The Arabic Verb

Studies in Functional and Structural Linguistics (SFSL)

Taking the broadest and most general definitions of the terms functional and
structural, this series aims to present linguistic and interdisciplinary research that
relates language structure — at any level of analysis from phonology to discourse —
to broader functional considerations, whether cognitive, communicative, pragmatic
or sociocultural. Preference will be given to studies that focus on data from actual
discourse, whether speech, writing or other nonvocal medium.
The series was formerly known as Linguistic & Literary Studies in Eastern Europe (LLSEE).

Founding Editor Honorary Editors

John Odmark Eva Hajičová Petr Sgall
Charles University Charles University

General Editors
Yishai Tobin Ellen Contini-Morava
Ben-Gurion University of the Negev University of Virginia

Editorial Board
Alexandra Y. Aikhenvald Jim Miller
La Trobe University University of Auckland
Joan Bybee Marianne Mithun
University of New Mexico University of California, at Santa Barbara
Nicholas Evans Lawrence J. Raphael
University of Melbourne CUNY and Adelphi University
Victor A. Friedman Olga Mišeska Tomić
University of Chicago Leiden University
Anatoly Liberman Olga T. Yokoyama
University of Minnesota UCLA
James A. Matisoff
University of California, Berkeley

Volume 63
The Arabic Verb. Form and meaning in the vowel-lengthening patterns
by Warwick Danks
The Arabic Verb
Form and meaning
in the vowel-lengthening patterns

Warwick Danks
University of St Andrews

John Benjamins Publishing Company

Amsterdam / Philadelphia
 TM
The paper used in this publication meets the minimum requirements of
8

American National Standard for Information Sciences – Permanence of

Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data

Danks, Warwick.
  The Arabic verb : form and meaning in the vowel-lengthening patterns / Warwick Danks.
       p. cm. (Studies in Functional and Structural Linguistics, issn 0165-7712 ; v. 63)
Includes bibliographical references and index.
1.  Arabic language--Verb. 2.  Arabic language--Morphology.  I. Title.
PJ6145.D36   2011
492.75’6--dc22 2011004873
isbn 978 90 272 1573 4 (Hb ; alk. paper)
isbn 978 90 272 8695 6 (Eb)

© 2011 – John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands
John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
 To my late mother
 Mavis Florence Danks (1922–1999)
 and my late grandmother
 Emma Louise Williams (1891–1982),
 for their strength, patience
 and unswerving faith in me
 Table of contents

Preface xi
Notes on symbols, abbreviations and other conventions
followed in examples xiii

chapter 1
Introduction 1
1.1 Saussurean structuralism  2
1.1.1 The linguistic sign  2
1.1.2 Language as a system  3
1.2 Beedham’s method of lexical exceptions  4
1.2.1 Principles  4
1.2.1.1 Unexplained exceptions are indicative
of incorrect analysis  4
1.2.1.2 From form to meaning  5
1.2.1.3 Synchronic basis  5
1.2.1.4 The Hegelian triad and scientific method  8
1.2.2 An attempt to apply the method  8
1.2.2.1 Phase 1: Choose a formal construction  9
1.2.2.2 Phase 2: Identify the problems, anomalies,
contradictions, etc.  9
1.2.2.3 Phase 3: Identify the unexplained lexical exceptions  10
1.2.2.4 Phase 4: Identify the properties of the exceptions  10
1.2.2.5 Phase 5: What might lead to the exceptions?  11
1.2.2.6 Phase 6: The semantic phase  11
1.3 A fresh approach to the problem  11
1.4 Further methodological considerations  13

chapter 2
Verbal morphology and the lexicon 15
2.1 Arabic verbal morphology  16
2.1.1 Inflectional morphology  16
2.1.2 Derivational morphology  17
2.1.2.1 ‘Biliteral’ verbs  17
2.1.2.2 Pattern I triliteral stems  18
viii The Arabic Verb

2.1.2.3 Derived or augmented triliteral stems  18

2.1.2.4 Quadriliteral stems  21
2.1.2.5 Morphological connections  21
2.2 Distribution of verbal patterns in the lexicon  23
2.2.1 Data collection  24
2.2.1.1 Method  24
2.2.1.2 Lexical issues  24
2.2.2 Comparison of data with existing sources  27
2.2.3 Quantitative analysis of the data  29
2.2.3.1 Productivity by root  29
2.2.3.2 Productivity by pattern  30
2.2.3.3 Co-occurrence of patterns  31
2.3 Summary  37

chapter 3
Alternative morphologies 39
3.1 Alternatives to the Arabic root as the primary basis of derivation  39
3.1.1 Word- and stem-based approaches  40
3.1.1.1 Is the concept of root necessarily excluded?  41
3.1.1.2 Specific arguments for a fully vocalised base  42
3.1.1.3 External evidence  46
3.1.1.4 Conclusion  50
3.1.2 Matrices, etymons and radicals  51
3.1.2.1 Bohas’s hypothesis  51
3.1.2.2 Assessment of the matrix/etymon model  54
3.1.2.3 Conclusion  55
3.2 Prosodic templatic morphology  55
3.2.1 Three morphemic tiers  56
3.2.2 Prosodic analysis  56
3.2.2.1 Syllable types in Arabic  56
3.2.2.2 Minimal stems  58
3.2.2.3 Further noun stems  59
3.2.2.4 Verbal stems  60
3.3 Summary  62

chapter 4
Understanding Arabic verbal semantics: Form and meaning 63
4.1 The grammarian’s dilemma  63
4.2 Specific approaches from the grammars  65
4.2.1 The ‘reductionist’ approach  66
 Table of contents ix

4.2.2 The ‘explicit specification’ approach  71

4.2.3 The approaches compared  73
4.3 The linguist’s contribution  73
4.3.1 Semitic comparative linguistics  73
4.3.2 The synchronic linguist’s perspective  78
4.4 Summary  80

chapter 5
Evaluating the pattern III – pattern VI semantic relationship 83
5.1 Selection of patterns III and VI  83
5.2 Mutuality and reciprocity  84
5.2.1 Data collection  84
5.2.2 Data interpreted  87
5.2.2.1 Pattern III mutuality  87
5.2.2.2 Pattern VI reciprocity  90
5.2.2.3 Pattern III – pattern VI correlations  92
5.2.2.4 Correlations with pattern I  93
5.3 Pattern III conativity  101
5.4 Summary  102

chapter 6
Transitivity and valency 103
6.1 Defining transitivity in Arabic  103
6.1.1 Transitive verbs  103
6.1.2 Intransitive verbs  104
6.1.3 Ambitransitivity  105
6.1.4 Transitivity through a preposition  105
6.1.5 Multi-transitive verbs  106
6.2 Valency  108
6.3 Data collection  110
6.4 Data analysis  112
6.4.1 Valency and the ta- prefix  112
6.4.1.1 A hierarchical approach to valency structures  113
6.4.1.2 Hierarchical exceptions  114
6.4.1.3 Summary  121
6.4.2 Valency and relationships with pattern I  122
6.4.2.1 Pattern I and pattern III valency  122
6.4.2.2 Pattern I and pattern VI valency  127
6.5 Summary  129
x The Arabic Verb

chapter 7
The pattern III template: From form to meaning 131
7.1 Formal characterisation of pattern III  132
7.2 Vowel lengthening and plurality  133
7.2.1 An introduction to verbal plurality  134
7.2.2 Mutuality and reciprocity as verbal plurality  137
7.2.2.1 Formal comparison of vowel lengthening in pattern III
and broken nominal plurals  137
7.2.2.2 Is plurality a valid interpretation of mutual action?  140
7.2.3 Summary – Plurality  142
7.3 The long ā in Arabic morphology  143
7.3.1 Nominal templates with long ā  145
7.3.1.1 Participles  145
7.3.1.2 Verbal nouns  149
7.3.1.3 Nouns of instrument  151
7.3.1.4 Nouns of occupation and intensity  152
7.3.2 Commonalities of meaning  153
7.3.2.1 Agency and patiency  154
7.3.2.2 Process and result  155
7.3.2.3 Temporal complexity  156
7.3.3 Summary – Hypothesis of long ā as an aspectual marker  156

chapter 8
An aspectual model for Modern Standard Arabic 157
8.1 Defining aspect  157
8.2 The Arabic s-stem and p-stem verb forms:
Tense or grammatical aspect?  159
8.3 Vendler’s aspectual categories  162
8.4 Olsen’s scheme  165
8.4.1 Overview  165
8.4.2 Application to MSA  169
8.4.2.1 States  169
8.4.2.2 Activities  172
8.4.2.3 Accomplishments  175
8.4.2.4 Achievements  179
8.4.2.5 Semelfactives  183
8.4.2.6 Stage-level states  185
8.5 Summary  187
 Table of contents xi

