘head’ (the head of the word is the single
 24.5 Why might we need to know? 543

morpheme that determines the syntactic category and morphosyntactic properties of the
word), we might hope to find a clear correlation: derivational affixes would always behave as
heads and inflectional affixes would never behave as heads. The reality is not nearly so simple,
however. For one thing, the data of English do not support this simplistic dichotomy and in
any case, it is not clear that any current theorist actually subscribes to this simplistic defini-
tion. It emerges that a consideration of headedness sheds little light on the status of affixes as
inflectional or derivational.
The facts about headedness in English morphology are well-known, so we only review
them briefly here. While it is usually the case that derivational suffixes act as heads,
prefixes are typically not heads (category-changing prefixes like de‑ do exist, though).
Further, it is clear that inflections, while they typically do not change syntactic category
(but see the discussion of participles above), do add or change morphosyntactic features.
So the correlation between headedness and the inflection/derivation divide is murky to
begin with.
Further, headedness in words has been a controversial topic for discussion for some time
now. It is in Marchand (1969), where the terms ‘determinans’ and ‘determinatum’ are used
and it was brought into the generative literature by Williams (1981). Part of the difficulty is
that there is no agreement about the relative importance of criteria for headedness, or what
criteria are being used. Thus Bauer (1990) argues that criteria for headedness do not pick out
a coherent element in the word for the head, while Lieber (1992) shows what can be achieved
by using a much more restricted definition of headedness.
In all of this discussion, the position of inflectional endings has been a particular problem.
Bauer (1990) argues that inflectional and derivational affixes do not provide a coherent picture
of what the head must be; Lieber (1992) argues that inflectional affixes need special treatment
in terms of headedness; Di Sciullo and Williams (1987) argue that we need a notion of ‘relativ-
ized head’ to deal with inflection—though not only with inflection. It seems that however we
try to define the head of a word, the definition does not apply easily to inflection.
It is hard to know what point the discussion on headedness has reached. Many scholars
argue that at least for derivational morphology, it is possible to point to a morph which is the
head of the word. Di Sciullo and Williams (1987) do not accept that the word necessarily has
a single head; it may have many of them. Di Sciullo and Williams (1987: 27) also say that their
approach allows them ‘to maintain that inflectional affixes are not separated from derivational
morphology in any way’, and this is consistent with the view from many scholars that there is
no split morphology (see Haspelmath 2002 for a neat summary). Those who see a distinction
between the headedness of inflection and derivation are consistent with the scholars who see
morphology as split.
Overall, the confusion about the headedness of inflectionally-marked words not only
detracts from the theoretical standing of the notion of head of a word, but also undermines
the distinction between inflection and derivation, particularly in a language which, like
English, has cases which seem to lie close to the inflection–derivation divide.
544 Inflection versus derivation

24.5.4 Conclusion
It seems that although we probably can maintain a distinction between inflectional and deri-
vational morphology relatively well in English—albeit with certain problematical cases
which do not invalidate the fundamental notion—the distinction is not helpful to us in
understanding any other aspects of the morphology of English. The classification might be
useful in terms of typology, but does not throw much light on the behaviour of English mor-
phological processes. On the other hand, the distinction between inflection and derivation in
English might be necessary for syntacticians, and indeed, as we suggested above, particular
syntactic frameworks might be forced to draw the line between inflection and derivation in
different places.
 chapter 25

The analysis and limits

of conversion

25.1 Prospectus
In this chapter we review the notion of conversion and its correlates. As noted in Chapter 2
we consider conversion to be a directional process which links an input and an output form
that are formally but not semantically identical. Since it is definitional that the output of this
process is formally but not semantically identical to the input, it is what Bauer calls an iden‑
tity operation (Bauer 1983: 32). It is a part of morphology by virtue of being an operation—
or, more accurately, a set of operations (there are different patterns of conversion)—which is
parallel to operations of affixation.
In this chapter we will principally be concerned with two points. The first is to investigate
which of the various types of systematic homonymy or polysemy that can be found in English
are instances of conversion. We approach this by considering the extent to which they match
the canonical pattern of conversion in English. The second point is to consider and evaluate
the various theoretical approaches to conversion (frequently called something other than
‘conversion’) which can be found in the literature. The various types of conversion in English
are dealt with in this volume where they correspond to various types of affixal derivation (e.g.
in deverbal nominalizations, in denominal verbs, and so on). Here we stand back and take an
overview.

25.2 The issues

The main issue raised here concerns which forms should be said to be linked by the process(es)
of conversion. English is full of homophonous forms which perform functions that range
from slightly different to extremely different. We use homophony here in its most literal
sense of ‘any two grammatically or semantically distinct forms that sound the same’. At one
end of a scale, we find forms that are accidentally homophonous, and therefore clearly syn‑
chronically unrelated. The insect a cricket and the game have no obvious semantic relation‑
ship, although they may go back to the same French word. Similarly, port ‘harbour’ and port
546 The analysis and limits of conversion

‘fortified wine’ are semantically unrelated, and apparently homophonous by chance. At the
other end of the scale, beaten the participle as used in the passive voice, as in Chelsea were
beaten in the semi-final, and beaten the participle as used, for example, in the present perfect,
as in Chelsea have beaten Manchester City, are systematically homophonous, and there is little
reason even to assume that there are two different morphosyntactic words in English.
In between we find a range of different types of homophony, treated in a range of different
ways by linguists and lexicographers. The most lexical of these are of little relevance here:
these are examples like cricket and port above. There are many other types, however, which
are potentially relevant to the study of morphology.
As a first example, consider homophonous affixal morphs. English is full of these. A few
examples are given in (1) for illustration.

(1) decency fatty <y>, /i/

friendly stupidly <ly>, /li/
killer smaller <er>, /ə ‖ ər/
teachers provides towards <s>, /z/
warmth tenth <th>, /θ/
vicarage coverage <age>, /ɪʤ/
wisdom kingdom <dom>, /dəm/

Most of the examples in (1) are probably uncontroversial instances of homophonous but
distinct affixes (for example ‑er, ‑ly, ‑s, ‑th, ‑y): cases of affixal homonymy. Some of them are
more often viewed as instances of polysemy (‑age, ‑dom). We have already noted in several
places in this volume that where there is doubt, we prefer the polysemy solution, but there
are places where polysemy is simply not tenable. The difficulty is in stating the conditions
under which homonymy has to be preferred to polysemy (see Bauer 2003b; Lieber 2004). It
may be that, rather than there being a set of necessary and sufficient criteria, each case has to
be treated on its merit—which may also be true with lexical morphemes.
Somewhere in the range between homonymy and polysemy is a set of homophonous
words which are generally taken to be derivationally related. The base forms or citation forms
of two lexemes are homophonous, but the two lexemes belong to different word classes (and
thus form parts of different inflectional paradigms where this is relevant), despite being
semantically related. We will take the examples in (2) as being clear-cut instances of what we
would consider to be conversion:

(2) a leaflet to leaflet

dirty to dirty
to spy a spy
 25.2 The issues 547

There are a number of reasons why examples like these are seen as a definable type of
homophony between homonymy and polysemy, and why the relationship between them is
seen as being derivational.
The first of these is that such examples fit into parallel sets where a similar relationship
between lexemes is marked by a derivational affix. This seems to imply that the relationship
in this type should also be considered derivational. Examples are given in (3).

a.
(3) leaflet to leaflet
fluoride to fluoridate
patina to patinize
beauty to beautify
b. dirty to dirty
legal to legalize
pretty to prettify
flat to flatten
rich to enrich
c. to spy a spy
to kill a killer
to reside a resident
to inhabit an inhabitant

This link with overt derivational affixes has led to analyses in which there is said to be an affix
linking these words, but it is an affix of zero form. This analysis will be taken up below
(Section 25.4.2).
In cases like those in (2) and (3) as well as in the overtly derived examples, there is also a
semantic relationship between base and derivative that can be teased apart into two factors.
First, there is semantic dependency, where an explanation of the meaning of the derivative
naturally uses the base lexeme, but not vice versa. Thus resident might be defined as ‘person
who resides’, and legalize as ‘make legal’. Legal, on the other hand, would not naturally or
usefully be defined as ‘what something becomes once it is legalized’ nor would reside be
defined as ‘to act as a resident’. Semantic dependency of this sort normally leads to lower
frequency for the derived word, as the added complexity of meaning narrows down the
­versatility of the word in discourse.
Second, the derivative frequently has a less broad semantic scope than the base. For exam‑
ple, the base rich can refer to something that causes indignation (as in It was a bit rich of her
to suggest that she was entirely innocent), but you cannot enrich a conversation by putting into
it comments which are seen as being rich in this sense. In other words, some semantic nuances
of the base are not carried over to the derivative. Or the opposite may occur, where
548 The analysis and limits of conversion

the derivative has nuances that could not be predicted from the semantics of the base; for
example, the noun fluoride simply refers to a particular chemical substance, but the derived
verb fluoridate is used exclusively for the addition of fluorides to consumables for the purpose
of improving dental health, not just any addition of fluorides to anything. The result of the
reduced semantic scope of derivatives is again that text frequency of the derived lexeme tends
to be lower than the text frequency of the base lexeme.
Note that semantic dependency may not always correlate with the relative age of the
base and the derivative (Bram 2011). That is, although typically, the base is older than
the derivative and the derivative also depends semantically on the base, this is not neces‑
sarily the case. Where this situation is reversed, we typically have cases of back-formation.
For example, babysitter can be defined as ‘one who babysits’, with the noun being seman‑
tically dependent on the verb. The verb babysit, however, is not attested until nine years
after babysitter (1946 versus 1937, according to the OED), thus pointing to this as a clear
case of back-formation (see also Chapter 23). However, the expected norm is that order
of coining in English correlates with semantic dependency, semantic scope, and text
frequency.
In all of these ways, therefore, we have pairs where it seems we can see a base and a corre‑
sponding derivative, and that the derivative behaves in ways consistent with those derivatives
that are overtly marked.
Moreover, the productivity of such cases appears to work like the productivity of other
lexical derivations: there are apparently unmotivated gaps in the paradigm, there is competi‑
tion with other affixal forms, the existence of a particular form to fill a slot in the semantic
paradigm is not entirely predictable, and there can be doublets (for example, teaser and tease).
Cases which meet these criteria we term instances of ‘conversion’, a term which we here
intend to be theory-neutral. Consider the list of agentive nouns from monosyllabic verbs
presented in (4).

(4) a. affixation
bake ◆ baker
hunt ◆ hunter
serve ◆ servant
stand ◆ standee
b. conversion
flirt ◆ flirt
guide ◆ guide
judge ◆ judge
spy ◆ spy

However, our working definition is not sufficient to delimit cases of conversion completely.
There are many instances of pairs of words in English which are close to conversion, but
 25.3 Cases for further consideration 549

which might not meet the definition entirely, and we consider some of these in the next sec‑
tion. There are also theoretical objections to the picture we have painted thus far, which will
be taken up later.

25.3 Cases for further consideration

Less central cases are less central for a number of reasons. In this section, a number of differ‑
ent types will be considered, and the extent to which they might be considered to belong
with the canonical cases of conversion.

25.3.1 Adjective to noun cases

Adjective to noun conversion was not included in the examples in (2) because there is more
than one type, and the status of some of the types is controversial. We review three cases in
this section: de-adjectival definite nouns, de-adjectival concretes, and plural nouns from
adjectives.

25.3.1.1 De-adjectival nouns preceded by the: the rich, the ridiculous

Here we consider a type where a word with the same form as an adjective can occur in the
position of a head noun in a definite phrase preceded by the. The specificity of the environ‑
ment is part of the argument for this not being straightforward conversion.
The productivity of this type is not usually fully recognized. Any pragmatically appropri‑
ate adjective may be used in this way. However, there is a difference between the case where
the phrase is understood as singular (whether or not singular concord is overtly used) (5)
and where it is understood as plural (6). The examples in (5) and (6) are from the BNC.

(5) New Statesman and Society 1985–1994: The old is dead. Where is the new?
Today 1985–1994: Now, however, the outrageous is expected.
Ring of Fire 1988: Neither the demonic nor the angelic is suppressed.
New Musical Express 1992: . . . we had grown to expect the remarkable as a matter of

course
London School of Economics lecture 1993: The good has to be imposed from outside, it’s

not, it’s not in human nature.

(6) The Daily Mirror: . . . a stone’s throw from the marina in Puerto Banus where the
rich and famous moor their luxury yachts.
Radical Approaches to Adult Education 1988: Thus, as we have argued at length in a recent

book, “doubly disadvantaged” sections of the working class—the unemployed, women,
550 The analysis and limits of conversion

black people, the retired, and the disabled—have especially acute educational
needs.
Introduction to Social Administration in Britain 1990: the old are more frequently ill than

the rest of the population
The Daily Mirror 1992: only the strong survive.
The Redundancy of Courage 1992: And so they processed, the defiant and the hopeless,

the resigned and those who pleaded with a silent desperation . . .

Where the phrase is understood as singular, it denotes something non-human, whereas when
it is understood as plural, it denotes something human. In neither case, note, is there any
feeling that a noun has been omitted (or is ‘understood’ in the terminology of traditional
grammar), as is the case in the examples in (7), also from the BNC.

(7) Still Life 1988: A pinker skin-tone under the brown . . .

Bible study group meeting 1993: The old nature has been crucified, the new has come.

The productivity of the construction under discussion is masked by the fact that tokens of
the relevant construction types are rare, and where they do occur are often not in a syntactic
context where number is apparent. It is also masked by the de-adjectival concrete type
­discussed in Section 25.3.1.2, which may sometimes provide tokens which appear to be coun‑
terexamples to the generalization stated here. There are occasional other apparent counterex‑
amples, like the accused, which is like the examples from Section 25.3.1.2 in that it can occur in
a full range of phrase types, but unlike it in that it takes no plural marking.
This type of formation is almost certainly syntactic rather than lexical: it is extremely pro‑
ductive, the output forms are not full nouns in that they do not inflect as nouns (they do not
show plural forms, for instance), and the syntactic context is a crucial part of the construc‑
tion. These instances are not usually treated as conversion, and it is no doubt for these
reasons.

25.3.1.2 De-adjectival nouns that take any determiner: an intellectual

Unlike the previous case, the nouns created by this process take part in the full inflection of
nouns and may occur in a full range of noun phrases (with all relevant determiners). Also
there are unpredictable gaps in the paradigm: there is no obvious reason why intellectual has
become a full noun meaning ‘intellectual person’ but poor has not become a full noun mean‑
ing ‘poor person’, but belongs to the category described in Section 25.3.1.1. Some established
examples of conversion are presented in (8).

(8) Arctic, Australian, bilingual, composite, daily, digestive, empty, executive, favourite, green,
imponderable, intellectual, interrogative, mural, parliamentarian, round, simian, variable
 25.3 Cases for further consideration 551

25.3.1.3 Adjectives with plural noun counterparts: news, burnables

We consider next examples in which we have nouns that appear to be derived from adjec‑
tives, but with the addition of a plural ‑s. This is a particularly murky area, as it is not clear that
we are dealing with a single, unified phenomenon. In a few cases there is no singular noun
corresponding to the plural. For example, although there is an adjective new and a noun news,
there is no intermediate noun new; the form news, furthermore, is semantically singular. In
other cases, we do have a corresponding singular. The noun good exists as a technical term in
the field of economics, alongside goods, and goods is treated as a plural noun. Perhaps most
perplexing is that we are not necessarily dealing here with a small number of discrete or mar‑
ginal examples. Consider the examples in (9):

(9) 
adjustables, allowables, affordables, burnables, buyables, chewables, drinkables,
durables, freezables, hand-washables

The formation of de-adjectival nouns in ‑ables is quite productive in contemporary English as

the examples above from COCA suggest, but it is far from obvious whether we should treat
them as examples of conversion. Indeed, in some or all of these cases we might be tempted to
treat the plural marker as a derivational suffix, which would rule these words out as potential
instances of conversion. In Chapter 7.2.1 we left this possibility open, without pursuing it, and
we will continue to leave it open here.

25.3.2 Mention versus use: ‘but me no buts’

Any word which is mentioned can appear as a noun and be inflected as such. Proverbial say‑
ings are full of admonitions such as

If ifs and buts were apples and nuts

all the wee laddies would fill their guts
If ifs and ands were pots and pans
there’d be no trade for tinkers

Google gives almost 2.5 million hits for the expression no more thank yous. In principle, even
whole sentences could be used this way: NDon’t give me any of your ‘I can’t do it’s.
The use of such items as verbs is rather less usual, but still not unusual. Effing and
­blinding (BrE ‘using swear words’) has become a fixed phrase, where the ‑ing seems to
indicate a verbal usage. Again, a phrase could provide the input to this process: NStop your
I‑can’t‑ing.
The fact that this is possible with virtually any base suggests that it is syntactic rather than
lexical, and thus should not count as conversion.
552 The analysis and limits of conversion

25.3.3 Formations related to prepositions: a down, to down,

the down train
Prepositions are rare as bases in derivation, and where they do occur tend to be sporadic: for
example, we can have a downer, but there are no established words overer, paster, througher,
underer, and so on.
With the case of the down train (also the up train, a through train) there is no particular
reason to believe that the relevant items have ‘become’ adjectives. Many things other than
adjectives can appear in attributive position in English, and these could simply be preposi‑
tions (or prepositional adverbs, or particles) in attributive position.
With other cases, it is noteworthy that the noun and the verb seem to be independent
derivatives from the preposition: to have a down on somebody (‘to be ill-disposed towards
someone’ chiefly BrE, AusE, or NZE) has nothing to do semantically with to down a drink or
to down tools (BrE ‘to go on strike’) except the independent relationship with down.
Nevertheless, this process is clearly lexical and not syntactic. It shows semantic depend‑
ency and the expected difference of semantic scope, with the lower frequencies in the deriva‑
tives. It therefore seems that the examples where a preposition becomes a noun or a verb
could reasonably come under the heading of conversion, albeit as a minor type.

25.3.4 Minor phonological modification

25.3.4.1 Devoicing of final obstruent: believe ◆ belief
This seems to be a non-productive class of related words, where the difference, synchroni‑
cally, between the two words lies in the final obstruent. In (10) there are examples of verbs
and nouns linked by the voicing of a final fricative.

(10) verb noun

abuse /z/ abuse /s/
advise advice
believe belief
calve calf
devise device
excuse /z/ excuse /s/
grieve grief
halve half
house /z/ house /s/
mouth /ð/ mouth /θ/
prove proof
relieve relief
 25.3 Cases for further consideration 553

sheathe sheath
sheave sheaf
shelve shelf
strive strife
teethe teeth
thieve thief
use /z/ use /s/
wive wife
wreathe wreath

In (11) the relationship is again one of voicing contrast between final fricatives, but this time
the pairs are verb and adjective.

(11) verb adjective

loathe loath ~ loth
save safe

In (12) there is a difference in voicing in the final obstruent, but this time the sounds are
alveolar plosives.

(12) verb noun

ascend ascent
bend bent
descend descent
extend extent
gild gilt
portend portent

Finally, in (13) there is a set which are linked by the final fricative, like those in (10), but there
is a concomitant vowel difference, which is synchronically unpredictable (though in many
instances it derives from the fact that one of the vowels was long, the other short, in Middle
English).

(13) verb noun

bathe bath
braze brass
breathe breath
choose choice
554 The analysis and limits of conversion

clothe cloth
glaze glass
graze grass
live life
lose loss

In these pairs of words there is not necessarily a consistent directionality: semantically, it

would seem that belief depends upon believe, but that sheathe depends upon sheath. In some
cases, such as clothe and cloth, the precise relationship at the time of formation is masked by
subsequent semantic changes. Nevertheless, these pairs are clearly more related than chance
near-homonyms.
Some authors (Bauer 1983; Štekauer 1996 by implication) include such cases among
instances of conversion. Others (Bauer and Huddleston 2002) do not. Since one of the cen‑
tral defining features of conversion is that the form remains constant, it seems that these
formations must be excluded from the domain of conversion.

25.3.4.2 Stress shift: frágment ◆ fragmént

Pairs of this type have been discussed elsewhere (see Chapter 10), and full exemplification
will not be provided here. The important point is that the noun and verb differ in stress:
sometimes this leads to differences in vowel quality (as in /ˈfræɡmənt/ versus /fræɡˈment/),
sometimes it does not (as in /ˈdaɪʤest/ versus /daɪˈʤest/).
On the basis of the discussion in Section 25.3.4.1, it might appear that these cannot be
instances of conversion (pace Bauer 1983; Štekauer 1996) because there is a difference in
form. However, there are differing theoretical interpretations of the facts which can lead to
diametrically opposed conclusions.
The first theoretical position looks at the surface forms of the nouns and the verbs related
in this way, and sees a difference in phonological form which is a concomitant of the change
of word class. Since the change of word class is marked by that difference in form, this fails
one of the crucial tests for conversion, and so must be some other kind of word-formation—
some kind of internal modification (Bauer 2003b).
An alternative view might consider the production of such forms as a linear process. For
example, in various generative models ranging from Chomsky and Halle (1968) to Distributed
Morphology (see, for example, Harley and Noyer 1999), there is a lexical entry in which there
is no stress marked. In Distributed Morphology the word class is attributed to the word by
virtue of its position in a syntactic tree. Once the morphosyntactic process has determined
whether the relevant phonological string is to be considered a noun or a verb (that is, once the
conversion process, however that is to be understood, has taken place), the phonology assigns
a stress pattern to the string, using its word class designation as part of the input. Once the
stress is determined, the precise nature of the vowels is determined by general reduction rules.
In such a model, the difference in stress is a result of the marking of word class, not an element
 25.3 Cases for further consideration 555

which in itself forces the interpretation of the derivative as a noun or a verb. At the point at
which word class is determined (Distributed Morphology) or the change from one word class
to another takes place (Chomsky and Halle 1968), there is no marker of the changed status,
and so this can be seen as conversion on a par with words such as control where there is no
stress difference and no segmental difference between the noun and the verb.
Despite the possibility of this sort of theoretical treatment, it seems that for the hearer,
stress must act as part of the signal that a verb or a noun is present, in more or less the same
way that an affix does. In other words, from the listener’s point of view, stress does function
as the marker of the changed status. That being the case, it seems that the conservative option
is not to treat stress shift as being in the same category as conversion.

25.3.5 Participles: his shooting, an interested party

The theoretical treatment of participial forms has always been a contentious matter. Solutions
from the literature include a conversion analysis, of course, as well as the postulation of
homophonous affixes (Allen 1978), the treatment of the categorial status of participles purely
as a matter of syntactic structure, as for example in Distributed Morphology, or as an example
of category-changing inflection (Haspelmath 1996). It is not our purpose here to review all of
the many theoretical approaches to participles, much less to decide among them; we merely
point out here that the problem has been dealt with in many ways in the literature. It should
also be noted that the processes dealt with here look more like conversion under some theo‑
retical approaches to conversion than under others. The question of theoretical approaches
will be taken up again in Section 25.4.
The ‑ing form appears as an inflectional affix (by definition—see Chapters 2 and 5) on
verbal bases in the progressive forms of the verb (14a). It may also occur in attributive posi‑
tion before a noun, where it may be sub-modified by so, very, and so on (14b). It may also be
used as the base for ‑ly adverbial suffixation (14c) and in various constructions in which it has
some nominal and some verbal features (14d, e). Morphologically, it is possible for such
forms to take a plural ‑s (14f).

(14) a. They were smiling at the camera.

b. They had a very convincing argument.
c. ‘How nice,’ she added smilingly.
d. Shooting elephants is not politically correct.
e. Their shooting of the elephants caused international headlines.
f. The recordings are archived on hard disk.

As the examples in (14) illustrate, the suffix ‑ing occurs in a range of places from those in
which it looks entirely verbal to those where it looks entirely nominal. There is no
556 The analysis and limits of conversion

allomorphy of this form, and it is as close to fully productive as any English morphology, so
that all non-modal verbs have an ‑ing form. While not all ‑ing forms are equally readily found
in all of these uses, it is not clear to what extent that is a syntactic matter, and to what extent
it is a pragmatic matter.
The thing that makes this look like a lexical process is the fact that different forms seem to
become integrated into the new word class in idiosyncratic ways. For example, interesting is
clearly an adjective, with a full range of adjectival uses and a meaning which is no longer
compositional, but it does not become so much of a noun that it takes a plural ‑s. Leaning, on
the other hand, can be used attributively as in the leaning tower, and has also become a noun
which can be pluralized: leanings.
The issues concerning past participial ‑ed (and its irregular congeners) are similar to those
concerning ‑ing. The ‑ed suffix is used inflectionally to mark past participles (15a). It can also
be used attributively before a noun, in which position it can be sub-modified (15b). Forms in
‑ly and ‑ness can be derived from such ‑ed forms (15c). Following the pattern discussed in
Section 25.3.1.1, such words can also surface as nouns (15d), and can also form plurals, although
far less easily than ‑ing forms (15e).

(15) regular irregular

a. He has visited the castle already. He has seen us.
b. a very isolated community a well trodden path
c. excitedly brokenness
d. help for the afflicted whiteness [. . .] ironically becomes the
hidden (BNC)
e. the undecideds (COCA) the factory-builts (COCA)

As is clear from (15), there is variation in the past participial forms, and the form of the more
adjectival or nominal uses is typically the same as for the inflectional verbal uses. There are
instances like drunken and sunken which have become fixed as adjectives and no longer have verbal
use in standard varieties, and these do complicate the issue, since the old forms may usurp some
of the functions that are normally held by the past participle form, but even then the current past
participle form may sometimes be used in the ways indicated: the sunk foundations (BNC).
The problems with including such historical developments as cases of conversion are first
that in the clear-cut cases conversion involves the transfer of a base belonging to a lexeme of one
word class to the base of a lexeme of a different word class. Here, it is arguable that the word
classes differ, but it is not an unmarked base which is adopted, but an overtly inflected form.
Semantically, the effects of that inflection have to be ignored, but the form remains as the word
moves to a different word class. Second, it is not always clearly the case that all the expected
features of the new word class are achieved: there seem to be degrees of verbiness and nouni‑
ness in different constructions with the ‑ing forms, for example, as illustrated in (14).
 25.3 Cases for further consideration 557

In looking at both types of participle, we must distinguish between those processes which
are genuinely automatic and those which are much more clearly idiosyncratic. The gerund
usage illustrated in (14) seems totally productive, and should probably be seen as syntactic in
some way. Perhaps the gerund could be considered a construction which uses the present
participle, a form generated inflectionally. Or the inflection could be seen as using a derived
‑ing form. The morphological extensions of ‑ing and ‑ed may behave like ‑ly, ‑ness, and the
plural markers. On this account, the lack of a word like connivingness in our corpora is simply
a gap. However, why drunk becomes a pluralizable noun while woken does not is left unex‑
plained: at a guess, it is something to do with the frequency of drunk in nominal environ‑
ments which leads speakers to expect full nominal behaviour. If this is true, there is no
morphological principle as such involved. The same principle could account for the fact that
some ‑ing and ‑ed forms feel like better adjectives than others. It is not a matter of grammar,
but a matter of semantics and frequency in particular environments, which allows extensions
into other adjectival environments by analogy.
There are instances of participles also appearing in environments that suggest that they are
prepositions: concerning, excepted, excepting, given, including, regarding, and so on. It is never‑
theless not clear that there is a process deriving prepositions from verbs: rather verbal con‑
structions become used so often that they become perceived as chunks which have the same
function as prepositions (Kortmann and König 1992).

25.3.6 Type coercion

In the discussion of derivational processes in this book we have frequently made use of the
notion of type coercion, or simply coercion (Pustejovsky 1995). This is a semantic pro­
cess by which a lexical item of one semantic type, say a count noun or an intransitive verb, is
forced by the syntactic context in which it is used into an interpretation as another semantic
type, say a non-count noun or a transitive verb. Since the input and output of type coercion
are semantically distinct but formally identical, the question arises whether they should be
seen as related by conversion.

25.3.6.1 Mass versus count: a cake ◆ some cake

The majority of common nouns in English can be used as either count nouns or as non-count
nouns. While there are a few like knowledge which seem extremely awkward in count envi‑
ronments, on the whole the shift between count and non-count uses is automatic. Even
count/non-count pairs like bread and loaf are subject to type coercion, as in advertisements
for Reizenstein’s breads or the phrase some loaf of bread (BNC). Here there is an argument
( Jackendoff 1990; Pustejovsky 1995) that it is not strictly the noun which is count or non-
count but the noun phrase in which it occurs: the reading is count or non-count because of
the syntactic context, not because of the particular noun chosen. If this argument is accepted,
558 The analysis and limits of conversion

the distinction between count and non-count nouns per se is purely a syntactic matter and
not a matter of lexis at all. That being the case, this is not a matter of conversion or of
­morphology; it may not even be a matter of polysemy.

25.3.6.2 Transitive versus intransitive: to walk ◆ to walk the dog

While there are instances in English where the difference between a transitive and an intran‑
sitive usage is marked, this is not the norm. Some examples of the marking are provided in
(16). The processes illustrated in (16) are not readily available to speakers of contemporary
English for new coinages.

(16) intransitive transitive

sit set
moan bemoan

As in the case of count versus non-count nouns considered in Section 25.3.6.1, it can be argued
that the verbs take on the requisite meaning from the syntactic constructions in which they
occur, and that this polysemy is thus a matter of type coercion.
There is also the question of semantic dependency. With smell, the intransitive seems to
depend on the transitive; with walk, transitive senses seem to derive from and postdate
intransitive senses. The lack of consistent semantic dependency also seems to suggest that we
are not dealing here with a derivational matter. We will therefore not include this type of
relationship under the heading of conversion.

25.3.6.3 Proper versus common nouns: Mary ◆ the four Marys

It seems that all proper nouns can be used as common nouns with the meaning ‘entities
which bear this name’. This is not only true of given names like Mary, but also of family
names (the Joneses, the Smiths—usually meaning the plural members of a single family) and
place names (there’s a Cambridge in Massachusetts and another one in New Zealand; the river
Avon I’m talking about is in Christchurch). Even where the name denotes a unique entity, a
plural common noun may occur either by ellipsis (17) or by virtue of a figurative splitting of
the original entity (18), as these examples from the BNC illustrate:

Country Living 1991: the Russells also breed Berkshires, Saddlebacks, Lops, Middle
(17) 
Whites, Large Blacks and Tamworths

The Savage and the City in the Work of T.S. Eliot 1991: Dawson emphasized the “two
(18) 
Englands” created in the nineteenth century . . .

The change of grammatical category here seems to be a matter of type coercion, and does not
appear to be a derivational relationship. We will not include it as conversion.
 25.3 Cases for further consideration 559

25.3.6.4 Non-gradable versus gradable: English ◆ very English

Given the right syntactic context, almost any non-gradable adjective can be made gradable
(examples from the BNC, see also Chapter 6):

(19) Central television news scripts: It’s more English than anywhere else.
New Statesman and Society 1992: The Frigido deep-freeze has an indefinably Antarctic

quality (unlike the more Arctic Norfrigge).
Adam’s Paradise 1989: Ruth rose too, suddenly held by a despair more absolute than

any she’d yet felt.

This seems to be rather like the changes between count and non-count nouns: a polysemy
which is forced by the syntactic context. That being the case, this is not conversion.

25.3.7 Adverb formation: real ale ◆ real good

There are a number of forms in English which are found functioning both as adjectives and
as adverbs (see Chapter 15). The number of forms concerned differs from variety to variety,
so that some speakers allow She ate her meal leisurely, while others require She ate her meal in
a leisurely manner. There are also differences between words that have an overtly distinct
adverbial form like late and lately, or real and really from those which have no such distinc‑
tions. As is clear from the examples, the overtly marked adverbs may be synonymous with
the unmarked adverbs, or semantically distinct from them. Finally, using real as an intensifier
carries clear social messages, being overtly proscribed by purists, while the dual use of lei-
surely (for those speakers who use it) appears to be socially neutral.
There are many forms which can be adjectival which are also used as intensifiers for adjec‑
tives. Some are listed in (20). Most of these are considered non-standard by purists, but are
extremely common in speech, where very (itself originally an adjective) is now rare in the
vernacular (see e.g. Bauer and Bauer 2002).

(20) blind, bloody (and euphemisms), dead, deadly, fucking (and euphemisms), jolly, mighty,
pretty, real, right, wicked

In (21) we list some of the forms which can be used as either adjectives or adverbs in standard
English. Those forms in (21b) avoid the repetition of adjectival ‑ly followed by adverbial ‑ly,
which is less and less acceptable in modern English. Some comparative forms such as better
(the comparative of both good and well) also fall into the category illustrated in (21a).

(21) a. early, fair, fast, hard, loud

b. daily, deadly, fortnightly (BrE), friendly (NAmE), kindly, leisurely, monthly, nightly,
quarterly, weekly, yearly
560 The analysis and limits of conversion

Others of this type are rather less clearly acceptable in formal language, and have overt
adverbial alternatives, but are widely found. Examples are given in (22).

(22) Kiss me quick. Kiss me quickly.

Go/drive slow. Go/drive slowly.
He did it wrong. He did it wrongly.

In (23) there are some examples where the meanings of the adjectival forms and the adverbial
forms are so different that it is probably safer to view the two as separate lexemes, whatever
the etymology.

(23) adjectival function adverbial function

I’m ill. We can ill afford the ticket.
(now old fashioned)
an even distribution She even ate it.
his late father He arrived late.
the round table We wandered round.
the very person a very interesting book

Part of the difficulty in discussing this type of relation is that many adverbs can be used in
attributive position, where they might be considered to be partly adjectival. We take the view
that use in attributive position is not sufficient to illustrate a recategorization of these words.
Some examples are given in (24).

(24) the off switch, the overnight parcel rate, overseas mail, the then king, a through train,
an upward(s) movement

For a fuller discussion of this area of grammar, see Valera Hernández (1996).
Most of these types appear not to be productive, or only marginally so. Which words are
found with both functions and which are not seems to be a matter of lexis, and therefore
this is a reasonable area for word-formation, even if not productive word-formation. If we
take the point of view that adjectives and adverbs in English are distinct categories (Payne
et al. 2010), then the change of category criterion is met. Directionality seems to go from
the adjective to the adverb, since the forms are nearly always clearly adjectival. All of this
seems to be compatible with there being conversion here, albeit limited and non-­productive
conversion.
 25.3 Cases for further consideration 561

25.3.8 Compounds and phrases

It is not unusual in contemporary English to find cases in which a compound of one word
class can be found without modification in the syntactic context of another word class. For
example, where stir-fry would usually be assumed to be a verbal compound, it can be found
in a context which is clearly nominal.

(25) Washington Post 2004: I’ve also used the leaves in a stir-fry with broad bean sauce

Similarly, although seatbelt and blowtorch are typically nominal compounds, they can appear
in verbal contexts as well, as the following examples from COCA suggest:

Virginia Quarterly Review 2004: Graves was seatbelted in the shotgun position next to
(26) 
Ahktar, jostling in the inert manner of a crash-test dummy.
Best Friends for Never 2004: What are you going to do tomorrow when I forget about

you again? Blowtorch my bedroom?

Such examples fairly straightforwardly meet the criteria for conversion set out in Section
25.2: the verbal and nominal forms are homophonous, they clearly belong to distinct word
classes, they are semantically related, and their semantic relationships are similar to those
found between bases and overtly affixed forms derived from them. Some examples of conver‑
sion from compounds are clearly lexicalized (to day dream, to pickpocket, to blackmail), but
the examples in (26) show that the process is productive. Examples of this sort are discussed
in more detail in the relevant chapters.
The same argument can be made for phrasal elements of certain sorts. For example, we
frequently find that phrasal verbs like blow up, break down, call back, give away, hang out,
put down can appear in nominal contexts. Idiomatic nominal phrases of certain sorts
can also appear in obviously verbal environments, as the examples in (27), from
COCA, show.

Fantasy and Science Fiction 1995: Behind Zane Gerard, Tyque Raymond was thumb-
(27) 
upping me.
Fantasy and Science Fiction 2002: The cameras were all installed to monitor the reactor,

so they faced the center of the room. Most of them close-upped on specific pieces of
equipment.

So from (give someone) a thumbs up we get the clearly verbal form thumb-upping (interest‑
ingly, with loss of the plural ‑s). The second example is more intricate: what we apparently
have here is a verb to close-up corresponding to a noun a close up (a kind of photograph) from
an adjective (we saw it close up). Again, although forms of this sort may be less frequent than
562 The analysis and limits of conversion

the compound cases in (25)–(26), they still appear to have some productivity in the contem‑
porary language (see also the examples in (27) discussed in Section 25.4.2).

25.4 Modelling conversion

The label ‘conversion’ has been used at least since Sweet (1891). In part it has remained a
useful label because it does not make many theoretical claims. One claim that it does seem to
make is that there is a process linking the words involved: conversion turns one thing into
another. What is more, if a word is the output of conversion, it would seem to imply that it
has all the relevant functions of the type to which it has been converted. Even these implica‑
tions might be controversial.
A number of different terminologies and approaches to conversion can be found in the
literature. To a large extent, it is not clear that they are much more than alternative nomencla‑
tures for the same idea. But some of them have different implications for how conversion
should be dealt with in a grammar, including whether they are a part of morphology at all. In
what follows we consider some of these alternative approaches, and ask whether there is any
evidence to support their predictions.

25.4.1 Conversion, narrowly defined

Thus far, ‘conversion’ has been used as a cover-term for the process under review. But we can
specify it more closely within the theoretical literature, as hinted above. Conversion is a proc‑
ess which links an input form to an output form: it is directional. Since it is definitional that
the output of this process is formally (but not semantically) identical to the input, it is an
identity operation (Bauer 1983). It is a part of morphology by virtue of being an operation—
or, more accurately, a set of operations (there are different patterns of conversion)—which is
parallel to operations of affixation. Because it is a morphological operation, we expect it to
show productivity (availability and profitability) in the same way as other morphological
operations. Because it is a derivational process, we expect to find gaps in the established
outputs of the process, and places where the process is in competition with other morpho‑
logical processes.
All of this provides a coherent picture in line with what has been illustrated so far. The
productivity of conversion is illustrated elsewhere in this book (see Chapters 10, 11, 13, 14, 15)
and the issue of competition was outlined in Section 25.2, example (3)). Nevertheless, it is
controversial, as will become clear when we consider alternative points of view.
 25.4 Modelling conversion 563

25.4.2 Zero-derivation
The zero-derivation approach takes the parallelism between conversion and other affixed
forms illustrated in (3) seriously. If we are dealing with a derivational process, and it is paral‑
lel to instances of affixation, then perhaps we should say that it is another kind of affixation,
one where the form of the affix is zero. Just as legalize is made up of legal+ize, so empty (the
verb) is made up of empty+Ø.
There are various standard objections to this analysis. The first is to question the prolifera‑
tion of zeroes this gives rise to. If empty (adjective) has no affix on it, but empty (noun) has a
zero on it, empty (verb) must contain a different zero, since it is a non-synonymous affix. We
find, therefore, not only various zeroes contrasting with each other, but contrasting with the
lack of zero, that is with nothing at all. While this may give rise to a system which could be
run on a computer, it is hard to see how it could be a learnable system for real speakers.
Next, it is queried how we know that there is only one zero-affix in all cases of, say, verb to
noun derivation. Given the parallelism with affixed forms like amendment, closure, confusion,
dismissal, hatred, laughter, marriage, and so on, why should we expect change, freeze, influence,
recompense, remark, and others to have the same affix? At the same time, if they do not have
the same affix, how can we tell which base has which affix? The obvious riposte to this argu‑
ment is that we assume a single zero because of Ockham’s razor, but it remains true that it is
hard to prove that a single zero is involved, or even that the zero is a suffix and not a prefix.
The third objection considers the nature of the output. If we consider the nouns oil, pack-
age, landmine we assume that these are simple, derivationally complex, and compound
respectively. What then about the corresponding verbs oil, package, landmine? If they have a
zero-affix, they should all have exactly the same status, they should all be derivatives, and the
different statuses of their bases should be irrelevant. But it is not clear that we have any real
reason to say that the verbs are anything other than simple, derivationally complex, and com‑
pound, and if this is true, the use of the zero-affix is masking something. Consider the two
examples below.

(28) Then he called me a week later and said never mind, so I never minded (Parker 2000: 65)

(29) Writing the soundtrack of /Crazy Heart/ with his Fort Worth childhood friend T Bone
Burnett gave Bruton the national profile in contemporary Americana that he’d earned
decades previously by right-hand-manning it with Kris Kristofferson and Bonnie
Raitt (http://www.austinchronicle. com/music/2010‑02‑19/967813).

In each case, an apparent phrase is turned into an inflected verb. We would presumably not
wish to say that the verbs never mind and right-hand-man are phrases, and yet the bases look
like phrases. So the application of the zero-affix could be analysed here as having the effect of
making the word-formation process a morphological one. This seems to contradict the argu‑
ment just above about oil, package, and landmine (that is, that we have no reason to believe
564 The analysis and limits of conversion

that the status of the base has changed under zero-affixation), because if the verb does not
always retain the nature of its base, it is not clear that it does in any of these instances.
None of these arguments against zero-derivation is completely convincing—and accord‑
ingly there are a number of scholars who use the zero-derivation analysis (see Bauer and
Valera 2005). If we reject it here it is largely because it seems to make assumptions that are
not necessary.
There is an alternative view of the zero-affix hypothesis, deriving from the notion pro‑
pounded by Beard (1995) that derivation and affixation are two separate processes. According
to this view there is a semantic derivation that, for instance, makes a verb into a noun, but no
affixal material to realize that semantic derivation. This view has the option of making the
parallel with other cases of affixation overt (they all take part in the same semantic process)
while at the same time not requiring a zero form to be part of the derived string: the semantic
process is just given no realization. Both Plag (1999: ch. 8) and Lieber (2004: ch. 3) point out
the pitfalls of this sort of analysis, however: noun to verb conversion in English actually forms
a wider variety of verb-types than any of the overt verbalizing affixes do and shows a much
wider variety of meanings than any of the overt affixes, so considering verbalization as a sin‑
gle process seems like an incorrect move.

25.4.3 Relisting
Another way of considering conversion is that the input word is simply relisted in the lexicon
with a new word class. Some scholars (e.g. Strauss 1982) see this as a rebracketing, so that the
verb empty has the structure [[empty]a]v. Others (e.g. Lieber 1980) do not use this formula‑
tion, and it is not clear whether this is or is not theoretically significant.
Unlike the Beardian analysis described in the previous subsection, the relisting option
seems to suggest that the semantic effects of conversion should be less constrained than those
of any overt affix, although they should not be entirely random. Since this sort of analysis
assumes that one entry is the basis for another, we would expect at least some minimal con‑
tent from one entry to be transferred to the new entry. Some scholars (e.g. Adams 1973)
comment on the semantic regularity of the conversion process, and give quite constrained
lists of potential meanings for conversion. Lieber (2004), however, finds a wider range of
meanings than can be found with, say, ‑ize or ‑ify, and concludes that the relisting analysis has
at least some advantage over a zero-affixation analysis.

25.4.4 Underspecification/multifunctionality
Some scholars deny that there is any process of conversion at all, but rather argue that roots
in English are underspecified or completely unspecified for word class. In this sort of analysis,
there is no derivational process relating a noun like hammer to the verb hammer. Instead,
 25.4 Modelling conversion 565

there is simply a single lexical item hammer whose category is determined by syntactic con‑
text. Similar analyses have been put forward both in generative frameworks (Hale and Keyser
1993) and in Cognitive Grammar (Farrell 2001).
Various versions of the underspecification analysis have in common the difficulty of
explaining intuitions that the relationship between noun and verb is often felt to be direc‑
tional. Generally this problem is approached by appeal to semantic idiosyncrasies of the
underspecified bases: although bases may lack syntactic category, they do have semantic
information that accounts for the intuition that, say, hammer is more fundamentally an entity
than an event, and kiss the opposite. It is unclear whether the need to appeal to semantic
representation undermines the basic claims of this sort of analysis.
Also problematic within underspecification analyses is the failure of many roots to appear
in a full range of word classes. Although such analyses tend to focus on noun–verb pairs,
there is no reason why truly categoryless roots should not occur in adjectival contexts as well.
Indeed, the root calm can appear in adjectival environments (a calm disposition), nominal
environments (a calm), and verbal environments (to calm). The problem is that true multi‑
functionality of this sort is too rare in English to attribute to accidental gaps.

25.4.5 The influence of pragmatics: contextuals

Clark and Clark (1979) argue that in many cases a new instance of conversion cannot be
understood without reference to the pragmatic environment in which the word is coined.
For example, in the example in (30), we need to understand what Groucho Marx typically
does with his eyebrows in order to be able to understand the conversion.