chapter 9
Aspectual categorisation of patterns III and VI 189
9.1 Data  189
9.1.1 Verbs of ACTIVITY and STATE  190
9.1.1.1 Corpus data counts – Methodology  191
9.1.1.2 Results  193
9.1.2 Other categories represented  194
9.1.2.1 Verbs of surprise  194
9.1.2.2 Verbs of giving  198
9.1.2.3 Verbs of inception  201
9.1.2.4 Other verbs  202
9.2 Summary  211

chapter 10
Inceptive aspect 213
10.1 The case for a lexical aspect category of INCEPTIVE   213
10.2 Extension of Olsen’s scheme for INCEPTIVES  220
10.2.1 Feature marking of the onset phase  224
10.2.2 Other feature marking combinations  225
10.2.3 Extended scheme for lexical aspect categories  228
10.3 Inceptive verbs in patterns III and VI  228
10.4 Summary  235

chapter 11
The passive in patterns III and VI 237
11.1 Passive formation by vowel melody change  237
11.1.1 Corpus data evidence  238
11.1.2 Native speaker evidence  240
11.1.3 Conclusion  241
11.2 Passive participle formation  241
11.2.1 Dictionary evidence  241
11.2.2 Native speaker evidence  243
11.2.3 Corpus data evidence  245
11.2.4 Conclusion  245
11.3 Verbal noun forms  246
11.4 Summary  246
xii The Arabic Verb

chapter 12
Conclusions 247
12.1 Overview  247
12.2 The characteristics of patterns III and VI  248
12.2.1 Form  248
12.2.2 Meaning  249
12.2.3 Relating form and meaning  249
12.3 Directions for further research  250
12.3.1 Nominal aspect  250
12.3.2 Aspect and passivisability  251
12.3.3 Inceptivity of state and of activity  251
12.3.4 Alternative verbal noun forms  252
12.3.5 Defining atelicity  252
12.4 Summary  255

Bibliography 257

appendix i.  Data tables 265

appendix ii.  Semantic category labels 267
appendix iii.  Native speaker questionnaire 269
appendix iv.  Passive participles in patterns III and VI 275

Name index 277

Subject index 279
 Preface

This book was originally a Doctoral thesis, so I would firstly like to thank my
supervisors at the University of St Andrews: Christopher Beedham (Dept. of Ger-
man, School of Modern Languages) and Catherine Cobham (Dept. of Arabic
& Middle East Studies, School of History). Each of them contributed a unique
perspective from their own discipline and patiently listened while I attempted
to explain some of the more esoteric areas of the other’s field of expertise. Their
guidance has been much appreciated and their input never burdensome.
The provenance of the book is thus partly responsible for its layout: building
upon the formal foundations of Arabic verbal morphology in the early chapters
as a prerequisite for elucidating the meaning of the verb patterns specified in
the title. However, my dual intentions remain to make the complexities of the
Arabic language accessible for specialists in linguistics and to present linguis-
tic theory comprehensibly to Arabists with no advanced linguistics training. As
such, those who already have a good working knowledge of Arabic may care to
pass over Chapters 2 and 4 and some of the other sections. Moreover the mate-
rial in Chapter 3 may also be familiar to those with a specialism in Arabic or
Semitic linguistics. Similarly, linguistics specialists may find some of the sections
which address theory and terminology superfluous. I trust, however, that all will
find sufficient material which is new, and perhaps surprising, to make examina-
tion of this book as a whole rewarding. The need to make the book accessible to
English speakers, written as it is in that language, together with the inescapable
reality that the majority of linguistic research has been undertaken in English,
accounts for my frequent recourse to English examples. However, I believe that
I have been careful not to unjustifiably transfer interpretations and terminology
from my native tongue.
I am grateful to my examiners, Clive Sneddon (Dept. of French, University
of St Andrews) and Janet Watson (Professor of Arabic Linguistics, University
of Salford), for their diligent reading of the thesis and consequent ­corrections
and suggested revisions, particularly as Professor Watson unusually bridg-
es the two specialist audiences for whom I am writing. Thanks also to the
Publisher’s two anonymous reviewers, especially for references to material in
French and German.
xiv The Arabic Verb

I also want to acknowledge the Honeyman Foundation for their Student-

ship award, Dilworth Parkinson (Brigham Young University) for assistance in
accessing his invaluable arabiCorpus, Mari Broman Olsen (Microsoft) for her
encouragement as I applied and developed her model of linguistic aspect, and
my anonymous native speaker informants for their time and willingness to share
their insights.
 Notes on symbols, abbreviations
and other conventions followed in examples

Arabic transliteration

Consonants
‫ء‬ ’ ‫ض‬ D
‫ب‬ b ‫ط‬ T
‫ت‬ t ‫ظ‬ DH
‫ث‬ th ‫ع‬ ‫ع‬
‫ج‬ j ‫غ‬ gh
‫ح‬ H ‫ف‬ f
‫خ‬ kh ‫ق‬ q
‫د‬ d ‫ك‬ k
‫ذ‬ dh ‫ل‬ l
‫ر‬ r ‫م‬ m
‫ز‬ z ‫ن‬ n
‫س‬ s ‫ه‬ h
‫ش‬ sh ‫و‬ w
‫ص‬ S ‫ي‬ y

Vowels etc.

َ ‫ــــ‬ a ‫ى‬,‫ا‬ ā
ُ ‫ــــ‬ u ‫و‬ ū
ِ‫ــــ‬ i ‫ي‬ ī
ً ‫ـا‬ -an ‫ــَة‬ -a, -at

Notes
1. As is common practice, the transliteration follows pronunciation rather than
a symbol-for-symbol substitution, allowing the transliterated text to be read
aloud. It is recognised that production of speech in MSA is subject to many
idiolectal variations and thus the transliterated text here employs minimal in-
flection except where a vowel must be supplied preceding a joining hamza.
xvi The Arabic Verb

2. Consequently, grammatical cases are rendered only when explicitly pro-

nounced or where necessary to the argument and are glossed accordingly.
3. The combinations Damma-waw and kasra-yā (‫ ـــُو‬and ‫ )ـــِي‬have been repre-
sented as ū and ī respectively, except where the weak consonant begins the
following syllable when uw and iy are preferred.
4. In certain proper names containing the /g/ phoneme, the ‫ ك‬or ‫ ج‬from the
Arabic orthography has been transliterated as g.
5. To reflect pronunciation in modern informal speech, tā’ marbūTa (‫ )ــَة‬has
been transliterated simply as -a, except in construct or preceding an object
pronoun when it is represented as -at.

Citations and translations

Each example is given with appropriate citations except where it is a dictionary

listing from Wehr (1994) or it is my own composition or rewording of a previ-
ously cited example. Furthermore, all questionnaire examples (see Appendix III)
and all translations of Arabic examples are my own, with the exception of diction-
ary definitions.

Morphemic glossing

1 first person neg negation

2 second person nom nominative
3 third person npst non-past
acc accusative obj object
apt active participle pass passive
cmpr comparative pfv perfective
coll collective pl plural
def definite poss possessive
dim diminutive ppt passive participle
du dual pst past
f feminine q question marker
fut future rel relative
gen genitive sbj subject
ind indicative sbjv subjunctive
indf indefinite sg singular
m masculine vn verbal noun
 Notes on symbols, abbreviations and other conventions xvii

Notes
1. The table above contains some non-standard abbreviations used for ease of
glossing Arabic.
2. As far as is reasonably practical, the conventions followed are those given in
The Leipzig Glossing Rules (Bickel et al. 2008).
3. Due to the complex nature of Arabic morphology, including broken plurals,
discontinuous morphs and portmanteau morphs, it has not always been pos-
sible to fully match morphs with glossed meanings. For example, in p-stem
verb forms it is difficult to specify exactly which morphs carry person, gender
and mood and, for this reason, I make no attempt to gloss p-stem prefixes as
separate morphs. However, since the purpose of the gloss here is largely to
clarify the examples for non-Arabists, the practice I have followed should not
be taken to represent any particular view regarding which morph carries a
specific burden of meaning.
4. Active verbs have not been glossed as such and the typologically unmarked
categories of indicative mood, masculine gender and singular number are
only glossed where explicit in morphology.

Event time diagrams

|-------- Bounded at start

--------| Bounded at end
<-------- Unbounded at start
--------> Unbounded at end
0-------- Implicit start
----|---- Internal phase boundary
|| Punctual phase
↑, ↓ Locations of points in time
ET Event time
RT Reference time
ST Speech time
X External events

Privative feature marking (Olsen 1997)

[+feature] Marked for feature

[Øfeature] Unmarked for feature
Ø Unmarked for all features
xviii The Arabic Verb

Other abbreviations

(+ d.o.) Followed by direct object

(+ i.o.) Followed by indirect object
(s.o.) Someone
(s.th.) Something
e.g.[III] Roman numeral (with or without vowel): triliteral pattern
(see Chapter 2)
e.g.[QII] Q with Roman numeral: quadriliteral pattern (see Chapter 2)
e.g.[MUT] Semantic category labels, with or without arithmetic symbols
(see Appendix II for complete listing)
e.g.[3A] Number-letter combination: valency structure code (see Table 41)
e.g.{1} In discussion of valency: numbers designating items contributing
to valency
e.g. {YN?} In questionnaire examples: responses of individual informants
(see Appendix III)
 chapter 1

Introduction

From our earliest times on the planet, we humans have searched for order in the
world around us, whether gazing at the night sky and defining the somewhat
fanciful patterns of the constellations or peering deep within the living cell to
discover the elegant double-helix structure of DNA. We recognise patterns in art,
patterns in music and even patterns in the physical sciences, such as the periodic-
ity of the elements or mathematical sequences. Moreover, that there is order, not
a mere amorphous sea of chaos, leads us to conclude that there is meaning to be
found in the patterns we observe. Recognising the beauty and regularity of the
movements of the stars and the bodies of the solar system, the Psalmist attributes
divine meaning:
The heavens declare the glory of God; the skies proclaim the work of his hands.
Day after day they pour forth speech; night after night they display knowledge.
There is no speech or language where their voice is not heard. Their voice goes
out into all the earth, their words to the ends of the world. (Psalm 19: 1–4, NIV)

Whether we agree with the Psalmist’s assessment or not is irrelevant. What is clear
is that he has observed order and structure and proceeded to interpret it: he has
progressed from form to meaning. In recent times, the Human Genome Project
has mapped the entirety of human DNA, but now the discipline of bioinformatics
is tasked with understanding what effect each gene has on us as organisms, once
again identifying form and progressing thence to meaning.
Language is fundamentally an ordered phenomenon. By this I mean not to
imply that words or sentence components are necessarily ordered in a particular
way, though this is often true of specific languages. Rather I am more generally
stating that language is not chaotic, but possesses discernible patterns and repro-
ducible structure. Were it not so, there would be little point in me writing this
paragraph, for in doing so I make a very basic assumption. I assume that the lan-
guage I employ is sufficiently close to that codified variety we call English that it
will be comprehensible to other users whose language competencies also include
that variety. That is, not only does the language I share with my readers have
the same form, but its lexicon and its grammatical constructions have the same
meanings for my readers as they have for me.
2 The Arabic Verb

This study has been inspired by the relationship between linguistic form and
meaning which characterises the view of language expounded in the foundation-
al lectures on linguistics given in the early twentieth century by Ferdinand de
Saussure at the University of Geneva, subsequently published posthumously by
his students as the Cours de Linguistique Générale and consulted here in transla-
tion (Saussure 1966). If language implies order, then the prescriptive grammarian
identifies the order he discerns within its patterns and imposes it, whereas the de-
scriptive linguist classifies those patterns and the Saussurean structuralist specifi-
cally relates the form evident within such patterns to meaning. Furthermore, the
research detailed in the following chapters results from taking up the challenge
of evaluating Beedham’s (2005) method of lexical exceptions, itself firmly rooted
in Saussurean structuralism, as an appropriate tool with which to examine the
verbal system of Modern Standard Arabic (MSA).
Many critiques and expositions of Saussure have been published and may be
consulted, thus it is my intention here only to provide a brief restatement of the
principles which have been formative in my approach. Beedham’s method will be
less familiar, thus I will seek both to summarise his methodology and to explain
to what extent it has or has not been applied in my own research.

1.1 Saussurean structuralism

Two fundamental themes of Saussurean structuralism will be seen to pervade

this present research: the indivisibility of the linguistic sign and the notion of a
language as a consistent and self-contained system.

1.1.1 The linguistic sign

Saussure (1966: 66) defines the linguistic sign as “a two-sided psychological entity

… [in which] … [t]he two elements are intimately united, and each recalls the
other”. These two elements are introduced initially as “concept” and “sound-im-
age”, but Saussure (1966: 67) proceeds in the original French to designate them
signifié and signifiant, or “signified” and “signifier”, on the basis that “the last two
terms have the advantage of indicating the opposition that separates them from
each other and from the whole of which they are parts”. Without wishing to di-
minish the indivisibility which Saussure establishes between these two facets of
the linguistic sign, I prefer the respective designations “meaning” and “form”
followed by Beedham (2005) as more appropriate in contemporary English. We
should note, as Beedham does, that:
 Chapter 1.  Introduction 3

the indivisible linguistic sign … does not of itself tell us whether we should start
with the signifiants – with form – and move from there to the signifiés – to mean-
ing – or do it the other way round and start with meanings and move from there
to forms.  (Beedham 2005: 3)

However, I have already commented in my opening paragraphs that advances

in diverse fields of knowledge typically involve observation of form and proceed
from there to meaning. We shall return to the methodological implications of this
in Section 1.2.
The other noteworthy property of the linguistic sign as defined by Saussure
(1966: 67ff.) is that it is arbitrary: there is no intrinsic or iconic connection be-
tween form and meaning, such that the link only exists within a particular lan-
guage community. Thus, for example, according to our language background we
may designate a canine quadruped as dog, chien, or kalb, but there is no absolute
sense in which any of these forms from English, French and Arabic respectively is
any more ‘canine-like’ than any other. Whilst the arbitrariness of the sign appears
a generally sound maxim, it may be an oversimplification. Although Saussure
(1966: 69–70) recognises and addresses the issues of onomatopoeia and interjec-
tions, he does not deal with the counterexample of reduplication present in many
languages and noted in Sections 4.2.2 and 7.2.1.

1.1.2 Language as a system

It is somewhat ironic that what might be regarded as the ultimate statement of

­Saussurean structuralism, that a language is “un système où tout se tient”, is actual-
ly not attributable to Saussure but to Meillet (1893: 318–319 in Koerner 1999: 26).
Nevertheless, it is clear that this quotation neatly expresses Saussure’s (1966: 114)
claim that “[l]anguage is a system of interdependent terms in which the value of
each term results solely from the simultaneous presence of the others”.
Thus, according to Saussure, although the linguistic sign is in itself arbitrary, its
place within the system which is language is anything but arbitrary. Furthermore,
Saussure makes two observations which may seem obvious to the modern trained
linguist but are anything but trivial. Firstly, although he initially introduces the lin-
guistic sign by relating a word form to the concept it signifies and presents the lan-
guage as a system consisting of these interdependent signs, he includes not only words
but equally grammatical devices as integral and systematically dependent parts:
Within the same language, all words used to express related ideas limit each other
reciprocally … [However] [e]verything said about words applies to any term of
language, e.g. to grammatical entities.  (Saussure 1966: 116)
4 The Arabic Verb

Thus in French (or English), for example, singular and plural stand in relation to
one another within the category of number, whereas in Sanskrit (or Arabic) sin-
gular and plural also stand in relation to the further designation of dual number.
This leads to the second observation that each individual language is a system
in its own right, within which its various components may relate in ways which
are quite unlike their apparent counterparts in other language systems, or may
simply be absent from those systems. Among his other examples he cites the lack
of tense as a deictic delineator of time in Hebrew and the aspects of the Slavic
verb as characteristics which are quite distinct from his native French (Saussure
1966: 116–117).
To sum up what we have gleaned from Saussure, if we observe a characteristic
form within a language system, we should expect that form to have meaning as-
sociated with it and, conversely, if a language makes a distinction of meaning we
should expect to see a corresponding distinction of formal expression. Moreover,
because language is a coherent system, every linguistic sign stands contrasted in
both form and meaning with those around it. Finally, whilst the specific language
system is in itself consistent and coherent, its categories and divisions of form and
meaning do not necessarily align with those encountered in other systems, intro-
ducing a cautionary note for the linguist who may prejudice his conclusions about
a language system by observing it through a lens tinted with the characteristics of
his own native language.

1.2 Beedham’s method of lexical exceptions

Originally presented as “a method of investigating grammar” in Beedham

(1982: 135ff.), the method of lexical exceptions is refined and explained more fully
in Chapter 7 of Beedham (2005). The research presented here began as an attempt
to apply Beedham’s method to Arabic, though it will become clear that I have
departed in significant ways from the prescribed methodology, such that I claim
only that this work is inspired by the method, not strictly an application of it.
However, whilst some specifics of Beedham’s methodology have been sacrificed,
the underlying principles largely remain intact.

1.2.1 Principles

1.2.1.1  Unexplained exceptions are indicative of incorrect analysis

It is evident from Beedham’s approach to exceptions to grammatical rules that he
takes Saussure’s view of language to its extreme:
 Chapter 1.  Introduction 5

According to Saussure a language is a structure or system. System implies regu-

larity implies rules …. If a language really is systematic it should not allow excep-
tions at all, i.e. items which stand outside the system …. Yet exceptions … arise to
the extent that we, the grammarians, have got it wrong. We introduce them from
outside with rules that are not quite right.  (Beedham 2005: 153)

Thus he argues that correct formulation of grammatical rules will necessarily lead
to the elimination of (unexplained) exceptions. Essentially this is a strong restate-
ment of the Saussurean maxim of un système où tout se tient: for Beedham, there
is no place for exceptions in a language system.