(30) Lisa popped a cherry tomato leftover from her salad in her mouth and Groucho Marxed
her blonde eyebrows. (http://1000days.douglasblaine. com/20110420/the‑sex‑is‑still‑
always‑just‑fine/)

In He telegraphed his intention we have to work out pragmatically what the relationship is
between what he does to his intentions and telegraphy. Clark and Clark call these contex‑
tuals (see also Aronoff 1980), because they become meaningless outside of a particular
context.
It is not clear that the requirement of pragmatic interpretation of new words is con‑
fined to contextuals in general or instances of conversion in particular. Phrased differ‑
ently, it may be the case that all neologisms are, to some extent, contextual. Given the
expression big benefactors receive no tax breaks in Britain; nor are they lionised at cocktail
parties (BNC) we must, if the item is not already item-familiar, determine the connec‑
tion between lions and what happens at cocktail parties. In the case of the BNC example
auditionee in (31), context may leave less room for choice, but the fundamental principles
seem to be the same.
566 The analysis and limits of conversion

So you want to be an actor 1991: As an auditionee, I knew I worked best if I attempted

(31) 
something outrageous (BNC)

The requirement for some pragmatic help in the interpretation of neologisms is not contro‑
versial. In this context, the question becomes whether instances of conversion require par‑
ticularly high levels of pragmatic information to be interpreted. Phrased this way, it is not
clear that the question is answerable. It seems likely that there is a cline of the amount of
pragmatic information required, since a synthetic compound like parrot smuggler, if literal,
requires relatively little information beyond what is present in the construction, for its inter‑
pretation. Parrot smuggler is invented (but see NZPA 2007); budgie smugglers, on the other
hand, is an AusE expression for a pair of skimpy men’s swimming trunks. In such cases, a
great deal of pragmatic information is required to arrive at the correct interpretation. The
example makes the point that even here, whether a smuggler is or is not human is a matter of
pragmatics.
Nevertheless, designating a particular subset of neologisms as contextuals, does not seem
to be particularly insightful; rather, to the extent that it makes sense to designate an entire
construction in this way, it denotes some vaguely determined point on a scale. Instances of
conversion such as to crimson from crimson or to carpet from carpet (at least in the sense ‘to
cover with carpet’) seem to require relatively little pragmatic information, and are on a differ‑
ent point of the cline. Not all cases of conversion are the same in this regard, and we therefore
do not view this as a feature which is distinctive in conversion.

25.4.6 Conversion as inflectional

Myers (1984) argues that conversion is not derivational but inflectional. The argument is
placed within a specific generative model, and it is not clear how far it can be separated from
that theory.
Like the analyses discussed in Section 25.4.4, Myers’ analysis assumes that lexemes have
no word class until they are entered into a syntactic tree, at which point they gain the word
class from the syntactic structure and may be obliged to take on a specific morphological
form to reflect that word class. Given an underlying form which we can characterize as hate
in a syntactic structure [DP[+genitive] hate PP[+of]] (leading to surface structures like their hatred
of the United States (COCA)), the underlying element hate must take on the surface form
hatred because it is in a nominal position. Hatred is clearly an unpredictable form, and so
must be a listed variant for hate as a base form that occurs under just such circumstances.
In instances like I hunger for your touch, on the other hand, hunger and touch require no
morphological modification, but pass straight into the appropriate form. Their verbal or
nominal character (respectively) is guaranteed by a zero-affix which carries a word class
marking and the required inflectional categories (in the case of hunger, 1st person, non-past
tense). Had the sentence been I hungered for your touches, there would be overt suffixation
 25.5 Summary 567

rather than the zeros required in the original version of the example. The benefit of this pro‑
cedure, according to Myers, is that it keeps the righthand head rule intact, and allows headed‑
ness to apply to inflectional as well as to derivational morphology in English. These are not
clearly benefits, however: it is clear that the righthand head rule does not work for English
(cf. exceptions such as the category-changing prefix de‑), and there are many issues with
allowing inflectional affixes to be heads (see Chapter 24). Further, this analysis appears to
combine the underspecification analysis (Section 25.4.4) with a variant of the zero-affixation
analysis (Section 25.4.2), and therefore carries with it the problematic elements of both.

25.4.7 Conversion as metonymy

If conversion is a case of metonymy (Schönefeld 2005), it is a very special case. Most instances
of metonymy maintain word class, so that the crown means ‘the monarch’ by metonymy, not
‘to rule’. But even if we relax the term metonymy, or simply say that pairs of words that are
linked by conversion are linked by a figure of speech and not by a morphological relationship,
we would have to explain why the relationship is so like derivational relationships, as well as
so unlike other figurative relationships.

25.4.8 Various other nomenclatures

Various other terms are suggested in the literature as a way of avoiding the term conver‑
sion. These include ‘transposition’ and ‘functional shift’. As far as we are aware, all of these
can be reduced to one of the other options we have already mentioned. The terms may
focus on one aspect of the process or downplay some aspect of the process, but none adds
anything new.
The same is true of the periphrasis that ‘word W is used as an item of word class Z’ (‘empty
is used as a noun’, for example). While this mode of expression is often tempting, it is not
clear what it means ‘to use something as Z’ if it doesn’t mean that the form has taken on the
functions typically associated with class Z. If that is an accurate gloss, then it is not distinct
from the other views which have been discussed above.

25.5 Summary
Conversion in English is defined as what we find when a cluster of canonical conditions are
met, and as we move away from those canonical conditions, we move more into lexical or
syntactic processes rather than strictly morphological ones. Although there are various
approaches to conversion, we prefer one which flows naturally from this cluster of condi‑
tions, and this is captured by the nomenclature of conversion rather than any of the other
terminologies that have been suggested.
 chapter 26

Blocking, competition,
and productivity

26.1 Prospectus
In this chapter we consider the three interlinked topics of competition, blocking, and pro‑
ductivity. We define morphological processes as being in competition when they share some
domain between them, producing outputs which, if acceptable, might fill the same functional
slot in a paradigm (derivational or inflectional). So, for instance ‑ness and ‑ity may be in com‑
petition with the result that in some cases only an ‑ity word is attested, in others only a ‑ness
word, and in yet others the two co-exist. One way to prevent competition at the level of the
individual word would be through blocking. Aronoff (1976: 43) defines blocking as ‘the non‑
occurrence of one form due to the simple existence of another’. Unfortunately, such a defini‑
tion raises as many questions as it resolves. Can thief block stealer (Bolinger 1975: 109), or
does blocking affect only words created from the same base by competing morphological
processes? How is ‘simple existence’ to be understood in this context? Similarly, what does
‘nonoccurrence’ mean, given that we now have ready access through the world-wide web to
billions of words of text from a wider range of writers than was available when Aronoff wrote
this definition? We will need to consider such questions.
Bauer (2001: 211) says that “ ‘productivity’ deals with the number of new words that can
be coined using a particular morphological process, and is ambiguous between the sense
‘availability’ and the sense ‘profitability’. ” Not only does such a definition require the isola‑
tion of ‘a particular morphological process’ and further definitions of ‘availability’ and
‘profitability’, but it also requires that it should be possible to count neologisms and it
requires a notion of a new word.
None of our fundamental topics for this chapter is thus straightforward, and their inter‑
action is correspondingly less straightforward. Some of these matters have already been
traversed relatively superficially in Chapter 2. Here we wish to focus on the theoretical con‑
structs that are the foundations of dealing with these aspects of the morphology of English.
In some places, this will involve moving away from our general synchronic approach to
morphological matters, and viewing things in a more diachronic way.
 26.2 New words and old 569

26.2 New words and old

Although there are many estimates of a speaker’s vocabulary size in the literature, they are not
reliable (see Nation 1993 on some of the reasons for this). Often there is not even agreement
on what kind of ‘word’ is being counted or what it means to ‘know’ a vocabulary item. One
of the fundamental assumptions of any such measure is that individual speakers do ‘know’ a
number of ‘words’. Since the numbers cited in any of the experiments on the numbers of
words speakers ‘know’ is considerably smaller than the number of words in the largest dic‑
tionaries of English, there must be a number of words of English which individual speakers
do not ‘know’. Beyond that, there are items which are not known to individuals, are not in the
largest dictionaries, but have the potential to be words.
The words known to an individual speaker are, presumably, of no particular interest to the
linguist except insofar as large numbers of speakers share the same words, and there is a
social contract on what those words mean and how they are used. However, there are cer‑
tainly many words that only a minority of speakers share, and that are unfamiliar to most
other speakers.
The set of words listed in dictionaries is always slightly out-of-date, chosen to meet crite‑
ria which are irrelevant to the linguist (such as having a fixed number of attestations in writ‑
ten English), and often underrepresents the most productive morphological processes. For
instance, at the time of writing, the OED does not have a listing for fashion crime, for ginga
(/ɡɪŋə/ ‘red-headed person’) or a ‑ly adverb corresponding to hippopotamic.

(1) The Fat Man yawned widely. Indeed hippopotamicly, thought Pascoe. If such a word
existed. (Hill 2001: 335)

Potential words are of interest to the linguist. As Aronoff (1976: 19) points out, it is ‘the task
of a morphology to tell us what sort of new words a speaker can form’. The difficulty is that
the main evidence the linguist has as to what is a potential word is the set of actually observed
words. Since productivity involves previously non-actual words that were nevertheless
potential words becoming actual, the distinction is vital to studies of productivity: vital, but
possibly not directly measurable, especially if we take into account that speakers may vary a
great deal in their knowledge of words.
The real problem is the gap between what dictionaries register and what is merely non-
actualized potential. In this book, we have filled that gap by looking at corpus examples. But
any corpus example could be innovative, coining a new form, or could be repeating a form
which is already item-familiar to some speaker or writer. All we can assume is that attested
words in this in-between stage are recently coined and relatively recent innovations. The
word ‘relatively’ here hides a potential problem that, again, comes in through between-
speaker variation.
570 Blocking, competition, and productivity

Speakers are aware of some innovations (if not all), and occasionally draw attention to
them in their speech, as in example (1) above, and as in the examples in (2).

(2) a. Could that polish have been tainted with cyanide? Could Susan have been the
tainterer? Was there such a word a tainterer? Maybe she was a tainteress? (Strohmeyer
2004: 73)
b. The houses on Devonshire Close weren’t castles . . . but there was nevertheless
something distinctly castleish—castlesque? castleine? castilian?—about them.
(Block 2004: 36)
c. It was a low-maintenance farm, . . . some fields rented out to the ‘horsiculture’—the
riding fraternity—and some set aside (Lovesey 1997: 244)
d. Do you use “cinnamony” to describe something with cinnamon? If not, what do you
use? (http://www.answerbag.com/q_view/1306672#ixzz1zPuSJ5cO)

While such examples may be indicative, they provide poor evidence as to the actual state of
the language. The form tainterer ought to be ungrammatical as it has two agentive suffixes ‑er,
where one is usually enough; castlesque gets only one hit on Google, and that’s a name—the
OED lists only castle-like; cinnamony gets hundreds of thousands of hits on Google. So we
know that there are new words, but we cannot be sure at any given time whether a word is or
is not new. In principle it is new or not new only with respect to a given speaker at a given
time. Accordingly, there can be no direct measure of new words being used in the language,
but we know there must be new words.
A distinct question is whether we can be sure about newly established words in the speech
community. Here we are on a slightly firmer footing. Large dictionaries such as the OED
provide a continually up-dated list of established words, and the internet allows us to see how
widespread a given word is. By comparing corpus data with a source such as the OED we can
see whether a particular morphological process continues to be productive, though it is
doubtful that we can turn that into any specific measure. To illustrate this point, consider the
suffix ‑fold (see Chapter 18.4.3). The OED provides a handful of examples, not in its lemmas
but in its explanation of the history of the form. COCA provides hundreds of forms. Yet
many of the forms found in COCA must have been used many times in the past; it is simply
that the OED does not note the most productive uses, because there is no lexicographical
benefit to doing so.
Given this state of affairs, it is sometimes an open question how far unique forms can be
trusted as good data (see Bauer 2001: 57–8 for some discussion). We have generally trusted
our corpus data, and where individual examples fit into widespread patterns there seems to
be little reason to doubt it (see Chapter 3 for more discussion). Rarer patterns always have to
be treated with a little more care. At the same time, we must be careful not to dismiss forms
on the basis of our own prejudices.
 26.3 Competition 571

26.3 Competition
26.3.1 Competition in inflection
The traditional point of view is that there is no competition in inflection: oxen blocks a
potential *oxes, stood blocks a potential *standed, and so on. Thus it is that the standard view
of level-ordered morphology puts these irregular forms at level I and then assumes that they
block the regular morphology at level II (e.g. Kiparsky 1982a).
We have seen that this is true only to a certain extent. Not only do forms like oxes and
standed arise in child-language (where we might be tempted to dismiss them as incompetent
formations), but they arise also where we are not dealing with literal uses or where we are
dealing with compounds, especially compounds which are not literal endocentric com‑
pounds. To cite a well-known example, the preterite of the verb grandstand is grandstanded,
not *grandstood in standard varieties of English. Not only is the number of incompetent regu‑
lar formations greater than is usually assumed, but variation with regular forms (and with
competing irregular forms) is also found in instances where the incompetence excuse does
not seem to be available. If this were not the case, the well-attested movement of irregular
forms to regular forms over time could not arise; the diachronic processes of holp > helped
and shoon > shoes depend upon variability between regular and irregular forms.
Accordingly, we must allow for competition in inflectional morphology, and we must
allow for competition between irregular forms as well as competition between regular and
irregular forms. We have distinctions such as spit and spat as the preterite of spit (this may
depend on dialect to a very large extent), between come and came as the preterite of come,
between forgot and forgotten as the past participle of forget, between tread and trod as the
preterite of tread, between proved and proven as the past participle of prove, between swam
and swum as the preterite of swim, and so on (see Chapter 5). If we do not include Latin
plurals for nouns as part of English here, it is only because there is an argument that they
involve code-switching rather than genuinely alternative English inflections.
Standard stories are that frequency protects irregularity in morphology, so that an infre‑
quent irregular form tends to become regularized (e.g. Bybee 1985), while words with a par‑
ticular pattern tend to create a template which can attract more words—whether this involves
movement to regular or irregular morphology (Bybee and Moder 1983). At the same time,
denominal verbs tend to be regular, even when homophonous with irregular verbs (the
grandstanded case and also He ringed/*rang the pigeon).
We do not attempt to apply these general formulations systematically to our data to pre‑
dict outcomes, but we do need to consider what kinds of pattern might be attracting new
forms. Anderwald (2009) suggests that the homophony of the preterite and the past partici‑
ple might be a principled pattern in modern English. This principle supports the non-standard
patterns in which come is the preterite of come and swum is the preterite of swim, for example.
It does not explain why spit ◆ spat ◆ spat should alternate with or turn into spit ◆ spit ◆ spit. It
572 Blocking, competition, and productivity

might explain a paradigm of tread ◆ trod ◆ trod, but not one of tread ◆ tread ◆ trod. Bybee and
Moder (1983) propose that a general template illustrated in (3) means that verbs like swim are
changing to the pattern of verbs like sink.

(3) . . . ʌ Cnasal-or-velar]verb + past

While we have evidence of the power of such a template, its precise status is in question given
that we have words like junk and, despite the spelling, monk, which are not (primarily) verbs,
and words like flunk which are not past tense (but see Albright and Hayes 2003 or Keuleers
2008 for discussion).
Another pattern which might be added to the list is a preference for monosyllabic verbs
with a final alveolar plosive to have the same form in all three parts of the verb (Bauer 1997b).
Again, we could set out a template, like that in (3), which might be something like (4) (see
Bauer 1997b on the vowels involved).

(4) [Onset V Calveolar plosive]verb (all forms)

This generalization covers standard and fixed forms like cut, hit, put, but also forms with vari‑
ation like fit, knit, rid, shit, spit, tread.
Part of the difficulty with such views of what is happening in competition between morpho‑
logical processes is that they seem to be teleological: they seem to assume a motivated move‑
ment in a particular direction towards a more regular distribution of the processes involved. In
many instances, all we can see is that there is variation, not which direction we are moving in.
For example, the use of dove as the preterite of dive started in the USA and is a more recent
innovation in CanE, AusE, and NZE, but appears to be gaining ground there, even though hav‑
ing dive ◆ dove ◆ dived not only breaks Anderwald’s predictions on patterning, but it also mixes
regular and irregular forms in the same paradigm, and having dive ◆ dove ◆ dove does not fit with
any other general pattern of vowel alternation. In instances like shit, we know what the variation
is (preterite shat ~ shit), but we do not, at the moment, know which form is likely to become the
new standard (if either). Templates like (3) or (4) might have to be reinterpreted as templates
about domains where there is variation, rather than as patterns promoting change.
From this brief discussion, we can abstract the following points:

• There can be competition between formal ways of filling the same inflectional slot in the
paradigm.
• Competition is seen in variable forms for the inflected word-form.
• Competition may indicate static variation, or a gradual acceptance of one of the variants
at the expense of the other(s).
• Individual instances of variation may be part of a general pattern of variation or may be
isolated.
 26.3 Competition 573

• Change based on such variation may lead to greater regularity (analogical levelling) or
may not; where there is regularity this may be defined in terms of predictability of para‑
digms or in terms of more widespread use of the default variant (‘the regular past tense
ending’) and the two may contrast.

26.3.2 Competition in derivation

Competition in derivation looks rather different in that the sources of competitors tend to be
rather different. Nevertheless, the same fundamental principles seem to apply.
There can be variation in the ways of filling the same slot in a derivational paradigm. In
most cases, though this is not variation in the particular derivative formed (there is some of
that, too), it is a matter of having two or more synonymous affixes. Thus ‑ness and ‑ity are
typically cited as affixes which both have the function of deriving abstract nouns from adjec‑
tives, and as such are in competition (Chapter 12). Similarly ‑ify and ‑ize (and possibly other
processes) compete in creating new verbs (Chapter 13), ‑ly, ‑ish, and ‑esque compete in pro‑
ducing adjectives of similarity from nouns (Chapter 14), ‑er and ‑ist compete in producing
words denoting people in a particular profession or people who (habitually) perform certain
actions (Chapter 11).
Some of this variation seems to be static, some of it is variation involved in change. The
variability between ‑ify and ‑ize is identified by Plag (1999) as being largely determined by the
phonology of the base (or the phonology of the output word). The suffix ‑ize requires an
unstressed syllable preceding it, whereas ‑ify requires a syllable with some degree of stress
preceding it (see Chapter 13). On the other hand, the variation we find between nominaliza‑
tion affixes in words like closure, laughter, marriage, texturization seems to have diminished
with time, so that now we can still use ‑ation, but rarely use the others to create new forms
(see Chapter 10).
Variation may be part of a general pattern or be limited to a particular set of lexemes. The
prefixes mono- and uni‑ are rarely established on the same base, but the Concise Oxford
Dictionary lists both monocycle and unicycle and monopod and unipod. On the other hand,
there is regular alternation between negative prefixes so that we find pairs like those in (5),
with or without meaning differences (see Chapter 14 for discussion).

(5) amoral immoral

apolitical unpolitical
astable instable
athematic unthematic
atypical untypical
incomputable uncomputable
inconspicuous unconspicuous
574 Blocking, competition, and productivity

indecisive undecisive
inessential unessential
intestate untestate (marked as obsolete in the OED)

In derivation, it can be difficult to decide whether there is a default affix for a particular mean‑
ing, although where there is competition between a non-native affix and a native affix it is
often the case that the native one has the wider distribution. Nevertheless, there are instances
of competition in derivation leading to greater paradigm uniformity and instances where
uniformity seems to be less of a driving force. For example, over the past two hundred years,
‑ity has gained ground vis-à-vis its rival ‑ness with bases ending in the suffix ‑al, making it
practically obligatory in this morphological context, while overall ‑ity has lost ground to ‑ness
in the formation of new words (Arndt-Lappe 2014, based on OED data). Sometimes any
uniformity is masked by the fact that words containing affixes which are no longer productive
(or whose productivity is marginal) tend to persist in the vocabulary of the speech
community.
An example of this point is the history of nominalization markers in English (see Bauer
2006a). Despite the existence of hundreds of nominalization words with a host of different
affixes, only ‑ation, ‑ing, and conversion are reliably productive today (and then in very
restricted domains, see Chapter 10). There is, in one sense, regularization, but no regularity
can be determined by looking at a list of established words.

26.3.3 A diachronic view of competition

In principle, competition can be absolutely stable, even over a long period of time. More
often, though, at least one of the processes is gaining or losing productivity. Again in princi‑
ple, we can have any combination of stability and gaining productivity or losing productivity.
The stability can be a lack of productivity or can be at some level of productivity. The stability
of laughter, hatred, or length is due to lack of productivity. The current stability of ‑ify and ‑ize
is based on the productivity of both.
Perhaps the expected picture is that a given morphological process with a certain function
is replaced with another process with the same meaning. Thus ‑ster with the meaning ‘femi‑
nine’ was replaced by ‑ess in Middle English (Bauer 2006a). The suffix ‑ster has persisted to
the present day, and is still productive, just not with that meaning. Words still in use that were
formed with the original ‑ster (baxter, seamster, songster, webster, some now only used as
names) no longer have the meaning ‘feminine’.
There are many other possibilities, though. Among others, there are instances where
two affixes are in competition only in a small part of their domain. The suffixes ‑ness
and ‑ity appear to be in particular competition following the suffixes ‑ive and ‑ous
(Aronoff 1976: 37–45; Arndt-Lappe 2012) and barely in competition following ‑able,
 26.4 Blocking 575

with only 69 types attested in COCA for forms in ‑ableness and more than 1000 for
forms in ‑ability.
The results of diachronic competition are also unpredictable. As we have just seen, the
feminine meaning of the suffix ‑ster has vanished, even though the affix has survived with
other meanings. The suffix ‑th, which once created de-adjectival or deverbal nominalizations
(warmth, birth), has left many words behind which still have their original meaning. Words
like to-stand (Anshen and Aronoff 1997) have vanished completely with the construction that
licensed them. The suffix ‑(e)rel mentioned by Marchand (1969) (see Chapter 18) in words
like mackerel and wastrel is no longer recognized as an affix.
Where there is stable variation or variation leading to change, it would be expected that
variationist studies of morphology would indicate the different options being preferred by
different parts of the population. We know only of studies that involved inflectional forms,
such as past tense ‑ed (e.g. Labov 1972; Cheshire 1982). The difficulty for such studies is that
often the variation is spread over large periods of time, typically with insufficient textual
material to allow good conclusions to be drawn.

26.4 Blocking
If there can be competition between morphological processes on the same base, there can be
no blocking. The two notions are mutually incompatible. It might be possible to say no more
than that, and to refer to the many examples of failed blocking that appear elsewhere in this
book for support. The notion of blocking is, though, widespread, and there may be value in
attempting to deconstruct it a little.
Blocking can be divided into two parts: blocking by synonymy and blocking by homon‑
ymy. Blocking by synonymy is a generalization of an older notion that there can be no syno‑
nyms (Bloomfield 1935: 145), or that one of a pair of synonyms tends to be ‘lost’ (Sturtevant
1917: 99; Ullmann 1957: 112 and works cited there; Paul 1995 [1920]: 251).
Blocking by homonymy is occasionally used in historical linguistics to explain the disap‑
pearance of one term in favour of another. The best examples involve some kind of embar‑
rassment for speakers: the replacement of ass by donkey, the replacement of coney
(pronounced /kʌni/ and thus having obscene connotations) by rabbit, the replacement of
cock by rooster. Unfortunately there are other examples where potentially embarrassing
homonymy has not led to the abandonment of a word. Hookers in rugby have not been
renamed because of street-walker hookers. And there are many expressions including cock
which have not been changed, including cock-a-doodle-do, cockatoo, cockeyed, cockpit, cock of
the walk. The embarrassment need not be on grounds of obscenity, but on grounds of con‑
flicting usage: French examples of verbs for ‘milk’ and ‘mill’ collapsing as moudre or for ‘cat’
and ‘rooster’ collapsing as gat in Gascony are widely cited (Orr 1962) as instances of homo­
nymy which have had to be repaired after the fact. Parallel examples are rarely cited in the
576 Blocking, competition, and productivity

history of English, which seems to accept homonymy. Indeed, non-embarrassing homo­

nymy is so widespread that its value as an explanation for ‘non-occurrence’ must be called
into question (see Nevalainen 1999: 453). Obvious homonymy cases are words such as
funny, hot, jolly, light, mash, mass, mast, retort, or liver (as in The Verne-Smiths were not long
livers. Lewis’s father had died at sixty and his grandfather at sixty-two. w_fict_prose 1987, from
COCA). In extreme cases, we find that forms like cleave and let can survive for many years
with two diametrically opposed meanings: ‘stick together’ and ‘split apart’, and ‘allow’ and
‘prevent’ (as in without let or hindrance), respectively. To the extent that there are good
examples of homonymy being a factor in preventing the acceptance of a particular word, it
must be a weak one.
Blocking by synonymy is thus potentially a much more powerful explanation than block‑
ing by homonymy. There are at least two ways in which this synonymy can arise (as men‑
tioned above). Either the supposed blocker can be a morphologically unrelated word, or the
supposed blocker can be a word based on the same root (perhaps even the same base) as the
supposed blockee.
In order to test either of these hypotheses, we have to know what will be taken as evidence
of blocking functioning. It seems that blocking fails systematically in a number of places.
These include the following.

• Instances where the expected blockee is formed with an extremely productive process.
Thus Aronoff (1976: 45) claims that ‑ness affixation is not blocked because derivatives in
‑ness are never listed in the lexicon. The implication is that it is not formations which are
blocked, but listing in the lexicon which is blocked.
• Instances where the expected blockee is used in a way which is not synonymous with
the blocker or where the connotations are markedly different from those of the blocker.
This allows, for example, committal and commitment to be used to mean something other
than commission. Such deliberate use of competing affixes is relatively common.
Denseness and density can mean the same thing, but the OED suggests that denseness,
when it occurs, is used mainly in non-technical areas, while density is mainly a technical
term. COCA and the BNC support this interpretation. Such subtle differences in mean‑
ing arise in individual pairs in specific contexts, as shown in Chapter 12.
• Instances where the speaker, either temporarily or permanently, is unaware of the earlier
usage. Failure of memory is often a factor here, but so too is the necessity for sophisti‑
cated vocabulary from someone who is not used to producing this level of language.
This also applies to some children’s utterances.

These conditions can be summarized by saying that blocking does not prevent the produc‑
tion of a word, it prevents the institutionalization of a word, and by saying that blocking does
not work unless the blocker and blockee are not only synonymous but also stylistically
equivalent (see also Aronoff 1976: 56; Rainer 1988).
 26.4 Blocking 577

Rainer (1988) distinguishes between token-blocking, of the type discussed above, and
type-blocking where affixes are used within different domains, and the domain for affix x is
not available for affix y and vice versa. An example from English would be nominalizations of
verbs in ‑ation as opposed to nominalizations using conversion. If ‑ation suffixation in the
contemporary language is restricted or virtually restricted to verbs ending in ‑ify and ‑ize, and
conversion to nouns does not apply to this set of verbs, then there is type-blocking between
the two. While we accept this general point, it seems to us that the mechanisms are so differ‑
ent that the use of the term ‘blocking’ in the type-blocking sense is misleading. Accordingly,
we focus on token-blocking here.
Even in the reduced sense of ‘blocking’ to which this leads, it is our view that we have
found too many examples of the failure of any such principle for us to be able to give it much
credence. Our evidence might appear in many instances to come from ad hoc neologisms
rather than from institutionalized words, but some of these apparent neologisms can be used
fairly widely and a clear-cut distinction between what is and what is not institutionalized can
be hard to establish.
Some instances are given below to illustrate our point (not, it must be stressed, an exhaus‑
tive coverage). Many others can be found in the pages of this book.
Orientate has replaced (or largely replaced) the verb orient in the sense of ‘position one‑
self or an object in respect of the compass or some other object’. The verb orient still remains,
but typically with more abstract types of location, as suggested by the examples from cor‑
pora (e.g. it has been oriented towards an academic hermit’s life, American Studies International
2003, COCA).
As was shown in (5), various negative prefixes, even the less productive ones, may be used
with a certain degree of variability. While these sometimes show (or are said to show) a differ‑
ence in meaning (amoral versus immoral, for example), this is not always the case, as illustrated
in (6) with examples from COCA (see Chapter 17 for more examples and discussion).

Esquire 2010: Spitzer was every bit as dishonest and amoral and greedy as the worst of
(6) 
his enemies
NPR_TalkNation 2008: But when I do mention it to them, they assume that we’re
immoral, that we don’t have a moral code.

The various prefixes glossed as ‘to remove ~ from’ such as de‑, dis‑, and un‑ in denature, dis‑
mast, uncap show some diachronic patterning among themselves, but in some instances are
found on the same base with the same meaning (e.g. dethrone, disthrone, unthrone; see Bauer
2006a).
Various adjective-forming suffixes may be found on the same base apparently without
semantic distinction: barbaric, barbarous; gigantean, gigantic; Greenlandic, Greenlandish; wom‑
anish, womanly. This is not to deny that the same pairs of suffixes have, on occasions, become
lexicalized with distinct meanings: cupric, cuprous; mannish, manly; official, officious.
578 Blocking, competition, and productivity

Regular plurals are used for nouns which normally have irregular forms when they are (a)
parts of names or (b) non-literal. This gives us Mickey Mouses, Mapleleafs, and their gooses are
cooked (non-standard) (see Chapter 7 for more discussion).
None of this provides a picture of blocking functioning as expected. That being the case,
the question of whether blocking involves thief and stealer or only examples like bequeathal
and bequeathement becomes moot, and need not be taken seriously.

26.5 Productivity
It appears to be uncontroversial to say that some morphological processes can be used in the
creation of new words while others are rarely or never used in this way. In recent years, fol‑
lowing Corbin (1987) it seems to have become generally accepted that questions about pro‑
ductivity can be split into two: can a given process be used at all, and if so, to what extent is it
used? These are distinguished as ‘availability’ and ‘profitability’, though as we mentioned in
Chapter 2.6 we do not use the terminology unless potential ambiguity forces us to.
Even at this point we may have strayed beyond the boundaries of the uncontroversial:
although we can say at a given point that we lack evidence of the productivity of a given
morphological process, this cannot be interpreted as meaning that it could not be resurrected
at some point. Only those processes which leave no traces in the vocabulary can be safely
assumed to be dead.
The discussion of productivity (sensu profitability) has been carried out in two distinct
terminological frameworks. On the one hand, there is current usage of questions of ‘con‑
straints’ on productivity. In this framework, the nominalizing suffix ‑al has as a constraint that
it must be added to a verb which carries stress on the final syllable. The alternative
­terminology—one we have employed freely in other places in the book—is in terms of
domains of productivity. In these terms, the nominalizing ‑al suffix is productive only in the
phonological domain which has stress on the final syllable of the base (or, equivalently, on
the penult of the derivative). We take it that these two views are equivalent, though one
focuses on the exclusion of impossible forms and the other focuses on potential sites of inclu‑
sion. It is for this reason that the ‘domain’ metaphor seems to be preferable in general.
Nevertheless, there are places (such as the reduced use of adverbial ‑ly after an adjective
which ends in ‑ly) where the constraint terminology seems reasonable, since the limitations
are more neatly expressible in such terms.
Some types of domain are well-established in the literature, others are less well developed.
Domains may be established in phonological terms, in morphological terms (the affix y may
be added to a word which ends in affix x), in terms of word class (the affix x may be added
only to bases of word class y), in terms of semantics (the affix x may be added to a
word which bears the semantic specification [y]), in terms of pragmatics (the affix x may be
added only to bases to which the speaker wishes to express a particular attitude or when a
 26.5 Productivity 579

word expressing a particular attitude is required); the affix x may be added only to create a
word in some technical area) or in terms of etymology (the affix x may added only to bases
with a certain etymology—though we have suggested that such domains are likely to be
weak in contemporary English). In principle a domain could be established by the intersec‑
tion of any set of these.
A full discussion of domains is still required (though see van Marle 1985). It seems that
some domain types are, if not impossible, then not found in English. We do not have a
domain for a particular suffix which demands a prefixed base, for instance, or a domain for a
particular prefix that demands a suffixed base. While we have some affixes whose domains
seem to be monomorphemic bases, it is not clear that we have any whose domains are specifi‑
cally and exclusively morphologically complex bases, or which attach only to compounds.
Similarly, there are domains referring to prosodic categories or structures, but it is unclear
whether there are any principled restrictions as to which types of prosodic domain would not
be be able to play a role. While we do not find in English affixes which attach only to words
denoting animals or trees (for example), there seems to be no reason why such processes
would not be possible (and some occur in other languages). The best constraint seems to be
that the same category is not marked tautologically twice on the same base, but even that is
not always true, for example with diminutives (see Chapter 18), agentive nominalizations
(Chapter 11), or historically as in children when an earlier marker loses its force. Arguably,
multiple prefixation of the type sub-sub-prime is not tautological.
It is not always clear at what level domains should be stated. If, for example, ‑ity attaches to
words ending in ‑al (and allomorphs), ‑able, ‑ic, and also ‑ive (see Chapter 12.2.1), it may be
that there is a point at which it should be determined that the domain of this suffix is non-
native adjectival bases. The fact that other non-native affixes such as ‑ous and ‑ian can be
added to the list makes this seem reasonable. It is, however, not clear to what extent the lack
of words in ‑esquity, ‑istity, ‑entity, or ‑antity disproves the wider claim. The more specific
claims seem less contentious.
Not only may various different affixes share domains, even contrasting affixes can share
domains. Thus, ‑ee and ‑able may both be added to transitive verbs (so they share a domain,
but do not compete), and ‑ness and ‑ity may both be added to adjectives in ‑ive (where they
do compete). So the core cases of competition are instances where synonymous affixes (or,
more generally, morphological processes) share domains. Again, there is some question as to
how the various domains are best formalized. If it is the case that ‑ness has any adjective as its
domain, the fact that it competes with ‑ity on adjectives in ‑ive is more or less accidental.
Nevertheless, the basic point holds.
Some more general constraints on productivity have been proposed in the literature. These
include the word base hypothesis (Aronoff 1976: 21), The Unitary Base Hypothesis (Aronoff
1976: 48), the Unitary Output Hypothesis (Scalise 1984: 137), level ordering constraints
(starting with Bloomfield 1935, but see also Kiparsky 1982a, and Plag 1999: 45, Bauer 2001:
128 for critical discussion). We mostly deal with these elsewhere in this book (see Chapter 29),
580 Blocking, competition, and productivity

but here simply need to repeat that where it is clear how such constraints are to be inter‑
preted, they do not appear to be strongly supported by our data.
If affixes can compete in the same domain, we can imagine various possible outcomes.
One is that all bases in the domain can be used with either affix. Another is that the affixes are
distributed more or less randomly among the bases, perhaps with some duplication. This is
what appears to happen with ‑ness and ‑ity on bases in ‑ive. We can have expensiveness but
?
expensivity seems odd (despite expansivity); ?evasivity seems not to be used (and is not in
COCA at the time of writing). The implication is that the affixes in such instances are not of
equal productivity in particular domains. A more general conclusion is that even in domains,
there are degrees of productivity, it is not simply a matter of availability or non-availability
within a domain.
Another factor which can make productivity appear to be scalar in this way is the number
of bases that are available for a particular morphological process to act upon. In Chapter 18.4.6
the example of the suffix ‑ton is given, which appears to occur only on one base in the estab‑
lished lexicon (singleton) but which can be attested on two other bases. It is an open question
as to whether Nquintupleton is a possible word of English or not, but even if it is, there are
only a handful of possible bases for formations with this suffix. There are more potential
bases for the prefixation of step‑ (see Chapter 11.2.5 and 11.3.2), but not huge numbers. On the
other hand, the bases for ‑er suffixation or un‑ prefixation are legion. It is not clear how the
productivity of ‑er and ‑ton can be compared. It seems that the number of formations pro‑
duced is one factor in determining productivity, so that even if ‑ton is found on all of the
words to which it can in principle attach, and ‑er is found on not quite all of the words to
which it can in principle attach, ‑er will be viewed as more productive. This is contrary to the
claim made by Aronoff (1976).
This example makes it clear that, even in simple cases, it can be difficult to determine the
productivity of individual morphological processes. The actual productivity of morphologi‑
cal processes is no less fraught. Although we can do calculations over established words from
a source such as the OED (Plag 1999), and this will give some kind of impression of produc‑
tivity, or deduce productivity from the number of hapaxes in a corpus (Baayen and Lieber
1991; Baayen 1992, 1993) in ways which give intuitively plausible results, we do not appear to
be able to give reliable and repeatable measures of productivity that allow cross-linguistic
comparisons to be made or even that allow comparisons between different corpora (see
Baayen 1993 and for some discussion, Bauer 2001). Moreover, dictionary counts and corpus
counts tend to provide quite distinct results, in many instances.
That is not to say that we do not have any clues. Following the work by Baayen (1992, 1993)
it is generally accepted that the more productive a particular morphological process is, the
more it will have outputs which are of low token frequency, while the products of non-­
productive morphology will tend to have high token frequency. In a large enough corpus, all
the examples of an unproductive morphological process repeat words from a closed list of
established words. With productive morphological processes, in contrast, at least some of the
 26.6 Conclusion 581

words found in a corpus are likely to be neologisms or relatively recent formations, which
will have low token frequency. While it is by no means the case that all hapaxes in a corpus
like the BNC will illustrate productive morphological processes at work, nevertheless, the
hapaxes will provide a proxy measure for the number of words of low frequency which are
being formed by the relevant morphological process. This correlation has been used in other
chapters in this book to deduce productivity in various morphological processes.
Accordingly, we have not attempted in this book to give accurate measures of productivity,
but have restricted ourselves to rather impressionistic statements of degree of productivity.
From a practical point of view, it is not entirely clear how accurate measures would be of
value. It is clear that, for instance, EFL learners and teachers need to be most aware of the
most productive morphological processes, and can ignore those processes which are of very
low productivity, at least as far as the productive skills of speaking and writing are concerned.
From a theoretical point of view, it is not clear how productivity should be seen as part of the
language system, or even whether it should be (Bauer 2001 provides some discussion and
references). It has been claimed that productive morphology is stored differently in the brain
from unproductive morphology and if this is true, then some linguists will want to model the
two in different ways (e.g. Pinker and Prince 1994; Pinker 1999); others, however, want to
model them in precisely the same way (Rumelhart & McClelland 1986; Skousen et al. 2002;
Keuleers 2008). We can observe different behaviours on the part of different morphological
processes, but the interpretation of those behaviours is still controversial.
One of the possibilities is that productivity is not a cause of morphological distinctions,
but the result of morphological distinctions (Hay 2003). It may be that differences in produc‑
tivity are caused by differentials in phonological and semantic transparency, parsability or
naturalness (see, e.g. Mayerthaler 1981) between affixes, and that affixes which are easier for
the speaker/listener to isolate because their meaning is clearer, because their form is more
consistent, and because they comply with general cognitive principles and expectations of
language structure are more productive than those which are more difficult to isolate.
Tempting though this hypothesis may be, there are some features of English morphology
that tend to cast it into doubt. One of those features is the productivity of conversion in
English, which is not easily accounted for given the assumptions of theorists of naturalness.
Second, we have seen that affixes are productive in particular domains, some of which force
lack of transparency (for example, the use of velar softening) at morpheme boundaries, with‑
out apparently losing productivity. Again, what we see is that the morphological facts are
open to several theoretical interpretations.

26.6 Conclusion
Competition, blocking, and productivity are interlocking concepts in the study of morphol‑
ogy. Our data shows so much competition in domains of productivity that we cannot find
582 Blocking, competition, and productivity

strong support for any strong notion of blocking, even if we try to restrict the notion. The
notion of productivity is important in morphological description, but its theoretical impor‑
tance is open to interpretation and controversial. As a result we describe some of the phe‑
nomena we observe in this area, but do not attempt to provide a theoretical interpretation of
these phenomena.
 chapter 27

The nature of stratification

27.1 Prospectus
Although the primary subject of this volume has been the contemporary state of English
morphology, we cannot completely ignore the historical circumstances that have given rise
to the exceptional richness and intricacy of present-day English. Marchand (1969) of course
provides us with details of the origins and development of particular affixes and types of
compounding and conversion, and the OED supplements this with etymological informa-
tion about affixes, but neither gives a complete overview. Further, theoretical developments
in the last forty years have drawn us to look at the intertwined strands of English morphology
in new ways, and our current study of contemporary corpora can be used to shed further
light on theoretical issues that arise with regard to the interaction of different strata in English
morphology: those aspects of morphology that are native to the language and those aspects
that come into English as a result of wholesale borrowing from French and the classical lan-
guages Latin and Greek.
The theoretical issues have hinged for the most part on the extent to which and the specific
ways in which these strata can interact with each other. What we will suggest in the following
is that theoretical claims have largely been based on an incomplete picture, and that there is
less in the way of stratification in English derivation than has been assumed. This is in fact a
good result: as Aronoff and Fuhrhop (2002: 469) have pointed out, it is unclear ‘how the
etymological distinction is reflected in the synchronic grammar of a naïve speaker of English,
who cannot be expected to know the origins of words’. The conclusion that we come to is that
the etymological distinction is only a faint one, if it exists at all in the mental lexicons of
contemporary native speakers of English.
In Section 27.2 we will summarize in tabular fashion a very large amount of data on the
combinatorial properties of both suffixes and prefixes. Some of this data can be found piece-
meal in previous chapters, but the bulk of it is drawn from a careful search of COCA. In
Section 27.3 we will assess what a close look at the data can tell us about the interactions of
native and foreign strands of morphology in contemporary English. Specifically, concentrat-
ing on combinations of suffixes, we will examine the extent to which contemporary English
can really be said to have separate strata, and the ways in which the strata behave differently. In
Section 27.4 we consider the theoretical consequences of stratification (or the lack thereof).
584 The nature of stratification