1.2.1.2  From form to meaning

Again, Beedham states his position forcefully and unequivocally:
It is important to emphasise that one starts with form and goes from there to
meaning, not the other way round …. The linguist who starts with meaning is
doomed to be trapped in the analysis which produced that meaning.
 (Beedham 2005: 156)
Our starting point is to be some property which is formally expressed and hence
objectively observable and quantifiable, whether morphological or syntactic. For
example, Beedham’s analysis of the passive proceeds from its formal expression
as be + V-ed, rather than from a meaning-based definition of the passive as equiv-
alent to an underlying active. His research on the strong verbs of English and
German (Beedham 2005: 107ff.) begins with the formal morphological distinc-
tion that they do not form their preterite and second (past) participles1 regularly
and asserts that this distinct expression of form must be accompanied by distinct
meaning, for which he then proceeds to search. Thus, in emphasising the indi-
visibility of the sign (the form-meaning relationship), he again presents a purist
view of Saussure but furthermore chooses to “align [himself] with one side of
the debate about whether to go from form to meaning or from meaning to form”
(Beedham 2005: 6), namely that form is the structuralist’s natural and logical em-
pirical starting point.

1.2.1.3  Synchronic basis

Beedham makes it clear that he works exclusively synchronically, viewing histori-
cal considerations as a distraction: “[i]f you are concerned only with the modern
language there is no diachronic part to your work to be distracted by” (Beedham
2005: 8). Again, he is motivated by a high view of Saussurean structuralism which

1. Beedham (2005: 175) “follow[s] Jespersen (1924: 283–284) in calling it the 2nd participle

[in these languages] in order not to prejudice the issue of its meaning”.
6 The Arabic Verb

will not allow any anomalous elements within the synchronic system, no matter
their provenance or however historically entrenched they may be. Thus while he
admits that the “[i]rregular verbs [of English and German] are a historical ves-
tige … that does not stop the irregular verbs from being rule-governed and mean-
ingful synchronically” (Beedham 2005: 112). He is thus motivated by this belief
to find the meaning associated with these irregular verbs within their respective
modern language systems.
I am compelled to comment at this point that I believe Beedham to be an ide-
alist. The strong view of the Saussurean principles which he holds fast to and em-
ploys enables him to pursue his goal of definitively relating meaning to form with
a commendable single-mindedness. However, whilst taking those same principles
as a useful and necessary foundation, I confess to less idealism and more of what
I would like to regard as realism. It is in the realm of synchronic integrity that my
appeal to realism is most evident. To justify this, I will return to Saussure.
We must recognise that Saussure himself was a historical linguist and that
his original lectures were delivered at a time when, with the exception of pre-
scriptive grammarians, linguists were, as he comments, “completely absorbed in
diachrony”, and he expresses the hope that “[l]inguistics, having accorded too
large a place to history, will turn back to the static viewpoint of traditional gram-
mar but in a new spirit” (Saussure 1966: 82–83). Whilst wanting to clearly delimit
synchronic study from diachronic in order to establish its validity as a discipline,
Saussure (1966: 74) is well aware that language change cannot be ignored and
comments that “[c]hange in time takes many forms, on any one of which an im-
portant chapter in linguistics might be written”. He continues:
Language is radically powerless to defend itself against the forces which from
one moment to the next are shifting the relationship between the signified and
the signifier. This is one of the consequences of the arbitrary nature of the sign.
 (Saussure 1966: 75)

In his treatment of synchrony and diachrony, Saussure (1966: 88–89) employs the

analogy of the chess board, likening the state of the board at any stage of a game to
a static situation in a language system. Each possible state of the board is governed
by the predetermined rules of the game and progression from one state to the
next involves moving a single chess piece, though this move has an effect on the
entire system. His view of diachrony is valid in that he observes that for any given
state of the board, the succession of moves which brought it about is irrelevant.
However, his analogy of change breaks down in one respect. It represents the real
development of language in much the same way that a cinematographer captures
real life: the movie camera cannot record movement, the illusion of movement
only being created when a succession of still frames is projected consecutively.
 Chapter 1.  Introduction 7

For Saussure, a chess piece moves and displaces another such that no intermediate
state exists, but I contend that the following illustration, whilst perhaps fanciful,
is more helpful.
Imagine a well-ordered dwelling. The occupant of the house has taken the
trouble to acquire all the items one needs to live comfortably and efficiently and
has assigned each object its place with care to maximise its effectiveness. If a new
item is acquired it presents no problem, so the brand-new DVD recorder, not
previously required by the household but now indispensable, takes its place natu-
rally alongside the wide-screen television and the recently added digibox with
only minor adjustments to make some space. Sometimes, however, room must be
found for a newly acquired object by displacing another. I remember as a child I
was left a rather elegant upright piano by my deceased cousin. Finding room for it
was made all the more difficult because we already possessed a perfectly service-
able piano, but rejecting the gift was unacceptable, so for some months our lounge
contained two pianos side by side. An external observer during this period would
have been hard-pressed to explain why a family with only one (would-be) pianist
needed two such instruments fulfilling the same function. Ultimately, the original
piano was evicted and the system returned to a stable state.
I believe that this picture better represents language change and I would draw
attention to the phrase “from one moment to the next” in the preceding quote from
Saussure. This is essentially the problem for us as empirical synchronic linguists:
we are aiming at a moving target. Ideally each element has a unique place syn-
chronically defined within the system, but realistically readjustment of the system
is a constant, dynamic process which takes time, not one which moves stepwise
from one state to the next. The system is therefore better characterised as a stable
equilibrium.2 Thus I do concur with the ideal of un système où tout se tient, but
contend that in taking a synchronic snapshot of a language at any point in time
we are actually capturing a static representation of a moving phenomenon. There-
fore, for example, in explaining some lexical exceptions, such as near synonyms
with apparently redundant morphology which emerge in Chapter 6, I take the
view that we cannot expect the system to be completely free of anomalies, but that
we should expect that the system will always be readjusting itself to resolve those
anomalies by expelling or re-assigning them. Furthermore, true synchronic study
is also realistically unattainable for practical reasons: any language, especially a
major international variety like MSA, is in use over a geographically widespread
area by speakers with a wide range of ages and social backgrounds. I consider that
anomalies in our research resulting from these sociolinguistic ­factors are to be

2. The children’s toy with a weighted base which wobbles when pushed but always rights itself
is a useful illustration.
8 The Arabic Verb

expected and, to a degree, welcomed: a study which has too narrow a focus will
be of limited interest; thus restricting our research to the language of a small ho-
mogeneous group of individuals is not an option. I will therefore treat a language
as if it is a self‑consistent, homogeneous system, but will not be unduly surprised
if a small number of anomalies defies synchronic explanation.

1.2.1.4  The Hegelian triad and scientific method

Beedham’s method seeks to challenge existing views and rules of grammar by ex-
amining “problems, contradictions, flaws, exceptions, etc.” and thus to discover a
fresh solution which is free from these inconsistencies, a process which he frames
as “the Hegelian triad of thesis, antithesis, and synthesis” (Beedham 2005: 153–
154). As an integral part of the antithesis, lexical exceptions provide effective am-
munition for attacking accepted grammatical wisdom. Beedham (2005: 155) also
views exceptions as a practical means of overcoming the infinity of language, al-
lowing the researcher to concentrate on constructions which do not occur instead
of having to examine the limitless possibilities of those that do.
In support of his method, Beedham (2005: 155) furthermore states that “lexi-
cal exceptions … render theoretical linguistics an empirical science” and thus
presents his analysis of the passive as a scientific discovery. Whilst I applaud his
intentions and support any attempt to bring empirical scientific principles to
bear on linguistic research questions, to imply that this method, and it alone, can
transform theoretical linguistics in this way is surely an overstatement. However,
its underlying principles of “new observations … new facts, which lead to new
insights and new analyses” (Beedham 2005: 160) are consistent with the following
definition of science:
A branch of study which is concerned either with a connected body of demon-
strated truths or with observed facts systematically classified and more or less
colligated by being brought under general laws, and which includes trustworthy
methods for the discovery of new truth within its own domain.  (OED)

1.2.2 An attempt to apply the method

Beedham (2005: 163–164) conveniently summarises his method in six distinct

phases. I will briefly restate each of these and use this framework to comment
upon how my own research detailed in the forthcoming chapters follows this
methodology or to what extent it departs from it.
 Chapter 1.  Introduction 9

1.2.2.1  Phase 1: Choose a formal construction

Beedham specifies that the researcher should choose a construction which has a
sufficiently large number of lexical exceptions. Moreover, for purposes of com-
parison, he requires study of the chosen construction in two or more languages in
which the researcher is fluent. The concept of fluency in Modern Standard Arabic
is somewhat nebulous in itself (see further discussion in Section 2.2.1.2), but I
will assume sufficient competency in the language to have the required degree of
intuitive insight. However, my fluency in English and in other European languag-
es will be of very little assistance in the present research, since my chosen con-
struction is morphological and thus language-specific, or at least peculiarly Se-
mitic. Whilst I have some familiarity with Hebrew, and pan-Semitic comparison
does provide some insight into the nature of the Arabic verbal system in general
(as evidenced in Chapters 3 and 4), the specific vowel lengthening verb patterns
of Arabic which become the focus of my study from Chapter 5 onwards have no
well-attested cognates in Hebrew. Thus, if I were to follow the requirements of
Beedham’s methodology rigidly, I would conclude that since no cross‑linguistic
element to the study is open to me, it is not suitable for application of the method.
However, it is my contention that while the benefits of insights from other
languages may be seen in Beedham’s own work, it is not a fundamental underly-
ing principle that the method must be applied simultaneously in more than one
language. I will therefore work here with Arabic alone. I will, however, delay sin-
gling out one particular morphological pattern as suitable for analysis until I have
surveyed the verbal system as a whole.