27.2 Data
Although a fully general discussion of stratification in English morphology might logically
include the behaviour of compounds—for example, the extent to which neoclassical com-
bining forms have come to combine with native bases (e.g. forkology, cardioglider, psycho-
babble)—in this chapter we will confine ourselves to a discussion of the extent of
stratification in affixation. In what follows, we will refer to native and non-native affixes (see
the discussion in Chapter 2.7). The reader is referred to the individual chapters for more
detail on the formal behaviour and semantics of individual affixes. In the tables below we
consider most of the productive affixes of English, but the reader should note that they are
not absolutely exhaustive. If an affix appears in one of the tables, but fails to occur in others
this absence is significant, as it indicates that pertinent combinations are not attested in
our data.
As can be seen from Table 27.1, all native affixes with even marginal productivity in the
contemporary language can attach to either native or non-native bases. In Table 27.1 and the
following tables, the symbol ‘✓’ indicates that a particular form or combination is attested in
our data, the symbol ‘!’ indicates that a particular form or combination appears in very few
types, and a blank indicates that we have not found the form or combination attested.
As Tables 27.2 and 27.3 suggest, non-native affixes vary in their ability to take native bases;
nevertheless many prefixes and also quite a few suffixes can be found on both non-native and
native bases:
It is fairly clear from Tables 27.2 and 27.3 that non-native suffixes are generally more
inclined to be selective in their bases than non-native prefixes are, but perhaps less so than
earlier literature suggests (cf. Giegerich 1999, for example).
We must also ask to what extent we find both native affixes attached to non-native affixes
and non-native attached to native ones. In a derivational system without stratification, we
might expect the two orderings to be equally likely, all other things being equal (that is, as
long as each affix can fulfil its other selectional requirements). Of course, all things are never
equal: as we have seen in Chapter 21 (and references cited there), accounting for the ordering
of affixes in English is quite a difficult issue as there are many factors that seem to be involved
in determining what can attach to what. Nevertheless, we might expect to find at least a fair
amount of interleaving of native and non-native affixes.
In Tables 27.4–27.7 we have put together examples of combinations of suffixes that we have
found attested. Many of these are taken from Plag and Baayen (2009) (marked with ‘p’ in the
Source column), who themselves have gathered examples from Hay and Plag (2004) and
from the OED. To these we have added our own examples of combinations we find attested
in COCA, and very occasionally from Google (marked with ‘g’ in the Source column).
Tables 27.8–27.11 show combinations of prefixes. Searching for combinations of prefixes
presents a problem that generally does not occur in searching for combinations of suffixes,
Table 27.1 Native affixes and the bases they select
Affix Native bases Examples Non-native bases Examples
after‑ ✓ afterbirth, afterchurch, afternoon ✓ aftercourse, after-effect, after-image
back‑ ✓ backboard, backflow, backland ✓ back-country, back-focus, back-office
be‑ ✓ befall, befriend, behead ✓ becalm, beglamored, besport
by‑ ✓ by-blow, by-catch, byway ✓ by-effect, by-election, bypass
‑dom ✓ beardom, kingdom, wisdom ✓ pariahdom, entreedom, marveldom
down‑ ✓ down-beat, downbound, downfall ✓ down-ballot, downcity, downcycle
‑er ✓ writer, thinker, sleeper ✓ employer, attacker, defaulter
‑fold ✓ sevenfold, twofold ✓ billion-fold
fore‑ ✓ forearm, forebode, foresee ✓ forecourt, foremastered, fore-thorax
‑fulA ✓ slothful, wakeful, rightful ✓ deceitful, resentful, purposeful
‑hood ✓ babehood, thinghood, fatherhood ✓ geniushood, mountainhood, gorillahood
‑ish ✓ babyish, goodish, doomish ✓ iconish, modernish, caricaturish
‑less ✓ eyeless, fatherless, houndless ✓ ageless, captionless, roseless
‑let ✓ eyelet, leaflet, ringlet ✓ chainlet, coverlet, statelet
‑ling ✓ darkling, earthling, lordling ✓ changeling, princeling, sharkling
‑ly ✓ nightly, fleshly, readerly ✓ actorly, comradely, musicianly
mid‑ ✓ mid-answer, midbreath, mid-glide ✓ midargument, midcareer, midevent
mis‑ ✓ mishit, misbelieve, misbirth ✓ misact, misdefine, mispredict
‑ness ✓ afraidness, awakeness, bigness ✓ abjectness, benignness, obtuseness
off‑ ✓ off-black, off-board, off-earth ✓ off-angle, off-campus, off-license
on‑ ✓ ongoing, onload, onset ✓ onmarch, onmoving, on-passing
out‑ ✓ out-bid, outbreak, outflow ✓ out-anticipate, outbalance, outdistance
over‑ ✓ overall, overanswer, overbid ✓ overabrade, overattend, overcapture
‑ship ✓ queenship, hateship, fathership ✓ consulship, tenureship, carpetship
‑some ✓ healthsome, darksome, mirthsome ✓ adventuresome, frolicsome, fruitsome
‑ster ✓ oldster, youngster, spinster ✓ fraudster, strumster, scenester
Affix Native bases Examples Non-native bases Examples
un‑ ✓ unfair, unlearn, undeath ✓ unable, undevout, unmanifest
under‑ ✓ underarm, underdrive, undereat ✓ underachieve, underclass, underemploy
up‑ ✓ upbound, upflow, uphold ✓ upcycle, upgrade, upmountain
‑ward ✓ deathward, downward, godward ✓ cityward, equatorward, mirrorward
‑wise ✓ babywise, drinkwise, healthwise ✓ agewise, caloriewise, fashionwise
Table 27.2 Non-native suffixes and the bases they select
Affix Native bases Examples Non-native bases Examples
‑able ✓ bakeable, bearable, doable ✓ admirable, cherishable, imaginable
‑acy ✓ advocacy, conspiracy, determinacy
‑age ✓ footage, gooseage, readage ✓ baronage, brokerage, symbolage
‑alN ! upheaval ✓ acquittal, disbursal, recital
‑alA ! bridal, tidal ✓ aerial, bronchial, electoral
‑anN ! Roughian ✓ centenarian, grammarian, musician
‑anA ! elvan ✓ apian, brontosaurian, carnivoran
‑ance ! riddance, hindrance ✓ abettance, conductance, utterance
‑ancy ✓ claimancy, hesitancy, occupancy
‑antA/N ! heatant, coolant, floatant ✓ accelerant, refrigerant, accountant
‑ary ✓ deputary, expansionary, sectary
‑ateV ✓ alienate, capsulate, assassinate
‑ation ! botheration, flirtation, starvation ✓ abbreviation, insinuation, revolution
‑ee ✓ askee, callee, helpee ✓ contactee, expellee, interpretee
‑ery ✓ cheesery, grainery, wifery ✓ museumery, riflery, distillery
‑ese ! motherese ✓ dissertationese, funeralese, computerese
‑esque ✓ girlesque, hell-esque, island-esque ✓ bordello-esque, divaesque, humanesque
‑ess ✓ witchess, falconress, folkstress ✓ composeress, mentoress, hauntress
‑ette ! wenchette, wifette, yeomanette ✓ conductorette, reporterette, chefette
‑ic ! folkloric, runic ✓ aquatic, iconic, synaptic
‑ical ! churchical, folklorical, raspberrical ✓ physical, exegetical, meteorological
‑ify ! churchify, mouse-ify, prettify ✓ electrify, liquify, purify
‑ine ✓ adamantine, estuarine, pantherine
‑ism ✓ motherism, foodism, leftyism ✓ ageism, careerism, abnegationism
‑ist ✓ leftist, hornist, womanist ✓ absolutist, biologist, colonist
‑ite ! foamite ✓ aluminite, calcite, dendrite
Affix Native bases Examples Non-native bases Examples
‑ity ! oddity ✓ absurdity, binarity, density
‑ive ! talkative, walkative ✓ adaptive, captive, substantive
‑ize ! ironize, leatherize, weaponize ✓ analogize, crystallize, unionize
‑ment ! unfoldment, upliftment, wonderment ✓ advisement, bombardment, incitement
‑or ! sailor ✓ abductor, counselor, descriptor
‑ory ✓ commendatory, divinitory, modulatory
‑ous ! heathenous, lumpous, righteous ✓ amorous, cancerous, frivolous
‑trix ✓ adminstratrix, coredemptrix, generatrix
Table 27.3 Non-native prefixes and the bases they select
Affix Native bases Examples Non-native bases Examples
a‑ ✓ acultural, asemantic, acephalous
ante‑ ! ante-buildings, anteroom ✓ antebellum, antecedent, antenatal
anti‑ ✓ anti-day, anti-bone, antichoice ✓ anti-abrasion, antichemistry, antinatural
arch‑ ! archdeacon, archbishop, archfriend ✓ arch-conservative, archheresy, archvillain
circum‑ ✓ circumambient, circumcontinental, circumlocution
cis‑ ✓ cisalpine, cislunar, cisplatinum
co‑ ✓ co-band, co-believer, co-father ✓ co-adapted, co-composer, co-suspect
contra‑ ! contraflow ✓ contra-causal, contradistinction, contrafactual
counter‑ ✓ counterblow, counter-death, counter-flow ✓ counteract, countereffect, countertheory
crypto‑ ! cryptochurches, crypto-free, cryptokeys ✓ crypto-analysis, cryptologic, crypto-racist
de‑ ✓ dehorn, debone, deice ✓ deautomate, decharge, deidealize
demi‑ ! demi-cheese, demigod, demi-island ✓ demi-celebrity, demifigure, demisphere
dis‑ ! disbelief, disband, dislike ✓ disaffect, disconfirm, disfunction
en‑ ✓ endear, enroll, enshroud ✓ enact, encapsulate, enchain
ex‑ ✓ ex-baker, ex-band, ex-boyfriend ✓ ex-assistant, ex-beauty, ex-chief
extra‑ ✓ extra-cold, extra-deep, extra-kin ✓ extraabdominal, extracreative, extradense
hemi‑ ! hemi-field ✓ hemicycle, hemiparasite, hemithyroid
hyper‑ ✓ hyperflow, hypercool, hyperbright ✓ hyperactive, hyper-complex, hyperplastic
hypo‑ ! hypospray, hypostick ✓ hypoacoustic, hypodermal, hypothermic
NEG
in‑ ✓ immortal, inactive, infertile
inter‑ ✓ interarm, interband, interbreed ✓ interaction, intercept, interpsychic
intra‑ ✓ intraisland, intraday, intra-earth ✓ intracerebral, intragranular, intra-item
mal‑ ! mal-feeling ✓ malaligned, malfunction, malpractice
mega‑ ✓ megabad, megacool, megadrive ✓ megachain, megamachine, megapopular
meta‑ ✓ meta-awareness, meta-beliefs, metaman ✓ meta-analysis, metalinguistic, metastructure
micro‑ ✓ microbeam, microcraft, microdrive ✓ microabrasion, microbiology, microunit
mini‑ ✓ minibreak, mini-foot, miniloaf ✓ mini-capital, mini-environment, mini-interval

(continued)
Table 27.3 Non-native prefixes and the bases they select (continued)
Affix Native bases Examples Non-native bases Examples
mono‑ ✓ monoband, monobrow, monoman ✓ monocausal, monopedal, monotheism
multi‑ ✓ multi-field, multi-board, multifight ✓ multi-agent, multicavity, multi-family
nano‑ ✓ nano-boat, nano-death, nanogear ✓ nanoarray, nanocycle, nanomagnet
neo‑ ✓ neo-cool, neodeath, neofold ✓ neoantique, neofeudal, neoliterate
non‑ ✓ non-arm, nonbedridden, nonchosen ✓ nonadept, noncapital, non-comic
para‑ ✓ parachurch, paraglide, parasail ✓ para-capitalist, paranationalist, paraphysical
peri‑ ! peri-workout ✓ periapical, perinatal, periurban
pico‑ ! picofast ✓ picogram, picoplankton, picowatt
poly‑ ✓ polywater, polybands, polycold ✓ polycarbon, polynuclear, polyvapor
post‑ ✓ post-afternoon, post-awakening, postbath ✓ post-acquittal, post-devolution, postmodern
pre‑ ✓ pre-answer, prebake, pre-cold ✓ pre-activity, precombat, prefuneral
proto‑ ✓ proto-board, proto-earth, protoheart ✓ proto-adult, protohuman, protomodern
pseudo‑ ✓ pseudo-cool, pseudo-friend, pseudo-men ✓ pseudo-abstract, pseudodistance, pseudo-legal
re‑ ✓ reheat, rewash, rewrite ✓ reaccept, reignite, repollute
retro‑ ✓ retrobreeding, retrocool, retroglide ✓ retrocession, retrosexual, retrovirus
semi‑ ✓ semi-clean, semi-death, semi-float ✓ semi-assault, semi-deviant, semiformal
sub‑ ✓ subband, sub-breed, subdeal ✓ subacute, subcategory, subdomain
super‑ ✓ superchild, superbright, superclean ✓ superabundant, supergenerous, superintense
supra‑ ! supra-church, supralife, ✓ supraannual, supraethnic, supra-state
supra-neighborhood
tera‑ ! terabeam ✓ terascale, terameter, terabase
trans‑ ! transcool, transearth, trans-light ✓ trans-bay, transcanal, transcontext
ultra‑ ✓ ultraaware, ultraclean, ultradark ✓ ultra-accurate, ultrafilter, ultrapure
uni‑ ! unibrow, unicorn, uni-tree ✓ uniblock, unilateral, uniport
 27.2 Data 591

Table 27.4 Non-native suffixes followed by native suffixes

Affixes Source Examples
‑able able+ness admirableness
‑age age+er packager
age+ful p carriageful
age+less p carriageless
age+ness p foolageness
age+wise p percentagewise
age+y cleavage-y
‑alA al+dom intellectualdom
al+hood tribalhood
al+ish normalish
al+less institutionalless
al+ship gubernatorialship
‑alN al+wise arousal-wise
al+er trialer
‑ance ance+er remembrancer
ance+less distanceless
ance+ship acquaintanceship
ance+wise appearance-wise
ance+y Renaissancy
‑ancy ancy+wise consistency-wise
‑antN ant+hood servanthood
ant+less servantless
ant+ship assistantship
‑ary ary+hood p secretaryhood
ary+less p dictionaryless
ary+ness p elementariness
ary+ship p judiciaryship
ary+wise culinary-wise
‑ation ation+er vacationer
ation+ful actionful
ation+hood nationhood
ation+less affectionless
ation+ship relationship
ation+wise admiration-wise
ation+y suctiony
‑ee ee+dom refugee-dom
ee+hood p employeehood
ee+ship p assigneeship
‑ery ery+dom p nurserydom
ery+less batteryless
ery+ship p ministryship
ery+wise jewelry-wise
‑ese ese+er Portugueser
ese+y Chinese-y

(continued)
592 The nature of stratification

Table 27.4 Non-native suffixes followed by native suffixes (continued)

Affixes Source Examples
‑esque esque+ness Freudianesqueness
‑ess ess+dom p princessdom
ess+hood p priestesshood
ess+less p governessless
ess+ly p princessly
ess+ship p governessship
ess+wise p huntresswise
ess+y princess-y
‑ette ette+ish p noveletteish
‑an ian+dom p Christiandom
ian+er p musicianer
ian+hood p roughianhood
ian+ish p christianish
ian+less p guardianless
ian+ly p guardianly
ian+ness p Victorianness
ian+ship p musicianship
ian+wise p Christianwise
‑ic ic+ish cinematic-ish
ic+ness democraticness
ic+wise economic-wise
‑ical ical+ness clericalness
ical+wise political-wise
‑ify ify+er acidifier
‑ine ine+er mariner
‑ist ist+dom p artistdom
ist+er p alchemister
ist+ly p artistly
ist+ship p evangelistship
ist+y dentist-y
‑ite ite+ish Israelitish
ite+ship Naziriteship
ite+wise favorite-wise
‑ity ity+dom celebritydom
ity+less gravityless
ity+wise coachability-wise
‑ive ive+ly p adaptively
ive+ness p addictiveness
ive+wise defensive-wise
‑ize ize+er neutralizer
‑ment ment+ee p experimentee
ment+er p documenter
ment+ful apartmentful
ment+ship p governmentship
ment+ward apartmentward
 27.2 Data 593

Affixes Source Examples

ment+wise assignment-wise
ment+y testament-y
‑or or+dom p protectordom
or+ful p traitorful
or+hood p traitorhood
or+ish p administratorish
or+less p connectorless
or+ling p professorling
or+ly p sailorly
or+ship p administratorship
‑ory ory+ness obligatoriness
ory+wise laboratory-wise
‑ous ous+ly p adventurously
ous+ness ambiguousness
‑ure ure+er sculpturer
ure+ful pleasureful
ure+hood creaturehood
ure+less pressureless
ure+ly creaturely
ure+ship tenureship
ure+some adventuresome
ure+wise disclosure-wise

so a word is in order about our data. For the reasons noted in Chapter 21, it is more likely for
prefixes to occur on already suffixed bases than for suffixes to occur on already prefixed bases.
Therefore, in looking at combinations of prefixes, care must be taken to examine the likely
bracketings of a complex word to determine whether a prefix is attached directly outside
another prefix, or outside a suffix that is itself outside a prefix. That is, strictly speaking prefix
combinations should show a structure like (1a), rather than (1b):

(1) a. prefix [ prefix [ base-suffix ] ]

b. prefix [ [ prefix-base] suffix ]

In the examples in Tables 27.8–27.11 we have confined ourselves to the bracketing in (1a),
where possible, and have marked cases where the bracketing is in doubt or is likely to be that
in (1b) with a following *. Especially problematic are forms in which a prefix is attached to a
deverbal form in ‑ing or ‑ed, as the analysis of such forms is highly theory dependent. We have
included such cases.
594 The nature of stratification

Table 27.5 Native suffixes followed by non-native suffixes

Affixes Source Examples
‑en en+able resharpenable
en+ee p flattenee
en+ment p lengthenment
en+ize blackenize
‑er er+age p farmerage
er+ese computerese
er+esque painteresque
er+ess p breweress
er+ette p sleeperette
er+ial managerial
er+ian g Quakerian
er+ism p boxerism
er+ist p consumerist
er+ite bleacherite
er+ous debaucherous
er+ize blenderize
‑fulA ful+ize awfulize
‑hood hood+ism p neighbourhoodism
‑ish ish+ian p Irishian
ish+ism p Britishism
ish+ist p Yiddishist
ish+ment p foolishment
ish+ery p Scottishry
‑let let+ist p novelettist
‑ly ly+ism p orderlyism
‑ship ship+ment p courtshipment
‑ster ster+age p hucksterage
ster+ess p hucksteress
ster+ette p rhymsterette
ster+ian p spinsterian
ster+ism p hucksterism
ster+ous p spinsterous
‑yA y+esque hippyesque
 27.3 Patterns 595

Table 27.6 Native suffixes followed by native suffixes

Affixes Source Examples
‑dom dom+ful p kingdomful
dom+less p kingdomless
‑en en+er p flattener
‑er er+dom p printerdom
er+ful p tumblerful
er+hood p loverhood
er+ish p robberish
er+less p leaderless
er+ling p preacherling
er+ly p loverly
er+ship p controllership
er+wise p loverwise
er+y slacker-y
‑fold fold+ness p tenfoldness
fold+wise p manifoldwise
‑fulA ful+ness p carefulness
ful+wise p despiteful-wise
‑hood hood+er neighborhooder
hood+less p childhoodless
hood+ly neighborhoodly
hood+wise sisterhood-wise
hood+y neighborhoody
‑ish ish+ly p babyishly
ish+ness p amateurishness
ish+wise g Englishwise
ish+y Englishy
‑less less+ness p aimlessness
less+wise p carelesswise
‑let let+hood triplethood
let+wise couplet-wise
‑ling ling+less p seedlingless
ling+ship p ducklingship
‑ly ly+hood p knightlihood
ly+ness p courtliness
ly+wise p beastlywise
‑ness ness+wise p businesswise
ness+y business-y
‑ship ship+er townshipper
ship+ful p kingshipful
ship+less p censorshipless
ship+wise citizenship-wise
ship+y hero worshipy
‑some some+ness fearsomeness
‑ster ster+dom p gangsterdom
596 The nature of stratification

Affixes Source Examples

ster+hood p spinsterhood
ster+ish p spinsterish
ster+ly p spinsterly
ster+ship p tapstership
ster+y gangster-y
‑th th+en p lengthen
th+less p depthless
th+wise p depthwise
‑ward ward+ness inwardness
ward+wise backward-wise
‑wise wise+ness p otherwiseness
‑yA y+ish p woollyish

Table 27.7 Non-native suffixes followed by non-native suffixes

Affixes Source Examples
‑able able+ism probabilism
able+ist compatibilist
able+ity absorbability
able+ize tangibilize
‑age age+able packageable
age+ery p plumagery
age+ette suffragette
age+ism suffragism
age+ist assemblagist
‑alA al+ese legalese
al+esque industrialesque
al+ian sesquipedalian
al+ic vocalic
al+ism behavioralism
al+ist environmentalist
al+ite socialite
al+ity eventuality
al+ize actualize
al+ous triumphalous
‑an an+age p guardianage
an+ary p physicianary
an+ate Americanate
an+ese lesbianese
an+esque Italianesque
an+ic Romanic
an+ism p Europeanism
an+ist p Europeanist
 27.3 Patterns 597

Affixes Source Examples

an+ous p ruffianous
an+ize Africanize
‑antN ent+al presidential
ant+ery tenantry
ant+ese consultantese
ant+ic cartomantic
ant+ism protestantism
ant+ist obscurantist
ent+ize immanentize
‑ary ary+ian p librarian
ary+ism p secretaryism
ary+ist p voluntaryist
ary+ize sedentarize
ary+ous alimentarious
‑ateV ate+able activatable
ate+al striatal
ate+ant hydratant
ate+ation activation
ate+ee allegatee
ate+ese administratese
ate+ic automatic
ate+ism privatism
ate+ive adulterative
ate+or activator
ate+ory assimilatory
ate+ous coruscatious
ate+ure implicature
‑ation ation+able actionable
ation+al educational
ation+ary inflationary
ation+esque blaxploitationesque
ation+ese educationese
ation+ette recessionette
ation+ism assimilationism
ation+ist abortionist
ation+ize televisionize
‑ee ee+ess p refugee-ess
ee+ism p absenteeism
‑ery ery+al cemeterial
ery+ian cemeterian
ery+ic imageric
ery-ist p effronterist
ery+ize batteryize
‑ess ess+ery witchessery
‑ette ette+age p briquettage
ette+ism p suffragettism
598 The nature of stratification

Affixes Source Examples

‑ic ic+ate authenticate
ic+ian statistician
ic+ism aestheticism
ic+ist geneticist
ic+ity electricity
ic+ize academicize
‑ical ical+ism biblicalism
ical+ist theatricalist
ical+ity cyclicality
ical+ize theatricalize
‑ify ify+able identifiable
ify+ance significance
ify+ant significant
ify+ate unificate
ify+ation electrification
ify+ee justifiee
ify+ic beatific
ify+ive significative
ify+ory amplificatory
‑ine ine+age libertinage
ine+al doctrinal
ine+ate glycerinate
ine+ic alkalinic
ine+ism alpinism
ine+ist alpinist
‑ism ism+al baptismal
ism+ic monopolismic
‑ist ist+ery p dentistry
ist+ess p artistess
ist+ette columnistette
ist+ic artistic
ist+ical egotistical
‑ite ite+ess Moabitess
ite+ic anchoritic
ite+ical cenobitical
ite+ism Jacobitism
ite+ist Maronitist
‑ity ity+ate debilitate
ity+ive capacitive
ity+ism ability-ism
ity+ist complicitist
ity+ize securitize
‑ive ive+al perspectival
ive+ate activate
ive+ism p activism
ive+ist p activist
 27.3 Patterns 599

Affixes Source Examples

ive+ity activity
ive+ize activize
‑ize ize+able actualizable
ize+ance cognizance
ize+ant cognizant
ize+ation actualization
ize+ee socializee
ize+ment aggrandizement
ize+or destabilizor
‑ment ment+al argumental
ment+ary testamentary
ment+ee experimentee
ment+ical testamentical
ment+ism good-governmentism
ment+ist p developmentist
ment+ive argumentative
ment+ite transvestmentite
ment+ous p medicamentous
‑or or+age p tutorage
or+ance doctorance
or+esque survivor-esque
or+ess p editoress
or+ette conductorette
or+ial editorial
or+ian p protectorian
or+ical oratorical
or+ism p actorism
or+ist p detectorist
or+ous p traitorous
or+ize factorize
‑ory ory+ial accusatorial
ory+ize accessorize
‑ous ous+ist seriousist
ous+ity pomposity
‑ure ure+able pleasurable
ure+al closural
ure+ess adventuress
ure+ous moisturous
600 The nature of stratification

Table 27.8 Non-native prefixes preceding native prefixes

Prefix combination Example
anti‑ anti+back antibackflow
anti+mis antimiscarriage*
anti+out anti-outsourcing
anti+over anti-overrun
anti+up antiuplander*
co‑ co+be co-belonging
co+under co-underwriter
counter‑ counter+out counter-outrage
ex‑ ex+be ex-betrothed
ex+fore ex-forehead
ex+mid ex-midshipman
ex+out ex-outlaw
ex+un ex-untouchable
hyper‑ hyper+be hyper-bedecked
hyper+over hyper-overarching
hyper+un hyper-unemployment
mega‑ mega+off mega-offspring
mega+out megaoutlet
mega+up megaupload
meta‑ meta+after metaafterphysics
micro‑ micro+mis micromismanagement*
mini‑ mini+down mini-downtown
mini+out mini-outbreak
mini+up mini-uproar
mono‑ mono+un mono-unsaturated
multi‑ multi+out multioutlet
multi+un multi-untalented
neo‑ neo+out neo-outlaw
non‑ non+after nonafterburning
non+be nonbelonging
non+by nonbypassable*
non+down nondowntown
non+fore non-foreclosed
non+mid non-midday
non+mis non-misleading
non+out non-outgassing
non+over nonoverweight
non+under non-underserved
non+up non-updatable*
para‑ para+mid paramidline
poly‑ poly+un polyunsaturated
post‑ post+be post-bewitched
post+down postdownsizing
post+fore post-foreclosure
post+mid postmidnight
post+out postoutbreak
 27.3 Patterns 601

Prefix combination Example

post+up post-uprising
pre‑ pre+down pre-downfall
pre+fore pre-foreclosure
pre+mid pre-midsentence
pre+on pre-onset
pre+out pre-outburst
pre+over pre-overhead
pre+under pre-understanding
pre+up pre-update
proto‑ proto+up proto-uproar
pseudo‑ pseudo+out pseudo-outcome
re‑ re+fore reforecast
re+out reoutline
re+over reoverlaying
re+un reunveil
re+under re-underwriting
re+up reuptake
semi‑ semi+out semi-outlaws
semi+un semi-unknown
semi+under semi-underground
semi+up semi-upscale
sub‑ sub+fore sub-foreman
super‑ super+out superoutburst
super+over super-overheated
super+un super-unfunny
super+under super-underdog
super+up super-upset
ultra‑ ultra+under ultra-underdog
ultra+up ultra-upscale

Table 27.9 Native prefixes preceding non-native prefixes

Combination Example
back‑ back+re back-reaction
back+trans back-translated
down‑ down+dis down-dislocation
mid‑ mid+inter mid-interview
mid+para midparagraph
mid+peri midperiphery
mid+pre mid-preview
mid+re mid-rewind
mid+sub mid-suburban
mis‑ mis+co miscoordination
602 The nature of stratification

Table 27.9 Native prefixes preceding non-native prefixes (continued)

Combination Example
mis+dis misdisunderstanding
mis+en misencoded
mis+inter misinterpret
mis+pre mispredict
mis+re misrepresent
mis+trans mistranslated
off‑ off+dis off-dissonant
off+micro off-microphone
out‑ out+mega out-megaphoned
out+re outrebounded
out+trans out-transmutation
over‑ over+co over-coordinated
over+de over-decentralized
over+dis overdiscounted
over+en overentangled
over+extra overextrapolate
over+inter over-interactive
over+meta over-metaphysical*
over+pre over-predict
over+re overreaction
over+sub oversubscribed
over+super oversupervise
un‑ un+circum uncircumscribed
un+co uncooperative*
un+dis undisguised
un+en unenlightened
un+ex unexcommunicated
un+extra unextraordinary
un+hypo unhypocritical
un+inter un-international
un+meta unmetaphysical
un+para unparagraphed
un+poly unpolychrome
un+post un-post-modern
un+pre unpredictable*
un+re unreburied
un+sub unsubtitled
un+super un-supervised
un+trans untransformed
under‑ under+dis underdisclosure
under+inter under-interpretation
under+micro undermicrowaved
under+pre underprescribed
under+re underrepresented
under+sub under-subscribed
under+super undersupervisor*
under+trans undertransmission
 27.3 Patterns 603

Table 27.10 Native prefixes preceding native prefixes

Prefix combination Example
after‑ after+mid after-midnight
after+un after-unfolding
mid‑ mid+after midafternoon
mid+fore mid-forearm
mis‑ mis+over mis-overheard
mis+under misunderattributed
over‑ over+out over-outrage
over+under over-underachiever
un‑ un+back unbackcountry
un+be unbewitch
un+down undowntrodden
un+fore unforetellable*
un+mid un-Midwest

Table 27.11 Non-native prefixes preceding non-native prefixes

Prefix combination Example
a‑ a+meta a-metaphysical
ante‑ ante+post ante-postmodern
ante+retro anteretrograde
anti‑ anti+ante anti-ante-bellum
anti+crypto anticryptography
anti+de anti-deforestation*
anti+dis antidisclosure
anti+en antientitlement*
anti+ex anti-ex-combatant
anti+hyper anti-hypertension
anti+in anti-independent
anti+inter anti-intercellular
anti+meta antimetaphysical
anti+micro antimicrobacterial
anti+multi anti-multicultural
anti+neo anti-neo-colonialist
anti+non anti-non-Anglo
anti+poly antipolygamy
anti+pre anti-premarital
anti+proto antiprotozoal
anti+pseudo anti-pseudojournalists
anti+re anti-replay
anti+sub anti-submissive
anti+trans anti-transplant
co‑ co+dis codiscoverer

(continued)
604 The nature of stratification

Table 27.11 Non-native prefixes preceding non-native prefixes (continued)

Prefix combination Example
co+en coenrollment
co+inter co-interventions
co+poly copolymer
co+re co-represent
co+sub co-sub-halos
co+super co-supervisor
counter‑ counter+dis counterdiscursive
counter+en counterenactment
counter+hemi counter-hemisphere
counter+inter counter-intervention
counter+mono countermonologue
counter+para counterparadigm
counter+pre counter-preemption
counter+re counterreactions
counter+sub countersubversive
counter+super counter-superpower
crypto‑ crypto+a crypto-atheistic
de‑ de+en de-entitle
de+inter de-interlacing
de+mono demonopolization
de+multi demultiplexing
de+poly depolymerized*
de+re derearrange
de+sub desublimate
de+super desuperheater
de+uni deunification
demi‑ demi+semi demi-semi-democratic
dis‑ dis+co discoordinated
dis+en disentangle
dis+in disinhibition
dis+inter disintermediate
ex‑ ex+anti ex-anti-kid
ex+counter ex-counterterror
ex+crypto excryptographer
ex+in ex-independent
ex+inter ex-intercontinental
ex+neo ex-Neo-Destour
ex+non ex-non-entity
ex+para ex-paracommando
ex+poly ex-polytechnics
ex+re ex-replacement
ex+semi ex-semi-permanent
ex+super ex-supermodel
extra‑ extra+anti extra-antiwear
extra+hemi extrahemispheric
 27.3 Patterns 605

Prefix combination Example

extra+hypo extrahypopharyngeal
extra+sub extra-subcutaneous
extra+super extra-super-sensitive
hemi‑ hemi+a hemianopia
hemi+crypto hemicryptophytes
hemi+demi hemidemisemiquaver
hemi+trans hemitransfixion*
hyper‑ hyper+circum hyper-circumspect
hyper+de hyperdeformed
hyper+en hyperenriched
hyper+in hyperinsecurity
hyper+inter hyperintervention*
hyper+meta hypermetabolic
hyper+multi hypermultilateral
hyper+para hyper-parathyroidism*
hyper+pre hyper-preparedness
hyper+re hyperreactive
hyper+sub hyper-suburban
hyper+uni hyper-unilateralism
hypo‑ hypo+para hypoparathyroid
hypo+re hyporeactivity
in‑
NEG
in+co incoordination
in+de indecomposable*
in+dis indissoluble*
in+pre inprescribable*
in+sub insubordination
in+trans intransmutable*
inter‑ inter+co intercooperation
inter+hemi interhemisphere
inter+in interindependent
inter+intra inter-intragroup
inter+meta intermetamorphosis
inter+micro intermicrotubule
inter+re inter-react
inter+sub intersuburban
inter+trans intertransverse
intra‑ intra+peri intrapericardial*
mega‑ mega+re megarenovation*
mega+sub megasubdivision*
mega+super megasuperstar
meta‑ meta+re metarepresentational
meta+sub meta-subtext
micro‑ micro+co micro-cogeneration
micro+dis microdisarmament
micro+en microencapsulated
micro+meta micrometastases
(continued)
606 The nature of stratification

Table 27.11 Non-native prefixes preceding non-native prefixes (continued)

Prefix combination Example
micro+mini microminidress
micro+nano micronanosecond
micro+para micro-parameters
micro+re microreform
micro+sub microsubmersibles*
micro+trans microtransactions*
mini‑ mini+de mini-defibrillator*
mini+dis mini-discovery*
mini+en mini-enclosure*
mini+micro mini-microwave
mini+mono mini-monograph
mini+pre mini-preview
mini+pseudo mini-pseudo-Medici
mini+re minireunion
mini+retro miniretrospectives*
mini+sub minisubmarine
mini+trans minitransmitters*
mono‑ mono+anti monoantiplatelet
multi‑ multi+anti multiantibiotic
multi+dis multidisabled
multi+inter multi-interdisciplinary
multi+mega multimegawatt
multi+para multiparagraph
multi+re multi-representational*
multi+tera multi-terawatt
multi+trans multi-transnational
nano‑ nano+en nano-enabled
nano+inter nano-interrogation
nano+re nano-reinforced
nano+trans nanotransformers
neo‑ neo+a neoatheistic
neo+counter neo-countercultural
neo+crypto neocryptopygus
neo+dis neodissociation*
neo+en neoenlightenment*
neo+inter neointerventionism*
neo+meta neometaphysical
neo+post neo-post-Marxists
neo+pre neopremodern
neo+re neo-restructuring
neo+retro neoretromaximinimalist
neo+ultra neo-ultramontanist
non‑ non+anti nonantisocial
non+circum noncircumstantial*
 27.3 Patterns 607

Prefix combination Example

non+co noncooperation*
non+contra noncontraceptive*
non+counter noncounterfeit
non+de nondecodable*
non+dis nondisordered
non+en non-entitlement*
non+hyper nonhyperactive
non+hypo non-hypoallergenic
non+in nonindependence
non+inter noninteractive*
non+intra non-intracranial
non+mal nonmalware
non+mega nonmegapixel
non+meta nonmetaphysical
non+micro nonmicrogravity
non+multi nonmultimedia
non+neo nonneoplastic
non+para nonparametric*
non+peri nonperinatal
non+poly non-polycarbonate
non+post nonpostdoctoral
non+pre non-pre-emptible*
non+proto nonprototypical
non+re nonrenewable*
non+sub nonsuburban
non+super nonsupernatural
non+trans nontransplanted
non+ultra non-ultra-orthodox
non+uni nonunidimensional
para‑ para+neo paraneoplastic
peri‑ peri+sub perisubmandibular
post‑ post+anti post-antibiotic
post+de post-deregulation*
post+dis post-discharge
post+en postenlightenment*
post+hemi post-hemiabdominal
post+inter post-interview
post+meta postmetamorphic*
post+mono post-monopolization*
post+multi post-multicultural
post+neo postneonatal
post+para post-paracentesis
post+re postrestructuring
post+retro post-retrofit
post+sub post-suburban
608 The nature of stratification

Prefix combination Example

post+super post-superpower
post+trans post-transplant
pre‑ pre+ante preantepenultimate
pre+anti pre-antibiotic
pre+co pre-coordination
pre+contra precontraception
pre+counter pre-counterculture
pre+de predecode
pre+dis predischarge
pre+en pre-enlistment
pre+in preindependence
pre+inter preinterview
pre+meta premetamorphosis
pre+micro pre-microchip
pre+mono pre-monotheistic
pre+neo pre-neoclassical
pre+para preparadigmatic*
pre+post prepostmodernists
pre+proto preprototype
pre+retro preretrofit
pre+trans pretransplant
proto‑ proto+de proto-deconstruction*
proto+demi proto-demi-Celts
proto+in proto-indeterminacy*
proto+inter proto-internet
proto+meta proto-metaphysical*
proto+post proto-postmodern
proto+pre proto-preservationist*
proto+sub protosuburban
proto+super proto-superstate
pseudo‑ pseudo+dis pseudo-disclosure
pseudo+en pseudo-enchantment*
pseudo+in pseudoindependents
pseudo+inter pseudo-interview
pseudo+meta pseudo-metathesis
pseudo+multi pseudo-multimedia
pseudo+neo pseudoneointima
pseudo+poly pseudo-polymaths
pseudo+re pseudo-review
pseudo+super pseudo-superheroes
re‑ re+co recoordinated
re+de redefruition
re+dis redislocation
re+en re-enforce
re+inter reinterview
 27.3 Patterns 609

Prefix combination Example

re+meta re-metamorphizes
re+mono remonopolization*
re+sub resubjugate
re+trans retransmit
retro‑ retro+trans retrotransposable
semi‑ semi+co semi-cooperative
semi+crypto semi-crypto-necrophiliac
semi+de semi-defrost
semi+dis semi-disassemble
semi+en semi-enclosed
semi+ex semi-ex-baseball fan
semi+hypo semi-hypochondriac
semi+in semi-invisible
semi+mono semi-monopoly
semi+non semi-nonsense
semi+para semi-parametric
semi+peri semi-peripheral
semi+pre semi-premeditated
semi+re semi-rebuilding
semi+sub semi-submarine
semi+super semi-superpower
semi+trans semi-transparent
sub‑ sub+anti sub-antarctic
sub+hyper sub-hypersonic
sub+nano sub-nanometer
sub+para sub-paragraph
sub+re sub-replacement*
sub+trans sub-transfer
sub+uni sub-universities*
super‑ super+anti super-antioxident
super+de super-dehydrated
super+dis super-discharge
super+en super-entrenched
super+hyper super-hyper-regional
super+micro super-microdots
super+mini superminimalists
super+mono super-monopolists
super+para superparamagnetic
super+poly super-polymer
super+post superpost-panamax
super+re super-reinforced
super+sub super-subsonic
super+extra superextraordinary
super+inter superinternational
super+multi supermultivitamins
610 The nature of stratification

Prefix combination Example

super+trans supertransport
supra‑ supra+re supra-representational*
supra+sub supra-subsonic
supra+uni supra-universal*
trans‑ trans+anti trans-antarctic
trans+peri transpericardial*
trans+sub trans-substantiate
ultra‑ ultra+dis ultradisposable*
ultra+meta ultrametabolism*
ultra+micro ultra-micro-engraving
ultra+mini ultraminiature
ultra+para ultra-paradoxical
ultra+re ultra-reactionary*
ultra+sub ultra-suburban
ultra+super ultra-supercritical

27.3 Patterns
If we look at the full range of data, what we see is that English derivational morphology actu-
ally displays very little in the way of stratification. We show this first by grouping affixes
according to their restrictiveness with regard to bases, that is, showing the extent to which
they allow bases that are both simple and already derived, and the extent to which those
bases can be both native and non-native. Example (2) ranks the suffixes in the tables from
most to least restrictive:

(2) a. Allow only non-native simplex bases; do not allow complex bases at all:
Non-native: ‑ancy, ‑ine
Native: none
b. Allow both non-native and native simplex bases; do not allow complex bases at all:
Non-native: none
Native: ‑enV, ‑fold, ‑let, ‑ster
c. Allow both non-native and native simplex bases; do not attach to complex bases with
native affixes:
Non-native: ‑ation, ‑alN, ‑ance, ‑ant, ‑ary, ‑ateV, ‑ic, ‑ical, ‑ity, ‑ive, ‑or, ‑ory, ‑ure
Native: none
d. Allow both non-native and native simplex bases; allow both non-native and native
complex bases
 27.3 Patterns 611

Non-native: ‑able, ‑age, ‑alA, ‑ee, ‑ery, ‑ese, ‑esque, ‑ess, ‑ette, ‑an, ‑ism, ‑ist, ‑ite, ‑ize,
‑ment, ‑ous
Native: ‑dom, ‑er, ‑fulA, ‑hood, ‑ish, ‑less, ‑ling, ‑ly, ‑ness, ‑ship, ‑some, ‑ward, ‑wise, ‑y

We note that while there are non-native suffixes that only attach to simplex non-native bases,
there are no native suffixes that are restricted with regard to the etymological origin of the
simplex bases they attach to, although there are a few that attach only to simplex bases (that
is, they are never found attached to already affixed bases). There is a small cohort of non-
native suffixes that attach at least sporadically to native simplex bases, but attach only to
complex bases ending in non-native suffixes. For native suffixes, on the other hand, if they
attach to complex bases at all, they tend to attach freely to those ending in both native and
non-native suffixes. The majority of native suffixes fall into this least restrictive of categories.
What is perhaps surprising is the number of non-native suffixes that fall into the least restric-
tive category as well: indeed, there are more non-native suffixes (N = 16) that attach freely in
this way than there are ones that fall into the more restrictive category (N = 13).
Prefixes show an even clearer pattern. All of the native prefixes attach indiscriminately to
both native and non-native bases, as do the vast majority of non-native prefixes. There are
only five non-native prefixes that seem never to attach to native bases (a‑, circum‑, cis‑, NEGin‑,
mal‑) and eight that attach to native bases only rarely (contra‑, hemi‑, peri‑, pico‑, supra‑, tera‑,
trans‑, uni‑). It is notable that none of these prefixes is particularly productive in contempo-
rary English. As for attaching to other prefixes, non-native prefixes of course attach freely to
other non-native prefixes, but they also seem to attach readily to native prefixes as well. If
anything is surprising at all in the data compiled in Tables 27.8–27.11, it is that native prefixes
are relatively unlikely to attach to any other prefix, whether native or non-native. We have far
more combinations of non-native attaching to native than the reverse.
What the data indicate, then, is that native and non-native affixes in English may not be
completely integrated into a single derivational system, but they are far more integrated than
might at first be imagined. We find near-complete integration among prefixes, and only a bit
less integration among suffixes. We offer this result with some caveats however. First, it must
be acknowledged that some of the examples in the tables in this chapter are derived on com-
plex bases that are rather lexicalized (wizardess, bastardize, courtshipment, dictionaryless, min-
istry-ship, celebritydom, co-belonging, ex-betrothed, megaoutlet, mini-uproar, pre-onset, etc.), and
therefore arguably are not perceived as complex. Second, some of these forms appear to
violate the selectional restrictions of affixes they contain: for example, we find forms like
tribalhood or gubernatorialship, where ‑hood and ‑ship generally attach only to nouns. And
most importantly, not all combinations are equally productive. Some are attested multiple
times, others a few times, several only once.
It is of course difficult to know what to make of those combinations that occur only once. On
the one hand, it is easy enough to dismiss the combinations that occur infrequently or only once;
Aronoff and Fuhrhop (2002) tend to dismiss occasional occurrences as not significant, for
612 The nature of stratification

example. On the other hand, as Hay and Plag (2004) point out, given two low-probability affixes,
the probability of their combination becomes extremely low indeed; they calculate that to attest
a single example of ‑ee-hood would require nearly 1.7 million types in a corpus. Correspondingly,
we cannot exclude the possibility of finding new combinations as the corpus grows larger. Given
these comments, the fact that we find even a single example of a particular combination should
be taken as significant. We have therefore not dismissed such isolated occurrences.
Taking these caveats into account, we can conclude that native affixes are more productive
on non-native bases (both simplex and complex) than non-native affixes are on native bases,
but that the latter pattern is not at all rare or anomalous. We might even speculate that the less
frequent occurrence of the non-native-on-native ordering is not to be attributed to any ves-
tige of stratification, but rather to the fact that the non-native affixes in question are less
generally productive or less separable than native ones. Certainly, in the case of prefixes,
where the level of separability is higher (see Chapter 21), we have a higher level of attachment
of non-native to native.
This being the case, we have to ask why the myth of stratification is so well-established in
the literature on English word-formation. We believe there are several reasons, all of which
contribute to some degree.

• Typology: It is often the case in other languages that native goes with native and non-
native with non-native (see, for example, Fleischer 1975 on German, and Townsend 1975
on Russian). These languages treat stratification rather more restrictively than is the
case in English. This may have established expectations about the way in which English
was likely to behave.
• Genesis: Non-native affixes made their way into English by being borrowed as parts of
non-native words, and the best parallels to bases in the non-native words were equally
non-native. Thus the earliest English formations using non-native affixes were probably
on non-native bases, but as is indicated by our data, that has changed in more modern
English.
• Item-familiarity: Even in the examples listed above, many of the mixtures of native and
non-native occur in words that are not item-familiar. Generalizations over item-familiar
words thus exaggerate the influence of stratification.

27.4 Theoretical consequences

The patterning of native and non-native affixes is interesting enough on its own, but it is
perhaps more interesting when put in the context of theoretical claims about the stratifica-
tion of English derivation made, for example, by the theory of Lexical Phonology and
Morphology (Siegel 1974; Allen 1978; Kiparsky 1982a; Giegerich 1999) or Aronoff and
Fuhrhop’s (2002) Monosuffix Constraint.
 27.4 Theoretical consequences 613

Lexical Phonology and Morphology, at least in its earliest form, is now largely discredited
as a theory of the interaction of morphology and phonology, but the notion that there are
two classes of affixes or two strata of derivation in English still persists to some extent. It has
never been explicitly claimed that the two strata align perfectly with the etymological origins
of particular affixes, but scrutiny of the most complete breakdown of affixes, that in Selkirk
(1982), suggests that the division falls to a large extent along etymological lines. Note that
Selkirk’s (1982) list of affixes overlaps to a large extent with those in the tables presented here,
but not entirely. For example, she treats ‑man (postman) as a suffix, whereas we do not. On
the other hand, she does not categorize ‑ite or ‑ess at all.

(3) Selkirk’s breakdown of suffixes:

a. Class 1: ‑ette, ‑ity, ‑th, ‑ance, ‑ee, ‑ation, ‑alA, ‑ic, ‑ary, ‑ous, ‑an, ‑ese, ‑esque, ‑ive, ‑ory,
‑ify, ‑ate V, ‑enV
Class 2: ‑hood, ‑ship, ‑dom, ‑er, ‑let, ‑ling, ‑ful, ‑age, ‑ness, ‑alN, ‑less, ‑ly, ‑y, ‑ish, ‑some
b. 
c. Both: ‑ist, ‑ism, ‑ment, ‑able, ‑ize

What this organization suggests is that with the exception of the suffixes ‑age and ‑alN, all
non-native affixes are Class 1, and with the exception of the unproductive native ‑th, all native
affixes are Class 2. Selkirk does not explicitly justify this division, referring to previous work
on the subject (Allen 1978; Siegel 1978), but some of the reasoning for including ‑alN is theory
internal, and would seem to argue for several other Class 1 affixes being Class 2 instead, a
conclusion that Giegerich (1999) develops in some detail. Assuming the division in (3),
however, Selkirk claims that Class 2 suffixes may follow Class 1 suffixes, but not vice versa.
The net effect is that for the most part native suffixes follow non-native, but non-native suf-
fixes usually do not follow native suffixes.
After Selkirk, there is remarkably little attention given to individual English affixes in the
literature on Lexical Phonology and Morphology, the assumption presumably having been
that once classified as stress-changing (Class 1) and stress-neutral (Class 2), there was little
more to be said about particular affixes. As can be seen from our observations in this volume,
however, this is by no means a sound assumption. We have seen that when non-native affixes
are stress-changing, they are rarely stress-changing in exactly the same way (see especially
Chapter 9, as well as Burzio 1995; Raffelsiefen 1999). The data in Section 27.2 suggest further
that there is far more interleaving of native and non-native affixes than Lexical Phonology
and Morphology would have predicted (see also Chapter 21).
We should point out that the strict stratal division of affixes is a tenet only of classical
Lexical Phonology and Morphology. A more recent version of the theory, that of Giegerich
(1999), advocates distinct lexical strata, but not on the basis of a division of affixes. Indeed,
Giegerich argues that quite a few affixes would have to belong to both Class 1 and Class 2, if
the behaviour of affixes were taken seriously. Giegerich instead advocates a model in which
614 The nature of stratification

bases rather than affixes are assigned to different strata. What is interesting about Giegerich’s
model for our purposes is that he claims that the native/non-native distinction does have a
place in English morphology. Specifically, he argues that non-native affixes (in his terms
[+Latinate]) are largely confined to non-native bases, while native affixes ([–Latinate] in his
terms) have no such restriction. In other words, he argues that regardless of how our theoreti-
cal model is constructed, the distinction between native and non-native affixes is still active
in the contemporary language. The data we have gathered here offer little support for
Giegerich’s claim: while there is a tendency for non-native suffixes to favour non-native bases,
there is far more mixing of the two systems of derivation than such theoretical claims might
have led us to expect.
As we have mentioned, the classic theory of Lexical Phonology and Morphology has been
largely discredited, and therefore our observations might be deemed unnecessary nails in an
already sealed coffin. The notion that there is something to be said about the differential
behaviour of native and non-native derivation persists, however. A prominent example of this
is Aronoff and Fuhrhop’s (2002: 473) Monosuffix Constraint, which they state as in (4).

(4) Suffixes that select Germanic bases select unsuffixed bases.

Indeed, Aronoff and Fuhrhop (2002: 475–6) make the stronger claim that English derivation
is strongly stratified on a native/non-native basis.
Furthermore, although Latinate suffixes may disobey the constraint, they attach to unsuf-
fixed words much less commonly than Germanic suffixes do, and they normally attach to
suffixes which also carry the feature Latinate, so the picture drawn here is that there are two
different word-formation systems, especially within the combination of suffixes, one
Germanic and one Latinate.
Aronoff and Fuhrhop’s claims seem weak even on the basis of their own evidence, which
is largely culled from the OED. Although they claim that ‑ness is the only suffix that attaches
freely to other Germanic affixes, their own charts show at least twenty other examples of
native-on-native suffixed forms. Their reason for dismissing these combinations is that they
are infrequent, sometimes attested more than a century ago and therefore not contemporary,
and presumably therefore not productive.
We have given good reasons in this chapter and elsewhere in this volume to suggest that
examples cannot be dismissed because they are infrequent. And oddly, not all of the combi-
nations that Aronoff and Fuhrhop dismiss really are infrequent. For example, they indicate
that there are thirty examples of er + y and twenty of er + ship. Given that they were working
with dictionary data, and that forms with these affixes are likely to be relatively transparent,
we find these patterns to be quite robust. If these combinations were semantically transpar-
ent, why would we expect them to be recorded in a dictionary at all? In any case, the examples
that we have cited above are largely contemporary (the ones we have added from COCA are
entirely contemporary), and although some of them are quite infrequent, we still consider
 27.5 Conclusion 615

them as significant. The conclusion that we draw is that the distinction between native and
non-native morphology is not strong in the contemporary language.

27.5 Conclusion
The data adduced here thus suggest that stratification is only faintly visible in contemporary
English derivation: although there is a tendency for non-native suffixes to be somewhat
selective, neither native affixation nor non-native prefixation seem to be stratified.
Nevertheless, we might rightly raise the question of how a naïve native speaker of English
would be able to maintain even this faint distinction without having any conscious knowl-
edge of the etymology of English morphemes. The best answer still seems to hinge on pho-
nological characteristics of native versus non-native derived words. With respect to bases,
Adams (2001) notes the following:

But the native–foreign distinction does correlate to some extent with differences in the
shapes of words. Native words are often monosyllabic, or have two syllables with the accent
on the first. Foreign words often have three or more syllables and are typically not accented
on the initial syllable (Adams 2001: 12).

Further clues might come from phonological coherence and segmentability. As has been
observed in several ways (Raffelsiefen 1999; Hay 2003), native suffixes by and large are more
likely to form distinct phonological words, or in psycholinguistic terms to be more easily
separable than non-native suffixes. Non-native prefixes often have the characteristics of form-
ing distinct phonological words and being easily separable as well. And affixes that constitute
distinct phonological words or are in other ways more separable would be expected to follow
those that are phonologically cohering and therefore less separable (see Chapter 21 for dis-
cussion). In other words, to the extent that both prefixes and native suffixes are more likely to
be less phonologically fused to their bases than non-native suffixes, the distinction between
native and non-native continues to maintain a sort of shadow existence psycholinguistically.
 chapter 28

English morphology in a
typological perspective

28.1 Prospectus
It is beyond the scope of the present volume to show where English stands typologically with
respect to the myriad types of morphology displayed in the languages of the world. For one
thing, such a comparison would require much better knowledge than we have of the cross-
linguistic parameters that characterize morphology. In addition, it is not yet clear what sorts
of insight are to be gained from such large-scale comparisons. We nevertheless begin with the
traditional way of characterizing morphology as isolating, agglutinating, fusional, or polysyn-
thetic that goes back to the work of Humboldt (1836) and Sapir (1921), as well as more mod-
ern attempts at large-scale classification, concluding in the end that such schemas do not tell
us much that is interesting or useful.
We argue, however, that typological comparisons on a more local level have the poten-
tial to be more illuminating, and therefore we opt for a finer-grained comparison of
English with Romance morphology on the one hand and Germanic on the other.
Specifically we will compare English inflection and word-formation to that of French and
German. French of course has had direct historical influence on English, but German has
not. Dutch or Frisian might be thought a more accurate point of comparison, or even
Danish, as its ancestor Old Norse had some influence up until the Norman invasion.
Nevertheless, German provides a familiar and convenient West Germanic language for
comparative purposes.
As a Germanic language with a huge non-native vocabulary made up largely of Romance
elements, English finds itself torn between Germanic and Romance types of structure.
Occasionally there are signs of Greek types, as well. On this more local level of comparison,
then, English is typologically heterogenous, having been influenced in different directions by
the patterns derived from different parts of its patrimony. This makes it interesting, and also
challenging from the point of view of typology.
 28.2 Broad classificatory schemes 617

28.2 Broad classificatory schemes

28.2.1 The traditional Humboldtian classification
Since the work of Humboldt (1836) and Sapir (1921), it has been traditional to classify mor-
phological systems as belonging to one of four types: isolating, agglutinating, fusional, or
polysynthetic. It is well-known, however, that morphological systems as a whole rarely fit
squarely into one of the four categories, and even looking at inflection and derivation sepa-
rately it is sometimes hard to apply a hard and fast classification. Comrie (1981: 51) therefore
suggests two intersecting scales, which he calls the index of synthesis and the index of fusion.
The index of synthesis gauges how many morphemes there typically are per word, with iso-
lating languages being low on the index and polysynthetic languages high. The index of fusion
gauges how many meanings are typically packed into a single morpheme in a language, with
fusional languages being higher on the index than agglutinating languages. We will look at
English briefly both in terms of the original classification and in terms of Comrie’s modifica-
tion of it.
We begin with English inflection. Old English clearly had fusional inflection, with, for
instance, case and number expressed in a single exponent (Kastovsky 2000). Contemporary
English, however, has little inflection left to begin with, and in terms of the inflection it does
have, very few traces of fusional inflection remain. First, irregular past tenses and past parti-
ciples like began and begun show the fusion of the root and grammatical information, as do
umlaut plurals like feet, mice. Second, the third-person non-past singular ‑s in carrie‑s,
reduce‑s, understand‑s is fusional to the extent that third-person and singular can be seen as
independent grammatical features of English verbs. This is slightly dubious, since apart from
the use of this ‑s, the only place where singular and plural are regularly distinguished on
verbs is with the verb be. The verb be is also the only place where person is regularly marked
on verbs apart from the use of this ‑s. In other places, fusion has vanished. Singular on nouns
is unmarked. Where plural is marked irregularly, plural and genitive show agglutination, not
fusion (as in ox‑en‑’s). In the regular cases we have suggested that there is haplology of one
of the markers of plural and genitive (see Chapter 7.3.3). Regular past tenses are agglutina-
tive, too.
We turn now to derivation. Oddly, English is frequently said to be an isolating type of
language. Although in less ornate styles English might tend towards simplex words, its mor-
phology nevertheless cannot be said to be isolating. Many English words of more than three
syllables turn out to be morphologically complex (Jansen-ism, mater-ial-ism, penny-pincher),
although this is certainly not always the case, as examples like peristroika or kangaroo show.
On the other hand, very little derivation in English is fusional, and the fusional derivation we
find is lexicalized, for example, the isolated pairs abide ◆ abode, sing ◆ song, and the very few
causative forms linked by vowel change like fall ◆ fell, lie ◆ lay. Of the morphological types
suggested by the four-fold classification, derivation in English comes closest to being
618 English morphology in a typological perspective

agglutinative. Even where there are instances like persuasion /pəˈsweɪʒən ‖ pərˈsweɪʒən/
where it can be difficult to say precisely where the morpheme boundary runs, there is no
doubt as to the fact that there is a string of elements. But calling English agglutinative is at
best a broad-brush characterization: English of course has abundant non-agglutinative deri-
vation in the form of conversion (Chapters 10, 11, 14, 25) and clipping (Chapter 18). We will
return to this point shortly.
Compounds are, by their very nature, agglutinative, at least those compounds that we have
referred to as canonical compounds (Chapter 19). However, non-canonical compounds—
those that seem to have been formed by back-formation or conversion—again raise ques-
tions for the Humboldtian classification.
We can conclude that in terms of the traditional classification, English morphology fits best
overall into the agglutinative type. There are two ways in which this conclusion is unsatisfying,
however. First, English morphology is rather different from that of other agglutinative languages
like Turkish or Finnish (but see Haspelmath 2009 for a cross-linguistic test of the notion of
agglutination). In this regard, we could perhaps look to Comrie’s indexes of fusion and synthesis
to provide some nuance. We might say, for example, that English would be ranked relatively low
on the index of fusion, with few examples where more than one meaning is packed into the
same morpheme, and somewhere midway on the index of synthesis, with words—at least mul-
tisyllabic ones—frequently having more than one morpheme. Comparatively, Turkish would
probably be lower than English in terms of fusion and higher in terms of synthesis.
But even with the more nuanced measure, our classification of English is still unsatisfying.
First, it is vexing that the classification as a whole does not tell us what to do about sorts of
morphology that are non-concatenative, for example, conversion, back-formation, and clip-
ping. Since these are significant processes in English derivation and compounding, classify-
ing English as agglutinative misses something important. Worse, however, is that we are left
with a feeling that this kind of classification does not give us any real insight into English in
comparison to other languages. In other words, once we have determined that English
belongs to this category or that, what does this lead us to expect in terms of correlated struc-
tural properties or historical developments? To our knowledge, nothing follows from these
classificatory schemes (see Haspelmath 2009; Bauer 2010a: 136–7 for a similar conclusion).

28.2.2 Head- versus dependent- marking

Nichols (1986) suggests that another way of looking at morphological systems—specifically
inflectional systems—is in terms of head- versus dependent-marking. In head-marking lan-
guages, relationships between the head of a phrase and its dependents (a term which Nichols
uses to cover arguments and modifiers), are signalled by inflection on the head, whereas in
dependent-marking languages inflection appears on the arguments or modifiers of the head.
In English, for example, possession is marked on the possessor rather than on the possessed
 28.3 A finer-grained view 619

noun (the dog’s bed), which would suggest that English is a dependent-marking language.
However, subject person is marked on the verb (at least for third-person singular in the
present tense), which might be interpreted as head-marking, similar to plural marking in
noun phrases. If this is correct, then English could not be regarded as strictly one or the other
type, not even necessarily within the same phrase types. Given the general dearth of inflec-
tion in English, it is not clear how useful this distinction is for the typological classification of
English, nor again does it allow us to predict other characteristics of English.

28.3 A finer-grained view

The failure of broad classificatory schemes to tell us anything satisfying about English does
not mean, however, that we have given up on typology. It just means that we find a finer-
grained approach to cross-linguistic comparison more telling. The strategy we opt for here is
to look at English inflection, derivation, and compounding in comparison to our closer
neighbours, that is, Germanic and Romance languages. For the reasons indicated in Section
28.1, we focus most closely on a comparison with German and French, although we will have
reason to refer to the classical languages on occasion as well.

28.3.1 Inflection
One typological parameter along which languages may differ is the distinction between
word-based and stem-based inflection. Inflection in contemporary English is overwhelm-
ingly word-based. That is, the inflectional affixes are attached to morphological entities which
can stand alone, at least in the regular instances, as shown in (1).

(1) cat cat‑s

walk walk‑s
walk walk‑ed
walk walk‑ing
big big‑ger big‑gest

This contrasts with Old English, which had a stem-based verbal system and a heterogeneous
nominal system (Kastovsky 2006: 245–6). The status of German is controversial. While Old
High German is regarded unanimously as stem-based, the modern language has been ana-
lysed as stem-based or word-based by different authors and across verbal, nominal, and
adjectival domains (see Harnisch 2001 for discussion). French also invites somewhat incon-
clusive analyses of this parameter. Its nominal and adjectival inflection are word-based, as
shown in (2a) (with the plural /s/ showing up in liaison contexts and in the orthography),
620 English morphology in a typological perspective

but verbal inflection seems unclear in its status, as what can be analysed at the phonological
level as a stem, as in (2b) can also appear as a free form, as illustrated in (2c). It is the orthog-
raphy that still reflects the earlier stem-based nature of the system.

(2) a. French gloss

enfant ‘child’
enfant-s ‘children’
petit ‘small (sg)’
petit-s ‘small (pl)’
b. /ɛm/ ‘love (stem)’
aim-er ‘love (infinitive)’
aim-e ‘love (1sg/3sg/imperative)’
aimes ‘love (2sg)’
aim-ons ‘love (1pl)’
c. Tu /ɛm/ tes enfants.
‘You (sg.) love your children’

Overall, it nevertheless seems to be the case that of the three languages, English is most
clearly word-based.
The English umlaut plurals (e.g. feet, geese, lice, men, mice, teeth) are a Germanic heritage.
The system of umlaut is still much more elaborate in modern German than in modern
English, and is widely found supporting affixes, as well as being the only marker of plural
(compare English brother ◆ brethr‑en). The English system shows only the remnants of umlaut,
most umlaut forms having become regularized (see Chapter 7.2.4). Part of the result of this
regularization is that there are no declension classes in English nouns, just a division between
regular and irregular forms (Kastovsky 2000: 220–1).
The only case marking that English retains is in the genitive, and as was shown in Chapter 7
this has moved a long way from the genitive marking in English’s Germanic neighbours. The
periphrastic genitive has parallels in French, and in many of the modern Germanic languages,
where they are, however, relatively recent innovations. The English third-person singular non-
past tense ‑s is typologically extremely odd. It is odd from a universalist point of view in being
the only marking in the present tense paradigm, and yet being on the third-person singular,
usually the person which is least likely to be marked. Even within the system of English it
is odd, since it means, as pointed out by Kastovsky (2000: 217), that—except in the verb
be—person and number are marked only in the non-past, where tense is not marked, while
tense is marked only in the past, where person is not marked.
The lack of a morphological future in English is another Germanic feature, and specifically
not Romance. The Germanic languages can all mark future time by using the non-past form
of the verb, or by the use of auxiliaries.
 28.3 A finer-grained view 621

Ablaut marking in preterites and past participles (often accompanied by suffixation) is again
a Germanic phenomenon. The system seems to have been predictable in the earliest stages of
Germanic (Mailhammer 2007), but is unpredictable in all the modern Germanic languages. All
the languages have lost verbs of this type to regular paradigms. This has certainly been the case
in English, although, it is not always the case: as we have shown in Chapter 5.3.5.2 certain pat-
terns continue to attract new members through the workings of analogy. English no longer has
the circumfixes that characterize the other West Germanic past participles, although in earlier
stages of English the past participle was frequently circumfixed (for example, Middle English
ydrunken, yhoped ‘drunk’, ‘hoped’ and even Early Modern English yclept ‘called’).
English adjectival inflection shows only marking for comparative and superlative, and
not gender (as French and German), number (as French and German), or definiteness
(as German). Overall, English inflectional morphology is fundamentally Germanic in
nature, albeit much simplified from its West Germanic origins. It has lost declension and
conjugation classes, and most of the Germanic case marking (there are a few remnants in
the morphology of pronouns). What is perhaps odd typologically, however, is that where
English inflection shows variation between analytic and synthetic forms (that is, in the
genitive or the comparative and superlative, as illustrated in Chapters 6 and 7), the distri-
bution of the analytic and synthetic forms is probabilistic rather than categorical in
nature.

28.3.2 Derivation
Just like English inflection, English derivation is very largely word-based, see (3), where we
find obligatorily bound roots, they are virtually always borrowed. This is illustrated in (4),
with unkempt being a form that is based on a native bound root.

(3) arriv-al, command-er, em-power, hope-less, hospital-ize, parent-al, pre-arrange, prefer-

ment, short-ish

(4) auth-or, bapt-ism, ed-ible, electr-ify, in-ept, leg-al, psych-ology, regul-ar, scient-ist,
un-kempt

There are a few instances of word-formation by internal modification left in the lexicon: abide
◆ abode, believe ◆ belief, ímport ◆ impórt, rise ◆ raise, sing ◆ song, but only those with stress shift
show any signs of being productive. Word-formation involving vowel change is far more
widespread in German (as, for example, in the noun–verb pairs Bruch ‘break’ ◆ brechen,
Schnitt ‘cut’ ◆ schneiden, Stand ‘stand’ ◆ stehen, etc., see Fleischer and Barz 2007: 218), but not
productive there, either.
Both German and French show both prefixation and suffixation (there is a minimal
amount of circumfixation in German and Dutch word-formation), and the English pattern of
622 English morphology in a typological perspective

prefixes looking rather like prepositions is common to all three languages. Both French and
German are right-headed in derivation, so the right-headedness of English is unsurprising.
One surprising thing, though, is that we find a few cases of left-headedness in English (for
example, de‑ and a few other prefixes discussed in Chapter 29.3.1), which are not supported
by models from these other languages. Various analyses that attempt to avoid the left-
headedness of such formations (e.g. Kastovsky 1999; Nagano 2011) are partly motivated by
the awkwardness of marginal left-headedness.
It might be interesting to see how English compares with other languages in terms of the
semantic categories of derivation that it exhibits. We expect that all languages will have
means for changing items from one lexical category into another (that is, noun to verb, verb
to adjective, etc.), as English certainly does. But a broader comparison of the derivational
categories that are available cross-linguistically (for example, agentive or instrumental
nouns, causative or inchoative verbs, ranges of evaluatives, and the like) is not yet possible.
The problem here is that there is as yet no coherent cross-linguistic inventory of derivational
categories that might serve to test English against. Beard’s (1995) attempt to list possible
derivational categories is highly dependent on the range of inflectional categories in Indo-
European, and wider typological surveys are not as yet available (but see, for example,
Štekauer et al. 2012). The notion that there might be a universal paradigm of derivational
categories to gauge English against is an intriguing one, but one that will have to await fur-
ther research.
This said, there are still certain observations that we may make. While English is rich in the
means it has to derive agents and instruments, we have observed that it is rather poor in the
means of deriving patient nouns, especially inanimate ones. As we saw in Chapter 11, we fre-
quently find affixes whose primary function lies elsewhere being stretched to produce patient
nouns when they are needed. And although it is sometimes said that English is poor in evalu-
ative morphology, at least by comparison to German, Dutch, Spanish, or Italian, we have
found to the contrary (see Chapter 18) that contemporary English has many ways of forming
diminutives and augmentatives.

28.3.3 Conversion
Conversion, as the notion is used in reference to English, is not a universal. In the first
instance it has to be distinguished from multifunctionality of the type reported for some
Polynesian languages, where virtually any word can be used in any function. Further, the
term conversion can be used of rather different types of process in English than can be
observed in French and German.
In French, for example, an inflected form can function as a word in a word class not sup-
ported by the inflection. This is most obvious with infinitives (which are, of course, very
nominal-like parts of verbs), so that we find, for example, pouvoir ‘to be able’ ◆ pouvoir ‘power’.
In French such usage tends to be sporadic, but in German it is systematic, and the infinitive
 28.3 A finer-grained view 623

of any verb can function as a neuter noun: das Bellen ‘barking’, das Essen ‘eating, food’, das
Rückwärtsfahren ‘reversing’ (Hammer 1983: 263). This is rather similar in some ways to the
nominal uses of ‑ing forms in English.
Typically, in German, though, conversion of verb to noun is based on the stem rather than
on the inflected verb, so that we get forms like laufen ‘to run’ ◆ der Lauf ‘the run’. In such
instances, no further morphology is required for the nominative singular form of the noun,
though further morphology is required for plural forms, for instance. In noun to verb conver-
sion, on the other hand, extra morphology is automatically required—in German or in
French. Thus, for the English noun–verb pair telephone ◆ to telephone we get French le télé‑
phone ◆ téléphoner, German das Telefon ◆ telefonieren (see Manova and Dressler 2005 for
further discussion). The result is that the paucity of inflectional morphology in English
appears to make the process of conversion marginally easier to manage. How significant a
factor this is in the growth of conversion in English is hard to determine.
Conversion is an all-pervasive word-formation process in English. However, from the
point of view of Natural Morphology (for example, Manova and Dressler 2005), conver-
sion ought to be relatively infrequent, since it is not structurally iconic: ‘more meaning’ is
not reflected by ‘more form’. It is also less natural than affixation in a number of ways,
including morpho-semantic transparency. Manova and Dressler (2005) admit this, and
admit that English shows greater productivity for conversion than would be expected.
They suggest that the pattern of conversion was established by the erosion of endings, and
that the preference for conversion arises from a language-specific pattern. This seems to
imply that there may be some correlation not just between lack of inflection and conver-
sion but specifically between loss of inflection and conversion. Any such correlation
remains to be established.

28.3.4 Compounds
It is not clear whether compounding exists in all languages (see Bauer 2009b), but it is cer-
tainly an extremely widespread form of word-formation, so that the existence of compounds
in English, in other Germanic languages, and in Romance is not surprising. However, the
types of compound that are widespread in Germanic and Romance are different, and English
appears to have been influenced to some extent by both sides. There is also influence from
Greek in neoclassical word-formation.

28.3.4.1 From Germanic

Compounding is an active means of word-formation in all Germanic languages, but the rest
of Germanic tends to have a far more predictable template for compounds than English does,
including lefthand stress and relatively stable orthographic representation. Germanic lan-
guages have a compounding pattern where uninflected words are put side-by-side: these are
sometimes referred to as echte or eigentliche ‘genuine’ compounds in the German literature
624 English morphology in a typological perspective

(Grimm 1826; Fleischer and Bartz 2007: 136). There are also patterns with linking elements,
mostly derived from genitive or plural markers from earlier stages of the languages. English
seems to have abandoned linking elements in compounds, except possibly in a very few sets
such as sportsman, swordsman, yachtsman. The status of the ‑s in compounds of this sort is
somewhat controversial, however; although these compounds might be analysed as having a
linking element, other forms (for example, arms trade, drinks cabinet, perennials catalogue, and
new tastes menu) appear to have an ‑s that is genuinely interpretable as a plural.
In what follows, we comment on correspondences between English patterns and patterns
in other Germanic languages. In some cases, the features discussed here are common to
compounding in many languages, and in some cases there seems to have been parallel devel-
opment across Germanic.
Binary-branching is one of those features that may be a much broader phenomenon.
Generally we do find binary-branching in compounding in English, as well as in other kinds
of word-formation, but it is also the rule in most other languages that have compounding.
Some linguists might even argue that it is a language universal. However, in English, as in
German, there are exceptions to the binary-branching rules in coordinate compounds, espe-
cially, but not exclusively, the type illustrated by singer-songwriter-recording artist. A similar
pattern exists in French, with le bleu-blanc-rouge (’the blue-white-red’, referring to the French
flag), but is not widespread there.
Connected with binary-branching is the question of recursion. In English, as in other
Germanic languages, compounds have the ability to be embedded in other compounds—a
feature so common in German and Danish that it is sometimes mocked by speakers of those
languages, with examples like German Donau-dampf-schiff-fahrts-gesellschafts-unter-offizier
‘Danube steamship line under officer’ or Danish spor-vogns-skinne-skidt-skraber-­fag-forenings-
magasins-beklædnings-forvalter ‘street car rail dirt scraper trade union store­house clothing
director’. English examples such as Auckland University Law Library shelving policy show simi-
lar trends. This is not a given, if we look on a wider scale. There are several languages, such as
Fongbe (Lefebvre and Brousseau 2002), Ngiti (Lojenga 1994), and Slovak (Pavol Štekauer,
p.c.) where compounds are either not recursive or show very limited recursiveness.
As we have shown in Chapter 20 the range of semantic relationships possible between
the first and second elements of English compounds is virtually unlimited, when all com-
pound types are taken into account. This is especially true with regard to attributive com-
pounds: although attempts have been made at various times to catalogue a fixed set of
semantic relationships, these have invariably proven unsuccessful (see Chapter 20.3.1).
Other Germanic languages seem to have equally large sets of relationships. Yet this, again, is
not something that can be presupposed. Hup is said (Epps 2008: 216) to encode just three
semantic relations in compounds. According to Heath (2008: 191) many compounds in
Jamsay are ‘equivalent to a possessive construction’—though see Bauer and Tarasova (2010)
for an alternative interpretation of such observations. Biber and Gray (2011) also claim that
 28.3 A finer-grained view 625

the number of relationships in English has increased since the seventeenth century as the
number of different classes of noun used as modifiers has increased. The examples they cite
from their earlier period are almost all locatives of some kind (particularly if the modifier in
a form like linen handkerchief is viewed as a source rather than as a material).
Historically, Germanic compounds are characterized by a specific, that is leftward, stress
pattern, and the fact that a language shows a specific stress pattern in its compounds is not
unusual, see also Finnish, Hup (Epps 2008: 217), or Lango (Noonan 1992: 115). In English the
inherited and rather strict compound-specific stress pattern has given way, however, to a
complex and highly variable pattern of leftward and rightward stresses today (see Chapter
19.3.3). Other Germanic languages, for example German, also show some limited degree of
stress variation, but English is certainly most advanced in this respect among its relatives.
It may be that English also inherits from Germanic the tendency for words formed by com-
pounding to be nouns. However, there are more compound verbs in other Germanic languages
than there are in English, where formation of canonical compound verbs (as opposed to the crea-
tion of things that look like compound verbs by back-formation or conversion) seems to be very
new (Kastovsky 2000: 216). English shares with other Germanic languages a tendency for coordi-
nate compounding to be a productive type, although less prominent than attributive compound-
ing, for example (Olsen 2001; Bauer 2010b). The low use of coordinate compounds is probably an
areal feature (Wälchli 2005), though, that stretches beyond Germanic and Romance.

28.3.4.2 From Romance

Despite a somewhat obscure history (see Kastovsky 2009: 336–7), the kind of compound
illustrated by cut-throat, pickpocket, scarecrow in English is widely considered to be supported
in English by parallel types in Romance, and specifically French. Certainly the type is far
more productive in French than it is in contemporary English. Though historically there was
some marginal productivity of the type in English, the type is likely no longer productive in
the contemporary language.
Superficially the prevalence of this type in Romance and its relative absence in English
might be said to show that English and Romance are typologically distinct. But this claim of
distinctness holds only if the cut-throat type compound is seen as an alternative in Romance
to the sort of synthetic compound that is typical in Germanic. An alternative analysis counts
the cut-throat type as the exocentric equivalent to blow torch, call girl, driftwood, hangman,
mincemeat, pay day, plaything, searchlight, washhouse, whetstone, and so on, where there is a
verb in the first element and a head noun which stands in various grammatical and/or seman-
tic relationships to that verb. Indeed, we have analysed the cutthroat type in this way as exo-
centric argumental compounds in Chapter 20.2.3 (see also Lieber 2010).
More interestingly, perhaps, French might be seen as the origin of structures like canine
tooth, domestic goddess, equine anemia, human genome, equinoctial gale, melodic minor, quantita‑
tive analysis, spiritual dimension, and the like that compete with compounds in English; indeed
we raised the question in Chapter 19 whether these were to be treated as compounds (see
626 English morphology in a typological perspective

Chapter 19.2.1.2 for discussion). Since French traditionally avoids nominal modification of
nouns but creates derivational adjectives from nouns, it provides a model for such structures,
while Germanic frequently uses nouns to modify other nouns, thus creating compounds. By
inheriting both systems, English ends up with a mixed system. Having a system that allows
both kinds of modification, apparently synonymously, English has the possibility of two
routes to express the same fundamental idea. In some cases, as in (5), such pairs differ stylisti-
cally, and in others, as in (6), we find semantic distinctions.

(5) atom bomb/atomic bomb, climate change/climatic change, dogtooth/canine tooth, lan-
guage development/linguistic development, noun compound/nominal compound, sex
therapy/sexual therapy, town planning/urban planning, weather map/meteorological
map, woman doctor/female doctor

(6) brute strength/brutal strength, country house/rural house/bucolic house, demon lover/
demonic lover, folk music/popular music, health farm/healthy farm, nature strip/natural
strip, rhythm section/rhythmical section

Interestingly, it seems to be the case that derived adjectives are not historically parts of
adjective–noun compounds (Bauer 2009c: 403–4 and sources there). A few foreign adjec-
tives with foreign affixes are now found in Danish and German in such compounds (Danish
centralvarme, German Zentralheizung ‘central heating’), but they are still relatively rare. It
seems that for naming, English still prefers noun–noun compounds with Germanic and early
Romance vocabulary and adjective–noun phrasal constructions with later Romance vocabu-
lary unless the Romance vocabulary is very recent or is employed mainly or exclusively in
scientific domains (see Schlücker and Plag 2011 for the treatment of phrases versus com-
pounds in naming in German). Thus we get eyebrow and not *ocular brow, table top not *tabu‑
lar top but descriptive passage not *description passage and parliamentary language not
*parliament language. There is enough lexicalization in the system to confuse matters from
time to time (e.g. popular no longer means simply ‘of the people’, infantile is disparaging so
that we do not get *infantile school), and there are also some examples which go against this
general trend (such as pole star rather than polar star). Overall, however, there seems to be an
effect of history here which affects the typological appearance of the language. We do not
wish to develop this idea here, since it overlaps very largely with the areas of syntax and lexi-
cography or lexicology, and certainly goes beyond morphology, but we point to it as being a
factor in the structure of current English.
Another set of constructions that come into English from French and are sometimes clas-
sified as compounds, is the set of constructions made up of a noun and a postposed adjective,
like attorney general, blood royal, court martial, governor general, lie direct, notary public, princess
royal. While these persist as lexical items, there is some evidence that they are no longer
perceived as being left-headed (as they are in French), as the plural marker is more and more
added to the righthand word (where relevant). Here, the English typological pattern appears
 28.4 Conclusion 627

to have over-ruled the French one, but this has happened as these items have moved from
being syntactically created expressions in the original French to fixed lexical expressions in
current English.

28.3.4.3 From Greek

English of course also has a large and to some extent productive cohort of neoclassical com-
pounds deriving for the most part from Greek. We will not reiterate the characteristics of
these compounds here. The reader is referred to Chapter 19.4.5 for detailed discussion of
them, as well as comparison of their analysis in contrast to Greek.

28.4 Conclusion
What we have suggested here is that English morphology is not homogeneous from a typo-
logical point of view, and that some, at least, of that heterogeneity derives from the mixed
history of English vocabulary. A typological approach of the kind that has been taken here
does not necessarily tell us a great deal about the operation of English morphology, but does
raise questions about processes of diachronic development and the way they interact with
typological tendencies.
 chapter 29

English morphology and

theories of morphology

29.1 Prospectus
We did not write this book with the primary aim of making a contribution to linguistic theory,
but rather with the intention of giving as theory-neutral and as comprehensive as possible a
description of the contemporary state of English inflection and word-formation. Our strategy
of relying on data from corpora has given us substantial insight into the workings of English
morphology though, and we have tried in Part VI to show how those data speak to a number
of recurrent themes in morphological theory. In this last chapter we take a broader view of
what the English data tell us about morphological theory in general. In Section 29.2 we con-
sider a range of theoretical models, concluding that the facts of English morphology do not
allow us to argue strongly for or against most extant frameworks. Section 29.3 looks at a vari-
ety of theoretical principles that are relatively independent of broader frameworks; here we
argue that the corpus data do provide strong arguments against many of the specific claims
that have been made in the literature. In Section 29.4 we turn to what we do know, and how
that knowledge should help us to frame new theoretical models.

29.2 Broad theoretical models

Complex and intricate as it is, English morphology vastly underdetermines several major points
in morphological theory. What we mean by this is that the facts of English inflection and word-
formation are compatible with many different frameworks that have been proposed. This is not
to say that all theoretical models are equal with respect to accounting for all aspects of English
morphology, but just that no single existing model seems to make substantially better predic-
tions about English morphology as a whole than any other. We cannot of course look in detail
here at every possible theory of inflection and word-formation that has been proposed over the
last six or so decades, but we will consider some of the major proposals. Following Stump’s
(2001) taxonomy of linguistic theories we start with two major issues, the status of the
 29.2 Broad theoretical models 629

morpheme as a theoretical construct (Section 29.2.1) and the nature of the mechanisms by
which morphosyntactic features are mapped onto phonological forms (Section 29.2.2). Finally,
we turn to a totally different set of theories, analogical and constructional models.

29.2.1 Are there morphemes? IA, IP, and WP

Perhaps the single most important point at the heart of any morphological theory is the treatment
of the morpheme. Starting with the American structuralist tradition and continuing with much of
the generative literature from the 1970s on, scholars have been engaged in lively debate as to
whether the morpheme is a legitimate unit of analysis, or whether it should be dispensed with
entirely as a theoretical construct. On the one hand we have theories in which morphemes are
taken to be the basic units of analysis that are put together into hierarchically organized structures,
much as syntactic rules generate hierarchically organized structures from words. In the American
structuralist tradition, such theories were dubbed Item and Arrangement (IA) models (Hockett
1954, see also Harris 1942; Bloch 1947; Nida 1948). This sort of model continues in the generative
tradition in the work of Lieber (1980, 1992), Selkirk (1982), Williams (1981), and Di Sciullo and
Williams (1987), and in Distributed Morphology in such works as Halle and Marantz (1993) and
Harley and Noyer (1999). On the other hand we have theories that generally eschew the mor-
pheme as a unit of analysis and cast morphology as a set of processes. Two sorts of theory belong
to this camp, Item and Process (IP) theories (Hockett 1954; Aronoff 1976 in the generative tradi-
tion, Booij 2010 in Construction Morphology) and Word and Paradigm (WP) theories (Robins
1959; Matthews 1972; Anderson 1992; Stump 2001). The differences between IP and WP models
will be taken up in the next section; for our purposes it is sufficient to say here that both IP and
WP frameworks eschew the morpheme as a theoretical construct. We note, however, that it is not
impossible to construct a theory that is basically processual in nature but that nevertheless coun-
tenances the morpheme as a theoretical construct. Chomsky and Halle (1968) is a candidate for
such a theory.
In analysing affixation in English (both inflectional and derivational), IA, IP, and WP
models are more similar to each other than might at first appear. Consider, for example, the
formal representations in (1), where (1a) uses an IP format, (1b) an IA format, and (1c) a WP
format:

(1) a. Word-formation Rule from Aronoff (1976: 63) (IP)

Rule of negative un#
[X]Adj → [un#[X]Adj]Adj
semantics (roughly) un#X = not X
b. Lexical representation from Lieber (1980: 65) (IA)
Negative in‑
(phonological representation)
630 English morphology and theories of morphology

semantic representation: negative

category/subcategorication [a __ [a
diacritics: Level I
c. Blevins (2003: 748) (WP)
Preterite, Perfect participle, and Passive participle
Operations Fd (X) = Xd
Preterite R ([past]) = Fd (X)
Perfect part R ([perf]) = Y: Fd (X)
Passive part R ([pass]) = Y

Although the notation and the nature of the operation involved in building or relating words differ
from one framework to another, in the end all three in some way relate a shorter form (call it a
base) with a longer form which is a base augmented by a string of phonemes (un‑, in‑, ‑d).
Other formal variations on these rule formats can be found as well. Haspelmath (2002: 50)
represents derivational rules as in (2), and Booij (2010: 8) as in (3):

(2) /X/V ↔ /Xər/N

‘dox’ ‘ a person who (habitually) doesx’

(3) ωi ↔ Ni ↔ [one who PREDj]i

| | \
[ ]j[ər]k Vj Affk

Again, the notation differs—Booij’s theory separates phonological, morphological, and

semantic information into three parts, linked by the double-headed arrow, Haspelmath’s
notation collapses the phonological and morphological information and seems to lack
internal bracketing in the affixed word, but the end result is not all that different. Any one
of these approaches can provide an account of the basic facts of English affixational
morphology.
Given that affixation per se presents no formal problems for any of these frameworks,
arguments over theoretical frameworks have come to hinge on types of morphological phe-
nomena that English lacks (root and pattern morphology, consonant mutation) or is very
poor in (infixation, reduplication, ablaut, and other sorts of internal modification). However,
with clipping (see Chapter 18) and blend formation (see Chapter 19), even English has some
non-concatenative morphology that poses a challenge to certain theories. In particular, the
argument is made that non-concatenative morphology is much less suited to formal treat-
ment in IA models. The question is, how much of the morphology of a given language must
 29.2 Broad theoretical models 631

be non-concatenative to discredit IA models. For English morphology, concatenation is

certainly more characteristic than non-concatenation.
To illustrate the potential inadequacy of IA approaches to non-concatenative morphology,
consider the ablaut that occurs in the irregular verbal and nominal paradigms (see also
Chapters 5 and 7). Ablaut in English at first glance does seem to provide prima facie evidence
against an IA model. To defend an IA analysis, Nida (1948), for example, is forced to postu-
late what he calls a ‘replacive morpheme’ represented as eɪ → ʊ in order to derive the past
tense took from the present tense take, an analysis which is not terribly attractive; the sequence
eɪ → ʊ certainly looks more like a rule or process than a morpheme. Blevins (2003: 246–7),
in defending a WP analysis of West Germanic verbal inflection, concedes that in English
many irregular verbs may be treated as listed patterns, an analytical option that does not
favour one framework over another. There are of course a few ablaut patterns in English that
still spread (see the discussion in Chapter 5), but these constitute at best weak evidence
against IA. It is, however, much less clear how the subtractive patterns we find in hypocoris-
tics, clippings, and blends can be accommodated in an IA approach.

29.2.2 Realizational versus non-realizational models

English does not seem to provide good evidence one way or the other in the debate between
realizational and non-realizational theories of morphology. Stump (2001: 2) distinguishes
between theories that are non-realizational or ‘incremental’ as he terms them and realiza-
tional theories. Non-realizational theories allow bases to be inherently specified for morpho-
syntactic features and morphemes or morphological rules to add morphosyntactic features
to bases. Realizational theories, on the other hand, conceive of word-forms as being inher-
ently devoid of morphosyntactic features; features and forms are associated to each other by
mapping procedures of various sorts where the mapping between features and word-forms
need not be (and typically is not) one to one. In the former sort of theory, ‘words acquire
morphosyntactic properties only as a concomitant of acquiring the inflectional exponents of
those properties’, and in the latter ‘a word’s association with a particular set of morphosyntac-
tic properties licenses the introduction of those properties’ inflectional exponents . . . ’ (Stump
2001: 2) In the literature (Anderson 1992; Stump 2001) realizational theories are almost
always supported by appeal to complex inflectional systems, something which English, of
course, lacks.
Realizational theories are often said to be superior to non-realizational ones in two
respects: they allow more elegant analyses of the phenomena of multiple exponence and of
syncretism. In multiple exponence, a morphosyntactic feature in some language may be
expressed by more than one form in a word. With syncretism, the same affix or base may
be associated with more than one set of morphosyntactic features. Both cases involve map-
pings between form and meaning that are not one-to-one. It is argued that non-realizational
632 English morphology and theories of morphology

models are well suited to account for morphological phenomena in which the relationship
between form and morphosyntactic features is one-to one, but ill-suited to deal with other
sorts of mappings.
English inflection is too impoverished to exhibit much, if any, multiple exponence, but it
does display at least one prominent case of syncretism. Specifically, the regular past tense,
past participle, and passive participle forms are all marked with ‑ed. We argued in Chapter 5
that what some people call the passive participle and the past participle are not two separate
forms in English. Rather the passive is a construction that makes use of the past participle.
Blevins (2003: 746–7) makes a convincing argument that regular verbal inflection in West
Germanic languages is best treated in a realizational framework, as only that framework
allows reference to what he calls the dental stem form, a form ending in d/t (depending on
the language in question) which is inherently meaningless, but is associated with either the
preterite or the past participle or passive participle by realizational rules. But Blevins’s argu-
ment is most convincing for West Germanic languages other than English; he himself notes
that,

If one were to restrict attention to English, it is not obvious what would count as decisive
evidence against a morphemic account that was prepared to countenance the affix- and
word-level homophony entailed by a ‘one meaning-one affix’ principle. How precisely does
one establish that preterites and participles are marked by the same exponent, and not
merely by homophonous suffixes? (Blevins 2003: 246–7)

So although the facts of English are consistent with a realizational model, the real argument
for this model comes from West Germanic languages that have more in the way of verbal
inflections.
Although realizational theories are primarily thought of in relation to inflectional systems,
they need not be. Stump himself (2001: 252) very briefly considers the extent to which deri-
vation can be treated in a realizational framework, and Beard’s (1995) Lexeme Morpheme
Base Morphology might be said to give a realizational account of derivation and conversion:
instead of derivation being a process of adding affixal morphemes with determinate mean-
ings onto bases, for Beard there are two processes, first derivation in his terms, which involves
the association of a semantic or grammatical feature with a base, and second the association
of that base with a specific derivational exponent (which may be an affix or nothing, in the
case of conversion). The separation of the semantic part of the process from the formal part
is what makes it possible to construe the theory as realizational.
The problems with Beard’s hypothesis are well-known, and here, data from English deriva-
tion do provide some insight (see, for example, Booij 1996b; Plag 1999: ch. 8). Generally, it is
not clear that the semantic effects of derivation or conversion can be reduced to a small set of
semantic functions that are parallel to grammatical or morphosyntactic functions, as Beard
suggests. Further, the widespread tendency of affixes to express a number of different related
 29.2 Broad theoretical models 633

semantic functions is not easily accommodated; for example, in English the polysemy of the
‑er that forms person, instrument, location, and sometimes patient nouns (see Chapter 11) is
in principle treated no differently than the homophony of that affix with respect to the com-
parative ‑er: everything is homophony in such a theory and in effect, we would be forced to
postulate five or six different ‑er suffixes. Further, it is also not clear how realization rules
effect the association between a form that has undergone derivation in Beard’s sense and the
affix that expresses that derivation. To mention just one problem, unlike inflection, where
each morphosyntactic configuration is normally mapped onto a single form, in derivation a
single base marked with a specific semantic/grammatical feature need not be associated with
a unique form (take the nominalizations commission, commital, commitment from the verb
commit, for example). We can conclude that to the extent that a realizational model can be
extended to word-formation, the English data suggest a number of problematic points.

29.2.3 Analogical models and Construction Morphology

In Chapter 23 we saw that there are phenomena in English derivation, inflection, and com-
pounding that are hard to accommodate in any theory that is restricted to the concatenation
of formatives. It was also argued in that chapter that these phenomena can be accounted for
if one assumes some kind of analogical mechanism. The status of such a mechanism is, how-
ever, largely unclear, as is its relation to the theories and problems discussed in the previous
two subsections.
There are two main approaches within this family of models. One is computational, the
other theoretical. The theoretical approach has been developed mainly by Booij (e.g. 2010)
and is known under the name of Construction Morphology. This model posits form–
meaning pairs of various levels of granularity (called ‘schemas’ and ‘subschemas’), which
can express different kinds of generalizations across morphologically related items in the
lexicon. In this model the existence of local analogical formations is acknowledged, but is
not formally integrated into the model architecture itself, which is restricted to schemas and
subschemas. Nevertheless, local analogy is seen as something that can gradually turn into a
schema (Booij 2010: 88–93). Non-concatenative processes can also be formalized in a way
that is consistent with the overall architecture of the model. The flexibility of schemas and
subschemas gives Construction Morphology a conceptual and empirical advantage over
many other morphological models. However, in addition to schemas and subschemas, anal-
ogy is needed as an additional and quite separate mechanism in order to account for the
manifold isolated analogical formations that seem to be quite common in English (and
other languages).
This brings us to the other family of approaches, those in which analogy is the only
morphological mechanism available. In such models, the notion of analogy receives a formal
definition in terms of the kinds of features over which similarity is systematically computed
over sets of words. This process of similarity computation is implemented in computational
634 English morphology and theories of morphology

algorithms and can be used to predict, generate, or select pertinent complex words as
the output of a morphological category. Three well-known such algorithms are the
Tilburg Memory Based Learner (TiMBL, Daelemans and van den Bosch 2005; Daelemans
et al. 2007), Skousen’s Analogical Model of Language (AM, Skousen 1989, 1992; Skousen
et al. 2002, Skousen and Stanford 2007), and the Generalized Context Model (Nosofsky
1986, 1990). Albright and Hayes’ (2003) Minimal Generalization Learner may also be
subsumed under this label, as there seems to be no principled difference between the
‘miminal generalizations’ these authors call ‘rules’ on the one hand, and the analogically
derived sets of bases with the same feature values one gets with models that call them-
selves ‘analogical’.
Analogical algorithms create or select new forms on the basis of the similarity of a given
base with existing forms in the lexicon. Consequently, analogies are based on sets of words of
varying sizes with varying degrees of similarity, where a ‘set’ may consist of only a single
word, or may have dozens, hundreds, or thousands of words, depending on how many words
share a given set of features or feature constellations under comparison. The algorithm thus
capitalizes on the multiple relationships that words in the lexicon may have. Consider for
instance the many different classifications of irregular verbs discussed in Chapter 5. Each of
these classifications is based on certain kinds of similarities shared by smaller and larger sets
of pertinent verbs, and we argued that the theoretical or practical value of such classifications
is questionable. With an analogical algorithm, all kinds of similarities between a given verb
and the verbs in the lexicon can be used to predict its inflected forms.
In this way, analogical algorithms can model both rule-like behaviour (based on very large
analogical sets) and the rather idiosyncratic local analogies (in which the set of analogues
may be as small as only a single item). Furthermore, the often quite variable behaviour that
seems characteristic of large parts of the morphology of English emerges naturally through
the variable choices that become available by the different kinds of similarities that different
analogues or analogical sets may share with a given base. In sum, such algorithms seem theo-
retically well equipped to deal with the wide range of categorical and variable phenomena we
find in English morphology, including especially seemingly problematic phenomena such as
regular versus irregular inflection (Chapter 5), split base effects (Chapter 23), the emergence
of splinters, or other paradigmatic morphology. But how do they fare empirically?
It has to be stated that both Construction Morphology and analogical approaches to mor-
phology are currently research programmes rather than established theories. Booij’s
Construction Grammar has been developed mostly with case studies of Dutch morphology,
and the above-mentioned analogical algorithms have been applied so far only to a small sub-
set of phenomena of English morphology: verbal inflection (Skousen 1989; Derwing and
Skousen 1994; Albright and Hayes 2003; Keuleers 2008), the history of some negative pre-
fixes (Chapman and Skousen 2005), compound stress (Plag et al. 2007; Arndt-Lappe 2011;
Arndt-Lappe and Bell 2012), the choice between ‑ity and ‑ness (Arndt-Lappe 2014), and
adjectival comparison (Elzinga 2006). All of these studies have provided very promising
 29.3 Specific claims 635

results, some of them actually showing the superiority of analogical models to competing
non-analogical models. It remains to be shown how analogical algorithms can deal on a larger
scale with the wealth of intricate problems English morphology has in store.