1.2.2.2  Phase 2: Identify the problems, anomalies, contradictions, etc.

From the outset, I have in mind the confusing, often anomalous semantics of the
Arabic verbal patterns. This, in general terms, becomes the focus of Chapter 4.
However, having in mind the morphological complexities of the verbal system,
and conscious of the methodological principle of proceeding from form to mean-
ing, I am led first to comprehensively survey all verbs in the language according
to their derivational patterns and the interactions between those patterns (Chap-
ter 2), and to attempt to gain a better understanding of the morphological system
(Chapter 3) which gives rise to verbal forms. Having established the formal basis
for the verb patterns, in Chapter 4 I explore the semantic issues involved in char-
acterising the various verbal patterns, surveying the problems, anomalies, etc. in
the system as a whole before tackling the form-meaning relationship for a specific
verb pattern (in fact, for two related patterns) in the next phase.
10 The Arabic Verb

1.2.2.3  Phase 3: Identify the unexplained lexical exceptions

In order to do this, I must formulate the thesis I wish to challenge. In Chapter 5,
I collect dictionary data for all verbs occurring in the vowel lengthening patterns,
as Beedham recommends, and establish that the dominant pattern III meaning
is mutual (or implicitly reciprocal) and, similarly, that pattern VI most often has
explicit reciprocal meaning. Thus in terms of Beedham’s method, the thesis to be
examined is that the vowel lengthening patterns give rise to mutual or reciprocal
meaning. Although no grammarian would claim this to be a prescriptive rule,
many identify it as the most characteristic tendency of these patterns and it thus
clearly merits closer scrutiny.
Note that I am proceeding from regularity of form to irregularity of meaning
directly, unlike Beedham’s applications, in which the lexical exceptions display
irregular formal properties in morphology or syntax. However, since almost one
third of all verbs (244 in total) in the patterns concerned do not fit the thesis, even
when both morphologically related patterns occur together, rather than proceed-
ing to examine the substantial set of exceptions more closely, I conclude that the
underlying basis of the thesis is flawed and turn to a different property in search
of consistent meaning.
In Chapter 6 the focus turns to the syntactic behaviour of the set of verbs in
question and especially the derivational meaning of the prefix which distinguish-
es pattern VI from pattern III. The thesis is now that the ta- prefix is detransitivis-
ing. However, even an approach to detransitivisation on the basis of numerical
valency is found to be inadequate to explain the antithetical set which constitutes
40% of the pattern VI verbs (total 154).3

1.2.2.4  Phase 4: Identify the properties of the exceptions

This phase concentrates on identifying formal properties, though Beedham recog-
nises that it may also be helpful to study the semantic properties of the exceptions.
Concerning detransitivisation, I observe in the course of compiling the data in
Chapter 6 that ta- prefixation often causes a verbal argument to change from be-
ing a direct object to an indirect object and that this phenomenon is common
amongst the exceptions. Although I do not formally compile a list of exceptions
at this stage, I have identified a property which allows me to theorise.

3. In Chapter 6 I will use the term ‘numerical valency’ to indicate the simple number of verbal
arguments contrasted with ‘hierarchical valency’ which takes account of the nature of those
arguments.
 Chapter 1.  Introduction 11

1.2.2.5  Phase 5: What might lead to the exceptions?

At this point, I depart significantly from Beedham: instead of concentrating now
on the exceptions themselves, I use what I have discovered about them to inform
my view of transitivity. Whereas Beedham looks at the set of exceptions and asks
why they are anomalous, I return to look at the entire population and ask how
I can reformulate the thesis to bring the exceptions into line and dispense with
the anomaly. In this case, I develop a hierarchical model of transitivity, which
redefines what we mean by detransitivisation in Arabic. In some respects, my
departure from Beedham’s methodology is not that great: I have arrived at the
same point as he does by a different route and have succeeded in reformulating a
grammatical rule to eliminate (most) unexplained exceptions.

1.2.2.6  Phase 6: The semantic phase

Beedham now asks whether the meaning corresponding with the formal charac-
teristic which has been identified in the exceptions is applicable to the construc-
tion as a whole. Although I have found a solution to the detransitivising prefix
which is applicable to the whole set of verbs I am interested in, this new view of
transitivity proves to be of no benefit in understanding the vowel lengthening
component of their morphology. I am no closer to discovering its meaning and
must therefore reassess the methodology.

1.3 A fresh approach to the problem

Although I have presented my work thus far as a linear process in accordance with
Beedham’s phases of research, in reality the manner in which it progressed was
more organic. Intuitive leaps do not happen to order and indeed may precede the
completion of data gathering and interpretation. Thus the investigations detailed
in Chapters 5 and 6 represent my attempt to apply Beedham’s method, which
was a significant component of my original research question, and are argued
through to conclusions, even though it became clear relatively early on that they
would not deliver the understanding of the form-meaning relationship which I
was pursuing.
With no further rule to examine, I had no means to identify corresponding ex-
ceptions in pursuit of my goal. However, returning to the principles of ­Beedham’s
methodology rather than its practical outworking, I decided to investigate where
else within the language system of MSA we might find the formal characteristic
of vowel lengthening. Chapter 7 thus begins by evaluating the formal similari-
ties between verbal pattern III (and its derivative) and broken nominal plurals
and raises the possibility of verbal plurality. I then move on to examine other
12 The Arabic Verb

nominal forms which contain the long ā vowel. Intuitively, the connection with
these nominal forms seemed promising and suggested an aspectual meaning. I
began to collect aspectual data, but soon realised that before I could analyse that
data quantitatively I had first to identify the specific aspectual property which ap-
peared to be shared by the majority of the verbs in patterns III and VI, so that I
could test for it empirically.
Chapter 8 details my search for a model of lexical aspect which could be log-
ically and consistently applied to MSA, which I discover in Olsen (1997). The
identification of the aspectual feature which characterises the majority of the
verbs in the vowel lengthening patterns is described in Chapter 9. I now find that
I have a new tentative thesis, based not on an existing rule of grammar but on my
own observations, that patterns III and VI are atelic. I therefore concentrate on
the apparent exceptions, eliminating many of them by careful application of the
aspectual model with reference to corpus examples and native speaker feedback.
A few verbs are found to be structurally anomalous and are explained accord-
ingly. However, there remains a substantial set of verbs which do not readily fit
the atelic aspectual categories and thus constitute exceptions to the thesis of atel-
icity. I draw a semantic link between these verbs in terms of their shared sense of
inceptivity. Formally, as demonstrated through corpus examples in Chapter 10,
what sets them apart is that they do not behave syntactically in combination with
the grammatical aspects in the manner we observe for other verbs. Thus I proceed
to reformulate the aspectual model, extending it to include a new category which
incorporates these inceptive verbs. In doing so, I also explain a similarly anoma-
lous set of verbs in English, demonstrating that my analysis has cross-linguistic
applicability.
I have therefore arrived in Beedham’s terms at a new synthesis which I for-
malise in the concluding chapter: that the vowel lengthening verbal patterns have
atelic aspectual meaning. In my journey I have not always adhered to Beedham’s
methodology. However, to the extent that I reached my conclusion by generating
new data, making an intuitive leap and employing lexical exceptions as a tool
towards gaining a formal understanding of what unites all instances of my cho-
sen grammatical construction, I have followed the spirit of Beedham’s method. I
should emphasise that I had no preconceived notion that the answer to my search
for meaning in these verbal patterns would be aspectual. Moreover, that I would
conclude that my research has potential application to passivisability in Arabic, as
discussed in Chapter 11, was far from my mind. It is perhaps fitting, however, that
the research which unfolds in the chapters which follow, founded on Saussure’s
principles and inspired by Beedham’s methodology, provides insight into the
workings of aspect and the passive in a language so different from the European
languages in which Beedham’s own research takes place.
 Chapter 1.  Introduction 13