29.3 Specific claims

While English morphology provides relatively little fodder for the big-picture differences
among frameworks that have dominated theoretical discussions for the last six or seven dec-
ades, it does provide clear evidence that bears on some of the more specific theoretical prin-
ciples that have been formulated at one time or another. It is of course not possible to cover
all theoretical proposals that have been made; we limit ourselves to a selection of the most
prominent ones here for which English data provide some insight.

29.3.1 The Righthand Head Rule

The Righthand Head Rule (Williams 1981) has been a remarkably influential proposal.
Assuming that the notion that words have heads is a coherent one (see Chapter 24),
Williams proposes that the rightmost morpheme in a word is its head. In effect this
predicts that suffixes should be word class changing but prefixes should not be. It is well-
known that while the Righthand Head Rule works for the bulk of English morphology,
it fails in a number of cases. Compounds are right-headed, if they are headed at all, and
inflection is suffixal (to the extent that inflections are said to have the properties of
heads), but derivation deviates from the Righthand Head Rule at least with respect to
prefixation. Specifically, English has several prefixes that change category: be‑, de‑, dis‑,
en‑, out‑, and un‑. Depending on how we analyse the locative prefixes that make nouns
into prenominal modifiers (e.g. post-war Europe, see Chapter 16), we might count them
among the category-changing prefixes as well. The majority of these affixes display at
least some degree of productivity, so it is not possible to dismiss them as relics of some
sort. It is of course possible to argue that the prefixes do not themselves change category
(Nagano 2011), but this argument is not terribly convincing.

29.3.2 The Unitary Base and Unitary Output Hypotheses

Aronoff (1976: 48) proposes the Unitary Base Hypothesis which states that ‘the syntacti-
cosemantic specification of the base, though it may be more or less complex, is always
unique’. In other words, Aronoff suggests that we should not find an affix that attaches to
both verbs and nouns, much less to all lexical categories. Aronoff dismisses potential
counterexamples like the suffix ‑able by arguing that ‑able is in fact two suffixes. The
636 English morphology and theories of morphology

argument is not a strong one, Aronoff fails to note cases like marriageable where the
deverbal and denominal ‑ables make the same semantic contribution to their base.
Positing natural classes such as [+ Noun] to cover cases where, for example, adjectives
and nouns can be selected as input to a word-formation rule does not help either. As
shown in Plag (2004: 201) no feature system is able to cope with the many different con-
stellations of different base categories one finds with the derivational affixes of English.
This point is further substantiated by the data presented in this volume. It would seem
that it is not at all exceptional for affixes to attach to bases of a number of different word
classes—indeed, it seems more to be the norm than the exception.
Scalise (1984: 137) proposes what he calls the Unitary Output Hypothesis which states
that a form cannot be analysed as ‘a single affix if it produces outputs with different category
labels or different semantics’. Under this hypothesis, we would be forced to treat an affix like
‑an in English as two separate affixes, one that produces nouns (magician) and another that
produces adjectives (suburban). The problem with this hypothesis is that many ‑an forms do
double duty as nouns and adjectives (Bolivian and other inhabitant names, amphibian, egali-
tarian, etc.). It would seem odd to maintain that the noun Bolivian and the adjective Bolivian
are derived by distinct affixes, and such data therefore suggest that the Unitary Output
Hypothesis is incorrect. The only alternative is to consider such cases as conversion, in which
case the hypothesis is unfalsifiable.

29.3.3 Blocking and the Elsewhere Condition

Blocking and the Elsewhere Condition are two theoretical principles that are proposed to
explain a purported avoidance of synonymy in the mental lexicon.
We have shown in some detail both in Chapter 26 and in the descriptive chapters of Part
III that blocking is at best a tendency and at worst a myth with respect to English morphol-
ogy. Synonymy in derived forms is not at all unusual. Corpus examples can be found through-
out this volume, for example, display ◆ displayal, disregard ◆ disregardance, disruption ◆
disrupture, omission ◆ omitment (in Chapter 10), educationalist ◆ educationist (in Chapter 11),
Barbie-ish ◆ Barbieesque ◆ Barbie-like (in Chapter 14). Indeed, we agree with Plag (2003: 63–8)
that rather than being a theoretical principle, blocking is at most an artefact of lexical retrieval:
the more frequent an established form is, the easier it is for speakers to access, and the less
likely they are to produce a synonymous derived form.
Blocking is a specific case of the Elsewhere Condition. The Elsewhere Condition (Kiparsky
1982a) has a long history in linguistics, dating back to the Sanskrit grammarian Panini.
Roughly, it ensures that when two rules can potentially apply to the same base, the more
specific rule preempts the more general one. We do not intend here to consider the merits of
the Elsewhere Condition with respect to phonological rules. It has, however, been invoked in
the analysis of inflection (see, for example, Xu and Aronoff 2011 for a recent treatment),
although here we again find that it cannot be taken as a hard and fast principle. We have seen
 29.3 Specific claims 637

much more variation in English inflection than the Elsewhere Condition would lead us to
expect; example (4), discussed below, is a fairly convincing case in point.

29.3.4 Lexical Phonology and Morphology: Level Ordering

and Bracket Erasure
Lexical Phonology and Morphology is a theoretical framework that had its heyday in the
early 1980s, and has since been largely discredited. Nevertheless here we mention two of the
important theoretical claims of the framework, and the sort of data from English that have
been used to call them into question.
Lexical Phonology and Morphology assumes first of all, that morphology may be strati-
fied into two or more levels, each of which is associated with a set of morphological rules and
an accompanying set of phonological rules. For English, Kiparsky (1982a), for example, pro-
poses three levels, roughly a level that contains derivational rules that involve stress shift and
allomorphy of base and affixes, a second level that contains derivational rules that do not
involve stress shift and allomorphy, and a level for inflection. Lexical Phonology and
Morphology therefore assumes that English word-formation is stratified. In Chapter 27 we
have shown in some detail that the notion of Level Ordering that is part of Lexical Phonology
and Morphology is incorrect for English, and in fact that English displays only the faintest
trace of stratification, so we will not repeat those arguments here.
Another prominent tenet of Lexical Phonology and Morphology is the Bracket Erasure
Convention, a notion that actually goes back at least as far as Chomsky and Halle (1968). In its
most prominent expression Kiparsky (1982d: 140) states that at the end of each level of the
phonology and morphology all word-internal structure is eliminated. In effect, Bracket Erasure
obliterates the derivational history of a form. Kiparsky argues that this principle allows us to
explain why, when a noun is derived from a strong verb, and then undergoes conversion back
to a verb, the result is a regular verb (the familiar example is that of grandstand, whose past
tense is grandstanded rather than grandstood). The reasoning goes that if internal brackets are
erased, it will not be possible to ‘see’ that the verbal base is strong, and the newly derived verb
will of necessity be inflected regularly. The problem is that such internal information is not
always eliminated. For example, although the verb to fly out (‘to hit a fly ball’) is usually given
a regular past tense (flied out) as predicted by the Bracket Erasure Convention, it is neverthe-
less possible to find the strong form, as the following example from COCA suggests.

(4) Chicago Sun-Times 1999: Mark Grace, with the bases loaded and two out in the ninth, flew
out to right field and furthered the theory he has lifted many more beer bottles than
barbells.

Clearly if such forms are possible, knowledge of derivational history cannot be completely
obliterated.
638 English morphology and theories of morphology

Comparative forms such as harder-line or lower-key are another problem for level order-
ing, as discussed in Chapter 6. If, as proposed in stratal approaches, compound-formation
is ordered at a later derivational stage than affixation, the insertion of an affix into a com-
pound is predicted to be impossible. Furthermore, formations such as bug-friendlier or
sugar-plummier constitute further instances of level ordering paradoxes, as the semantic
scope of the comparative suffix is the whole compound, while comparative suffixation to
such long polysyllabic adjectives should be impossible for prosodic reasons.

29.3.5 The Monosuffix Constraint

Aronoff and Fuhrhop (2002: 473) state what they call the Monosuffix Constraint for English,
which says that ‘Suffixes that select Germanic bases select unsuffixed bases’. This says in effect
that we should never find a native suffix attached outside another native suffix. The sole
exception they allow is the suffix ‑ness. However, as demonstrated by Plag (2002), Hay and
Plag (2004: 592-3) and in Chapter 27 of this book, it is not unusual to find native suffixes
attached outside native suffixes; Table 27.6 shows many such combinations.

29.3.6 The First Sister Principle and related proposals

Proposed by Roeper and Siegel (1978: 208) in an influential analysis of English synthetic com-
pounding in early generative morphology, the First Sister Principle states that ‘All verbal com-
pounds are formed by the incorporation of a word in first sister position of the verb’. The first
sister of the verb is generally the verb’s object, sometimes what would be considered an adjunct
in contemporary linguistic theory, but never a subject. Compounds like girl-swimming would
therefore be ruled out. Selkirk’s First Order Projection Condition (1982: 37) and Lieber’s
(1983) Argument-Linking Principle differ in detail but have similar effects in that both rule out
compounds in which the first element has a subject-referencing interpretation.
We have shown extensively in Chapter 20, however, that it is not at all unusual to find
English argumental compounds in which either the first or the second element is subject-
referencing. Such readings are sometimes, but not always, context-dependent, but they can
and do occur with some frequency. It must be concluded that the First Sister Principle and
related proposals cannot be correct.

29.3.7 The Lexical Integrity Hypothesis and related proposals

This hypothesis has been framed in a number of ways over the years: the Generalized
Lexicalist Hypothesis (Lapointe 1980), the Word Structure Autonomy condition (Selkirk
1982), the Atomicity Thesis (Di Sciullo and Williams 1987) (see Lieber and Scalise 2006 for
an overview). Bresnan and Mchombo (1995) give this definition:
 29.4 Implications 639

A fundamental generalization that morphologists have traditionally maintained is the lexical

integrity principle, which states that words are built out of different structural elements and by
different principles of composition than syntactic phrases. Specifically the morphological
constituents of words are lexical and sublexical categories—stems and affixes—while the syn-
tactic constituents of phrases have words as the minimal unanalyzable units; and syntactic
ordering principles do not apply to morphemic structures. (Bresnan and Mchombo 1995: 181)

As we have shown extensively in the descriptive chapters of Part III, as well as in the general
discussion in Chapter 22, the Lexical Integrity Hypothesis (at least in this version) cannot be
said to hold of English. English indeed provides two sorts of evidence against the strongest
forms of the Lexical Integrity Hypothesis.
First, we have shown that many productive processes of affixation and compounding
incorporate phrases, and not just lexicalized phrases. However we are to conceive the rela-
tionship between syntax and morphology, there can be no strict prohibition on interaction,
as proposed under, for example, Botha’s (1983) No-Phrase Constraint. The term itself is a
counterexample to its own claim, as has frequently been noticed. Indeed, as we have shown
in Chapter 22 the ability to attach to phrases goes hand in hand with high productivity.
One of the consequences of the Lexical Integrity Hypothesis is the prediction that it
should be impossible to use a pronoun or reflexive to refer either to the base of a derived
word or to the first element of a compound, a phenomenon generally referred to as Anaphoric
Islandhood in the literature (Postal 1969). It is well-known, however, that many English
speakers allow sublexical reference, and it is not difficult to find evidence for this claim in the
corpora, as the example in (5) illustrates:

USAToday 1993: The story of Jessie Lee (a highly Clintonian figure–as in Eastwood, not
(5) 
Bill–played with steely-eyed grit by Van Peebles) and his wild bunch may be the stuff of
pulp fiction.

What we see in example (5) is that the base of Clintonian is available for reference, but sur-
prisingly that the complex word is subject to two separate parses: Clinton + ian for Bill Clinton
and Clint + onian for Clint Eastwood. We do not take a stand on whether this phenomenon
is to be treated as a matter of syntax (Lieber 1992) or pragmatics (Sproat and Ward 1987); we
merely point out that cases of sublexical reference are not difficult to come by, and as such
provide more reason to doubt the Lexical Integrity Hypothesis, at least in any strong form.

29.4 Implications
Thus far, the contents of this chapter might seem disappointing. The facts of English inflec-
tion and word-formation do not argue strongly for any particular (current) theoretical frame-
work and are largely compatible with many frameworks. To the extent that English data bear
640 English morphology and theories of morphology

clearly on specific theoretical proposals, in every case we find that English provides strong
counter-exemplification.
Still it would be a mistake to conclude that the vast amount we have learned about English
morphology is good only for casting doubt on all extant morphological theory, and disas-
trous to assume that the study of English morphology has little to contribute to theory-con-
struction. While the conclusions we have drawn so far are largely negative, the authors of this
book are not merely hunters and gatherers, but also avid theorists; although we have tried to
be as theoretically neutral as possible in describing the phenomena of English, that does not
mean that we do not see (at least some of) the implications of those data for theory. Our
views on morphological theory have inevitably come to be shaped by the knowledge of
English we have gained in writing this book. We do not intend to provide an over-arching
theory of morphology here (indeed, we doubt that we would ever reach complete agreement
on a theoretical framework!). Rather, based on the detailed study of English morphology we
have undertaken, we make a number of points that we take to be foundational for any mor-
phological theory.
The following are claims that we take to be uncontroversial, but nevertheless must form
critical axioms of a theory of morphology:

(6) a. Morphology is highly computational, but not necessarily computational in the

same way as syntax. That is, a theory of syntax must account for both linear order
and hierarchical structure, and within most contemporary theoretical frameworks,
must allow for some mechanisms (e.g. rules of movement) that account for
discontinuous dependencies. Rules of morphology must minimally allow for the
generation of an infinite number of new words and therefore must also allow for
extensive computation. This must surely involve something that results in synthesis
(making longer forms from shorter ones) and possibly also in analysis (in the sense
of analysing longer forms into shorter ones). However, there is no convincing
evidence that the theoretical correlates of movement rules are needed for
morphology.
b. Morphological computation can involve recursion, but in a more limited way than in
syntax. It is an empirical matter whether all languages have recursive morphology, but
it is clear that at least some (among them English) do, and morphological theory
must allow for that possibility.
c. Synonymy is rampant in word-formation, and our theory must be able to account for
it. We must build theories that do not rule out the generation of potentially
synonymous forms.
d. 
The mental lexicon is subject to paradigmatic pressure which may result in
morphological processes being stretched beyond their usual boundaries. Looked at
as an overall system, the morphology of a language makes certain formal means
available for inflection and word-formation. Especially with respect to word-
formation, if formal means of creating a particular sort of word is lacking, formal
means usually used for other purposes can be recruited to cover what’s lacking.
 29.4 Implications 641

e. Models of inflection must be able to deal with variant morphological and periphrastic
expression of morphosyntactic categories. Moreover, such models need an
architecture that can accommodate the probabilistic distribution of such competing
inflectional variants, instead of the apparently more common inflection-class-based
or grammatically-governed distribution.

Other assumptions that we make might be more controversial:

(7) a. Morphological rules, however they are formalized, cannot be hard and fast, but rather
must in some way be violable. Put in a slightly different fashion, word-formation
allows for a particular sort of coercion. This point follows from (6d): if word-
formation rules that typically derive one sort of word can nevertheless be deployed to
form another sort of word, our theory must have some way of allowing this to
happen.
b. Morphology is not syntax or phonology. We must resist the urge to build a theory
of morphology on the analogy to models that are created for the phenomena
of syntax or phonology. Although morphology interfaces in complex ways
with syntax and phonology, it cannot be subsumed under either syntax or
phonology.
c. Productivity is uncontroversially a gradient matter, but even minimally productive
processes of word-formation may give rise to new forms. We should not be too quick
to declare that a process of word-formation is dead. On the other hand, we should
probably allow for the possibility that new forms can be influenced by the existence
of specific, already established forms, for example, house husband on the basis of
housewife, or fruitarian on the basis of vegetarian.
d. Whatever our theoretical model, we must have some way of accounting for not only
the formal side of morphology, but also the semantic side. Specifically we must be
able to account for the substantial evidence that affixes (or morphological processes,
if the theorist prefers) are frequently semantically underspecified, and subject to
polysemy and meaning extensions of various sorts.

Finally, there is one point that we feel the need to emphasize, even though it does not directly
arise out of our study of English.

(8) Storage and computation are not mutually exclusive. That is, even relatively low
frequency complex forms can be and often are stored in the mental lexicon. If we take
this finding seriously, we must conclude that theories that claim that productively
derived words are always created on line cannot be correct.

What does this tell us about the structure of a morphological theory? Given the discussion
in Section 29.2, we do not take a stand on whether morphological theory should be mor-
pheme-based or process-based, realizational or non-realizational, although we suspect that
642 English morphology and theories of morphology

the emphasis on these issues over the years has been misguided. Somewhat in the spirit of
Optimality Theory, we might think about diverting our attention from the question of how
forms get generated and focus instead on how well potential outputs conform to canonical
patterns, rules, sets of morphologically related words, or constraints (or whatever we choose
to call them). Sometimes the need to find a word in the course of speaking or writing trumps
the rules. The best output may not be perfect, but it beats no word at all. And maybe this is
the most important theoretical insight we have gained in the process of writing this book.
Any theory of morphology worth its salt cannot ignore this fact.
In the end, we come back to a methodological point. Our vision of theory in morphology
has been strongly influenced by our initial decision to look at the data available in corpora
and to take seriously complex forms that might sometimes rub against the grain of prescrip-
tivist sensibilities. Theory-building would certainly be easier if the data were neater. But
English morphology is vastly more complex than we anticipated at the outset of this project
and theory must respond by being far more nuanced than it has been thus far. It is our hope
that what is to be found in this volume will be instrumental in influencing the next genera-
tion of morphological theory.
 REFERENCES

Acquaviva, Paolo. 2008. Lexical plurals: A morphosemantic approach. Oxford: Oxford University
Press.
Adams, Valerie. 1973. An introduction to modern English word formation. London: Longman.
Adams, Valerie. 2001. Complex words in English. Harlow, Essex: Longman.
Ainsworth, Helen. 1992. The mark of possession or possession’s mark? A case study. New Zealand
English Newsletter 6: 17–20.
Albright, Adam and Bruce Hayes. 2003. Rules vs. analogy in English past tenses: A computational/
experimental study. Cognition 90: 119–61.
Alexiadou, Artemis. 2001. Functional structure in nominals. Amsterdam: John Benjamins.
Algeo, John. 1971. The voguish uses of non. American Speech 46: 87–105.
Allen, Margaret R. 1978. Morphological investigations. Storrs: University of Connecticut dissertation.
Altenberg, Bengt. 1982. The genitive v. the of-construction: A study of syntactic variation in 17th century
English. Malmö: CWK Gleerup.
Amiot, Dany and Georgette Dal. 2007. Integrating neoclassical combining forms into a lexeme-
based morphology. In Geert Booij, Luca Ducceschi, Bernard Fradin, Emiliano Guevara, Angela
Ralli, and Sergio Scalise (eds), Online proceedings of the fifth Mediterranean morphology meeting
(MMM5), Fréjus, 15–18 September 2005. 323–36.
Anderson, Stephen R. 1992. A-morphous morphology. Cambridge: Cambridge University Press.
Anderson, Stephen R. 2005. Aspects of the theory of clitics. Oxford: Oxford University Press.
Anderwald, Lieselotte. 2009. The morphology of English dialects: Verb-formation in non-standard.
English. Cambridge & New York: Cambridge University Press.
Anshen, Frank and Mark Aronoff. 1997. Morphology in real time. Yearbook of Morphology 1996:
9–12.
Arndt-Lappe, Sabine. 2011. Towards an exemplar-based model of stress in English noun–noun
compounds. Journal of Linguistics 47(3): 549–85.
Arndt-Lappe, Sabine. 2013. Word-formation and analogy. In Peter O. Müller, Ingeborg Ohnheiser,
Susan Olsen, and Franz Rainer (eds), Word-formation: An international handbook of the lan-
guages of Europe. Berlin: Mouton de Gruyter.
Arndt-Lappe, Sabine. 2014. Analogy in suffix rivalry: the case of English –ity and –ness. English
Language and Linguistics 18(03): 497–548.
Arndt-Lappe, Sabine and Melanie Bell. 2012. Explaining compound stress analogically. Paper
presented at the 15th International Morphology Meeting, Vienna, 9–12 February.
Arndt-Lappe, Sabine & Ingo Plag. 2013. The role of prosodic structure in the formation of blends.
English Language and Linguistics 17(3): 537–563.
Aronoff, Mark. 1976. Word formation in generative grammar. Cambridge, MA: MIT Press.
Aronoff, Mark. 1980. Contextuals. Language 56: 744–58.
Aronoff, Mark and Nanna Fuhrhop. 2002. Restricting suffix combinations in German and English:
Closing suffixes and the Monosuffix Constraint. Natural Language and Linguistic Theory 20(3):
451–90.
644 References

Auwera, Johann van der. 1985. Relative that—a centennial dispute. Journal of Linguistics 21(1):
149–79.
Ayto, John (ed.). 1990. The Longman register of new words. London: Longman.
Baayen, R. Harald. 1992. Quantitative aspects of morphological productivity. Yearbook of
Morphology 1991: 109–49.
Baayen, R. Harald. 1993. On frequency, transparency and productivity. Yearbook of Morphology
1992: 181–208.
Baayen, R. Harald and Fermín Moscoso del Prado Martín. 2005. Semantic density and past-tense
formation in three Germanic languages. Language 81: 666–98.
Baayen, R. Harald and Rochelle Lieber. 1991. Productivity and English word-formation: A corpus-
based study. Linguistics 29: 801–43.
Baayen, R. Harald, Richard Piepenbrock, and Leon Gulikers. 1995. The CELEX Lexical Database
(CD-ROM). Philadelphia: Linguistic Data Consortium, University of Pennsylvania.
Baeskow, Heike. 2004. Lexical properties of selected non-native morphemes of English. Tübingen:
Narr.
Baker, Mark C. 2003. Lexical categories: Verbs, nouns, and adjectives. New York: Cambridge
University Press.
Bardsley, Dianne and Jane Simpson. 2009. Hypocoristics in New Zealand and Australian English.
In Pam Peters, Peter Collins, and Adam Smith (eds), Comparative studies in Australian and New
Zealand English, 46–69. Amsterdam & Philadelphia: Benjamins.
Barker, Chris. 1998. Episodic -ee in English: A thematic role constraint on new word formation.
Language 74(4): 695–727.
Barnhart, Robert K., Sol Steinmetz, and Clarence L. Barnhart (eds). 1990. The third Barnhart dic-
tionary of new English. N.p.: Wilson.
Bat-El, Outi. 2006. Blend. Encyclopedia of language and linguistics, 2nd edn, vol. 2, 66–70. Oxford:
Elsevier.
Bat-El, Outi and Evan-Gary Cohen. 2012. Stress in English blends: A constraint-based approach.
In Vincent Renner, François Maniez and Pierre Arnaud (eds), Cross-disciplinary perspectives on
lexical blending. Berlin & New York: Mouton de Gruyter.
Bauer, Laurie. 1978. The grammar of nominal compounding with special reference to Danish, English
and French. Odense: Odense University Press.
Bauer, Laurie. 1983. English word-formation. Cambridge: Cambridge University Press.
Bauer, Laurie. 1988. Introducing linguistic morphology. Edinburgh: Edinburgh University Press.
Bauer, Laurie. 1990. Be-heading the word. Journal of Linguistics 26: 1–31.
Bauer, Laurie. 1993. Manual of information to accompany the Wellington Corpus of Written New
Zealand English. Wellington: Victoria University, Department of Linguistics.
Bauer, Laurie. 1994. Watching English change. London & New York: Longman.
Bauer, Laurie. 1997a. Evaluative morphology: A search for universals. Studies in Language 21: 533–75.
Bauer, Laurie 1997b. A class of English irregular verbs. English Studies 78: 545–55.
Bauer, Laurie. 1998a. When is a sequence of two nouns a compound in English? English Language
and Linguistics 2(1): 65–86.
Bauer, Laurie. 1998b. Is there a class of neoclassical compounds, and if so is it productive?
Linguistics 36(3): 403–22.
Bauer, Laurie. 2000. Word. In Geert Booij, Christian Lehmann and Joachim Mugdan (eds),
Morphology: An international handbook of inflection and word-formation, 247–57. Berlin &
New York: Mouton de Gruyter.
 References 645

Bauer, Laurie. 2001. Morphological productivity. Cambridge: Cambridge University Press.

Bauer, Laurie. 2002. Why are we linguists and not linguisticians? In Katja Lenz and Ruth Möhlig
(eds), Of dyuersitie & chaunge of langage; Essays presented to Manfred Görlach on the occasion of
his 65th birthday, 27–32. Heidelberg: Winter.
Bauer, Laurie. 2003a. English prefixation: A typological shift? Acta Linguistica Hungarica. 50(1–2):
33–40.
Bauer, Laurie. 2003b. Introducing linguistic morphology, 2nd edn. Edinburgh: Edinburgh University
Press.
Bauer, Laurie. 2004a. Adjectives, compounds and words. Nordic Journal of English Studies 3(1)
(Worlds of words: A tribute to Arne Zettersten): 7–22.
Bauer, Laurie. 2004b. A glossary of morphology. Edinburgh: Edinburgh University Press.
Bauer, Laurie. 2006a. Competition in English word-formation. In Ans van Kemenade and Bettelou
Los (eds), The handbook of the history of English, 177–98. Malden, MA: Blackwell.
Bauer, Laurie. 2006b. Compounds and minor word-formation types. In Bas Aarts and April
McMahon (eds), The handbook of English linguistics, 483–506. Malden, MA: Blackwell.
Bauer, Laurie. 2008. Dvandva. Word Structure 1: 1–20.
Bauer, Laurie. 2009a. Facets of English plural morphology. In The annual of texts by foreign guest
professors, vol. 3. Prague: Univerzita Karlova v Praze & TOGGA.
Bauer, Laurie. 2009b. Typology of compounds. In Rochelle Lieber and Pavol Štekauer (eds), The
Oxford handbook of compounding, 343–56. Oxford: Oxford University Press.
Bauer, Laurie. 2009c. IE, Germanic: Danish. In Rochelle Lieber and Pavol Štekauer (eds), The
Oxford handbook of compounding, 400–16. Oxford: Oxford University Press.
Bauer, Laurie. 2010a. An overview of morphological universals. Word Structure 3(2): 131–40.
Bauer, Laurie. 2010b. Co-compounds in Germanic. Journal of Germanic Linguistics 22(3): 201–19.
Bauer, Laurie. 2010c. The typology of exocentric compounding. In Sergio Scalise and Irene Vogel
(eds), Cross-disciplinary issues in compounding, 167–76. Amsterdam & Philadelphia: Benjamins.
Bauer, Laurie and Ingrid Bauer. 1996. Word-formation in the playground. American Speech 71: 111–12.
Bauer, Laurie and Winifred Bauer. 2002. Adjective boosters in the speech of New Zealand chil-
dren. Journal of English Linguistics 30: 244–57.
Bauer, Laurie and Rodney Huddleston. 2002. Lexical word formation. In Rodney Huddleston
and Geoffrey K. Pullum (eds), The Cambridge grammar of the English language, 1621–722.
Cambridge: Cambridge University Press.
Bauer, Laurie and Antoinette Renouf. 2001. A corpus-based study of compounding in English.
Journal of English Linguistics 29: 101–23.
Bauer, Laurie and Elizaveta Tarasova. 2010. The meaning link in nominal compounds. Paper pre-
sented at Copenhagen Symposium: Approaches to the Lexicon, Copenhagen, 8–10 December.
Bauer, Laurie and Salvador Valera (eds). 2005. Approaches to conversion/zero-derivation. Münster:
Waxmann.
Beard, Robert. 1982. The plural as a lexical derivation. Glossa 16: 133–48.
Beard, Robert. 1995. Lexeme morpheme base morphology: A general theory of inflection and word.
Albany: State University of New York Press.
Becker, Thomas. 1993. Back-formation, cross-formation and ‘bracketing paradoxes’ in paradig-
matic morphology. Yearbook of Morphology 1993:1–25.
Bell, Melanie. 2011. At the boundary of morphology and syntax: Noun noun constructions in
English. In Alexandra Galani, Glynn Hicks, and George Tsoulos (eds), Morphology and its
interfaces, 137–68. Amsterdam & Philadelphia: Benjamins.
646 References

Bell, Melanie and Ingo Plag. 2012. Informativeness is a determinant of compound stress in English.
Journal of Linguistics 48(3): 485–520.
Berg, Thomas. 1998. The (in)compatibility of morpheme orders and lexical categories and its
historical implications. English Language and Linguistics 2; 245–62.
Berg, Thomas. 2003. Right-branching in English derivational morphology. English Language and
Linguistics 7(2): 279–307.
Berg, Thomas. 2011. The clipping of common and proper nouns. Word Structure 4: 1–19.
Bergen, Benjamin K. 2004. The psychological reality of phonaesthemes. Language 80(2): 291–311.
Biber, Douglas and Bethany Gray. 2011. Grammatical change in the noun phrase: The influence of
written language use. English Language and Linguistics 15(2): 223–50.
Bierwisch, Manfred. 1988a. Tools and explanations of comparison: Part 1. Journal of Semantics
6(1): 57–93.
Bierwisch, Manfred. 1988b. Tools and explanations of comparison: Part 2. Journal of Semantics
6(1): 101–46.
Bierwisch, Manfred. 1989. The semantics of gradation. In Manfred Bierwisch and Ewald Lang
(eds), Dimensional adjectives 71–261. Berlin: Springer-Verlag.
Bisetto, Antoinetta and Sergio Scalise. 2005. The classification of compounds. Lingue e linguaggio
4(2): 319–32.
Blevins, James P. 2003. Stems and paradigms. Language 79(4): 737–67.
Bloch, Bernard. 1947. English verb inflection. Language 23(4): 399–418.
Bloomfield, Leonard. 1935. Language. London: Allen & Unwin.
Bock, Kathryn, Sally Butterfield, Anne Cutler, J. Cooper Cutting, Kathleen M. Eberhard, and
Karin R. Humphreys. 2006. Number agreement in British and American English: Disagreeing
to agree collectively. Language 82(1): 64–113.
Bolinger, Dwight. 1968. Aspects of language. New York: Harcourt.
Bolinger, Dwight. 1975. Aspects of language. 2nd edn. New York: Harcourt.
Booij, Geert. 1986. Form and meaning in morphology: The case of Dutch ‘agent’ nouns. Linguistics
24: 503–17.
Booij, Geert. 1993. Against split morphology. Yearbook of Morphology 1993: 27–49.
Booij, Geert. 1996a. Inherent versus contextual inflection and the split morphology hypothesis.
Yearbook of Morphology 1995: 1–16.
Booij, Geert. 1996b. Review of Beard (1995). Language 72(4): 812–16.
Booij, Geert. 2007. Grammar of words: An introduction to linguistic morphology. 2nd edn. Oxford &
New York: Oxford University Press.
Booij, Geert. 2010. Construction morphology. Oxford: Oxford University Press.
Booij, Geert and Jerzy Rubach. 1984. Morphological and prosodic domains in lexical phonology.
Phonology Yearbook 1: 1–27.
Börger, Claudia. 2007. Word-formation processes from a cognitive perspective: An analysis of complex
prepositional lexemes. Hamburg: Kovač.
Botha, Rudolf P. 1983. Morphological mechanisms. Oxford: Pergamon Press.
Boyé, Gilles. 2006. Suppletion. In Keith Brown (ed.), Encyclopedia of language and linguistics 2nd
edn, vol 12, 297–99. Oxford: Elsevier.
Bram, Barli. 2011. Major total conversion in English: The question of directionality. Wellington:
Victoria University of Wellington thesis.
Brekle, Herbert E. 1970. Generative Satzsemantik und transformationelle Syntax im System der eng-
lishen Nominalkomposition. Munich: Fink.
 References 647

Bresnan, Joan. 1982. Polyadicity. In Joan Bresnan (ed.), The mental representation of grammatical
relations, 149–72. Cambridge, MA: MIT Press.
Bresnan, Joan and Sam Mchombo. 1995. The Lexical Integrity Principle: Evidence from Bantu.
Natural Language and Linguistic Theory 13: 181–254.
British National Corpus, The, version 3 (BNC XML Edition). 2007. Distributed by Oxford
University Computing Services on behalf of the BNC Consortium. <http://www.natcorp.
ox.ac.uk/>, accessed 1 November 2012.
Bulloch, John Malcolm. 1924. The delight of the Doric in the diminutive. In William A. Craigie,
John Buchan, Peter Giles, John Malcolm Bulloch, and Burns Club of London. The Scottish
tongue; A series of lectures on the vernacular language of Lowland Scotland delivered to the members
of the Vernacular Circle of the Burns Club of London, 125–51. London & New York: Cassell.
Burnard, Lou (ed.). 2007. Reference guide for the British National Corpus (XML Edition) <http://
www.natcorp.ox.ac.uk/XMLedition/URG/> (1 February, 2007.)
Burzio, Luigi. 1994. Principles of English stress. Cambridge: Cambridge University Press.
Burzio, Luigi. 1995. On the metrical unity of Latinate affixes. In Héctor Campos and Paula Marie
Kempchinsky (eds), Evolution and revolution in linguistic theory, 1–24. Washington, DC:
Georgetown University Press.
Burzio, Luigi. 1998. Multiple correspondence. Lingua 104(1): 79–109.
Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form. Amsterdam &
Philadelphia: Benjamins.
Bybee, Joan and Carol L. Moder. 1983. Morphological classes as natural categories. Language 59:
251–70.
Bybee, Joan and Dan I. Slobin. 1982. Rules and schemas in the development and use of the English
past tense. Language 58: 265–89.
Cannon, Garland. 1986. Blends in English word formation. Linguistics 24: 725–53.
Cannon, Garland. 1987. Historical change and English word-formation. New York: Lang.
Carney, Edward. 1994. A survey of English spelling. London & New York: Routledge.
Carnie, Andrew. 2006. Syntax: A generative introduction. Oxford: Blackwell.
Carstairs-McCarthy, Andrew. 2002. An introduction to English morphology. Edinburgh: Edinburgh
University Press.
Chambers dictionary, The. 2008. 11th edition. Vivian Marr (ed.). Edinburgh: Chambers Harrap.
Chapman, Don and Royal Skousen. 2005. Analogical modeling and morphological change:
The case of the adjectival negative prefix in English. English Language and Linguistics 9(2):
333–57.
Cheshire, Jenny. 1982. Variation in an English dialect: A sociolinguistic study. Cambridge: Cambridge
University Press.
Chomsky, Noam. 1970. Remarks on nominalization. In Roderick A. Jacobs and Peter S. Rosenbaum
(eds), Readings in English Transformational Grammar. 184–221. Waltham, MA: Ginn.
Chomsky, Noam and Morris Halle. 1968. The sound pattern of English. New York: Harper & Row.
Clark, Eve V. and Herbert H. Clark. 1979. When nouns surface as verbs. Language 55. 767–811.
Coates, Jennifer. 1983. The semantics of the modal auxiliaries. London: Croom Helm.
Collie, Sarah. 2008. English stress preservation: The case for ‘fake cyclicity’. English Language and
Linguistics 12(3): 505–32.
Collins English dictionary, 8th edition. 2006. Glasgow: HarperCollins.
Comrie, Bernard. 1981. Language universals and linguistic typology. Chicago: University of Chicago
Press.
648 References

Concise Oxford English dictionary. 2001. 10th edition. Judy Pearsall (ed.). Oxford: Oxford University Press.
Cook, Paul and Suzanne Stevenson. 2010. Automatically identifying the source words of lexical
blends in English. Computational Linguistics 36(1): 129–49.
Corbin, Danielle. 1987. Morphologie dérivationelle et structuration du lexique, 2 vols. Tübingen:
Niemeyer.
Covington, Michael A. 1981. Evidence for lexicalism: A critical review. Bloomington: Indiana
University Linguistics Club.
Croft, William. 1991. Syntactic categories and grammatical relations: The cognitive organization of
information. Chicago, IL: The University of Chicago Press.
Daelemans, Walter and Antal van den Bosch. 2005. Memory-based language processing. Cambridge:
Cambridge University Press.
Daelemans, Walter, Jakub Zavrel, Ko van der Sloot, and Antal van den Bosch. 2007. TiMBL:
Tilburg Memory Based Learner, version 6.0, Reference Guide (LK Technical Report 04-02).
Tilburg: ILK.
Dalton-Puffer, Christiane and Ingo Plag. 2000. Categorywise, some compound type morphemes
seem to be rather suffix-like: On the status of ‑ful, ‑type, and ‑wise in Present Day English. Folia
Linguistica 34(3–4): 225–45.
Davies, Mark. 2008. The Corpus of Contemporary American English (COCA): 400+ million words,
1990–present. <http://www.americancorpus.org/>, accessed 1 November 2012.
Davies, Mark. 2010. The Corpus of Historical American English (COHA): 400+ million words,
1810–2009. <http://corpus.byu.edu/coha/>, accessed 1 November 2012.
Davies, Mark. 2011. Google Books (American English) Corpus: 155 billion words, 1810–2009. <http://
googlebooks.byu.edu/>, accessed 1 November 2012.
Davis, Stuart. 1988. Topics in syllable geometry. New York: Garland Press.
Derwing, Bruce I. and Royal Skousen. 1994. Productivity and the English past tense: Testing
Skousen’s analogical model. In Susan D. Lima, Roberta Corrigan, and Gregory K. Iverson
(eds), The reality of linguistic rules, 193–218. Amsterdam & Philadelphia: Benjamins.
Di Sciullo, Anna Maria. 2005. Asymmetry in morphology. Cambridge, MA: MIT Press.
Di Sciullo, Anna-Maria and Edwin Williams. 1987. On the definition of word. Cambridge, MA: MIT
Press.
Dixon, Robert M. W. 1977. A grammar of Yidiny. Cambridge: Cambridge University Press.
Dixon, Robert M. W. 2005. A semantic approach to English grammar. 2nd edn. Oxford: Oxford
University Press.
Dixon, Robert M. W. 2010. Basic linguistic theory: Methodology. Oxford: Oxford University Press.
Dixon, Robert M. W. and Alexandra Y. Aikhenvald. 2002. Word: A typological framework. In
Robert M. W. Dixon and Alexandra Y. Aikhenvald (eds), Word: A cross-linguistic typology, 1–40.
Cambridge: Cambridge University Press.
Dollinger, Stefan. 2008. N/V(ing) + N compounds in North American English: On the trail of
the S-curve? Paper presented at the 1st Triennial Conference of the International Society for the
Linguistics of English, Freiburg, 8–11 October 2008.
Downing, Pamela. 1977. On the creation and use of English compound nouns. Language 77:
810–42.
Dressler, Wolfgang U. 1985. Suppletion in word-formation. In Jacek Fisiak (ed.), Historical semantics—
historical word-formation, 97–112. Berlin: Mouton de Gruyter.
Dressler, Wolfgang U. 1989. Prototypical differences between inflection and derivation. Zeitschrift
für Phonetik, Sprachwissenschaft und Kommunikationsforschungen 42: 3–10.
 References 649

Dressler, Wolfgang U. 2000. Extragrammatical vs. marginal morphology. In Ursula Doleschal and
Anna M. Thornton (eds), Extragrammatical and marginal morphology, 1–10. Munich: Lincom.
Dressler, Wolfgang U. and Lavinia Merlini Barbaresi. 1994. Morphopragmatics. Diminutives and
intensifiers in Italian, German, and other languages. Berlin: Mouton de Gruyter.
Earle, John. 1890. English prose. London: Smith, Elder.
Elzinga, Dirk. 2006. English adjective comparison and analogy. Lingua 116(6): 757–70.
Embick, David. 2007. Blocking effects and analytic/synthetic alternations. Natural Language and
Linguistic Theory 25(1): 1–37.
Emonds, Joseph. 1976. A transformational approach to English syntax. New York: Academic Press.
Epps, Patience. 2008. A grammar of Hup. Berlin & New York: Mouton de Gruyter.
Erdmann, Peter. 1999. Compound verbs in English: Are they pseudo? In Guy A. J. Tops, Betty
Devriendt and Steven Geukens (eds), Thinking English grammar, 239–52. Leuven: Peeters.
Fabb, Nigel. 1988. English suffixation is constrained only by selectional restrictions. Natural
Language and Linguistic Theory 6: 527–39.
Faiss, Klaus. 1981. Compound, pseudo-compound and syntactic group especially in English. In
Peter Kunsmann and Ortwin Kuhn (eds), Weltsprache Englisch in Forschung und Lehre: Festschrift
für Kurt Wächtler, 132–50. Berlin: Schmidt.
Farnetani, Edda, Carol Taylor Torsello, and Piero Cosi. 1988. English compound versus non-
compound noun phrases in discourse: An acoustic and perceptual study. Language and Speech
31: 157–80.
Farrell, Patrick. 2001. Functional shift as category underspecification. English Language and
Linguistics 5: 109–30.
Filppula, Markku. 2004. Irish English: Morphology and syntax. In Bernd Kortmann, Kate
Burridge, Rajend Mesthrie, Edgar W. Schneider, and Clive Upton (eds), A handbook of varieties
of English, vol. 2: Morphology and syntax, 73–102. Berlin & New York: Mouton de Gruyter.
Firth, John Rupert. 1964. The tongues of men, and speech. Oxford: Oxford University Press.
Fischer, Olga. 1992. Syntax. In Norman Blake (ed.), The Cambridge history of the English language,
vol. 2: 1066–1476, 207–408. Cambridge: Cambridge University Press.
Fischer, Olga and Wim van der Wurff. 2006. Syntax. In Richard Hogg and David Denison (eds),
A history of the English language, 109–98. Cambridge: Cambridge University Press.
Fleischer, Wolfgang. 1975. Wortbildung der deutschen Gegenwartssprache. Tübingen: Niemeyer.
Fleischer, Wolfgang and Irmhild Barz. 2007. Wortbildung der deutschen Gegenwartssprache.
Tübingen: Niemeyer.
Fraser, Bruce. 1974. The verb–particle construction in English. New York: Academic Press.
Fudge, Erik. 1984. English word-stress. London: Allen & Unwin.
Funk, Wolf-Peter. 1986. Towards a definition of semantic constraints on negative prefixation in
English and German. In Dieter Kastovsky and Aleksander Szwedek (eds), Linguistics across
historical and geographical boundaries: In honour of Jacek Fisiak on the occasion of his fiftieth birth-
day, 877–89. Berlin: Mouton de Gruyter.
Gagné, Christina. 2001. Relation and lexical priming during the interpretation of noun–noun com-
binations. Journal of Experimental Psychology: Learning, Memory, and Cognition 27: 236–54.
Gagné, Christina and Edward J. Shoben. 1997. The influence of thematic relations on the compre-
hension of non-predicating conceptual combinations. Journal of Experimental Psychology:
Learning, Memory, and Cognition 23: 71–87.
Ghomeshi, Jila, Ray Jackendoff, Nicole Rosen, and Kevin Russell. 2004. Contrastive focus redupli-
cation in English (the salad-salad paper). Natural Language and Linguistic Theory 22: 307–57.
650 References

Giegerich, Heinz J. 1999. Lexical strata in English: Morphological causes, phonological effects.
Cambridge: Cambridge University Press.
Giegerich, Heinz J. 2004. Compound or phrase? English noun plus noun constructions and the
stress criterion. English Language and Linguistics 8(1): 1–24.
Giegerich, Heinz J. 2005. Associative adjectives in English and the lexicon–syntax interface.
Journal of Linguistics 41(3): 571–91.
Giegerich, Heinz J. 2006. Attribution in English and the distinction between phrases and com-
pounds. In Petr Rösel (ed.), Englisch in Zeit und Raum—English in time and space: Forschungsbericht
für Klaus Faiss, 10–27. Trier: Wissenschaftlicher Verlag Trier.
Giegerich, Heinz J. 2009a. The English compound stress myth. Word Structure 2: 1–17.
Giegerich, Heinz J. 2009b. Compounding and lexicalism. In Rochelle Lieber and Pavol Štekauer
(eds), The Oxford handbook of compounding, 178–200. Oxford: Oxford University Press.
Giegerich, Heinz J. 2012. The morphology of -ly and the categorial status of ‘adverbs’ in English.
English Language and Linguistics 16(3): 341–59.
Glowka, A. Wayne. 1985. The continuing story of ‑ola. American Speech 60(2): 150–6.
Gnutzmann, Claus, Robert Ilson, and Joy Webster. 1973. Comparative constructions in contem-
porary English. English Studies 54(5): 417–38.
Goldsmith, John A. 1990. Autosegmental and metrical phonology. Oxford & Cambridge, MA: Blackwell.
Górska, Elzbieta. 1994. Moonless nights and smoke-free cities, or what can be without what? A
cognitive study of privative adjectives in English. Folia Linguistica 28: 413–45.
Green, Jonathon (ed.). 1991. Neologisms. London: Bloomsbury.
Green, Lisa. 1998. ApSec and predicate phrases in African-American vernacular English. In
Salikoko S. Mufwene, John R. Rickford, Guy Bailey, and John Baugh (eds), African-American
English: Structure, history and use, 37–68. London: Routledge.
Gries, Stefan Th. 2001. A corpus-linguistic analysis of English ‑ic vs ‑ical adjectives. ICAME Journal
25: 65–107.
Gries, Stefan Th. 2003. Testing the sub-test: A collocational-overlap analysis of English ‑ic and
‑ical adjectives. International Journal of Corpus Linguistics 8(1): 31–61.
Gries, Stefan Th. 2004a. Shouldn’t it be breakfunch? A quantitative analysis of the structure of
blends. Linguistics 42(3): 639–67.
Gries, Stefan Th. 2004b. Isn’t that fantabulous? How similarity motivates intentional morphologi-
cal blends in English. In Michel Achard and Suzanne Kemmer (eds), Language, culture, and
mind, 415–28. Stanford, CA: CSLI.
Gries, Stefan Th. 2004c. Some characteristics of English morphological blends. In Mary
A. Andronis, Erin Debenport, Anne Pycha, and Keiko Yoshimura (eds), Papers from the 38th
Regional Meeting of the Chicago Linguistics Society. Vol. II: The Panels, 201–16. Chicago: Chicago
Linguistics Society.
Gries, Stefan Th. 2006. Cognitive determinants of subtractive wordformation processes: A corpus-
based perspective. Linguistics 17(4): 535–58.
Gries, Stefan Th. 2011. Phonological similarity in multi-word symbolic units. Cognitive Linguistics
22(3): 491–510.
Grimm, Jacob. 1826. Deutsche Grammatik, vol. 2. Göttingen: Theil.
Grimshaw, Jane B. 1990. Argument structure. Cambridge, MA: MIT Press.
Guevara, Emiliano. 2007. Binary branching and linguistic theory: Morphological arguments. In
M. Cecilia Picchi and Alan Pona (eds), Proceedings of the XXXII Incontro di Grammatica
Generativa, Firenze, 2–4 March 2006, 93–106. Alessandria: Edizione dell’ Orso.
 References 651

Gussenhoven, Carlos. 2004. The phonology of tone and intonation. Cambridge: Cambridge
University Press.
Gussenhoven, Carlos and A. Broeders. 1981. English pronunciation for student teachers. Groningen:
Wolters-Noordhoff-Longman.
Gussmann, Edmund. 1987. The lexicon of English de-adjectival verbs. In Edmund Gussmann
(ed.), Rules and the lexicon, 79–101. Lublin: Catholic University.
Hale, Kenneth and Samuel Jay Keyser. 1993. On argument structure and the lexical expres-
sion of syntactic relations. In Kenneth Hale and Samuel Jay Keyser (eds), The view from
building 20: Essays in linguistics in honor of Sylvain Bromberger, 53–109. Cambridge, MA:
MIT Press.
Halle, Morris and Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In
Kenneth Hale and S. Jay Keyser (eds), The view from building 20, 111–76. Cambridge, MA: MIT
Press.
Halliday, Michael A. K. 1966. The concept of rank: A reply. Journal of Linguistics 2(1); 110–18.
Hamawand, Zeki. 2007. Suffix rivalry in adjective formation. London: Equinox.
Hamawand, Zeki. 2009. The semantics of English negative prefixes. London: Equinox.
Hammer, Alfred Edward. 1983. German grammar and usage, 2nd edn, revised by Martin Durrell.
London: Arnold.
Hammond, Michael. 1999. The phonology of English. Oxford: Oxford University Press.
Hanssen, Esther, Arina Banga, Robert Schreuder, and Anneke Neijt. To appear. Semantic and
prosodic effects of Dutch linking elements. Morphology.
Harley, Heidi. 2006. English words: A linguistic introduction. Malden, MA & Oxford: Blackwell.
Harley, Heidi and Rolf Noyer. 1999. Distributed morphology. Glot International 4(4): 3–9.
Harnisch, Rüdiger. 2001. Grundform- und Stamm-Prinzip in der Substantivmorphologie des
Deutschen. Synchronische und diachronische Untersuchung eines typologischen Parameters.
Heidelberg: Winter.
Harris, Zellig S. 1942. Morpheme alternants in linguistic analysis. Language 18: 169–80.
Haspelmath, Martin. 1996. Word-class-changing inflection and morphological theory. Yearbook of
Morphology 1995: 43–66.
Haspelmath, Martin. 2002. Understanding morphology. London: Arnold & New York: Oxford
University Press.
Haspelmath, Martin. 2007. Coordination. In Timothy Shopen (ed.), Language typology and syn-
tactic description. Vol. II: Complex Constructions, 2nd edition, 1–51. Cambridge: Cambridge
University Press.
Haspelmath, Martin. 2009. An empirical test of the Agglutination Hypothesis. In Sergio Scalise,
Elisabetta Magni, and Antonietta Bisetto (eds), Universals of language today, 13–29. Dordrecht:
Springer.
Hatcher, Anna G. 1960. An introduction to the analysis of English noun compounds. Word 16:
356–73.
Hawkins, John. 1994. A performance theory of order and constituency. Cambridge: Cambridge
University Press.
Hay, Jennifer. 2000. Causes and consequences of word structure. Evanston, IL: Northwestern
University dissertation.
Hay, Jennifer B. 2002. From speech perception to morphology: Affix-ordering revisited. Language
78(3): 527–55.
Hay, Jennifer. 2003. Causes and consequences of word structure. London: Routledge.
652 References

Hay, Jennifer and Ingo Plag. 2004. What constrains possible suffix combinations? On the interac-
tion of grammatical and processing restrictions in derivational morphology. Natural Language
and Linguistic Theory 22(3): 565–96.
Hayes, Bruce. 1982. Extrametricality and English stress. Linguistic Inquiry 13(2): 227–76.
Heath, Jeffrey. 2008. A grammar of Jamsay. Berlin & New York: Mouton de Gruyter.
Heyvaert, Liesbet 2003. A cognitive-functional approach to nominalization in English. Berlin: Mouton
de Gruyter.
Hilpert, Martin. 2008. The English comparative—language structure and language use. English
Language and Linguistics 12(3): 395–417.
Hinrichs, Lars and Benedikt Szmrecsanyi. 2007. Recent changes in the function and frequency of
Standard English genitive constructions: A multivariate analysis of tagged corpora. English
Language and Linguistics 11(3): 437–74.
Hockett, Charles Francis. 1947. Problems of morphemic analysis. Language 21(4): 321–43.
Hockett, Charles Francis. 1954. Two models of grammatical description. Word 10: 210–31.
Hockett, Charles Francis. 1987. Refurbishing our foundations: Elementary linguistics from an advanced
point of view. Amsterdam & Philadelphia: Benjamins.
Hohenhaus, Peter. 2004. Identical constituent compounding—a corpus-based study. Folia
Linguistica 38(3–4): 297–331.
Holmes, Janet, Bernadette Vine, and Gary Johnson. 1994. The Wellington Corpus of Spoken New
Zealand English (WSC): 1 million words, 1988–1994. Wellington: Victoria University of
Wellington.
Horn, Laurence R. 1989. A natural history of negation. Chicago: University of Chicago Press.
[Reissued: Stanford: CSLI, 2001].
Horn, Laurence R. 2002. Uncovering the un-word: A study in lexical pragmatics. Sophia Linguistica
49: 1–64.
Huddleston, Rodney. 2002a. The clause: Complements. In Rodney Huddleston and Geoffrey K.
Pullum (eds), The Cambridge grammar of the English language, 213–321. Cambridge: Cambridge
University Press.
Huddleston, Rodney. 2002b. The verb. In Rodney Huddleston and Geoffrey K. Pullum (eds), The
Cambridge grammar of the English language, 71–212. Cambridge: Cambridge University Press.
Huddleston, Rodney and Geoffrey K. Pullum (eds). 2002. The Cambridge grammar of the English
language. Cambridge: Cambridge University Press.
Huddleston, Rodney, Geoffrey K. Pullum, and Peter Peterson. 2002. Relative constructions and
unbounded dependencies. In Rodney Huddleston and Geoffrey K. Pullum (eds), The
Cambridge grammar of the English language, 1031–96. Cambridge: Cambridge University Press.
Hudson, Richard. 1995. Does English really have case? Journal of Linguistics 31: 375–92.
Hudson, Richard. 2006. Wanna revisited. Language 82: 604–27.
Humboldt, Wilhelm von. 1836. Über die Verschiedenheit des menschlichen Sprachbaues und ihren
Einfluss auf die geistige Entwicklung des Menschengeschlechts. Berlin: Königliche Akademie der
Wissenschaften.
Jackendoff, Ray. 1975. Morphological and semantic regularities in the lexicon. Language 51:
639–71.
Jackendoff, Ray. 1977. Constraints on phrase structure rules. In Peter Culicover, Thomas Wasow
and Adrian Akmajian (eds), Formal syntax, 249–83. New York: Academic Press.
Jackendoff, Ray. 1990. Semantic structures. Cambridge, MA: MIT Press.
Jackendoff, Ray. 1991. Parts and boundaries. Cognition 41(1–3): 9–45.
 References 653

Jackendoff, Ray. 2009. Compounding in the parallel architecture and conceptual semantics. In
Rochelle Lieber and Pavol Štekauer (eds), The Oxford handbook of compounding, 105–28.
Oxford: Oxford University Press.
Jakobson, Roman, C. Gunnar M. Fant, and Morris Halle. 1951. Preliminaries to speech analysis: The
distinctive features and their correlates. Cambridge, MA: MIT Press.
Jespersen, Otto. 1914. A modern English grammar on historical principles. Volume II: Syntax. London:
George Allen & Unwin & Copenhagen: Munksgaard.
Jespersen, Otto. 1917. Negation in English and other languages. Copenhagen: Bianco Lunos
Bogtrykkeri.
Jespersen, Otto. 1942. A modern English grammar on historical principles. Part VI: Morphology.
London: George Allen & Unwin & Copenhagen: Munksgaard.
Johansson, Stig. 1979. American and British English grammar: An elicitation experiment. English
Studies 60: 195–215.
Jones, Daniel. 2003. Cambridge English pronouncing dictionary, 16th edn. Cambridge: Cambridge
University Press.
Jurafsky, Daniel. 1996. Universal tendencies in the semantics of the diminutive. Language 72(3):
533–78.
Kaisse, Ellen M. 1983. The syntax of auxiliary reduction in English. Language 59: 93–122.
Kaisse, Ellen M. 2005. The interface between morphology and phonology. In Pavol Štekauer and
Rochelle Lieber (eds), Handbook of English word-formation, 25–47. Dordrecht: Springer.
Kastovsky, Dieter. 1999. English and German morphology: A typological comparison. In Wolfgang
Falkner and Hans-Jörg Schmidt (eds), Words, lexemes, concepts—approaches to the lexicon,
39–51. Tübingen: Niemeyer.
Kastovsky, Dieter. 2000. English morphology: A typological reappraisal. In Chris Schaner-Wolles,
John Rennison, and Friedrich Neubarth (eds), Naturally! Linguistic studies in honour of Wolfgang
Ulrich Dressler presented on the occasion of his 60th birthday, 215–24. Torino: Rosenberg & Sellier.
Kastovsky, Dieter. 2006. Vocabulary. In Richard Hogg and David Denison (eds), A history of the
English languages, 199–270. Cambridge. Cambridge University Press.
Kastovsky, Dieter. 2009. Diachronic perspectives. In Rochelle Lieber and Pavol Štekauer (eds),
The Oxford handbook of compounding, 323–40. Oxford: Oxford University Press.
Katamba, Francis. 1993. Morphology. London: Macmillan.
Katamba, Francis and John Stonham. 2006. Morphology. 2nd edn. London: Macmillan & New
York: St. Martin’s Press.
Kaunisto, Mark. 2007. Variation and change in the lexicon: A corpus-based analysis of adjectives in
English ending in -ic and -ical. Amsterdam: Rodopi.
Kettemann, Bernhard. 1988. Die Phonologie Morphologischer Prozesse im Amerikanischen Englisch.
Tübingen: Narr.
Keuleers, Emmanuel. 2008. Memory-based learning of inflectional morphology. Antwerpen:
University of Antwerpen dissertation.
Killgariff, Adam. 2006. BNC data base and word frequency lists. <http://www.kilgarriff.co.uk/bnc-
readme.html>, 15 November 2011.
Kim, John J., Steven Pinker, Alan Prince, and Sandeep Prasada. 1991. Why no mere mortal has ever
flown out to center field. Cognitive Science 15: 173–218.
Kingdon, Roger. 1958. The groundwork of English stress. London: Longman.
Kiparsky, Paul. 1982a. Lexical phonology and morphology. In The Linguistic Society of Korea
(eds), Linguistics in the morning calm: Selected papers from SICOL-1981, 3–91. Seoul: Hanshin.
654 References

Kiparsky, Paul. 1982b. Word formation and the lexicon. In Frances Ingemann (ed.), Proceedings of
the 1982 Mid-America Linguistics Conference, Lawrence, University of Kansas, 3–29. Lawrence:
Department of Linguistics, University of Kansas.
Kiparsky, Paul. 1982c. Word formation and the lexicon. In Frances Ingemann (ed.), Proceedings of
the 1982 Mid-America Linguistics Conference, Lawrence, University of Kansas, 3–29. Lawrence:
Department of Linguistics, University of Kansas.
Kiparsky, Paul. 1982d. From cyclic phonology to lexical phonology. In Harry van der Hulst and
Norval Smith (eds), The structure of phonological representations, vol. I, 131–75. Dordrecht:
Foris.
Kiparsky, Paul. 2005. Blocking and periphrasis in inflectional paradigms. Yearbook of Morphology
2004: 113–35.
Kjellmer, Göran. 2005. Negated adjectives in Modern English: A corpus-based study. Studia
Neophilologica 77(2): 156–70.
Klinge, Alex. 2005a. The origin of Weapons of Mass Destruction. Investigating traces of lexical for-
mation patterns in the (linguistic) history of Europe. In Henning Nølke, Irene Baron, Hanne
Korzen, Iørn Korzen, and Henrik Høeg-Müller (eds), Grammatica. Festschrift in Honour of
Michael Herslund, 233–247. Pieterlen: Lang.
Klinge, Alex. 2005b. The structure of English nominals. Copenhagen: Copenhagen Business School
dissertation.
Knowles, Elizabeth (ed.). 1997. The Oxford dictionary of new words. Oxford & New York: Oxford
University Press.
Kortmann, Bernd and Ekkehard König. 1992. Categorial reanalysis: The case of deverbal preposi-
tions. Linguistics 30: 671–97.
Kortmann, Bernd, Edgar W. Schneider, Kate Burridge, Rajend Mesthrie, and Clive Upton (eds).
2004. A handbook of varieties of English, vol. 2. Morphology and syntax. Berlin: Mouton de Gruyter.
Koshiishi, Tetsuya. 2002. Collateral adjectives, Latinate vocabulary, and English morphology.
Studia Anglica Posnaniensia 37: 49–88.
Kösling, Kristina. 2011. Stress assignment in triconstituent compounds. Siegen: University of Siegen
dissertation.
Kösling, Kristina and Ingo Plag. 2009. Does branching direction determine prominence assign-
ment? An empirical investigation of triconstituent compounds in English. Corpus Linguistics
and Linguistic Theory 5(2): 205–43.
Kreyer, Rolf. 2003. Genitive and of-construction in modern written English. Processability and
human involvement. International Journal of Corpus Linguistics 8: 169–207.
Kruisinga, Etsko 1932 [1911]. A handbook of present-day English. Part II, 5th edn. Groningen:
Noordhoff.
Kubozono, Haruo. 1990. Phonological constraints on blending in English as a case for phonology–
morphology interface. Yearbook of Morphology 1990: 1–20.
Kunter, Gero. 2011. Compound stress in English. The phonetics and phonology of prosodic prominence.
Tübingen: Niemeyer.
Kunter, Gero and Ingo Plag. 2007. What is compound stress? In Proceedings of the International
Congress of Phonetic Sciences, University of Saarbrücken, 6–10 August 2007. Saarbrücken:
Universität Saarbrücken.
Kuryłowicz, Jerzy. 1936. Dérivation lexicale et dérivation sémantique. Bulletin de la Société de
Linguistique de Paris 27: 79–92.
Kuryłowicz, Jerzy. 1964. The inflectional categories of Indo-European. Heidelberg: Winter.
 References 655

Kytö, Merja and Suzanne Romaine. 1997. Competing forms of adjective comparison in Modern
English: What could be more quicker and easier and more effective? In Terttu Nevalainen and
Leena Kahlas-Tarkka (eds), To explain the present: Studies in the changing English language in
honour of Matti Rissanen, 329–352. Helsinki: Société Néophilologique.
Kytö, Merja and Suzanne Romaine. 2000. Adjective comparison and standardization processes in
American and British English from 1620 to the present. In Laura Wright (ed.), The development
of standard English 1300–1800: Theories, descriptions, conflicts, 171–94. Cambridge: Cambridge
University Press.
Labov, William. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.
Lapointe, Steven G. 1980. A theory of grammatical agreement. Amherst: University of Massachusetts
dissertation.
Lapointe, Steven G. 1981. A lexical analysis of the English auxiliary verb system. In Michael Moortgat,
Harry van der Hulst, and Teun Hoekstra (eds), The scope of lexical rules, 1–78. Dordrecht: Foris.
Lappe, Sabine. 2007. English prosodic morphology. Dordrecht: Springer.
Lass, Roger. 1976. English phonology and phonological theory. Cambridge: Cambridge University
Press.
Leech, Geoffrey N. and Jonathan Culpeper. 1997. The comparison of adjectives in recent British
English. In Terttu Nevalainen and Leena Kahlas-Tarkka (eds), To explain the present: Studies in
the changing English language in honour of Matti Rissanen, 353–74. Helsinki: Société
Néophilologique.
Lees, Robert 1960. The grammar of English nominalizations. Bloomington: Indiana University.
Lefebvre, Claire and Anne-Marie Brousseau. 2002. A grammar of Fongbe. Berlin & New York:
Mouton de Gruyter.
Lehnert, Martin. 1971. Rücklaufiges Wörterbuch der englischen Gegenwartssprache. Leipzig: VEB.
Lehrer, Adrienne. 1995. Prefixes in English word formation. Folia Linguistica 29(1–2): 133–48.
Levi, Judith N. 1978. The syntax and semantics of complex nominals. New York: Academic Press.
Liberman, Mark and Alan Prince. 1977. On stress and linguistic rhythm. Linguistic Inquiry 8:
249–336.
Liberman, Mark and Richard Sproat. 1992. The stress and structure of modified noun phrases in
English. In Ivan A. Sag and Anna Szabolcsi (eds), Lexical matters, 131–81. Stanford: CSLI.
Lieber, Rochelle. 1980. On the organization of the lexicon. Cambridge, MA: MIT dissertation.
Lieber, Rochelle. 1983. Argument linking and compounding in English. Linguistic Inquiry 14:
251–86.
Lieber, Rochelle. 1992. Deconstructing morphology. Chicago & London: University of Chicago Press.
Lieber, Rochelle. 1998. The suffix ‑ize in English: Implications for morphology. In Steven G.
Lapointe, Diane K. Brentari, and Patrick M. Farrell (eds), Morphology and its relation to phonol-
ogy and syntax, 12–34. Stanford, CA: CSLI.
Lieber, Rochelle. 2004. Morphology and lexical semantics. Cambridge: Cambridge University Press.
Lieber, Rochelle. 2009. Introducing morphology. Cambridge: Cambridge University Press.
Lieber, Rochelle. 2010. On the lexical semantics of compounds: Non-affixal (de)verbal com-
pounds. In Sergio Scalise and Irene Vogel (eds), Cross-disciplinary issues in compounding,
127–144. Amsterdam & Philadelphia: Benjamins.
Lieber, Rochelle and Sergio Scalise. 2006. The Lexical Integrity Hypothesis in a new theoretical
universe. Lingue e Linguaggio 5: 7–32.
Lieber, Rochelle and Pavol Štekauer. 2009. Introduction. In Rochelle Lieber and Pavol Štekauer
(eds), The Oxford handbook of compounding, 3–18. Oxford: Oxford University Press.
656 References

Lindquist, Hans. 1998. The comparison of English disyllabic adjectives in ‑y and ‑ly in present-day
British and American English. In Hans Lindquist, Staffan Klintborg, Magnus Levin, and Maria
Estling (eds), The major varieties of English. Papers from MAVEN (The international conference
The Major Varieties of English) 97, 205–12. Växjö: Acta Wexionensia.
Ljung, Magnus. 1970. English denominal adjectives: A generative study of the semantics of group of
high-frequency denominal adjectives in English. Lund: Studentliteratur.
Löbner, Sebastian. 1985. Definites. Journal of Semantics 4: 279–326.
Lojenga, Constance K. 1994. Ngiti. Cologne: Köppe.
Longman dictionary of contemporary English. 1978. Paul Procter (ed.). Harlow, Essex: Longman.
Lowth, Robert. 1762. A short introduction to English grammar. London: Hughes [Reprinted 1967,
Menston: Scolar Press].
Lyons, John. 1966. Toward a notional theory of the parts of speech. Journal of Linguistics 2:
209–36.
Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge University Press.
Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.
McCray, Alexa T., Allen C. Browne, and Dorothy L. Moore. 1988. The semantic structure of neo-
classical compounds. In Robert A. Greenes (ed.), SCAMC’88—Proceedings of the 12th Annual
Symposium on Computer Applications in Medical Care, 165–8. Washington & New York: EEE
Computer Society Press.
McMillan, James B. 1980. Infixing and interposing in English. American Speech 55: 163–83.
Macquarie Dictionary, The. 1997. 3rd edn. Sydney: Macquarie Library.
Mailhammer, Robert. 2007. The Germanic strong verbs: Foundations and development of a new sys-
tem. Berlin & New York: Mouton de Gruyter.
Manova, Stela and Wolfgang U. Dressler. 2005. The morphological technique of conversion in the
inflecting-fusional type. In Laurie Bauer and Salvador Valera (eds), Approaches to conversion/
zero-derivation, 67–101. Münster: Waxmann.
Marchand, Hans. 1960. The categories and types of present-day English word-formation: A synchronic–
diachronic approach. Wiesbaden: Otto Harrasowitz.
Marchand, Hans. 1969. The categories and types of present-day English word-formation, 2nd edn.
Munich: Beck.
Marcus, Gary F. 1995. The acquisition of the English past tense in children and multilayered con-
nectionist networks. Cognition 56; 271–9.
Marcus, Gary F. 1999. Do infants learn grammar with algebra or statistics? Response to Seidenberg
& Elman, Eimas, and Negishi. Science 284: 436–37.
Marle, Jaap van. 1985. On the paradigmatic dimension of morphological creativity. Dordrecht: Foris.
Marsden, Peter H. 1985. Adjective pairs in ‑ic and ‑ical. Lebende Sprachen 1: 26–33.
Matthews, Peter Hugoe. 1972. Inflectional morphology. Cambridge: Cambridge University Press.
Matthews, Peter Hugoe. 2007. Syntactic relations: A critical survey. Cambridge & New York:
Cambridge University Press.
Matthewson, Lisa C. 1991. An application of autosegmental morphology to some nonconcatenative
phenomena in the Germanic languages. Wellington: Victoria University of Wellington MA thesis.
Mayerthaler, Willi. 1981. Morphologische Natürlichkeit. Wiesbaden: Athenaion.
Melloni, Chiara. 2007. Polysemy in word formation: The case of deverbal nominals. Verona: University
of Verona dissertation.
Meys, Willem J. 1975. Compound adjectives in English and the ideal speaker-listener. Amsterdam:
North Holland.
 References 657

Miner, Kenneth L. 1975. English inflectional endings and unordered rules. Foundations of Language
12: 339–65.
Mish, Frederick C. 1983. 9,000 words. Springfield MA: Merriam-Webster.
Mohanan, Karuvannur Puthanveettil. 1982. Lexical phonology. Cambridge, MA: MIT dissertation.
Mohanan, Karuvannur Puthanveettil. 1986. The theory of lexical phonology. Dordrecht: Reidel.
Mondorf, Britta. 2000. Wider-ranging vs. more old-fashioned views on comparative formation in
adjectival compounds/derivatives. In Bernhard Reitz and Sigrid Rieuwerts (eds), Anglistentag
1999 Mainz proceedings, 35–44. Trier: Wissenschaftlicher Verlag Trier.
Mondorf, Britta. 2003. Support for more-support. In Günter Rohdenburg and Britta Mondorf
(eds), Determinants of grammatical variation in English, 251–304. Berlin: Mouton de Gruyter.
Mondorf, Britta. 2006. Rewriting English grammar books: Factors constraining the choice
between synthetic and analytic comparative forms. In Christoph Houswitschka, Gabriele
Knappe, and Anja Müller (eds), Proceedings of the Anglistentag 2005, Bamberg (Proceedings of the
Conference of the German Association of University Teachers of English), 587–607. Trier:
Wissenschaftlicher Verlag Trier.
Mondorf, Britta. 2009a. More support for more-support: The role of processing constraints on the
choice between synthetic and analytic comparative forms. Amsterdam: Benjamins.
Mondorf, Britta. 2009b. How lexicalization reflected in hyphenation affects variation and word-
formation. In Andreas Dufter, Jürg Fleischer, and Guido Seiler (eds), Describing and modeling
variation in grammar, 361–88. Berlin: Mouton de Gruyter.
Montgomery, Michael. 1999. A superlative complex in Appalachian English. Southeastern Journal
of Linguistics 23: 1–14.
Myers, Scott. 1984. Zero-derivation and inflection. MIT working papers in linguistics 7: 53–69.
Nagano, Akiko. 2008. Conversion and back-formation in English: Toward a theory of morpheme-based
morphology. Tokyo: Kaitakusha.
Nagano, Akiko. 2011. The right-headedness of morphology and the status and development of
category-determining prefixes in English. English Language and Linguistics 15(1): 61–83.
Nation, I. S. Paul. 1993. Using dictionaries to estimate vocabulary size: Essential, but rarely fol-
lowed, procedures. Language Testing 10(1): 27–40.
Nelson, Nicole. 1998. Mixed anchoring in French hypocoristic formation. RuLing Papers 1: 185–99.
Nevalainen, Terttu. 1999. Early Modern English lexis and semantics. In Roger Lass (ed.), The
Cambridge history of the English language: Vol. III—1476 to 1776, 332–458. Cambridge: Cambridge
University Press.
Nguyên, T. Anh-Thu’, C. L. John Ingram, and J. Rob Pensalfini. 2008. Prosodic transfer in
Vietnamese acquisition of English contrastive stress patterns. Journal of Phonetics 36: 158–90.
Nichols, Johanna. 1986. Head-marking and dependent-marking in grammar. Language 62(1):
56–119.
Nida, Eugene A. 1948. The identification of morphemes. Language 24: 414–41.
Noonan, Michael. 1992. A grammar of Lango. Berlin & New York: Mouton de Gruyter.
Nosofsky, Robert M. 1986. Attention, similarity, and the identification-categorization relationship.
Journal of Experimental Psychology: General 115: 39–57.
Nosofsky, Robert M. 1990. Relations between exemplar similarity and likelihood models of clas-
sification. Journal of Mathematical Psychology 34: 393–418.
OED Online. Oxford University Press. <http://www.oed.com/>, accessed 1 November 2012.
Ohala, John J. 1974. Experimental historical phonology. In John M. Anderson and Carles Jones
(eds), Historical linguistics, vol. 2, 353–89. Amsterdam & Oxford: North Holland.
658 References

Olsen, Susan. 2001. Copulative compounds: A closer look at the distinction between morphology
and syntax. Yearbook of Morphology 2000: 279–320.
Orr, John. 1962. Three studies on homonymics. Edinburgh: University Press.
Orton, Harold, Stewart Sanderson, and John Widdowson (eds). 1978. The linguistic atlas of
England. London: Croom Helm & Atlantic Highlands NJ: Humanities Press.
Ostendorf, Mari, Patti Price, and Stefanie Shattuck-Hufnagel. 1996. Boston University Radio Speech
Corpus. Philadelphia: Linguistic Data Consortium, University of Pennsylvania.
Oxford advanced learner’s dictionary of current English. 1974. A. S. Hornby (ed.). Oxford: Oxford
University Press.
Oxford English dictionary, The. 1933. James Augustus Henry Murray, Henry Bradley, William A.
Craigie, and Charles Talbut Onions (eds). Oxford: Clarendon Press.
Oxford English dictionary, The. 1989. 2nd edn. Robert William Burchfield (ed). Oxford: Oxford
University Press.
Palmer, Frank. R. 1974. The English verb. London: Longman.
Palmer, Frank R. 1990. Modality and the English modals, 2nd edn. London & New York: Longman.
Palmer, Frank R. 2001. Mood and modality, 2nd edn. Cambridge: Cambridge University Press.
Palmer, Frank, Rodney Huddleston, and Geoffrey K. Pullum. 2002. Inflectional morphology and
related matters. In Rodney Huddleston and Geoffrey K. Pullum (eds), The Cambridge grammar
of the English language, 1565–1620. Cambridge: Cambridge University Press.
Pater, Joe. 2000. Non-uniformity in English secondary stress: The role of ranked and lexically
specific constraints. Phonology 17(2): 237–74.
Pater, Joe. 2004. Bridging the gap between receptive and productive development with minimally
violable constraints. In René Kager, Joe Pater, and Wim Zonneveld (eds), Constraints in phono-
logical acquisition, 219–44. Cambridge: Cambridge University Press.
Paul, Hermann. 1995 [1920]. Prinzipien der Sprachgeschichte. 5th edn. Tübingen: Niemeyer.
Pawley, Andrew. 2002. Using he and she for inanimate objects in English: Questions of grammar
and world view. In Nick Enfield (ed.), Ethnosyntax: Explorations in language and culture, 110–37.
Oxford: Oxford University Press.
Pawley, Andrew. 2008. Australian vernacular English: some grammatical characteristics. In Kate
Burridge and Berndt Kortmann (eds), Varieties of English 3: The Pacific and Australasia, 362–97.
Berlin & New York: Mouton de Gruyter.
Payne, John. 2011. Genitive coordinations with personal pronouns. English Language and Linguistics
15(2): 363–85.
Payne, John and Rodney Huddleston. 2002. Nouns and noun phrases. In Rodney Huddleston
and Geoffrey K. Pullum (eds), The Cambridge grammar of the English language, 323–524.
Cambridge: Cambridge University Press.
Payne, John, Rodney Huddleston, and Geoffrey K. Pullum. 2010. The distribution and category
status of adjectives and adverbs. Word Structure 3: 31–81.
Peters, Pam. 1999. Paradigm split. In Christian Mair and Marianne Hundt (eds), Corpus linguistics
and linguistic theory: Papers from the Twentieth International Conference on English Language
Research on Computerized Corpora (ICAME 20), 301–12. Amsterdam: Rodopi.
Pietsch, Lukas. 2005. Variable grammars: Verbal agreement in northern dialects of English. Tübingen:
Niemeyer.
Piñeros, Carlos Eduardo. 1998. Prosodic morphology in Spanish: Constraint interaction in word for-
mation. Columbus: Ohio State University dissertation.
Pinker, Steven. 1999. Words and rules. London: Weidenfeld & Nicolson.
 References 659

Pinker, Steven and Alan Prince. 1994. Regular and irregular morphology and the psychological
status of rules of grammar. In Susan D. Lima, Roberta L. Corrigan, and Gregory K. Iverson
(eds), The reality of linguistic rules. Amsterdam & Philadelphia: Benjamins.
Plag, Ingo. 1996. Selectional restrictions in English suffixation revisited. A reply to Fabb (1988).
Linguistics 34(4): 769–98.
Plag Ingo. 1998. Morphological haplology in a constraint-based morpho-phonology. In Wolfgang
Kehrein and Richard Wiese (eds), Phonology and morphology of the Germanic languages, 199–215.
Tübingen: Niemeyer.
Plag, Ingo. 1999. Morphological productivity: Structural constraints in English derivation. Berlin &
New York: Mouton de Gruyter.
Plag, Ingo. 2002. The role of selectional restrictions, phonotactics and parsing in constraining suf-
fix ordering in English. Yearbook of Morphology 2001: 285–314.
Plag, Ingo. 2003. Word-formation in English. Cambridge: Cambridge University Press.
Plag, Ingo. 2004. Syntactic category information and the semantics of derivational morphological
rules. Folia Linguistica 38(3–4): 193–225.
Plag, Ingo. 2006. The variability of compound stress in English: Structural, semantic and analogi-
cal factors. English Language and Linguistics 10(1): 143–72.
Plag, Ingo. 2010. Compound stress assignment by analogy: The constituent family bias. Zeitschrift
für Sprachwissenschaft 29(2): 243–82.
Plag, Ingo and R. Harald Baayen. 2009. Suffix ordering and morphological processing. Language
85(1): 109–52.
Plag, Ingo, Gero Kunter, and Sabine Lappe. 2007. Testing hypotheses about compound stress
assignment in English: A corpus-based investigation. Corpus Linguistics and Linguistic Theory
3(2): 199–233.
Plag, Ingo, Gero Kunter, Sabine Lappe, and Maria Braun. 2008. The role of semantics, argument
structure, and lexicalization in compound stress assignment in English. Language 84(4):
760–94.
Plank, Frans. 1981. Morphologische (Ir-)Regularitäten. Tübingen: Narr.
Plank, Frans. 1994. Inflection and derivation. In Ron E. Asher (ed.), The encyclopedia of language
and linguistics, 1671–8. Oxford: Pergamon.
Poser, William J. 1992. Blocking of phrasal constructions by lexical items. In Ivan Sag and Anna
Szabolsci (eds), Lexical matters, 111–30. Stanford: CSLI Publications.
Postal, Paul M. 1969. Anaphoric islands. In Rob I. Binnick, Alice Davison, Georgia M. Green, and
Jerry L. Morgan (eds), Papers from the 5th regional meeting of the Chicago Linguistic Society,
209–39. Chicago: Chicago Linguistic Society.
Poutsma, Hendrik. 1916. A grammar of late modern English. Groningen: Noordhoff.
Prasada, Sandeep and Steven Pinker. 1993. Generalization of regular and irregular morphological
patterns. Language and Cognitive Processes 8: 1–56.
Prćić, Tvrtko. 2005. Prefixes vs. initial combining forms in English: A lexicographic perspective.
International Journal of Lexicography 18(3): 313–35.
Prćić, Tvrtko. 2008. Suffixes vs. final combining forms in English: A lexicographic perspective.
International Journal of Lexicography 21: 1–22.
Pullum, Geoffrey K. 1982. Syncategorematicity and English infinitival to. Glossa 16: 181–215.
Pullum, Geoffrey 1991. English nominal gerund phrases as noun phrases with verb-phrase heads,
Linguistics 29(5): 763–99.
Pullum, Geoffrey K. 1997. The morpholexical nature of English to-contraction. Language 73: 79–102.
660 References

Pullum, Geoffrey K. and Rodney Huddleston. 2002. Adjectives and adverbs. In Rodney
Huddleston and Geoffrey K. Pullum (eds), The Cambridge grammar of the English language,
525–95. Cambridge: Cambridge University Press.
Pustejovsky, James. 1995. The generative lexicon. Cambridge, MA: MIT Press.
Quinn, Heidi. 2005. The distribution of pronoun case forms in English. Amsterdam & Philadelphia:
Benjamins.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A comprehensive
grammar of the English language. London & New York: Longman.
Radford, Andrew. 1988. Transformational grammar: A first course. Cambridge: Cambridge
University Press.
Raffelsiefen, Renate. 1993. Relating words: A model of base recognition. Linguistic Analysis 23(1–2):
3–159.
Raffelsiefen, Renate. 1999. Phonological constraints on English word formation. Yearbook of
Morphology 1998: 225–87.
Raffelsiefen, Renate. 2004. Absolute ill-formedness and other morphophonological effects.
Phonology 21: 91–142.
Rainer, Franz. 1988. Towards a theory of blocking: The case of Italian and German quality nouns.
Yearbook of Morphology 1988: 155–85.
Ramscar, Michael J. A. 2002. The role of meaning in inflection: Why the past tense doesn’t require
a rule. Cognitive Psychology 45: 45–94.
Ramson, William Stanley (Bill). (ed.). 1988. The Australian national dictionary. Melbourne: Oxford
University Press.
Randle, William. 1961. Payola. American Speech 36(2): 104–16.
Rappaport Hovav, Malka and Beth Levin. 1992. ‑er nominals: Implications for a theory of argu-
ment structure. In Tim Stowell and Eric Wehrli (eds), Syntax and Semantics 26: Syntax and the
Lexicon, 127–53. New York: Academic Press.
Reischer, Jürgen. 2005. WordNexus—Ein Datenretrieval-System für Blends und Kontaminationen
mit umfangreichem deutschen und englischen Korpus. <http://www.lingua-ex-machina.de/>,
accessed 5 February 2011.
Reischer, Jürgen. 2008. Die Wortkreuzung und verwandte Verfahren der Wortbildung: eine korpus-
basierte Analyse des Phänomens ‘Blending’ am Beispiel des Deutschen und Englischen. Hamburg:
Kovač.
Renner, Vincent. 2008. On the semantics of English coordinate compounds. English Studies 89:
606–13.
Renner, Vincent. 2013. English cum, a borrowed coordinator turned complex-compound marker.
Morphology 23(1): 57–66.
Rice, Keren. 2000. Morpheme order and semantic scope: Word formation in the Athapaskan verb.
Cambridge: Cambridge University Press.
Rickford, John. 1986. Social contact and linguistic diffusion: Hiberno-English and New World
Black English. Language 62: 245–89.
Riddle, Elizabeth. 1985. A historical perspective on the productivity of the suffixes ‑ness and ‑ity.
In Jacek Fisiak (ed.), Historical semantics, historical word-formation, 435–61. New York: Mouton
de Gruyter.
Robins, Robert H. 1959. In defense of WP. Transactions of the Philological Society 58.1, 116–44.
Roeper, Thomas. 1988. Compound syntax and head movement. Yearbook of Morphology 1988:
187–228.
 References 661

Roeper, Thomas and Muffy E. A. Siegel. 1978. A lexical transformation for verbal compounds.
Linguistic Inquiry 9: 199–260.
Rohdenburg, Günter. 1996. Cognitive complexity and increased grammatical explicitness in
English. Cognitive Lingustics 7: 149–82.
Rosenbach, Anette. 2002. Genitive variation in English: Conceptual factors in synchronic and dia-
chronic studies. Berlin: Mouton de Gruyter.
Rosenbach, Anette. 2005. Animacy versus weight as determinants of grammatical variation in
English. Language 81(3): 613–44.
Rosenbach, Anette. 2006. Descriptive genitives in English: A case study on constructional gradi-
ence. English Language and Linguistics 10(1): 77–118.
Rumelhart, David E. and James L. McClelland. 1986. On learning the past tenses of English verbs. In
David E. Rumelhart, James L. McClelland, and the PDP research group (eds), Parallel distributed
processing: Explorations in the microstructure of cognition, 216–71. Cambridge, MA: MIT Press.
Rusiecki, Jan. 1985. Adjectives and comparison in English: A semantic study. London: Longman.
Ryder, Mary Ellen. 1994. Ordered chaos: The interpretation of English noun–noun compounds.
Berkeley, Los Angeles & London: University of California Press.
Ryder, Mary Ellen. 1999. Bankers and blue-chippers: An account of ‑er formations in present-day
English. English Language and Linguistics 3(2): 269–97.
Sapir, Edward. 1921. Language. New York: Harcourt, Brace & Jovanovich.
Scalise, Sergio. 1984. Generative morphology. Dordrecht: Foris.
Scalise, Sergio. 1988. Inflection and derivation, Linguistics 26: 561–81.
Scalise, Sergio and Antoinetta Bisetto. 2009. Classification of compounds. In Rochelle Lieber and Pavol
Štekauer (eds), The Oxford handbook of compounding, 49–82. Oxford: Oxford University Press.
Scalise, Sergio and Antoinetta Bisetto. 2012. Morbo-Comp. <http://morbocomp.sslmit.unibo.it>,
accessed 1 November 2012.
Schlücker, Barbara and Ingo Plag. 2011. Compound or phrase? Analogy in naming. Lingua 121:
1539–51.
Schneider, Klaus P. 2003. Diminutives in English. Tübingen: Niemeyer.
Schönefeld, Doris. 2005. Zero-derivation—functional change—metonymy. In Laurie Bauer and
Salvador Valera (eds), Approaches to conversion/zero-derivation, 131–59. Münster: Waxmann.
Schreuder, Robert and R. Harald Baayen. 1997. How complex simplex words can be. Journal of
Memory and Language 37: 118–39.
Selkirk, Elisabeth O. 1982. The syntax of words. Cambridge. MA: MIT Press.
Sepp, Mary. 2006. Phonological constraints and free variation in compounding: A corpus study of
English and Estonian noun compounds. New York: New York City University dissertaion.
Shannon, Claude E. 1948. A mathematical theory of communication. The Bell System Technical
Journal, 27: 379–423 & 623–56. <http://cm.bell-labs. com/cm/ms/what/shannonday/shan-
non1948.pdf>, accessed 3 November 2010.
Siegel, Dorothy C. 1974. Topics in English morphology. Cambridge, MA: MIT dissertation.
Siegel, Dorothy C. 1978. The adjacency constraint and the theory of morphology. In Mark Stein
(ed.), Proceedings of the North Eastern Linguistic Society, 1977: 189–97. Amherst: University of
Massachusetts.
Siemund, Peter. 2007. Pronominal gender in English: A study of English varieties from a cross-linguistic
perspective. London: Routledge.
Silva, Penny (ed.). 1996. A dictionary of South African English. New York: Oxford University
Press.
662 References

Sinclair, John M. (ed.). 1987. Looking up. An account of the Cobuild Project in lexical computing.
London: Collins.
Skousen, Royal. 1989. Analogical modeling of language. Dordrecht: Kluwer.
Skousen, Royal. 1992. Analogy and structure. Dordrecht: Kluwer.
Skousen, Royal and Thereon Stanford. 2007. AM: Parallel. Provo, UT: Brigham Young University.
Skousen, Royal, Deryle Lonsdale, and Dilworth B. Parkinson (eds). 2002. Analogical modeling.
Amsterdam & Philadelphia: Benjamins.
Slotkin, Alan R. 1990. Adjectival ‑less and ‑free: A case of shifting institutional currency. American
Speech 65(1): 33–49.
Smith, Carlotta. 1997. The parameter of aspect, 2nd edn. Dordrecht: Kluwer.
Southerland, Ronald H. 1994. Derivatives in pre‑: A persuasive morphological resource? American
Speech 69(2): 168–76.
Spencer, Andrew. 2003. Does English have productive compounding? In Geert E. Booij, Janet
DeCesaris, Angela Ralli, and Sergio Scalise (eds), Topics in morphology. Selected papers from the
Third Mediterranean morphology meeting, Barcelona, 20–22 September 2001, 329–41. Barcelona:
Institut Universitari de Lingüística applicada, Universitat Pompeu Fadra.
Sproat, Richard and Gregory Ward. 1987. Pragmatic considerations in anaphoric island phenom-
ena. In Barbara Need, Eric Schiller, and Anna Bosch (eds), Papers from the 23rd Annual Regional
Meeting of the Chicago Linguistic Society, 321–35. Chicago: Chicago Linguistic Society.
Štekauer, Pavol. 1996. A theory of conversion in English. Frankfurt-am-Main: Lang.
Štekauer, Pavol. 2005. Meaning predictability in word-formation. Novel, context-free naming units.
Amsterdam & Philadelphia: Benjamins.
Štekauer, Pavol, Salvador Valera & Lívia Körtvélyessy. 2012. Word-formation in the world’s ­languages:
A typological survey. Cambridge: Cambridge University Press.
Stemberger, Joseph Paul. 1981. Morphological haplology. Language 57(4): 791–817.
Steriade, Donca. 1999. Lexical conservatism in French adjectival liaison. In Barbara E. Bullock,
Jean-Marc Authier, and Lisa A. Reed (eds), Formal perspectives in Romance linguistics, 243–70.
Amsterdam & Philadelphia: Benjamins.
Steriade, Donca. 2000. Paradigm uniformity and the phonetics–phonology boundary. In Michael
Broe and Janet Pierrehumbert (eds), Papers in laboratory phonology V: Acquisition and the lexi-
con, 313–34. Cambridge: Cambridge University Press.
Strang, Barbara M. H. 1968. Modern English structure. 2nd edn. London: Arnold.
Strauss, Steven. 1982. On ‘relatedness paradoxes’ and related paradoxes. Linguistic Inquiry 13: 694–700.
Stump, Gregory T. 1998. Inflection. In Andrew Spencer and Arnold M. Zwicky (eds), The hand-
book of morphology, 13–43. Oxford: Blackwell.
Stump, Gregory T. 2001. Inflectional morphology. Cambridge: Cambridge University Press.
Sturtevant Edgar H. 1917. Linguistic change. Chicago & London: University of Chicago Press.
Sussex, Roland. 2004. Abstand, Ausbau, creativity and ludicity in Australian English. Australian
Journal of Linguistics 24: 3–19.
Swart, Henriëtte de. 2011. Mismatches and coercion. In Claudia Maienborn, Klaus von Heusinger,
and Paul Portner (eds), Semantics, 574–97. Berlin & Boston: Mouton de Gruyter.
Sweet, Henry. 1891. A new English grammar: Logical and historical. Oxford: Clarendon Press.
Sweet, Henry. 1892. A short historical English grammar. Oxford: Clarendon Press.
Sweet, Henry. 1898. A new English grammar. Part II: Syntax. Oxford: Clarendon Press.
Szmrecsanyi, Benedikt. 2005. Language users as creatures of habit: A corpus-based analysis of
persistence in spoken English. Corpus Linguistics and Linguistic Theory 1(1): 113–50.
 References 663

Szmrecsanyi, Benedikt. 2006. Morphosyntactic persistence in spoken English: A corpus study at the
intersection of variationist sociolinguistics, psycholinguistics, and discourse analysis. Berlin: Walter
de Gruyter.
Szymanek, Bogdan. 1985. On intermorphic extensions in English and Polish. In Edmund
Gussmann (ed.), Phono-morphology. Studies in the interaction of phonology and morphology,
177–91. Lublin: Catholic University.
Thomas, Macklin. 1983. Interchangeable pairs in un‑, in‑ and en‑, etc. American Speech 58: 78–80.
Townsend, Charles E. 1975. Russian word-formation. Columbus, OH: Slavica.
Trips, Carola. 2009. Lexical semantics and diachronic morphology: The development of ‑hood, ‑dom
and ‑ship in the history of English. Tübingen: Niemeyer.
Trudgill, Peter. 1974. The social differentiation of English in Norwich. Cambridge: Cambridge
University Press.
Tulloch, Sarah (ed.). 1991. The Oxford dictionary of new words. Oxford & New York: Oxford
University Press.
Ullmann, Stephen. 1957. Principles of semantics. Glasgow: Jackson & Oxford: Blackwell.
Urban dictionary. <http://www.urbandictionary.com/>, accessed 1 November 2012.
Urdang, Laurence, Anne Ryle, and Tanya H. Lee. 1986. Ologies and isms. 3rd edn. New York:
Gale.
Vaan, Laura de, Robert Schreuder, and R. Harald Baayen. 2007. Regular morphologically complex
neologisms leave detectable traces in the mental lexicon. The Mental Lexicon 2: 1–23.
Valera Hernández, Salvador. 1996. Adjetivos y adverbios en inglés. La relación de homomorfia.
Granada: Servicio de Publicaciones de la Universidad de Granada.
Vogel, Irene. 2006. Phonological words. In Keith Brown (ed.), Encyclopedia of language and lin-
guistics. Vol. 9: 531–34. Oxford: Elsevier.
Wagner, Thomas. 2010. Interlanguage morphology: Irregular verbs in the mental lexicon of German-
English interlanguage speakers. Tübingen: Narr.
Wälchli, Bernard. 2005. Co-compounds and natural coordination. Oxford: Oxford University Press.
Wales, Katie. 1996. Personal pronouns in present-day English. Cambridge: Cambridge University
Press.
Ward, Gregory, Richard Sproat, and Gail McKoon. 1991. A pragmatic analysis of so-called ana-
phoric islands. Language 67: 439–74.
Ward, Gregory, Betty Birner, and Rodney Huddleston. 2002. Information packaging. In Rodney
Huddleston and Geoffrey K. Pullum (eds), The Cambridge grammar of the English language,
1363–447. Cambridge: Cambridge University Press.
Weber, Andrea. 2002. Assimilation violation and spoken-language processing. Language and
Speech 45(1): 37–46.
Webster’s new international dictionary of the English language. 1957. 2nd edn. Springfield, MA:
Merriam-Webster.
Webster’s third new international dictionary of the English language, unabridged. 1993. Springfield,
MA: Merriam-Webster.
Wells, John C. 2008. Longman pronunciation dictionary, 3rd edn. Harlow: Longman.
Wentworth, Harold. 1941. The allegedly dead suffix -DOM in modern English. Publications of the
Modern Language Association 56: 280–306.
Wentworth, Harold. 1944. American dialect dictionary. London: Constable.
Wiese, Richard. 1996a. Phrasal compounds and the theory of word syntax. Linguistic Inquiry 27:
183–93.
664 References

Wiese, Richard. 1996b. Prosodic alternations in English morphophonology: A constraint-based

account of Morpheme Integrity. In Jacques Durand and Bernard Laks (eds), Current trends in
phonology: Models and methods, vol. 2, 731–56. Salford: European Studies Research Institute,
University of Salford.
Williams, Darrell. 1992. English comparative compounds with OVER, UNDER and OUT. In
Michael Bernstein (ed.), Proceedings of the Ninth Eastern States Conference of Linguistics
(ESCOL) 1992: 272–81. Ithaca, NY: Cornell University.
Williams, Edwin. 1981. On the notions ‘lexically related’ and ‘head of a word’. Linguistic Inquiry 12:
245–74.
Williams, Theodore. 1965. On the ness peril. American Speech 40: 279–86.
Wordspy. The word lover’s guide to new words. <http://wordspy.com/>, accessed 1 November
2012.
Wright, Joseph. 1905. The English dialect grammar. Oxford: Frowde.
Xu, Fei and Steven Pinker. 1995. Weird past tense forms. Journal of Child Language 22: 531–56.
Xu, Zheng and Mark Aronoff. 2011. A Realization Optimality Theory approach to blocking and
extended morphological exponence. Journal of Linguistics 47: 673–707.
Yoon, James Hye Suk 1996. Nominal gerund phrases in English as phrasal zero derivations.
Linguistics 34(2): 329–56.
Zamma, Hideki. 2003. Suffixes and stress/accent assignment in English and Japanese: A survey. In
Takeru Honma, Masao Okazaki, Toshiyuki Tabata, and Shin-ichi Tanaka (eds), A new century
of phonology and phonological theory: A Festschrift for Professor Shosuke Haraguchi on the occasion
of his sixtieth birthday, 456–69. Tokyo: Kaitakusha.
Zandvoort, R. W. 1962. A handbook of English Grammar. 2nd edn. London: Longmans.
Zimmer, Karl E. 1964. Affixal negation in English and other languages. An investigation of
restricted productivity. Word 20(5): 21–45.
Zirkel, Linda. 2010. Prefix combinations in English: Structural and processing factors. Morphology
20(1): 239–66.
Zwicky, Arnold. 1985. Heads. Journal of Linguistics 21(1): 1–29.
Zwicky, Arnold. 1989. Quicker, more quickly, *quicklier. Yearbook of Morphology 1988: 139–73.
Zwicky, Arnold. 1995. Why English adverbial ly is not inflectional. Papers from the 31st regional
meeting of the Chicago Linguistic Society 31(1): 523–25.