1.4 Further methodological considerations

Before proceeding to examine language data, it will be helpful to clarify some mat-
ters of methodology. It will be noted in the early chapters that my method begins
with collecting and analysing dictionary data, consistent with Beedham’s own
approach. The limitations of such a data source are discussed in Section 2.2.1.2.
However, as my research progresses and begins to concentrate on specific exam-
ples, it becomes necessary to examine and verify the actual contemporary usage
of certain verbs in some detail and specifically in contexts which provide unam-
biguous evidence.
Data are obtained in two ways: questionnaires which present a range of ex-
amples to native speaker informants,4 an approach also used by Beedham, and
corpus examples of actual language. Native speaker input was found to have its
own limitations, especially as the number of informants was small and the vol-
ume of data which may reasonably be tested on volunteers is limited. The re-
sponses obtained were not always consistent and the feedback from informants
that the examples are presented without context is valid. However, question-
naires did provide the opportunity to test a small number of verbs in a range of
aspectual settings.
Thus, in order to verify contemporary usage of a wide range of verbs, I have
extensively consulted arabiCorpus, a web-based resource developed and main-
tained by Dilworth Parkinson at Brigham Young University. I have principally
used the corpus in two ways: to obtain counts of verbs in order to present data
which reflect their current frequency and to identify specific examples in context
which demonstrate aspectual and other lexical properties of the verbs under in-
vestigation. Thus many of the examples cited are from sources contained within
arabiCorpus, though I also draw upon other sources of actual modern Arabic us-
age including web pages located by searching for specific strings in Google. In this
way, I have been able to isolate examples which contextually demonstrate verbal
properties, the subtleties of which would be difficult to elicit from native speakers
using artificial sentences.

4. See Appendix III for full methodology.

 chapter 2

Verbal morphology and the lexicon

As discussed in Chapter 1, the starting point in seeking to apply Beedham’s

method of lexical exceptions is to look for irregularity in the language system,
identifying sub-classes of lexical items which do not conform to the normal rules
or patterns. Arabic is certainly a language of patterns: indeed, in common with
other Semitic languages, it is notable for the regularity of its morphology, in
which the verbal patterns are the keystone. Foreign learners of Arabic will attest
to the observation that mastering the verbs is key to learning the language and
most Arabic scholars view the verbal system as the framework around which the
language is built, as the ordering of entries in Hans Wehr’s Dictionary of Mod-
ern Arabic (Wehr 1994) and other dictionaries confirms. Arabic also strikes the
linguist as a language rich in verbs, both numerically and semantically. Even the
constituent-order typology of Classical Arabic, which is strongly VSO, points to
the prominence of the verb. However, whereas Beedham (2005: 107ff.) singles
out the irregular verbs of English, German or Russian as obvious candidates for
his method, the Arabist has no comparable set of verbs to examine. Simply put,
there are no irregular verbs in Arabic. I will discount as morphologically trivial
and extremely rare such apparent anomalies as Example (1), a common verb
subject to changes in certain parts of its paradigm motivated purely by phono-
tactic convenience, in this case the avoidance of an awkward, syllable-initial glot-
tal stop following syllable-final /r/ leading to elimination of the middle radical
(hamza) and resyllabification as in (1b):

(1) a. ‫رأى‬
ra’ā
‘he saw’
b. ‫يـرى‬
yarā
‘he sees’

Thus, while it is proposed to examine the verbal system of MSA for exceptions,
it will not be morphologically irregular verbs which will form the basis of the
study. Instead, we must look to those verbs which are anomalous in other ways.
The aim of the present chapter is to explain and document much of the lexico-
logical groundwork, thus it will serve as an introduction to the morphological
16 The Arabic Verb

structure of the Arabic verb patterns and examine how these formal patterns
are realised and distributed in the lexicon. It will examine verbal morphology
as represented throughout the lexicon both qualitatively and quantitatively. As
such, I will postpone critical examination of some of the assumptions until later
chapters: in particular, Chapter 3 will explore different approaches to Arabic
morphology, and Chapter 4 will concentrate on the semantics of the derived
verbal patterns.

2.1 Arabic verbal morphology

A brief overview of the verbal system of MSA is presented here for the benefit
of non-Arabists. Full paradigms may be consulted elsewhere: for example Reig
(1983) presents conjugations for 184 different example verbs. It is important to
realise, however, that the variations represented in these tables are almost com-
pletely predictable from sets of rules which cover the behaviour of the patterns
when weak consonants, emphatic phonemes, reduplication and the glottal stop
(hamza) are present in all the attested combinatorial possibilities, giving rise
to phonotactically motivated changes, or “morphophonological adjustments”
(Holes 2004: 110ff.). Thus, conjugations in the sense of groups of verbs which are
inflected differently, as in Latin or Greek for example, do not exist in MSA.

2.1.1 Inflectional morphology

Table 1 demonstrates the main verbal inflections using the paradigmatic triliteral
pattern I verb ‫( فـعـل‬fa‫ع‬ala – ‘to do’).5 Although I have included the labels ‘perfect’
and ‘imperfect’ for ease of reference to traditional grammars, the designations
‘s-stem’ and ‘p-stem’ are preferred by Holes (2004) and have the merit of labelling
the forms without prejudging their meanings, referring respectively to suffixed
and prefixed stems. It should be noted that the so-called prefixed stem paradigm
actually consists of the stem inflected with both prefixes and suffixes, otherwise
regarded as circumfixes (Bauer 2003: 263–264) or “discontinuous bound affixes”
(Holes 2004: 106). In fact, the paradigm given in Table 1 only shows one set of
p-stem affixes, modification to the suffixed portion giving rise to two further par-
adigms traditionally designated subjunctive and jussive.
Those unfamiliar with Arabic may wish to note that the least morphologically
complex form in the paradigms is the third-person, masculine singular of the per-
fect or s-stem, represented in standard unvowelled orthography by the root letters

5. The numerical designations of the verbal patterns are treated in Section 2.1.2.3.

 Chapter 2.  Verbal morphology and the lexicon 17

Table 1.  Perfect and imperfect inflectional paradigms (active)

Person/gender/ Perfect/s-stem Imperfect/p-stem
number
1sg fa‫ع‬al-tu ُ
‫فـعـلـت‬ ’a-f‫ع‬al-u ‫أفـعـل‬
2msg fa‫ع‬al-ta َ‫فـعـلـت‬ ta-f‫ع‬al-u ‫تـفـعـل‬
2fsg fa‫ع‬al-ti ‫ت‬
ِ ‫فـعـلـ‬ ta-f‫ع‬al-īna ‫تـفـعـلـيـن‬
3msg fa‫ع‬al-a ‫فـعـل‬ ya-f‫ع‬al-u ‫يـفـعـل‬
3fsg fa‫ع‬al-at ْ‫فـعـلَـت‬ ta-f‫ع‬al-u ‫تـفـعـل‬

2du fa‫ع‬al-tumā ‫فـعـلـتـمـا‬ ta-f‫ع‬al-āni ‫تـفـعـالن‬

3mdu fa‫ع‬al-ā ‫فـعـال‬ ya-f‫ع‬al-āni ‫يـفـعـالن‬
3fdu fa‫ع‬al-atā ‫فـعـلـتـا‬ ta-f‫ع‬al-āni ‫تـفـعـالن‬

1pl fa‫ع‬al-nā ‫فـعـلـنـا‬ na-f‫ع‬al-u ‫نـفـعـل‬

2mpl fa‫ع‬al-tum ‫فـعـلـتـم‬ ta-f‫ع‬al-ūna ‫تـفـعـلـون‬
2fpl fa‫ع‬al-tunna ّ
‫فـعـلـتـن‬ ta-f‫ع‬al-na ‫تـفـعـلـن‬
3mpl fa‫ع‬al-ū ‫فـعـلـوا‬ ya-f‫ع‬al-ūna ‫يـفـعـلـون‬
3fpl fa‫ع‬al-na ‫فـعـلـن‬ ya-f‫ع‬al-na ‫يـفـعـلـن‬

alone. Whereas speakers and scholars of Indo-European languages are most often
used to referring to verbs by their infinitives, no direct equivalent exists in Arabic
and thus the least complex form is used for citation purposes. Hence in succeed-
ing chapters, and in common with Wehr (1994), Wright (1967) and many others,
when Arabic verbs are cited as examples the English translation will be given in
the infinitive. Only when a distinction is necessary, such as in translating verbs
in context, will the person, number, gender and tense/aspect of the citation form
be explicitly rendered. Attention should also be drawn to the transliterations, in
which the full classical pronunciations are given. However, modern speakers of
the standard language will often not pronounce all the word-final short vowels
and this will vary in context according to the onset of the word which follows the
verb, but also stylistically as a matter of register and idiolect.

2.1.2 Derivational morphology

2.1.2.1  ‘Biliteral’ verbs

Although some authors, for example McCarthy & Prince (1990a), prefer to iden-
tify and enumerate verbs derived from a separate class of biliteral roots, they will
be treated here as triliteral roots having the same second and third radical conso-
nants (C2 = C3). Al-Qahtani (2005: 58–60) elaborates further upon why this as-
sumption is reasonable diachronically, citing phonological processes. Synchronic
observation reveals that while the citation form (s-stem, 3msg) of the verb in (2a)
18 The Arabic Verb

differs formally from a triliteral verb in which C2 ≠ C3, when a suffix having an
initial consonant is added to the s-stem, such as the first person singular in (2b),
the familiar s-stem sequence is evident. This rule applies throughout the s-stem
and p-stem paradigms and it could therefore be argued that identification of such
verbs as biliteral is merely an artefact of choosing a paradigmatic form with a
vowel-initial suffix as the citation form.

(2) a. ّ‫ حـب‬ ‫ضـ َرب‬

Habba (C2 = C3) Daraba (C2 ≠ C3)
‘he loved’ ‘he hit’
ُ
b. ‫ حـبـبـت‬ ُ
‫ضـربـت‬
Hababtu Darabtu
‘I loved’ ‘I hit’

Thus, for the purposes of this chapter, verbal derivational morphology will be
assumed, as is traditional, to be based upon triliteral and quadriliteral roots. The
data referred to in Appendix I are also arranged accordingly.

2.1.2.2  Pattern I triliteral stems

In Table 1 the triliteral pattern I paradigm verb ‫( فـعـل‬fa‫ع‬ala) was used to exemplify
inflectional morphology. Other pattern I variants exist having different vowelling
on the middle radical in the s-stem. In verbs where this vowel is ‘u’ (‫ – ض ّمة‬Damma),
the p-stem middle radical vowel is also ‘u’ and, where it is ‘i’ (‫ – كسرة‬kasra), the
p-stem always has ‘a’ (‫ – فتحة‬fatHa). The possible combinations are shown in Table 2
with examples. While there is clearly some systematicity relating the p-stem vowel
to its s-stem counterpart, opinions differ as to whether there is good synchronic
evidence of syntactic or semantic consistency to the different vowelling schemes,
suggesting a meaningful classification on this basis as per Wright (1967: I.30), or
whether, as Badawi et al. (2004: 60) claim, they are “best treated as a lexical feature”.
However, for the purposes of this chapter it is sufficient to note their existence and
to be aware that the vowelling may have some syntactic and/or semantic signifi-
cance. In tabulated data (see Appendix I), I have used the labels Ia, Ii and Iu to dis-
tinguish triliteral pattern I verbs according to their s-stem vowelling.

2.1.2.3  Derived or augmented triliteral stems

Grammars of Classical Arabic, for example Wright (1967), identify fourteen de-
rived triliteral patterns distinguished by Roman numerals II to XV, all of which
exhibit modifications beyond that of pattern I, which we have already met. This
system of numerical designations is generally favoured outside the Arab world,
though many Arab grammarians continue to refer to the patterns by their
 Chapter 2.  Verbal morphology and the lexicon 19

Table 2.  Pattern I variants (active)

s-stem p-stem s-stem p-stem
middle middle
vowel vowel
a a ‫ذهَـب‬ dhahaba ‫يذهَـب‬ yadhhabu to go
a i ‫ضـ َرب‬ Daraba ‫يضـرب‬
ِ yaDribu to hit
a u ‫كـتـَب‬ kataba ‫يكـتـُب‬ yaktubu to write
i a ‫شـرب‬ِ shariba ‫يشـ َرب‬ yashrabu to drink
u u ‫حـسُـن‬ Hasuna ‫يحـسُـن‬ yaHsunu to be good

t­ raditional labels, which consist of the citation forms resulting from insertion of
the consonants of the paradigm root ‫( فـعـل‬f‑‫‑ع‬l) into the s-stem derivational tem-
plates. Table 3 presents all fifteen patterns, including their numeric and traditional
word-form designations; templatic representations of the stems are also provided.
It should be noted that some forms are effectively obsolete in MSA: Wehr (1994)
does not identify any examples of patterns XIII or XV, while XI, XII and XIV are
also extremely rare in modern Arabic (see Table 10). These patterns are therefore
excluded from the analysis and discussion of data which follows, in which refer-
ence to pattern IX is also limited. Although in common use, pattern IX verbs are
few in number and highly restricted semantically to colours and defects. Note in
Table 3 that ‘(i)’ represents a vowel which may either be regarded as present in
the pattern but elided when pronounced postvocalically, or as an epenthetic vow-
el supplied together with a leading glottal stop (hamza) to enable pronunciation
when no vowel precedes, since all syllables must begin with a CV sequence and
syllable-initial CC is disallowed. McCarthy & Prince (1990a: 11–12) discuss this
further, treating this phenomenon as an epenthetic syllable which is not properly
part of the template. Comparison of these s-stem forms with their corresponding
p-stems, where the leading syllable of the s-stem is absent both in speech and in or-
thography, lends credence to their position. Compare, for example, pattern VII in
Tables 3 and 4. Attention is drawn in these tables to the vowelling of the augmented
patterns II–XV: unlike pattern I which exhibits three variants of the middle radical
vowel in both s- and p-stems, each of the augmented patterns has only one vowel-
ling scheme in the s-stem and one in the p-stem; also, whereas all augmented pat-
tern s-stems are vowelled throughout with ‘a’, p-stem vowelling, including that of
the inflectional prefix, varies from pattern to pattern. For sake of completeness, it
must also be noted that one form of expression of the passive in MSA involves pre-
dictable stem-internal vowel changes in both s- and p-stem forms. This will not be
elaborated upon here, except to highlight that this supports the notion that vowel
melody in Arabic is potentially morphemic i.e. meaning-bearing (McCarthy 1981;
1985; McCarthy & Prince 1990a), as explained more fully in Chapter 3.
20 The Arabic Verb

Table 3.  Triliteral patterns I–XV (s-stem, active)

Number Pattern designation Template (stem)
(citation form: 3msg from s-stem)
I ‫فـعـل‬ fa‫ع‬ala, fa‫ع‬ila, fa‫ع‬ula C1aC2aC3, C1aC2iC3, C1aC2uC3
II ‫فـعّـل‬ fa‫ع ع‬ala C1aC2C2aC3
III ‫فـاعـل‬ fā‫ع‬ala C1āC2aC3
IV ‫أفـعـل‬ ’af‫ع‬ala ’aC1C2aC3
V ‫تـفـعّـل‬ tafa‫ع ع‬ala taC1aC2C2aC3
VI ‫تـفـاعـل‬ tafā‫ع‬ala taC1āC2aC3
VII ‫انـفـعـل‬ (i)nfa‫ع‬ala (i)nC1aC2aC3
VIII ‫افـتـعـل‬ (i)fta‫ع‬ala (i)C1taC2aC3
IX ‫افـعـ ّل‬ (i}f‫ع‬alla (i)C1C2aC3C3
X ‫اسـتـفـعـل‬ (i)staf‫ع‬ala (i)staC1C2aC3
XI ‫افـعا ّل‬ (i)f‫ع‬ālla (i)C1C2āC3C3
XII ‫افـعـوعـل‬ (i)f‫ع‬aw‫ع‬ala (i)C1C2awC2aC3
XIII ‫افـعـوّل‬ (i)f‫ع‬awwala (i)C1C2awwaC3
XIV ‫افـعـنـلـل‬ (i)f‫ع‬anlala (i)C1C2anC3aC3
XV ‫افـعـنـلى‬ (i)f‫ع‬anlā (i)C1C2anC3ā

Notes:
C1, C2 and C3 (designated as f , ‫ ع‬and l by Arab grammarians) represent any of the 28 consonants of MSA,
(with some phonotactic and combinatorial limitations).
(i): epenthetic vowel.
Other characters in template representations are fixed consonants and vowels as in transliterated text.