Sources of data

Amman, George A. 1944. Determining the age of pinnated and sharp-tailed grouses. Journal of
Wildlife Management 8: 170–71.
Anderson, Kevin J. and Doug Benson. 1995. Ill wind. New York: Tom Doherty.
Baum, Frank L. 1900. The wizard of Oz. Chicago: George M. Hill.
Block, Lawrence. 2004. The burglar on the prowl. Harpenden: No Exit Press.
Cannell, Stephen J. 2003. Hollywood tough. New York: St Martin’s.
Dekker, Diana. 2007. Obituary for Margaret Neave. New Zealand Medical Journal 23 March 2007,
120(1251), <http://www.nzma.org.nz/journal/120-1251/2475/>, accessed 3 February 2010.
Eskin, Frada. 1995. Values and ideals, isms and ologies: Making contact across a desert of differ-
ences. Public Health 109: 1–5.
 References 665

Hill, Reginald. 2001. Dialogues of the dead. London: HarperCollins.

Kernick, Simon. 2003. The murder exchange. London: Bantam.
Lodge, David. 2001. Thinks. London: Secker & Warburg.
Lovesey, Peter. 1997. Upon a dark night. London: Little, Brown.
McClister, Michael. 1999. Victim’s choice. New York: St Martin’s.
McNab, Andy. 2008. Cross fire. London: Corgi.
Men’s Health. 2001. Foods that you must must eat before you die. <http://thebestlist.menshealth.
com/list/3787>, accessed 23 February 2012.
Meynell, Laurence. 1978. Papersnake. London: Macmillan.
Middleton, Sue. 2006. Research Assessment as a pedagogical device: A Bernsteinian exploration
of its impact on New Zealand’s subject/s of Education. Paper presented at the Australian
Association for Research in Education Conference, Adelaide, 27–29 November 2006 <http://
www.aare.edu.au/06pap/mid06131.pdf>, accessed 1 November 2012.
Newbie. 2001. A painless transition, please. <http://www.mail-archive.com/newbie@linux-man-
drake.com/msg58073.html>, accessed 23 September 2011.
NZPA. 2007. Parrot smuggler fined $20, 000. New Zealand Herald, 9 November 2007. <http://
www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=10475065>, accessed 7 January 2011.
Parker, Barbara. 2000. Suspicion of malice. New York: Dutton.
Rathbone, Julian. 2001. Homage. London: Allison & Busby.
Sandford, John. 2001. Chosen Prey. London: Simon & Schuster.
Strohmeyer, Sarah. 2004. Bubbles a broad. New York: Dutton.
Tolkien, John R. R. 1954. The lord of the rings. Part I: The fellowship of the ring. London: George
Allen & Unwin.
Trevanian. 1998. Incident at Twenty Mile. New York: St Martin.
Vittachi, Nury. 2008. Mr Wong goes west. Crows Nest, NSW: Allen & Unwin.
Wallace, Danny. 2008. Yes man. N.p.: Ebury Press.
Waugh, Alexander. 2008. The house of Wittgenstein. New York: Doubleday. Excerpt from <http:
//www.amazon.com/House-Wittgenstein-Family-War/dp/product-description/0385520603>,
accessed 3 February 2010.
White, Stephen. 2002. Warning signs. London: Little, Brown.
INDEX OF AFFIXES AND OTHER FORMATIVES

a‐ (adjective) 303–6 ‐antN 216, 217–21, 231, 233, 519, 523, 587, 591, 597
a‐ (adverb) 331–2 ant‐ 179, 417 see anti‐
a‐ (negative) 179, 354, 355, 356, 358, 360, 361, 367, ante‐ 333–9, 345, 349–50, 510, 589, 603, 608
368, 369, 377, 378, 379, 510, 515, 516, 589, 603, anti‐ 55, 179, 183, 417, 440, 511, 514, 589, 600, 603,
604, 605, 606, 611 604, 606, 607, 608, 609, 610
a‐ (verb) 267, 268, 282 ‐ar 181, 621 see ‐alA
‐a 386, 392 see ‐erDIMINUTIVE arch‐ 4, 333–9, 344, 345, 348, 513, 589
‐able 33, 124, 167, 178, 181, 186, 221, 222, 246, 247, ‐arian 216, 217, 225–6, 241, 242, 288, 320
248, 254, 289, 290, 291–303, 307–8, 362, 494, ‐aryA 181, 185, 285, 290, 291–303, 309, 311, 494, 510,
501, 502, 512, 515, 542, 574, 579, 587, 591, 594, 587, 591, 596, 597, 599, 610, 613
596, 597, 598, 599, 611, 613, 635 ‐aryN 185
‐ables 551, 636 astro‐ 454
‐acy 510, 587 ‐at‐ 181, 197, 220, 255, 270, 299, 301
ad‐ (Latin prefix) 180 ‐ateA 114–6, 183, 185, 228, 254
after‐ 333–4, 340–4, 345, 350, 513, 514, 585, 603 ‐ateN 4, 276–7
‐age 189, 202, 222, 231, 240, 245, 250–3, 256, ‐ateV 42, 179, 186, 200, 219, 221, 274–7, 280, 284,
262–6, 494, 501, 510, 516, 546, 587, 591, 594, 596, 287, 293, 297, 300, 363, 366, 494, 503, 510, 522,
597, 598, 599, 611, 613 523, 587, 596, 597, 598, 610, 613
‐al‐ 201 ‐ati‐ 301, 526, 530
‐alA 114–6, 163–4, 166, 167, 168, 169, 171, 172, 175, ‐ation 166, 167, 172, 175, 179, 181, 184, 185, 187, 195,
178, 181, 182, 189, 222, 246, 248, 253, 254, 289, 196, 201–2, 213, 221, 231, 236, 239, 240, 253, 264,
290, 291–303, 309, 311, 314–7, 320, 324, 362, 494, 302, 324, 466, 467, 468, 469, 471, 494, 508, 510,
508, 510, 515, 574, 579, 587, 591, 594, 596, 597, 515, 516, 521, 538–40, 541, 574, 577, 587, 591, 597,
598, 599, 611, 613, 621 598, 599, 610, 613
‐alN 33, 164, 165, 167, 189, 192, 195, 196, 198–9, 199, ‐ative 185, 247
204, 213, 239, 240, 242, 466, 467, 468, 469, 510, ‐atory 185
515, 541, 578, 587, 591, 610, 613, 621 atto‐ 426
ambi‐ 427 auto‐ 513
an‐ (negative) 179, 360 see a‐ (negative)
‐anA 114–6, 172, 179, 181, 188, 252, 254, 288, 289,
290, 291–303, 311, 314–7, 320, 510, 516, 579, 587, back‐ 333–4, 340–4, 345, 347, 510, 585, 600,
592, 594, 611, 613, 636 601, 603
‐anN 216, 217, 221–5, 222, 231, 235, 241, 504, 587, be‐ 267, 268, 282, 506, 510, 585, 600, 603, 635
596, 636 bi‐ 179, 426
‐ana 184, 245, 250–3, 262–6 biblio‐ 454
‐ance 195, 196, 197, 199, 204, 231, 240, 510, 515, 523, bin‐ 179 see bi-
587, 591, 598, 599, 610, 613 bio‐ 441, 454
‐ancy 587, 591, 610 see ‐ance ‐bot 303, 526, 528
‐antA 114–6, 178, 220–1, 222, 254, 288, 289, 290, ‐burger 529
291–303, 309, 494, 515, 587, 598, 599, 610 by‐ 333–4, 340–4, 345, 346, 350–1, 510, 585, 600
668 Index of affixes and other formatives

‐c‐ 181, 220, 255, 256, 276 ‐edPAST 66, 67, 69, 76, 471, 505, 537, 556, 557, 575,
‐cation see ‐ation 593, 630, 632
‐cator 220 ‐eeDIMINUTIVE 191
‐ce see ‐ance ‐eeN 38, 166, 184, 189, 213, 216, 217, 221, 226–8, 231,
centi‐ 426 237–8, 254, 388, 467, 468, 495, 512, 515, 579, 587,
‐cide 18, 302, 455 591, 592, 594, 597, 598, 599, 611, 613
circum‐ 333–9, 345, 346, 510, 589, 602, ‐een 386, 399
605, 606, 611 ‐eer 181, 184, 189, 192, 193, 216, 225–6, 231,
cis‐ 333–9, 346, 510, 589, 611 235, 510, 515
co‐ 333–9, 346, 511, 514, 515, 589, 600, 601, 602, electro‐ 454
603, 604, 605, 606, 608, 609 em‐ see en‐
con‐ (Latin prefix) 180 en‐ 180, 200, 267, 268, 282, 510, 515, 589, 602, 603,
contra‐ 333–9, 346, 510, 589, 607, 608, 611 604, 605, 606, 607, 608, 609, 621, 635
counter‐ 333–9, 346, 510, 515, 516, 589, 600, 604, ‐enA 303–6
606, 607, 608 ‐enDIMINUTIVE 399
‐cracy 455 ‐enPAST_PARTICIPLE 66, 71, 73, 75, 76
‐crat 303 ‐enPLURAL 133
crypto‐ 513, 589, 603, 604, 605, 606, 609 ‐enV 192, 193, 497, 510, 594, 595, 596, 610, 613
‐cum‐ 438 ‐ence see ‐ance
‐cy 171, 181, 256 see ‐ance ‐ency see ‐ance
end‐ 180 see endo-
endo‐ 180
de‐ 183, 203, 267, 268, 354, 355, 356, 358, 363, 367, enne‐ 427
369, 376, 380, 440, 501, 503, 513, 515, 543, 577, 589, ‐ent 172, 466 see also ‐ant
602, 603, 604, 605, 606, 607, 608, 609, 622, 635 epi‐ 333–9, 345, 346
dec‐ 427 ‐er‐ 201
deca‐ 427 ‐erCOMPARATIVE 103, 104, 105, 119, 323, 324,
deci‐ 426 522, 546, 633
‐delic 526, 530 ‐erDIMINUTIVE 386, 392, 393–4
demi‐ 426, 510, 589, 604, 605, 608 ‐erN 38, 53, 167, 169, 178, 181, 213, 216, 217–21,
di‐ 179 see dia- 222, 229, 231, 232–3, 238, 239, 240, 241, 254,
di‐ (number) 426 442, 446, 466, 467, 468, 471, 472, 494, 504,
dia‐ 179 510, 513, 515, 516, 519, 546, 570, 573, 580, 585,
dis‐ 183, 267, 354, 355, 356, 358, 361, 362, 367, 371, 591, 592, 593, 594, 595, 611, 613, 614, 621,
372, 373, 375, 376, 380, 440, 510, 515, 516, 577, 630, 633
589, 601, 602, 603, 604, 605, 606, 607, 608, ‐erel see ‐rel
609, 610, 635 ‐ería (Spanish suffix) 240 see ‐teria
‐dom 32, 245, 248–50, 256, 257, 259–61, 324, 495, ‐ern 303–6
510, 513, 514, 515, 516, 546, 585, 591, 592, 593, 595, ‐ery 178, 206, 231, 240, 245, 250–3, 256, 262–6,
611, 613 494, 510, 513, 514, 515, 538–40, 587, 591, 594, 596,
down‐ 267, 333–4, 340–4, 345, 510, 585, 601, 603 597, 598, 611
dys‐ 357 see dis‐ ‐ese 135, 166, 167, 168, 181, 184, 188, 216, 217, 228,
229, 241, 510, 587, 591, 594, 596, 597, 611, 613
‐esque 114–6, 184, 289, 290, 291–303, 311, 313, 320,
‐e‐ 181, 299, 301 416, 510, 514, 515, 516, 573, 587, 592, 594, 596,
‐edA 114, 303–6, 307, 311, 313, 506 599, 611, 613
 Index of affixes and other formatives 669

‐ess 169, 184, 189, 216, 217, 242–3, 494, 496, 510, hyper‐ 183, 267, 404–5, 409–11, 513, 589, 600, 603,
574, 587, 592, 594, 597, 598, 599, 611, 613 605, 607, 609
‐est 103, 105, 323, 324, 522 hypo‐ 180, 421–2, 510, 589, 602, 604, 605,
‐et‐ 299 607, 609
‐et 387, 399
‐ette, 184, 216, 218, 242–3, 386, 395–6, 399, 510,
587, 592, 594, 596, 597, 598, 599, 611, 613 ‐i‐ 181, 182, 219, 252, 276, 297–8, 299, 301
ex‐ 55, 333–9, 347–8, 511, 514, 589, 600, 602, ‐i 216, 217, 228, 241
603, 604, 609 ‐ia 245, 248, 250–3, 262–6
exa‐ 427 ‐ial 174, 320, 599 see ‐alA
exo‐ 180 ‐ian 229, 247, 596, 597, 598, 599 see ‐an
extra‐ 333–9, 345, 346, 347, 351, 513, 514, 589, 602, ‐iard 425
604, 609 ‐ible 290, 621 see also ‐able
‐ey see ‐y -ic- 270
-ic 16, 114–6, 124, 172, 174, 179, 181, 222, 228, 246,
248, 254, 255, 289, 290, 291–303, 311, 314–7,
femto‐ 426 318–20, 326, 362, 494, 496, 508, 510, 515, 523,
‐fold 420, 537, 570, 585, 595, 610 579, 587, 592, 596, 597, 598, 610, 613
for‐ 267, 268, 282 see also fore ‐ical 289, 290, 291–303, 311, 314–7, 318–20, 510,
fore‐ 267, 333–4, 340–4, 345, 349–50, 513, 585, 515, 587, 592, 598, 599, 610
600, 601, 603 ‐ie 191, 386, 388–92, 394, 400, 402
‐free 354, 367, 368 ‐ify 33, 172, 175, 179, 186, 192, 193, 200, 202, 220,
‐fulA 114–6, 167, 289, 290, 303–6, 311, 494, 496, 264, 276, 277, 282–3, 284, 287, 363, 366, 494,
510, 515, 585, 591, 592, 593, 594, 595, 611, 613 510, 515, 516, 523, 564, 573, 577, 587, 592, 598,
‐fulN 418–9, 512, 515 613, 621
il‐ see in‐ (negative)
‐ile 114–6, 185, 254
‐gate 19, 525 ill‐ 354
geo‐ 440, 454 ‐illion 526
giga‐ 427 im‐ see in‐ (negative)
grand‐ 216, 243–4 ‐im 139
graph 303 ‐in‐ 181, 270, 276, 301
‐graphy 455 in‐ (locative) 333–4, 340–4, 345, 346, 351, 510, 515
great‐ 216, 243–4 in‐ (negative) 18, 163–4, 180, 183, 354, 355, 356,
357, 360, 361, 367, 368, 371, 372, 377, 378, 379,
510, 589, 603, 604, 605, 607, 608, 609, 611,
hecto‐ 427 621, 629
hemi‐ 426, 510, 589, 604, 605, 607, 611 ‐ine 254, 289, 290, 291–303, 311, 314–7, 510, 587,
hepta‐ 427 592, 598, 610
hexa‐ 427 ‐ing A 289, 290, 306
‐holic 526, 530 ‐ingDIMINUTIVE 386
‐hood 245, 248–50, 256, 257, 259–61, 495, ‐ingN 195, 196, 202–3, 206, 207, 214, 231, 239,
512, 513, 515, 585, 591, 592, 593, 594, 595, 452, 466, 467, 468, 469, 471, 472, 473, 477,
611, 613 574, 623
hydro‐ 454 ‐ingPARTICIPLE 14, 69, 70, 71, 169, 307, 505, 537,
hyp‐ 180 see hypo‐ 538–40, 541, 555, 556, 557, 593
670 Index of affixes and other formatives

inter‐ 203, 333–9, 498, 513, 589, 601, 602, 603, ‐less 11, 114–6, 167, 169, 289, 354, 355, 357, 359, 363,
604, 605, 606, 607, 608, 609 366, 367, 368, 371, 494, 495, 510, 514, 515, 585,
intra‐ 333–9, 346, 347, 351, 513, 589, 605, 607 591, 592, 593, 595, 596, 611, 613, 621
‐ion 174, 222 see ‐ation ‐let 386–7, 512, 585, 594, 595, 610, 613
ir‐ see in‐ (negative) leuko‐ 485
‐ishA 9, 11, 19, 55, 114–6, 254, 289, 290, 303–6, ‐licious 19, 527, 530
311, 312–3, 324, 416, 421, 510, 513, 514, 515, ‐like 289, 290, 312, 441
516, 573, 585, 591, 592, 593, 594, 595, 596, ‐ling 25, 386, 387–8, 495, 510, 585, 593, 595, 611, 613
611, 613, 621 ‐logy 19, 303, 441, 455, 621
‐ishN 189, 216, 217, 228, 241, 494, 495, 496 ‐lyA 33, 113, 114, 169, 289, 290, 303–6, 311, 314–7,
‐ism 19, 172, 175, 181, 222, 234, 245, 253–5, 256, 494, 510, 546, 559, 573, 578, 585, 592, 593, 594,
266, 321, 494, 510, 514, 515, 523, 587, 594, 596, 595, 596, 611, 613
597, 598, 599, 611, 613, 621 ‐lyADV 104, 107, 163, 165, 167, 168, 171, 172, 183, 322,
‐istA 114–6, 288, 494 323–7, 495, 505, 507, 536, 542, 546, 555, 556, 557,
‐istN 181, 216, 217, 221–5, 231, 234–5, 241, 321, 455, 559, 568, 578
504, 508, 510, 514, 515, 519, 523, 573, 587, 592,
594, 596, 597, 598, 599, 611, 613, 621
‐istic 289, 321, 523 macro‐ 397
‐it‐ 181, 182, 197, 220, 270, 300, 301 mal‐ 354, 355, 357, 358, 364, 367, 368, 510, 515,
‐ite 183, 216, 217, 228, 241, 242, 253, 510, 587, 592, 589, 607, 611
594, 596, 598, 599, 611, 613 ‐man 613
‐iti‐ 301 ‐matic 527, 530
‐ition see ‐ation maxi‐ 405–6, 409–11
‐itis 455 mega‐ 18, 19, 406–7, 409–11, 427, 511, 514, 515, 589,
‐ity 32, 33, 165, 167, 171, 172, 175, 179, 187, 188, 189, 600, 602, 605, 606, 607
206, 245–8, 253, 254, 256, 257, 261, 455, 510, 515, megalo‐ 406
568, 573, 574, 579, 580, 587, 592, 596, 598, 599, ‐meister 216, 217, 225–6, 231, 236
610, 613, 634 ‐ment 32, 34, 195, 196, 198, 199–201, 206, 213, 222,
‐ive 114–6, 175, 181, 222, 246, 254, 289, 290, 231, 239, 240, 254, 324, 466, 467, 468, 510, 515,
291–303, 309, 324, 362, 466, 494, 510, 515, 523, 541, 588, 592, 599, 611, 613, 621
574, 579, 580, 588, 592, 597, 598, 599, 610, 613 meta‐ 333–9, 344, 345, 348, 510, 515, 589, 600, 602,
‐ize 29, 32, 33, 42, 165, 167, 168, 172, 175, 179, 603, 605, 606, 607, 608, 610
181, 182, 183, 184, 186, 189, 200, 202, 222, micro‐ 397, 398, 426, 511, 589, 600, 602, 603, 605,
264, 276, 277, 282–3, 284, 287, 363, 366, 606, 607, 608, 609, 610
494, 501, 508, 510, 515, 516, 523, 564, 573, mid‐ 333–4, 340–4, 345, 501, 511, 514, 585, 600,
577, 588, 592, 594, 596, 597, 598, 599, 611, 601, 603
613, 621 midi‐ 405–6
milli‐ 426, 427
mini‐ 18, 19, 183, 396–7, 405, 498, 511, 589, 600,
kilo‐ 427 605, 606, 609, 610
‐kin 386, 394–5 mis‐ 183, 354, 355, 357, 358, 363–4, 367, 368, 440,
510, 516, 585, 600, 601, 603
‐mo 4
‐l‐ 181, 229 mono‐ 426, 510, 573, 590, 600, 604, 606, 607,
‐leN 386 608, 609
‐leV 4 ‐morph 303, 455
 Index of affixes and other formatives 671

morpho‐ 454 ‐ous 114–6, 169, 181, 228, 246, 247, 248, 289, 290,
multi‐ 427, 511, 515, 590, 600, 603, 604, 605, 606, 291–303, 309, 311, 314–7, 362, 494, 510, 574, 579,
607, 608, 609, 610 588, 593, 594, 596, 597, 599, 611, 613
out‐ 267, 333–4, 340–4, 345, 346, 347, 351, 352–3,
513, 514, 585, 600, 601, 602, 603, 635
‐n‐ 16, 181, 223–4, 229, 298, 299 over‐ 4, 55, 203, 333–4, 340–4, 345, 347, 348–9,
nano‐ 4, 397–9, 426, 510, 590, 606, 609 351–2, 441, 513, 514, 585, 600, 601, 602, 603
naut 303
neo‐ 333–9, 347, 511, 590, 600, 603, 604,
606, 607, 608 pan‐ 427
‐ness 18, 32, 167, 171, 172, 183, 245–8, 256, 257, para‐ 333–9, 345, 346, 510, 590, 600, 601, 602,
259, 261, 323, 494, 495, 496, 501, 511, 514, 604, 605, 606, 607, 608, 609, 610
515, 556, 557, 568, 573, 574, 576, 579, 580, past‐ 55
585, 591, 592, 593, 595, 596, 611, 613, 614, ‐pegs 386, 399
634, 638 penta‐ 427
neuro‐ 18 peri‐ 333–9, 345, 346, 350–1, 510, 590, 601, 605,
‐nik 216, 217, 225–6 607, 609, 610, 611
nomin‐ 18 peta‐ 427
non‐ 55, 354, 355, 356, 357, 358, 362, 365, 367, 370, ‐phage 303
371, 377, 378, 379, 501, 512, 514, 515, 590, 600, ‐phile 303, 455
603, 606, 609 philo‐ 303, 454
nona‐ 427 ‐phobe 303, 454
‐n’t 64, 86, 87 photo‐ 441
‐nym 303 pico‐ 426, 510, 590, 611
poly‐ 427, 510, 590, 600, 602, 603, 604, 607,
608, 609
‐oDIMINUTIVE 386, 392–3, 402 ‐poo(h) 386, 399
‐oSPLINTER 527, 530 ‐pops 386
octa‐ 427 post‐ 4, 183, 333–9, 347, 350, 440, 512, 514, 515,
octo‐ 427 590, 600, 602, 603, 606, 607, 608, 609
‐od‐ 181 pre‐ 4, 14, 18, 183, 333–9, 345, 347, 349–50, 440,
‐odynia 485 498, 512, 514, 590, 601, 602, 603, 604, 605, 606,
off‐ 333–4, 340–4, 345, 346, 510, 585, 600, 602 607, 608, 609, 621
‐oid 185, 285, 289, 290, 291–303, 311, 313, 320, 416 pro‐ 418
‐ola 399, 411 proto‐ 333–9, 347, 513, 515, 590, 601, 603, 607, 608
omni‐ 427 pseudo‐ 414–6, 590, 601, 603, 606, 608
on‐ 333–4, 340–4, 345, 346, 510, 585, 601
‐or 181, 219, 302, 320, 588, 593, 597, 599, 610, 621
see also ‐erN quadr‐ 427
‐orama see ‐rama quasi‐ 415–6
‐ory 172, 185, 289, 290, 291–303, 309, 510, 515, 521, quin‐ 427
588, 593, 597, 598, 599, 610, 613
‐oseA 185
‐oseN 4, 185 ‐rama 19, 411, 527
‐osis 485 re‐ 18, 267, 419–20, 499, 513, 515, 590, 601, 602,
‐ot 399 603, 604, 605, 606, 607, 608, 609, 610
672 Index of affixes and other formatives

‐red 540 sur‐ 333–9, 345, 347, 348–9

‐rel 575 syn‐ 180
retro‐ 333–9, 347, 454, 513, 590, 603, 606, 607,
608, 609
‐ric 4 ‐t‐ 181, 197, 226, 255, 274, 299
‐rific 19, 527, 530 ‐tainment 528
‐ry see ‐ery ‐tarian 526, 529, 530
‐teen 184, 422–3, 425, 507, 537
tele‐ 455
‐s‐ 181 ‐ter 4
‐s3SG 69, 75, 505, 540, 541, 542, 546, 620 tera‐ 427, 513, 590, 606, 611
‐sADV 328, 546 ‐teria 240
‐sDIMINUTIVE 386, 394, 402 tetra‐ 427
‐‘sGENITIVE 53, 144, 506, 521 ‐thN 31, 172, 206, 496, 507, 546, 575,
‐sLINK 624 596, 613
‐sPLURAL 123, 125, 404, 506, 507, 538–40, 546, 551, ‐thORDINAL 425–6, 536–7, 546
555, 556, 561 theo‐ 455
‐‘sPLURAL 128 ‐tic‐ 276
‐sPLURAL_LIKE 124 ‐tion see ‐ation
sarco‐ 485 to‐ 31, 32
‐scape 19, 527 ‐ton 421, 523, 580
‐scope 303, 454 trans‐ 333–9, 346, 513, 590, 601, 603, 605, 606,
‐self 152–3 607, 608, 609, 610, 611
semi‐ 55, 183, 426, 501, 512, 514, 590, 601, tri‐ 426
604, 609 ‐trix 216, 217, 588
sept‐ 427 ‐tronic 528, 530
sex‐ 427 turbo‐ 409
‐ship 245, 248–50, 256, 257, 259–61, 511, 514, 515, ‐tyN 256
585, 591, 592, 593, 594, 595, 596, 611, 613, 614 ‐tyNUMBER 421–2, 425, 537
‐sion see ‐ation
‐someA 114–6, 289, 290, 303–6, 311, 314–7, 420–1,
501, 510, 585, 593, 595, 611, 613 ‐u‐ 181, 182, 276, 297–8, 301
‐someAPPROXIMATOR 420–1 ‐ual see ‐alA
‐someN 420 ultra‐ 409–11, 512, 514, 590, 601, 606,
‐someN /sǝʊm || soʊm/ 302 607, 610
‐stan 528 ump‐ 423
step‐ 34, 216, 243–4, 580 un‐ 180, 203, 267, 306, 354, 355, 356, 357, 358,
‐ster 216, 217, 231, 235–6, 254, 574, 575, 585, 594, 360, 361, 362, 363, 365, 367, 371, 372, 373,
595, 610 374, 375, 376, 377, 378, 379, 380, 382, 512,
sub‐ 333–9, 345, 349, 421, 498, 512, 514, 515, 515, 577, 580, 585, 600, 601, 602, 603, 621,
590, 601, 602, 603, 604, 605, 606, 607, 608, 629, 635
609, 610 under‐ 333–4, 340–4, 345, 347, 349, 351,
super‐ 18, 19, 333–9, 345, 347, 348–9, 385, 407–8, 352, 441, 513, 514, 586, 600, 601,
409–11, 512, 590, 601, 602, 604, 605, 607, 608, 602, 603
609, 610 uni‐ 426, 510, 573, 590, 604, 605, 607, 609,
supra‐ 333–9, 345, 348–9, 510, 515, 590, 609, 611 610, 611
 Index of affixes and other formatives 673

up‐ 267, 340–4, 345, 513, 515, 586, 600, ‐wise 167, 322, 329–31, 586, 591, 592, 593,
601, 603 595, 596, 611
‐ure 174, 175, 181, 195, 196, 197–8, 199, 204, 231, with‐ 31
302, 466, 467, 468, 494, 510, 515, 588, 593, 597,
599, 610
‐ut‐ 300 ‐yA 55, 113, 114, 169, 251, 289, 290, 303–6, 309,
‐ution see ‐ation 311, 314–7, 494, 496, 511, 515, 546, 586, 591,
592, 593, 594, 595, 596, 611, 613
‐yDIMINUTIVE 25, 251 see also ‐ie
‐v‐ 181 ‐yN 175, 206, 219, 245, 251, 254, 255–6, 259, 260,
vice‐ 216, 244, 501 455, 546, 614
viges‐ 427 yotta‐ 427

‐ward / ‐wards 328–9, 512, 514, 586, 592, 596, 611 zetta‐ 427
‐ware 459, 528, 529 ‐zilla 528, 530
‐ways 322, 329 ‐zza 402
 INDEX OF NAMES

Acquaviva, Paolo 123, 125 340, 354, 361, 378, 379, 394, 402, 405, 413, 431,
Adams, Valerie 3, 221, 340, 441, 463, 465, 467, 432, 435, 439, 452, 454, 456, 458, 460, 463, 465,
469, 564 466, 467, 470, 479, 480, 481, 486, 508, 533, 534,
Aikhenvald, Alexandra 7 537, 543, 545, 546, 554, 559, 562, 564, 568, 570,
Ainsworth, Helen 149 572, 574, 577, 579, 580, 581, 618, 623, 624, 625,
Albright, Adam 78, 572, 634 626
Alexiadou, Artemis 195 Bauer, Winifred 559
Algeo, John 371, 372 Baum, Frank L. 394
Allen, Margaret R. 296, 503, 555, 612, 613 Beard, Robert 28–9, 306, 317, 534–5, 564, 622,
Altenberg, Bengt 149 632–3
Amiot, Dany 486 Becker, Thomas 519, 520
Amman, George A. 135 Bell, Melanie 433, 434, 435, 446, 447, 448, 450,
Anderson, Kevin J. 132 525, 634
Anderson, Stephen R. 17, 28, 142, 534, 629, 631 Benson, Doug 132
Anderwald, Lieselotte 71, 72, 74, 76, 78, 571, 572 Berg, Thomas 401, 402, 442, 501, 502
Anshen, Frank 575 Bergen, Benjamin K. 24
Arndt-Lappe, Sabine 447, 459, 460, 462, 519, 525, Biber, Douglas 624
574, 634 see also Lappe Bierwisch, Manfred 104
Aronoff, Mark 16, 34, 177, 182, 285, 413, 503, 565, Birner, Betty 79, 433
568, 569, 574, 575, 576, 579, 580, 583, 611, 612, Bisetto, Antoinetta 465
614, 629, 635, 636, 638 Blevins, James P. 630, 631, 632
Auwera, Johann van der 153 Bloch, Bernard 541, 629
Ayto, John 29, 30, 41 Block, Lawrence 570
Bloomfield, Leonard 444, 575, 579
Bock, Kathryn 122
Baayen, R. Harald 40, 67, 68, 73, 76, 164, 270, Bolinger, Dwight 568
433, 450, 494, 495, 497, 519, 580, 583 Booij, Geert 103, 117, 233, 474, 534, 629, 630,
Baeskow, Heike 454, 456 632, 633–4
Baker, Mark C. 195, 322 Börger, Claudia 333
Banga, Arina 272 Bosch, Antal van den 634
Bardsley, Dianne 389, 393 Botha, Rudolf P. 639
Barker, Chris 26, 237, 239 Boyé, Gilles 22
Barnhart, Clarence L. 41 Bram, Barli 279, 548
Barnhart, Robert K. 41 Braun, Maria 117, 439, 446, 447, 450, 471, 634
Barz, Irmhild 621, 624 Brekle, Herbert E. 463, 474
Bat-El, Outi 460 Bresnan, Joan 351, 489, 638
Bauer, Ingrid 394 Broeders, A. 447
Bauer, Laurie 7, 18, 28, 32, 41, 110, 116, 122, 134, Brousseau, Anne-Marie 624
169, 202, 226, 229, 235, 279, 281, 298, 308, 333, Browne, Allen C. 485–6, 487
 Index of names 675

Bulloch, John Malcolm 389 Downing, Pamela 474

Burnard, Lou 41 Dressler, Wolfgang U. 22, 28, 190, 460, 533, 623
Burzio, Luigi 187, 496, 520, 613
Butterfield, Sally 122
Bybee, Joan 68, 78, 322, 497, 571, 572 Earle, John 36
Eberhard, Kathleen M. 122
Elzinga, Dirk 634
Cannell, Stephen J. 411 Embick, David 110
Cannon, Garland 389, 460 Emonds, Joseph 322
Carney, Edward 46 Epps, Patience 624, 625
Carnie, Andrew 38, 153 Erdmann, Peter 451, 452, 453
Carstairs-McCarthy, Andrew 3, 12, 142 Eskin, Frada 19
Chambers Dictionary 55, 450
Chapman, Don 634
Cheshire, Jenny 71, 575 Fabb, Nigel 493, 494
Chomsky, Noam 171, 182, 195, 296, 444, 554–5, Faiss, Klaus 445
629, 637 Fant, Gunnar M. 166
Clark, Eve V. 285, 565 Farnetani, Edda 444
Clark, Herbert H. 285, 565 Farrell, Patrick 565
Coates, Jennifer 80, 81 Filppula, Markku 151
Cohen, Evan-Gary 460 Firth, John Rupert 24
Collie, Sarah 171, 188 Fischer, Olga 157
Collins English Dictionary 450 Fleischer, Wolfgang 612, 621, 624
Comrie, Bernard 617–8 Fraser, Bruce 351
Concise Oxford Dictionary 12, 573 Fudge, Erik 446
Cook, Paul 460 Fuhrhop, Nanna 583, 611, 612, 614, 638
Corbin, Danielle 578 Funk, Wolf-Peter 381
Cosi, Piero 444
Covington, Michael A. 376
Croft, William 27 Gagné, Christina 476, 525
Culpeper, Jonathan 113, 114 Ghomeshi, Jila 457, 490
Cutting, J. Cooper 122 Giegerich, Heinz J. 19, 168, 170, 296, 322, 323–4,
Cutler, Anne 122 435, 446, 447, 449, 456, 536, 583, 612, 613–4
Glowka, A. Wayne 411
Gnutzmann, Claus 118
Daelemans, Walter 634 Goldsmith, John A. 11
Dal, Georgette 486 Górska, Elzbieta 368
Dalton-Puffer, Christiane 289, 330, 331, 340, Gray, Bethany 624
354, 418 Green, Jonathon 41
Davies, Mark 40, 41 Green, Lisa 62, 71
Davis, Stuart 461 Greenbaum, Sidney 26, 27, 63, 66, 72–8, 80, 103,
Dekker, Diana 135 108, 109, 110, 124, 134, 141, 142, 148, 151, 394,
Derwing, Bruce L. 634 438, 439
Di Sciullo, Anna-Maria 12, 155, 157, 543, 629, 638 Gries, Stefan Th. 318, 458, 459, 460, 462
Dixon, Robert M. W. 7, 110, 162, 381, 383 Grimm, Jacob 624
Dollinger, Stefan 476 Grimshaw, Jane B. 195, 207
676 Index of names

Guevara, Emiliano 503 Jackendoff, Ray 322, 452, 457, 463, 470, 474, 490,
Gulikers, Leon 41, 450 536, 557
Gussenhoven, Carlos 445, 447 Jakobson, Roman 166
Gussmann, Edmund 278 Jespersen, Otto 114, 140, 280, 354, 360, 361, 364,
372, 381, 463, 467, 470
Johansson, Stig 122
Hale, Kenneth 565 Johnson, Gary 41
Halle, Morris 166, 171, 182, 296, 444, 554–5, 629, 637 Jones, Daniel 188
Halliday, Michael A. K. 19 Jurafsky, Daniel 385
Hamawand, Zeki 306, 316–7, 354, 378
Hammer, Alfred Edward 623
Hammond, Michael 192 Kaisse, Ellen M. 86, 162, 163
Hanssen, Esther 272 Kastovsky, Dieter 617, 619, 620, 622, 625
Harley, Heidi 3, 554, 629 Katamba, Francis 76, 439
Harnisch, Rüdiger 619 Kaunisto, Mark 306, 318
Harris, Zellig S. 629 Kernick, Simon 56
Haspelmath, Martin 28–9, 103, 323, 439, 474, 536, Kettemann, Bernhard 186
538, 543, 555, 618, 630 Keuleers, Emmanuel 78, 572, 581, 634
Hatcher, Anna G. 463, 474 Keyser, Samuel Jay 565
Hawkins, John 148 Killgariff, Adam 67
Hay, Jennifer 164, 493, 494, 495, 496, 497, 581, Kim, John J. 68
583, 612, 638 Kingdon, Roger 446
Hayes, Bruce 78, 184, 296, 572, 634 Kiparsky, Paul 110, 117, 205, 296, 509, 571, 579, 612,
Heath, Jeffrey 624 636, 637
Heyvaert, Liesbet 195 Kjellmer, Göran 377, 381, 382
Hilpert, Martin 110–14, 120 Klinge, Alex 155, 157
Hill, Reginald 569 Knowles, Elizabeth 41
Hinrichs, Lars 147 König, Ekkehard 557
Hockett, Charles Francis 158, 541, 629 Kortmann, Bernd 557
Hohenhaus, Peter 457, 490 Körtvélyessy, Lívia 622
Holmes, Janet 41 Koshiishi, Tetsuya 23, 289
Horn, Laurence R. 354, 363, 364, 365 , 372, 373 , Kösling, Kristina 449, 451
374, 375, 376, 377, 379, 381, 382 Kreyer, Rolf 146, 147
Huddleston, Rodney 3, 26, 27, 63, 65, 66, 72, 73, Kruisinga, Etsko 152
76, 79, 80, 83, 84, 85, 88, 103, 105, 107, 108, 110, Kubozono, Haruo 459, 460
111, 122, 123, 124, 141, 142, 143, 150, 151, 153, 154, Kunter, Gero 117, 439, 444, 445, 446, 447,
214, 298, 308, 317, 322, 323, 326, 327, 332, 354, 450, 471, 634
361, 378, 379, 413, 431, 432, 433, 436, 466, 467, Kuryłowicz, Jerzy 113, 322
470, 554, 560 Kytö, Merja 110, 119
Hudson, Richard 89, 150
Humboldt, Wilhelm von 616, 617
Humphreys, Karin R. 148 Labov, William 575
Lapointe, Steven G. 509, 638
Lappe, Sabine 26, 117, 190, 296, 389, 391, 393, 401,
Ilson, Robert 448 403, 439, 446, 447, 450, 460, 471, 634 see also
Ingram, John 444 Arndt-Lappe
 Index of names 677

Lass, Roger 166 Mayerthaler, Willi 581

Lee, Tanya H. 455 McClelland, James L. 581
Leech, Geoffrey 26, 27, 63, 66, 72–8, 80, 103, 108, McClister Michael 411
109, 110, 113, 114, 124, 134, 141, 142, 148, 151, 394, McCray, Alexa T. 485–6, 487
438, 439 Mchombo, Sam 489, 638
Lees, Robert B. 195, 463, 467, 474 McKoon, Gail 433
Lefebvre, Claire 624 McMillan, James B. 26, 413, 414
Lehnert, Martin 41 McNab, Andy 132
Lehrer, Adrienne 333, 340 Melloni, Chiara 212
Levi, Judith N. 435, 463, 467, 474 Merlini Barbaresi, Lavinia 190
Levin, Beth 233, 239 Meynell, Laurence 56, 489
Liberman, Mark 184, 185, 186, 187, 446, 449 Meys, Willem J. 12, 30
Lieber, Rochelle 28, 30, 233, 259, 264, 270, Middleton, Sue 294
282, 283, 285, 351, 354, 419, 433, 434, 441, 446, Miner, Kenneth L. 171
457, 463, 465, 466, 467, 470, 471, 539, 543, Mish, Frederick C. 41
546, 564, 580, 625, 629, 638, 639 Moder, Carol L. 78, 571, 572
Lindquist, Hans 118 Mohanan, Karuvannur Puthanveettil 162, 296
Ljung, Magnus 306, 314 Mondorf, Britta 110–19
Löbner, Sebastian 38, 472 Montgomery, Michael 114
Lodge, David 395 Moore, Dorothy L. 485–6, 487
Lojenga, Constance K. 624 Moscoso del Prado, Martin F. 67, 68, 73, 76
Longman Dictionary of Contemporary Myers, Scott 566–7
English 41
Lonsdale, Deryle 634
Lovesey, Peter 570 Nagano, Akiko 278, 279, 280, 281, 286, 287,
Lowth, Robert 74 622, 635
Lyons, John 322, 433 Nation, I.S. Paul 569
Neijt, Anneke 272
Nelson, Nicole 403
Macklin, Thomas 376 Nevalainen, Terttu 576
Macquarie Dictionary 55 Nguyên, T. AnThu’ 444
Mailhammer, Robert 621 Nichols, Johanna 618–9
Manova, Stela 623 Nida, Eugene A. 629, 631
Marantz, Alec 629 Nixon, Richard 525
Marchand, Hans 3, 28, 40, 206, 234, 246, 250, Noonan, Michael 625
274, 278, 279, 289, 295, 318, 321, 340, 354, 365, Nosofsky, Robert M. 634
372, 374, 381, 399, 402, 440, 441, 445, 452, Noyer, Rolf 554, 629
460, 463, 467, 470, 473–4, 504, 506, 520, 543, NZPA 566
575, 583
Marcus, Gary F. 78
Marle, Jaap van 579 Ohala, John J. 174
Marr, Vivian 450 Olsen, Susan 444, 446, 479, 625
Marsden, Peter H. 318 Orr, John 575
Marx, Groucho 565 Orton, Harold 151
Matthews, Peter H. 28, 629 Ostendorf, Mari 41, 432, 450
Matthewson, Lisa C. 412 Oxford Advanced Learner’s Dictionary 41
678 Index of names

Oxford English Dictionary, The 12, 16, 30, 40, Pustejovsky, James 536, 557
43, 55, 133, 199, 200, 202, 219, 220, 221, 225, 228, Python, Monty 348
232, 237, 240, 247, 250, 252, 255, 266, 268, 270,
271, 274, 276, 277, 278, 281, 284, 285, 286, 302,
324, 325, 329, 332, 350, 358, 359, 361, 363, 371, Quinn, Heidi 151
372, 387, 388, 393, 396, 397, 399, 411, 415, 419, Quirk, Randolph 26, 27, 63, 66, 72–8, 80, 103,
421, 438, 452, 506, 522, 523, 524, 548, 569, 570, 108, 109, 110, 124, 134, 141, 142, 148, 151, 394,
574, 576, 580, 583, 614 438, 439

Palmer, Frank 72, 73, 76, 80, 84, 85, 88, Radford, Andrew 322
108, 111 Raffelsiefen, Renate 11, 164, 167, 170, 178, 187, 193,
Panini 636 251, 272, 273, 274, 520, 613
Parker, Barbara 563 Rainer, Franz 576–7
Parkinson, Dilworth B. 634 Ramscar, Michael J. A. 68
Pater, Joe 188, 193 Ramson, William Stanley (Bill) 389
Paul, Hermann 575 Randle, William 411
Pawley, Andrew 150 Rappaport Hovav, Malka 233, 239
Payne, John 122, 123, 124, 141, 142, 143, 150, 151, Rathbone, Julian 132
153, 154, 322, 323, 432, 433, 560 Reischer, Jürgen 459, 460
Pensalfini, J. Rob 444 Renner, Vincent 439, 480
Peters, Pam 110 Renouf, Antoinette 470
Peterson, Peter 63, 153 Rice, Keren 497, 542
Piepenbrock, Richard 41, 450 Rickford, John 62
Pietsch, Lukas 62 Riddle, Elizabeth 257–8
Piñeros, Carlos Eduardo 403 Robins, Robert H. 629
Pinker, Steven 68, 78, 131, 443, 581 Roeper, Thomas 467
Plag, Ingo 3, 30, 33, 117, 164, 174, 186, 189, 192, 193, Rohdenburg, Günter 119
198, 251, 270, 272, 275, 276, 277, 278, 282, 283, Romaine, Suzanne 110, 119
284, 285, 287, 289, 296, 317, 318, 322, 324, 330, Rosen, Nicole 457, 490
331, 340, 354, 413, 418, 434, 436, 439, 440, 441, Rosenbach, Anette 147, 149, 436, 437
444, 446, 447, 448, 449, 450, 451, 459, 462, Rubach, Jerzy 117
463, 465, 470, 471, 493, 494, 495, 496, 497, 502, Rumelhart, David E. 581
503, 519, 525, 564, 573, 579, 580, 583, 612, 625, Rusiecki, Jan 104
632, 634, 636, 638 Russell, Kevin 457, 490
Plank, Frans 28–9, 274, 533, 538, 540 Ryder, Mary Ellen 240, 463, 467, 474
Poser, William J. 110 Ryle, Anne 455
Postal, Paul M. 639
Poutsma, Hendrik 152
Prasada, Sandeep 68, 78 Sanderson, Stewart 151
Prćić, Tvrtko 486 Sandford, John 132
Price, Patti 41, 432, 450 Sapir, Edward 616, 617
Prince, Alan 184, 185, 186, 187, 449, 581 Scalise, Sergio 28, 365, 533, 579, 636, 638
Pullum, Geoffrey 3, 26, 27, 63, 72, 73, 76, 84, 85, Schlücker, Barbara 625
88, 90, 103, 105, 107, 108, 110, 111, 153, 195, 214, Schneider, Klaus P. 386, 389, 394
317, 322, 323, 326, 327, 436, 560 Schönefeld, Doris 567
 Index of names 679

Schreuder, Robert 272, 433, 519 Tolkein, John R.R. 423

Selkirk, Elisabeth, O. 117, 467, 469, 613, Townsend, Charles E. 612
629, 638 Trevanian 54
Sepp, Mary 117, 446, 450 Trips, Carola 259
Shannon, Claude E. 448 Trudgill, Peter 69
Shattuck-Hufnagel, Stefanie 41, 432, 450 Tulloch, Sarah 41
Shoben, Edward J. 476, 525
Siegel, Dorothy C. 296, 509, 612
Siegel, Muffy E. A. 467 Ullmann, Stephen 575
Siemund, Peter 150 Urdang, Laurence 455
Silva, Penny 389
Simpson, Jane 389, 393
Sinclair, John M. 270 Vaan, Laura de 433
Skousen, Royal 581, 634 Valera Hernández, Salvador 560 see also Valera
Slobin, Dan I. 68 Valera, Salvador 564, 622 see also Valera
Sloot, Ko van der 634 Hernández,
Slotkin, Alan R. 368 Vendler, Zeno 375
Smith, Carlotta 419 Vine, Bernadette 41
Southerland, Ronald H. 333 Vittachi, Nury 55, 56
Spencer, Andrew 435 Vogel, Irene 11
Sproat, Richard 433, 446, 639
Stanford, Thereon 634
Steinmetz, Sol 41 Wagner, Thomas 78
Štekauer, Pavol 434, 446, 476, 554, 622, 624 Wälchli, Bernard 482, 625
Stemberger, Joseph Paul 145 Wales, Katie 62, 151
Steriade, Donca 187, 520 Wallace, Danny 524
Stevenson, Suzanne 460 Ward, Gregory 79, 433, 639
Stonham, John 439 Waugh, Alexander 135
Strang, Barbara M.H. 389 Weber, Andrea 179
Strauss, Steven 564 Webster, Joy 118
Strohmeyer, Sarah 570 Webster’s Third New International Dictionary
Stump, Gregory T. 533, 628, 629, 631, 632 55, 133
Sturtevant Edgar H. 575 Wells, John C. 45, 171, 339, 344, 360, 423
Styne, Jule 132 Wentworth, Harold 31–2, 151
Sussex, Roland 389, 393 White, Stephen 54
Svartvik, Jan 26, 27, 63, 66, 72–8, 80, 103, 108, Widdowson, John 151
109, 110, 124, 134, 141, 142, 148, 151, 394, Wiese, Richard 171, 489
438, 439 Williams, Darrell 333, 351
Swart, Henriëtte de 365 Williams, Edwin 12, 28, 443, 543, 629, 635, 638
Sweet, Henry 135, 140, 148, 448, 562 Williams, Theodore 246
Szmrecsanyi, Benedikt 110, 119, 147, 149 Wright, Joseph 151
Szymanek, Bogdan 182 Wurff, Wim van der 157

Tarasova, Elizaveta 624 Xu, Fei 78

Torsello, Carol Taylor 444 Xu, Zheng 636
680 Index of names

Yoon, James Hye Suk 195 Zavrel, Jakub 634

Zimmer, Karl E. 354, 361, 362, 364, 372,
381, 382
Zamma, Hideki 186 Zirkel, Linda 494, 497–8
Zandvoort R.W. 389 Zwicky, Arnold 322, 324, 437
 INDEX OF SUBJECTS

ablaut 21, 22, 32, 66, 412, 521, 541, 621, 630, 631; analogy 71, 78, 281, 421, 445, 447, 454, 503, 519,
classical 177 520, 521, 529, 557, 573, 621, 629, 633–5
accidence 533 see inflection Analogical Model of Language 634
acronym 25 analysability 35, 358, 581, 615
adjective 27, 322; as base 197, 198, 201, 221, 222, analysis 640
225, 226, 245, 249, 251, 256, 257, 268, 269, 275, anapest 301
276, 278, 284, 290, 294, 304, 314, 328, 334–7, anaphoric island 433, 639
340–3, 348, 350, 351, 353, 355, 356–7, 362, 367, anaphoric one 433
371, 377, 387, 391, 396, 397, 402, 404–5, 406, anchoring 391, 401, 402, 403
407, 409, 414, 415, 417, 421, 579; ascriptive 317, antonym 362, 372, 381–3
447 see also qualitative; associative 317, 447 apophony 20; see ablaut, consonant alternation,
see also relational; attributive 118; stress shift, umlaut
collateral 23, 288; colour 381, 383; comparison apposition 434
of 103–20; compound 113, 448; derived 113, 216, apostrophe 10, 128; greengrocer’s 129
288–321; directional 328; gradable Arabic 140
288, 306, 443, 559 see also gradability; inflection argument 36–7, 482
of 103–20; predicative 118; qualitative 104, Argument Linking Principle 638
288, 306; relational 104, 288, 306, 348, 435, 476 argument structure 351–3, 445
adjunct 36–8, 482 ascriptive see adjective
adverb 27, 38, 306, 322, 559–60; as base 218, 254, assimilation 15, 89, 163, 172, 496; obligatory vs.
278, 415; derived 322–32; directional 328; optional 179–80
manner/dimension 330 associative see adjective
affix 17–9; cohering and non-cohering 162, 163, atomicity, syntactic 432
169, 615; combination of 493–508, 540, 542, Atomicity Thesis 638
583–615 attitude 385–427 see also evaluation
affixoid 354, 440–1 augmentative 385, 404–11, 535, 622
agglutination 616, 617–8 auto-stress 168, 183–4, 226, 227, 252, 395, 422,
agreement, subject-verb 62, 122, 124 494, 515
alliteration 412 auxiliary see verb
allomorph 14 availability 32, 34, 562, 568, 577, 578, 580
allomorphy 164–82, 305, 339, 427, 556, 637; in
affixes 16, 45, 163, 164, 178–80, 197, 268,
296–302, 344, 359–61; in bases 16, 67, 71, 72, back-derivation see back-formation
129, 164, 165–77, 271, 274, 281, 296–302, 344, back-formation 20, 72, 275, 280–1, 284, 286, 287,
423, 425, 520 398, 442, 452, 519, 521, 525, 617
allophone 323 bahuvrihi see compound
American structuralist tradition 629 base 17–8, 391, 431, 441; adjectival see adjective, as
amphibrach 301 base and so on; allomorphy see allomorphy,
682 Index of subjects

base (cont.) circumfix 503, 621

base; bound (obligatorily bound) 42, 198, 201, citation form 8, 65
219, 221, 222, 224, 225, 246, 251, 268, 276, 290, classical language 167, 179, 619 see also Greek,
293, 299, 304, 334, 337, 342, 355, 357, 358, 359, Latin
360, 395, 402, 405, 415, 441; complex 225, 226, classifying 437
227, 245, 382, 387, 418; form of a verb 65; free clipping 25, 190–2, 267, 281, 287, 386, 396, 400–4,
(potentially free) 42, 219; indeterminate 523; 408, 418, 431, 618, 630, 631; back~ and fore~
phrasal 218, 223, 225, 246, 250, 252, 255, 270, 402; complex 458
278, 287, 290, 304, 305, 306, 330, 334, 337–8, 342, clitic 9, 84–6, 142
363, 391, 393, 396, 409, 437, 509–17, 639; see also Cobuild 41, 270, 377
compound base, split base code-switching 139, 140, 571
binary branching 443, 503, 624 coercion see type coercion
bipartite 122 Cognitive Grammar 565
blend 267, 458–62, 463, 483–5, 525, 630, 631; coinage 30
attributive 483; appositive 483–4; Collins English Dictionary 450
compromise 483–4; coordinative combining form 303, 405, 406, 417, 441, 455, 584
460, 483 see also neoclassical word-formation
blocking 568–82, 636; by homonymy 575, 576; comparative and comparison 103–4, 105–6,
by synonymy 34, 110, 287, 324, 326, 377, 381, 107–20, 317, 324, 507, 522, 559, 638; of adverbs
504, 541, 575, 576; type versus token 577 107–9; double 106–7
borrowing 23, 140, 194, 201, 223, 229, 240, 293, competition among patterns 33, 77, 204, 259–61,
357, 358, 394, 399, 521, 540, 583, 621 287, 311–3, 315–6, 319–20, 330, 377–81, 393, 406,
Boston University Radio Speech Corpus 409–11, 416, 519, 548, 562, 568–82; in
(BURSC) 40, 432, 450, 471, 473 derivation 573–4; in inflection 571–3; see also
bound see base, morph, root synonymy
Bracket Erasure Convention 637–8 complementary distribution 14, 33, 269, 287, 322,
British National Corpus (BNC) 33, 34, 40, 323, 541
43, 45, 67, 79, 80, 86, 87, 88, 119, 130, 138, complementizer 153
156, 199, 208, 216–44, 252, 268, 270, 272, Complexity Principle 119
283, 311, 325, 326, 328, 329, 331, 332, 386, 394, complexity-based ordering 496, 502
395, 396, 402, 409, 418, 421, 441, 470, 477, compositionality 30, 463–4, 556
479, 499, 542, 549, 550, 556, 557, 558, 559, compound and compounding 12, 24, 72, 289,
565, 576, 581 354, 414, 429–90, 506, 524–5, 529, 571, 618,
623–7; additive coordinate 480; adjectival 56,
57, 116, 448, 453, 472; adjective-noun 434–6,
cardinal 422–5 445, 448, 451, 452, 476, 626; adjective-
case 620 see also genitive preposition 454; adverb-adjective 438; affixal
category; canonical 28, 533, 537; change see argumental 467; appositional 56, 141, 479;
word-class; classical 533; morphological 19, argumental 446, 448, 453, 463, 465, 466–74,
519, 521, 541; preservation see word-class; 484, 625; attributive 446, 453, 466, 474–9, 489,
see also word-class 624; bahuvrihi 465, 478; bases 218, 223, 225,
CELEX 41, 73, 450, 471, 495 246, 252, 254, 270, 278, 281, 287, 295, 304, 305,
Chambers Dictionary, The 55, 450 326, 330, 334, 337–8, 342, 363, 395, 409, 415, 413,
change, linguistic 149, 153, 194, 205, 556, 571, 572, 419, 509–17; canonical 442, 443, 451, 618;
574–5, 625, 627 clipped 458; compromise coordinative 480;
child-language 571 conjunctive 481; coordinate 423, 436, 439,
 Index of subjects 683

443, 444, 446, 447, 453, 466, 474, 479–82, 489, 203–4, 275, 277, 286, 287, 452, 477, 564, 623;
624, 625; co-participant 56, 481; derivative- preposition and adverb 332, 552; verb to
particle 56; determinative 154; deverbal 467; noun 208, 214, 231, 236, 239, 240, 454, 465,
disjunctive 481; dvandva 482; endocentric 435, 466, 473, 577
452, 465, 474–8, 489, 490; exocentric 131, 133, coordination 434, 437
423, 435, 438, 465, 473–4, 478–9, 625; corpus 3, 40–2, 277, 361, 363, 396, 445, 498, 516,
genuine 623; identical constituent 457; 569, 570, 580, 642
left-branching 449; neoclassical see Corpus of Contemporary American English
neoclassical word-formation; nominal 451–2; (COCA) 33, 40, 43, 45, 67, 76, 77, 79, 80, 83,
non-affixal (de)verbal 466, 470; 86, 88, 90, 104, 106, 107, 110, 111, 112, 114, 118,
non-argumental 463, 465, 474; 122, 133, 141, 145, 149, 156, 180, 192, 195–215,
non-canonical 154, 438, 442, 451, 452, 453, 525, 216–44, 245, 247, 251, 252, 253, 257, 268, 270,
618; noun-noun 431–4, 445, 452; noun- 271, 272, 273, 275, 277, 279, 280, 288–321, 325,
preposition 453; phrasal 56, 437, 451, 456–7, 326, 328, 329, 330, 331, 332, 334–9, 348, 350, 353,
463, 488; possessive see bahuvrihi; 354–84, 387, 388, 395, 396, 397, 399, 402, 405,
prepositional 453; primary 466; 406, 407, 408, 409, 413, 414, 415, 417, 418, 419,
reduplicative 457, 463; rhyming 413; 420, 421, 433, 434, 436, 437, 438, 439, 441, 443,
right-branching 449; root 466; spelling 444, 451, 453, 457, 468, 470, 475, 476, 478, 488,
of 55–6, 445, 451 see also concatenation, 489, 490, 494–5, 497, 498, 499, 501, 506, 507,
hyphen, solid, spaced; synthetic 442, 466, 508, 509–17, 523, 524, 525, 529, 536, 539, 542,
467, 625; title 140; translative 56, 481; 551, 556, 561, 566, 570, 575, 576, 577, 580,
triconstituent 449, 451; verbal 56, 448, 451, 583–615, 637
452, 525, 625 see also synthetic; verb-noun Corpus of Historical American English
56, 448, 452; verb-preposition 454 (COHA) 40
compound stress 434, 444–9, 451, 525 countability 214–5 see also noun, count
computation 641 counterfactual 62, 70
concatenation in spelling 432, 434, 446, 516
conjugation class 621
conjunction 153, 155 dactyl 112, 165, 186, 193, 252, 256, 271, 276, 301
connotation 576 see evaluation Danish 616, 624, 626
consonant; alternation 21, 163, 172–6, 404; date 425
cluster 113; double and doubling 48, 49–51, declension class 620, 621
69, 125; mutation 630 default 573, 574
constituent family 447, 450, 453 degemination 168–9
constraint-based approaches 578 degree 103–5
Constructionism 503, 629, 632, 633, 634 deixis 155, 156
constructional model 629 demonstrative 153
contextual 285, 565–6 dependent-marking languages 618–9
contradictory see negative derivation 28, 72, 159–427, 431, 522, 547, 548,
contrary see negative 621–2 see also inflection; et passim
contrast 16, 458, 541 determinans and determinatum 543 see head
conversion 9, 27, 45, 213, 288, 341, 431, 442, 493, determiner 107, 153, 225; determiner phrase 457
503–5, 538, 541, 545–67, 574, 581, 617, 618, 622–3, devoicing see voicing
636; adjective and adverb 332, 559–60; diachrony see change, linguistic
adjective and noun 216, 549–51; adjective and dictionary 12, 203, 381, 445, 471, 488, 537, 569,
verb 275, 452; noun to verb 570, 580
684 Index of subjects

diminutive 25, 243, 385–404, 535, 579, 622 disparagement, pejoration) 135, 366, 368,
diphthong 47, 183 371, 372, 378, 379, 385, 390, 395, 409, 413, 416,
distal 152, 153, 155 426, 626; neutral 371, 381, 396, 409, 417–8;
Distributed Morphology 554–5, 629 positive (appreciation, endearment) 381, 385,
ditransitivity 37 390, 407, 409
domain 32–3, 34, 572, 579, 581 event, simple vs complex 207
doubt 66 excess 404, 407 see also evaluation
Dutch 165, 388, 389, 394, 616, 621, 622, 634 expletive insertion 9, 26, 55, 191–2, 413–4
dvandva see compound exponence, multiple 631
extender 16, 23, 45, 165, 181–2, 197, 201, 219, 223–4,
229, 255, 256, 270, 273, 276, 297, 298, 299, 300,
elision 223, 230, 255, 269, 273 301, 314, 425, 427
ellipsis 403, 479, 486, 536, 550, 558 extragrammaticality 460
Elsewhere 129, 636–7
English; African-American 62, 71;
Appalachian 114; Australian 12, 51, 127, 136, familiarity 385 see also item-familiarity
389–91, 392, 400, 402, 403, 552, 566, 572; feature 10, 631
British 12, 26, 31, 45, 49, 52, 54, 80, 86, 88, figurative language 558, 567, 571 see also
119, 122, 131, 134, 149, 166, 204, 205, 218, 225, metaphor, metonymy
247, 276, 296, 299, 329, 360, 390, 394, 402, Finnish 618, 625
403, 404, 412, 426, 445, 448, 478, 540, 551, First Order Projection Condition 638
552, 559; Canadian 51, 572; Caribbean 62, First Sister Principle 638
184; dialectal 62, 71, 81, 86, 90, 130, 149, 151, flectional see fusion
152, 153, 168, 445, 571; Early Modern 621; Fongbe 624
Irish 62, 63, 88, 151, 184, 399; Middle 130, foot 168
133, 156, 553, 621; New Zealand 51, 63, 134, formative 16, 181, 281, 440
136, 140, 154, 389–91, 392, 394, 400, 403, 552, fraction 425–6
572; non-standard 78, 90, 136, 137, 559, 571; free, form 219; morph 14; variation 323
North American 45, 49, 54, 80, 87, 89, 90, French 35, 48, 130, 139, 140, 157, 175, 256,
114, 119, 122, 149, 156, 204, 218, 225, 236, 299, 293, 357, 387, 395, 403, 426, 439, 474,
329, 360, 390, 401, 403, 413, 445, 448, 478, 545, 575, 583, 616, 619, 620, 621, 622–3, 624, 625
479, 541, 559, 572; Old 259, 268, 357, 617, 619; frequency 34, 67, 74, 118, 130, 202, 204, 281, 319,
poetic 130; Received Pronunciation 344, 368, 435, 446, 448, 451, 457, 488, 496, 548,
(RP) 48; Scottish 63, 64, 88, 95, 129, 153, 557, 571, 580
156, 389, 392, 399, 423; South African Friesian 616
51, 389–90, 392; Southern US 71; standard 90, functional shift 567
129, 132, 151, 152, 153, 571, 572; et passim fusion 533, 616, 617–8; index of 617
epenthesis 15, 106, 165, 170, 171, 172, 425
established 199, 351, 421, 464, 580
etymology 16, 33, 48, 130, 137, 167, 171, 174, geminate 15, 87, 163, 168–9
177, 196, 332, 387, 388, 426, 434, 498, gendered terms 216, 217, 230, 241, 242–3,
514, 515, 521, 541, 545, 560, 579, 583, 611, 396, 574
612, 613, 615 Generalized Context Model 634
euphemism 413 Generalized Lexicalist Hypothesis 638
evaluation 231, 235, 236, 243, 366, 385–427, 391, genitive 103, 119, 141–9, 324, 506, 617, 620;
411, 535, 622; negative (depreciation, descriptive 436; determiner 436; group 141;
 Index of subjects 685

head 141; nested 149; oblique 142; of 142, imperative 62, 64, 79
146–9, 521; partitive 123, 146; phrasal 141; Indo-European 535
<’s> 141, 146–9, 521; semantics of 142–3 infinitival to 88–90
German 140, 165, 389, 612, 619, 620, 621, 622, 624, infinitive 62, 64, 79
625, 626; Old High 619 infix 630
Germanic 35, 131, 388, 434, 435, 456, 614, 616, 619, inflection 59–158, 436, 493, 505–7, 519, 521–2, 566,
620, 622–3, 623–5, 631; North 35; West 35, 619–21; class-changing 555; contextual 534–5,
616, 621, 632 539; inherent 103, 534–5, 539; semantics
Google 499, 551, 570, 584 in 68; stem-based 619; vs. derivation 8, 28–9,
Google Book Corpus 40, 132, 135 322–3, 440, 505–6, 533–44; word-based 619; et
gradability 103–5, 365, 368, 378 passim
grammar; generative 565; traditional 550 inflectional; base 19, 29; language see fusion
grammaticalization 259, 440 informativity 445, 447, 448
Greek 35, 48, 136–7, 177, 182, 256, 295, 357, 397, initialism 25, 396
426, 441, 456, 498, 583, 616, 627 institutionalization 30, 437, 489, 576, 577
internal modification 621, 630 see ablaut,
apophony, consonant-alternation, stress-shift,
habitual 62 vowel-alternation, umlaut
hapax 580, 581 interrogativity 156
haplology 112–13, 145, 189–90, 197, 230, 251, 273, intransitivity 36–7
326, 396, 456, 506, 617 inversion 442, 452, 453
head 28, 437, 442, 443, 453, 454, 465, 472, 474, IP see Item and Process
476, 540, 542–3; left-hand 344, 434, 439, 622, irregularity see regularity
626; right-hand 442, 443, 464, 486, 487, 567, isolating 616, 617–8
622; relativized 543 Italian 128, 139–40, 536, 622
head-marking languages 618–9 item and arrangement (IA) 629–31
heavy syllable 296, 298, 301 item and process (IP) 629–31
Hebrew 139 item-familiarity 12, 30, 202, 306, 319, 344, 368,
homography 54 394, 396, 464, 516, 565, 569, 612
homonymy 17, 545, 546, 575
homophony 66, 82, 279, 288, 289, 340, 412, 425,
545, 571; of affixes 42, 216, 323, 344, 347, 391, Jamsay 624
399, 499, 546, 555, 633
Hup 624, 625
hyphen and hyphenation 10, 54–7, 337, 342, Lango 625
396, 407, 419, 420, 423, 432, 438, 449–51, 452, Latin 15, 35, 136–7, 176, 180, 182, 206, 256, 297,
516 409, 418, 426, 438, 441, 498, 521, 571, 583;
hypocoristic 134, 144, 191, 389, 394, 631 Latin plurals in English 137–9, 536
hyponymy 464, 465, 474, 483, 486 Latinate 614 see non-native
lax 166
learned see non-native
IA see Item and Arrangement lemma 537
iamb 186, 198, 271, 297, 301 level ordering 164, 493–508, 509, 571, 579, 637–8;
iconicity 623 see also stratification
idiom 12, 64, 89, 124, 489, 561 lexeme 8, 533, 560
idiomatization 30, 259 Lexeme Morpheme Base Morphology 632–3
686 Index of subjects

Lexical Category Prominence Rule 449 language 228–9; location 209, 210, 213, 216,
lexical density 18 231, 240–1, 250, 256, 258, 262–6, 282, 283, 285,
lexical gap 29, 189–90, 273–4, 278, 548, 550, 331, 352, 353, 407, 477, 633; manner 207, 326;
557, 562 means 207, 231, 240–1, 326; measure 209, 211,
Lexical Integrity Hypothesis 509, 638–9 213, 231, 240–1; modality 308, 327; mode of
lexical item 12, 626 belief 256; move 285; ornative 268, 282, 283,
Lexical Phonology and Morphology 164, 296, 284, 285, 306, 313, 477; path 209, 211, 308;
612–4, 637–8 patient 209, 211, 212, 216, 231, 237–40, 633;
lexicalization 12, 29–33, 117–8, 215, 233, 282, 300, pattern of action 266; payment
316, 319, 331, 360, 366, 368, 371, 378, 379, 396, 264; performative 283, 285, 286, 477;
402, 408, 436, 440, 445, 446, 450, 451, 457, 463, person 258; prejudice 266; prejudiced
474, 489, 516, 540, 577, 617; of phrases person 234–5; privative 282, 285;
435, 438, 439 product 207, 209, 210, 212; punctual 213;
lexicography 546, 570, 626 quality 257, 259, 352; quantitative 333, 347,
lexicology 626 352, 353; resultative 195–215, 230–1, 259, 262,
light syllable 296, 298 264, 282, 283, 284, 477; scientific study 266;
linking element 165, 455, 624 similative 283, 285, 311; speech act relation
listedness 433, 576 327; speech form 253, 256, 258, 266; state 257,
listeme 12 259; stative 195–215, 230–1, 262, 264, 285, 308;
literalness 571 stimulus 231, 232; theme 216, 231, 239;
loan see borrowing viewpoint 331; see also gendered term,
locative, spatial 333, 346–8; temporal transposition
333, 346–8 mental lexicon 12, 67, 496, 640
mention 9, 128, 551
meronomy 286, 478
Maori 136, 140 metaphor 346, 421, 465, 478
meaning, lexical 14, 539 metonymy 264, 283, 423, 465, 479
meaning of derivatives; agent 211, 212, 216, 220, Metrical Phonology 296
225, 231, 232–7, 238, 579, 633; agent Minimal Generalization Learner 634
collective 212; approximative 313, 420, 421; modal see verb
aspectual 310; behaviour 263; causative 282, modification 433, 437
283, 284, 285, 310, 366, 373; collective 250, 256, monomorphemic 31, 32, 112, 381, 496
258, 260, 262–6; comparative 347; monophthong 47
completive 308; concrete object 258; Monosuffix Constraint 612, 614
degree 326–7; direction 282; disease 266; Morbocomp Corpus 41, 470
doctrine 253; domain 327; duration 326–7; morph 13–14, 24, 25; bound (obligatorily
evaluation 327; eventive 195–215, 230–1, 258, bound) 14, 19, 441, 525; empty 16; free
262, 264, 308, 309, 310; experiencer 231, 232–7; (potentially free) 14; replacive 631
factive 207; field of study 253, 256; follower/ morpheme 13–20, 25, 30, 541, 629–31, 631
proponent/adherent of 226, 228, 234, 241; morpheme-based morphology 20
frequency 326–7; having 311; hierarchy 349; morphological complexity 112
inchoative 282, 283, 285, 310, 373; morphological family 67, 519, 521, 525
inhabitant 216, 224, 228–9, 241; instance 209, morphophonemic variation 163
212, 213; instrument 210, 216, 220, 231, 232–7, morphophonology 164, 177
241, 285, 286, 476, 633; intensification 282; morphosyntactic feature 10
iteration 419; kinship 241, 243, 244; motivation, loss of 30
 Index of subjects 687

multifunctionality 564–5, 622 Norse, Old 616

multiple correspondence 187, 520, 521 northern subject rule 62
multiple exponence 631 Norwegian 140
noun 27; abstract 255, 256, 257–61, 348;
agent 141, 467, 519, 579, 633; as base 198,
name 52, 127, 131, 135, 295, 401; personal 224, 201, 217, 219, 220, 225, 227, 230, 238, 246, 247,
283; place 224, 229; see also noun, proper 248, 249, 251, 253, 257, 260, 264, 268, 275,
native 35–6, 46; affix 168, 303–6, 333, 340–4, 276, 281, 284, 290, 291, 294, 304, 308, 314,
348–51, 357, 386–95, 493, 498, 514, 574, 584–615; 328, 334–7, 340–3, 348, 350, 353, 355, 358,
base 170, 198, 201, 223, 227, 229, 245, 252, 268, 359, 362, 363, 367, 369, 370, 371, 387, 391, 396,
269, 305, 326, 351, 355, 358, 396, 584; 397, 402, 404–5, 406, 407, 409, 414, 416,
lexeme 203 421, 571; collective 122, 540; common 218,
Natural Morphology 581, 623 221, 224, 229, 253, 558; count 264, 369,
negative 86–8; base 381–4; 557–8; derived 195–266, 288; ethnic 136;
contradictory 364–5, 366, 367, 368, 370, 371, individual level 348; inhabitant 135;
372, 379; contrary 364–5, 366, 367, 368, 370, instrument 467, 633; kinship 475;
371, 372, 379, 417; pejorative 366, 367, 368, non-count 124, 139, 214–5, 557–8;
375; privative 354, 366, 367, 368, 369, 370, patient 468, 633; personal 348, 633;
372, 374, 375, 381; reversative 354, 362, 365, proper 52, 127, 218, 222, 224, 252, 253, 270,
366, 367, 369, 370, 374, 375, 381, 417; 290, 295, 299, 303, 347, 400, 434, 464, 529,
scale-external 365, 366, 367, 368, 379; 558 see also name; regular and irregular 620;
standard 354, 370; stereotype 365, 366, 367, relational 38, 465, 472, 473, 474, 486;
371, 372, 373, 417 sortal 38, 472, 473, 475; stage-level 347;
negative contraction 71, 88–90 tribal name 136
neo-classical word-formation 18, 254, number 57, 306, 385, 422–7; as base 218, 290,
256, 303, 330, 339, 396, 405, 406, 415, 417, 421, 304, 420
441–2, 454–6, 463, 485–7, 584, 627;
additive 487; compromise 487; see also
combining form obligatorily bound see bound
neologism 30, 270, 284, 285, 303, 350, 361, 362, Ockham’s Razor 563
363, 364, 395, 565–6, 568, 577, 580 onomatopoeia 25, 278, 285
Ngiti 624 opacity; phonological 42, 162, 168, 496;
nominalization 16, 143, 540, 541 semantic 42, 162, 168, 256, 328, 360, 396, 440,
nonce 30 463, 485, 520; see also transparency
non-concatenative morphology 618, 630, 633 Optimality Theory 642
see also back-formation, blend, clipping, ordinal 425–6
paradigmatic morphology orthography 12–13, 25, 46–57, 69, 79, 83, 90,
non-native 35–6, 493; affix 33, 166, 196–202, 106, 118, 125–9, 130, 132, 144, 170, 180, 248,
291–303, 333, 334–9, 348–51, 395–400, 498, 514, 326, 357, 359–61, 388, 401, 406, 407, 420,
574, 584–615; base 170, 197, 208, 219, 221, 223, 226, 422, 423, 434, 445, 446, 461, 516, 620;
227, 229, 245, 252, 268, 269, 300, 305, 326, 351, 355, see also concatenation in spelling, hyphen
356–7, 358, 362, 396, 579, 584; formative and hyphenation, solid spelling, spaced
301; lexeme 48, 203, 204, 206, 288, 361 spelling
non-past see tense output-orientation 112, 164, 193, 272, 296, 298,
non-referentiality 437, 464 391, 393, 451, 461
No-Phrase Constraint 639 overgenerating morphology 503
688 Index of subjects

palatalization 175, 179, 224, 298, 299 priming effect 149

paradigm 9, 435, 518, 520, 546, 568; defective 82, probabilistic structure 119, 147, 460, 641
83, 110, 534; derivational 568; ~ gap 548, 549; productivity 29–33, 34, 76–8, 221, 229, 252, 288,
inflectional 61, 568; lexical 22; ~ pressure 305–6, 323, 324, 327, 329, 332, 339, 344, 348,
640; ~ uniformity 187, 520, 574 349, 350, 358, 361–4, 388, 389, 395, 396, 399,
paradigmatic morphology 20, 197, 222, 271, 451–2, 459, 465, 487, 488, 496–7, 515, 519, 525,
518–30, 634 534, 537, 540–2, 549, 550, 556, 561, 562,
paradox, ordering 117, 638 568–82, 583, 641; see also availability,
parasynthesis 500–1, 502–3, 506 profitability
participle 113, 453, 537–8, 543, 555–7; as profitability 32–3, 34, 405, 562, 568, 578
adjective 306, 310; passive 632; past 10, 62, projection of an argument 220, 221, 227
68, 73, 75, 290, 466, 470, 471, 541, 556, 571, 617, pronominalization 122
621, 632; present 62, 69, 73, 116, 538 pronoun 150–3; absolute 152; as base 290;
partitive see genitive default 151; determinative 152;
passive 64, 79 emphatic 152–3; impersonal 151;
past see tense interrogative 153; personal 150–1;
pejorative 349 possessive 151–2; reflexive 152–3;
periphrasis 641; in comparison 103, 110–20, 324, relative 153; sole subject 151
507; in modality 81 prosodic morphology 163, 190–2, 460–2
phoneme 323 prosody 182–8, 638
phonestheme 24 prototypical meaning 457, 490
phonology 161–94, 515; generative 296; proverb 12
lexical 162; post-lexical 162; weakening 440 proximal 152, 153, 155
phonotactics 391, 456, 515 punctuation mark 10
phrasal verb see verb
plain form 64, 68, 73, 75
plural 121–41, 206, 304, 438, 506, 507, 535–6, 541, quantification 418–22
548, 561, 578, 617; foreign 136–40; quantifier 420
unmarked 134–6 quantity 385–427
pluralia tantum 123, 124
politeness 66
Polynesian languages 622 rankshift 19
polysemy 17, 42, 230, 256, 259, 265, 282, 344, 347, realizational and non-realizational model 631–3
354, 365, 366, 397, 464, 529, 539, 545, 546, 558, recursion 396, 417, 424, 432, 499–500, 508,
559, 633, 641 624, 640
polysynthesis 616, 617–8 redundancy 376–7
possessive see genitive reduplication 386, 411–3, 439, 630;
pragmatic information 286, 477, 478, 481, contrastive 457
565–6, 639 referencing 38–9, 212–3; adjunct 213;
prefix 17, 54 see also affix argument 212–3, 307–11; event 310–1;
premodifier 123 object 38–9, 143, 148, 307–9, 388, 467–9, 471, 473;
preposition 107, 333, 340, 438, 442, 452, 552, 557, 622; prepositional-object 39, 467–9, 471;
as base 218, 278, 391, 552 subject 38–9, 143, 148, 217–26, 233, 239, 307,
prescriptivism 44, 69, 105, 106, 130, 136, 140, 150, 309, 446, 468, 470–1, 473
276, 326, 418, 559, 642 referentiality 464
preterite 65, 68, 70, 73, 75, 571 register see style
 Index of subjects 689

regularity 34–5, 323, 574, 634; formal 534, 535, split morphology 543
537, 571; semantic 534, 537, 539; see also verb, stem 19, 623; ~ selection 187, 520
irregular; verb, regular storage 641
regularization 574 stratification 583–615 see also level-ordering
relative clause 154 stress 111, 113, 223, 288, 461, 494; alternation see
relevance 497 shift; assignment 300, 396, 423, 451, 525;
relisting 564 attraction 360; clash 183; contrastive 445,
Renaissance 35 458; lapse 167; left 437, 446, 448, 625 see also
resonance 158, 519 compound stress; level or double 445; main
resyllabification 163, 298 or primary 166; neutrality 515, 613; noun vs.
rhyme 412, 413 verb 279, 361; preservation 166, 184, 187–8,
Righthand Head Rule 635 248, 422; reduction 271, 272–3; retraction
rivalry see competition (long, strong and weak) 184, 185; right 434,
Romance languages 474, 616, 619, 625–7 444, 445, 446; secondary 166, 183, 184, 275,
root 17–8; ~ and-pattern 630; bound 276, 297, 408; shift 21–2, 163, 164, 165, 170,
(obligatorily bound) 18, 19, 197, 441, 455, 621 184–7, 201, 204–6, 224, 229, 256, 271, 272, 276,
rule-governedness 34 296, 297, 298, 299, 300, 301, 305, 360, 361, 415,
Russian 612 417, 442, 515, 554, 613, 621, 637; see also
auto-stress, compound stress
strong form 84
Sanskrit 465 style, register, or domain 149, 168, 169, 235, 243,
scalarity 580 see also profitability 248, 250, 260, 303, 306, 313, 349, 350, 351, 356,
schemas and sub-schemas 633 372, 388, 398, 409, 412, 413, 416, 421, 422, 425,
scientific vocabulary 228, 233, 242, 275, 284, 409, 426, 576, 626; see also written and spoken
415, 427, 626 subjunctive 62, 64, 79
selection point 462 sub-morphemic elements 158; final <m> 157;
selectional restriction 164, 192–3, 271, 296, 494, final /n/ 152, 157; final <r> 157; final
501, 515, 584 <s> 152, 157; final <t> 157; initial /h/ 24, 156;
semantic extension 239 initial <th> 24, 156; initial <wh> 24, 156
semantics 118; bleaching 440; dependency subtraction 25–6, 460, 631
547–8; relations in compounds 446, 624; suffix 17, 54; consonant-initial 168; vowel-initial
scope 348, 497, 542, 547–8 168, 169; see also affix
semi-suffix 289, 440, 441 see also affixoid superlative 103–4, 105–6, 110–20, 317, 324, 522
separability 495–6 suppletion 22–3, 66, 70, 88, 107, 109, 423,
silent <e> 49; loss of 51–2, 69 425, 537
singular 121, 548 syllabicity 105, 168
size 25, 385–427 syllabification 11, 164, 168–9
Slovak 624 syllable 111; boundary 162; structure 391, 461
solid spelling 432, 449–51 syncretism 61, 65, 71, 72, 631, 632; base 75
sound symbolism 24–5 synonymy 268, 326, 328, 378, 456, 559, 573, 575,
spaced spelling 449–51 579, 636, 640 see also blocking
Spanish 128, 240, 403, 622 syntax 29, 36, 61, 70, 103, 118–9, 414, 424, 467,
spelling see orthography 477, 534, 544, 555, 558, 639, 640
spirantization 174, 175, 298 syntactic category see word-class
splinter 19, 411, 459, 525–30, 634 synthesis 640; in comparison 103, 107, 110–20;
split base 187, 520, 634 index of 617
690 Index of subjects

tautology 579 verb 27; as base 197, 198, 201, 217, 219, 220, 222,
tense; future 620; past 10, 62, 65, 541, 575, 617, 225, 226, 249, 251, 254, 268, 270, 290, 291,
620, 632; non-past 62, 65, 620; sequence 292–3, 300, 304, 307, 309, 334–7, 340–3, 349,
of ~s 82 352, 356–7, 359, 362, 367, 369, 371, 373, 377,
tense (of vowels) 166, 183 380, 388, 402, 404–5, 419, 465; auxiliary
Tilburg Memory Based Learner (TiMBL) 634 61–4, 71, 80–90; causative 217, 220;
title 140, 244 defective 541; denominal 571; derived
to-infinitive 83 267–87, 398; class 213–4, 374, 375;
transitivity 37 ditransitive 217, 220, 227, 238, 292–3; double
transparency 32, 34, 42, 581, 614; modal 81; intransitive 280, 282, 353, 419, 558;
phonological 162, 165, 581; semantic 371, 440, irregular 65, 66–8, 70–80, 519, 572, 617, 631;
447, 448, 464, 623; see also opacity lexical 61–4, 64–80; modal 63, 80–3;
transposition 27–8, 314–7, 324, 567 phrasal 12, 205, 218, 442, 453, 454;
trisyllabic laxing 164, 170, 171, 172, 248, 298, primary 63; prepositional object 217, 219,
299, 301 220, 227, 292–3, 353, 473; regular 65, 66–8,
trochaic laxing 170, 172 519, 572; with sentential complements 218,
trochee 165, 171, 172, 186, 193, 198, 271, 276, 297, 219, 220, 353; transitive 217, 219, 220, 226, 238,
299, 301, 391, 402 280, 282, 292–3, 352, 353, 419, 542, 558, 579;
truncation 163, 165–6, 168, 276, 301, 386, see also unaccusative, unergative
391, 393 verbal diathesis 37, 79, 351, 352, 467, 473
Turkish 618 verb-particle constructions 4
type coercion 104–5, 114, 125, 244, 260, 308, 318, violability 641
348, 365, 374, 536, 557–9, 641 vocabulary size 569
type-familiarity 30 voicing 68, 174, 206, 301, 392, 423, 552, 553
typology 544, 612, 616–27 vowel; alternation 109, 131–3, 163, 169–73, 423,
572, 621 see also ablaut, umlaut;
coalescence 224, 271; hiatus 223, 229, 255, 298,
umlaut 20, 22, 131–3, 161, 617, 620 299, 450; laxing 170, 171, 172, 301; length 46–8;
unaccusative 37, 219, 221, 227, 292–3, 307, reduction 170–1, 248, 272–3, 300;
308, 310 strengthening 170, 172, 248
unanalysability 32
underspecification 564–5, 567, 641 see also
polysemy weak form 84–6
understood 550 Wellington Corpora of Written and Spoken
unergative 37, 219, 221, 227, 292–3, 307, New Zealand English 41, 202
308, 310 wh-words 153–4, 156–7
Unitary Base Hypothesis 579, 635–6 Wikipedia 484
Unitary Output Hypothesis 579, 635–6 word 8–13, 569; actual 30; function 150–8;
univerbation 391, 442 grammatical 8, 10, 11, 27; head 9;
unproductive 32 lexical 10, 27; morphosyntactic 10;
Urban Dictionary 44, 244, 524 orthographic 10, 432; phonological 11, 615;
potential 30, 569
Word and Paradigm (WP) 629–31
variability 641 Word Based Hypothesis 579
velar softening 53, 164, 172, 175, 179, 224, 248, word-based morphology 20, 621
255, 581 Word Structure Autonomy Condition
 Index of subjects 691

word-class 18, 26–8, 456, 515, 546; change of 28, <y> to <i> rule 52–3, 69, 126
124, 323, 338, 347, 355, 358, 359, 371, 376, 537, 543, Yiddish 413
554, 555, 635; closed 27; major 534; open 27;
preservation of 338, 357, 358
word-form 8, 533 zero affix 563
Wordspy 44, 270, 278, 483, 484, 524 zero-derivation 563–4, 567 see conversion
Word Structure Autonomy condition 638
WP see Word and Paradigm
written and spoken 272 3rd singular non-past 69, 73, 82, 617