Table 4.  Triliteral patterns I–XV (p-stem, active)

Number Pattern 3msg from p-stem
designation
I ‫فـعـل‬ ‫يَـ ْفـ َعـ ُل يَـ ْفـ ِعـ ُل يَـ ْفـعُـ ُل‬ yaf‫ع‬alu, yaf‫ع‬ilu, yaf‫ع‬ulu
II ‫فـعّـل‬ ‫يُـفَـعِّـل‬ yufa‫ع ع‬ilu
III ‫فـاعـل‬ ‫يُـفَـا ِعــ ُل‬ yufā‫ع‬ilu
IV ‫أفـعـل‬ ‫يُـ ْفـ ِعـ ُل‬ yuf‫ع‬ilu
V ‫تفـعّـل‬ ‫يَـتَـفَـعَّـ ُل‬ yatafa‫ع ع‬alu
VI ‫تـفـاعـل‬ ‫يَـتَـفَـاعَـ ُل‬ yatafā‫ع‬alu
VII ‫انـفـعـل‬ ‫يَـ ْنـفَـ ِعـ ُل‬ yanfa‫ع‬ilu
VIII ‫افـتـعـل‬ ‫يَـ ْفـتَـ ِعـ ُل‬ yafta‫ع‬ilu
IX ‫افـعـ ّل‬ ُّ‫يَـ ْفـ َعـل‬ yaf‫ع‬allu
X ‫اسـتـفـعـل‬ ‫يَـسْـتَـ ْفـ ِعـ ُل‬ yastaf‫ع‬ilu
XI ‫افـعـا ّل‬ ُّ‫يَـ ْفـ َعـال‬ yaf‫ع‬āllu
XII ‫افـعـوعـل‬ ‫يَـ ْفـعَـوْ ِعـ ُل‬ yaf‫ع‬aw‫ع‬ilu
XIII ‫افـعـوّل‬ ‫ـو ُل‬ ِّ ‫يَـ ْفـ َع‬ yaf‫ع‬awwilu
XIV ‫افـعـنـلـل‬ ‫يَـ ْفـ َعـ ْنـلِـ ُل‬ yaf‫ع‬anlilu
XV ‫افـعـنـلى‬ ‫يَـ ْفـ َعـ ْنـلِي‬ yaf‫ع‬anlī
 Chapter 2.  Verbal morphology and the lexicon 21

2.1.2.4  Quadriliteral stems

Similarly, roots comprising four consonants give rise to a base pattern and three
augmented patterns, also designated by Roman numerals. In order to distinguish
these clearly from the triliteral patterns I have adopted the convention of prefixing
the numeral with ‘Q’. Pattern QIV is rare and QIII probably obsolete in MSA. In
Tables 5 and 6, the practice of traditional grammars, such as Wright (1967: I.48),
in using ‫  فـعـلـل‬as the paradigmatic base pattern is retained, though it should not
be inferred that this necessarily implies that the third and fourth radicals are iden-
tical, as the templatic representation in Table 5 makes clear.

Table 5.  Quadriliteral patterns I–IV (s-stem, active)

Number Pattern (citation form) Template (stem)
QI ‫فـعلـل‬ fa‫ع‬lala C1aC2C3aC4
QII ‫تـفـعلـل‬ tafa‫ع‬lala taC1aC2C3aC4
QIII ‫افـعـنـلـل‬ (i)f‫ع‬anlala (i)C1C2anC3aC4
QIV ‫افـعـلـ ّل‬ (i)f‫ع‬alalla (i)C1C2aC3aC4C4

Table 6.  Quadriliteral patterns I–IV (p-stem, active)

Number Pattern designation 3msg from p-stem
QI ‫فـعلل‬ ‫يُـفَـعْـلِـ ُل ز‬ yufa‫ع‬lilu
QII ‫تـفـعلل‬ ‫يَـتَـفَـعْـلَـ ُل ؤ‬ yatafa‫ع‬lalu
QIII ‫افـعـنـلل‬ ‫يَـ ْفـ َعـ ْنـلِـ ُل ز‬ yaf‫ع‬anlilu
QIV ‫افـعـل ّل ء‬ ‫يَـ ْفـ َعـلِـ ُل ر‬ yaf‫ع‬alillu

2.1.2.5  Morphological connections

It will be useful to observe at this stage that certain triliteral and quadriliteral pat-
terns share morphological features. Some of the issues raised here bear further
investigation.
2.1.2.5.1  The putative ‘t’ affix. Patterns V, VI, VIII, X and QII share the consonant
‘t’ as part of their template, distinct from the root consonants. Application of con-
ventional concatenative morphology to the pairs II and V, III and VI and QI and
QII would suggest that ‘t’ or ‘ta-’ is a prefix which derives the second member of
the pair from the first and is thus morphemic in nature, raising a number of ques-
tions. Does prefixed ‘t’ consistently carry a semantic burden or fulfil a syntactic
function across these patterns as well as being formally consistent, such that it
is identifiably morphemic? Is ‘t’ in patterns VIII and X an infixed version of the
same morpheme and, if so, to which base patterns do these relate? Alternatively,
is infixed ‘t’ a different morpheme altogether?
22 The Arabic Verb

2.1.2.5.2  Shared vowel melody. Observe again the pairs II and V, III and VI and
QI and QII: the unprefixed members of each pair all share the p-stem vowel mel-
ody ‘u‑a‑i’, while the corresponding ‘t’ prefixed patterns share the vowel melody
‘a‑a‑a‑a’. It has already been noted, following McCarthy (1981) and McCarthy &
Prince (1990a), that vowel melody in Arabic may be morphemic. This may there-
fore suggest that II, III and QI and their ‘t’ prefixed counterparts may share com-
mon semantic features. Likewise, VII, VIII and X (and indeed many of the higher
forms) also share the p-stem vowel melody ‘a‑a‑i’, suggesting that there may be a
common semantic component here too.
2.1.2.5.3  Alternative classification scheme. It is possible to reclassify the Arabic
verbal patterns on the basis of their morphological features. This is particularly
helpful for Semiticists working cross-linguistically, facilitating reference to cog-
nate forms. Table 7 is based on such a scheme by MacDonald (1963), classifying
patterns I to X (excluding IX). MacDonald describes G, D, C and N as the basic
patterns from which other Semitic patterns derive. The L pattern is treated as a
special case of D and hence represented as subsidiary to it, following O’Leary
(1969: 217), who views the lengthened vowel as resulting from failed consonantal
gemination. Observe that the right-hand section of the table contains the patterns
modified with ‘t’. It should be noted that certain other Semiticists, such as Ryder
(1974), refer to G (Grund) as the B (base) stem.

Table 7.  Patterns classified according to morphology (after MacDonald 1963: 100)

ground G I Gt VIII
doubled D II Dt V
[lengthened] [L] III [Lt] VI
causative C IV Ct X
na-reflexive N VII Nt ?

MacDonald, a university lecturer, saw didactic advantages in this classification,

though it is not unproblematic. Firstly, as a matter of consistency in nomencla-
ture, C is a semantic label, not a morphological one, and risks prejudging the
meaning associated with the form, as discussed more fully in Chapter 4. Use of
the designation S (or Š) for pattern IV is probably preferable on these grounds,
though this is morphologically more typically Eastern Semitic and results in ter-
minological opaqueness for pattern IV, though not for X, which would become
St in this scheme. Leemhuis (1977) prefers to designate IV as H which is typically
Western Semitic, but has the disadvantage of opaqueness for both patterns IV and
X. It is not appropriate to discuss the Semitic origins of pattern IV here, nor the
 Chapter 2.  Verbal morphology and the lexicon 23

question of whether pattern X is its ‘t’ derivative: I explore these issues further in
Danks (2007). Another consequence of arranging the patterns in this scheme is
that there is an assumption that all the ‘t’ derivatives share a morpheme, whether
prefixed or infixed, i.e. that they share a common semantic component realised by
‘t’. MacDonald (1963: 104) refers to it as “reflexive ta-”, which is indicative that his
scheme classifies patterns V, VI, VIII and X not purely according to morphologi-
cal form but also on semantic grounds, the validity of which is at best question-
able, as the diversity of tabulated meanings in Chapter 4 will testify. However,
perhaps the most obvious flaw in this scheme is to regard N as a basic pattern,
leading to the notable absence of any pattern Nt in Classical or Standard Arabic,
though O’Leary (1969: 226) identifies a pattern in the dialect of Tlemcen (North-
Western Algeria) which is formally, if not necessarily semantically, a candidate for
Nt. Indeed, MacDonald’s designation of N and ‘t’ as both reflexive highlights the
difficulty and there may be grounds to suggest that pattern VII might more help-
fully be included in this scheme as a derivative of G, hence Gn.

2.2 Distribution of verbal patterns in the lexicon

The hypothesis that certain verb patterns in MSA are morphologically related was
introduced above. This is relatively uncontroversial inasmuch as it is self-evident
from examination and comparison of formal realisations. Moreover, it is a key
concept in morphology that a morpheme is a meaning-bearing unit of language,
an indivisible unit of morph and seme: Saussure’s signifiant and signifié, or form
and meaning.
It has already been stated that, subject to certain phonological and combinato-
rial constraints, a root consisting of any combination of three or four consonants
can combine with any of the corresponding verbal pattern templates to produce
a verb. In practice, however, no root gives rise to verbs on all possible patterns,
not even on each of the nine most common. It is conceivable that the distribution
of verb forms within the lexicon is entirely random. If, however, morphological
relationship implies not just commonality of form but also commonality of se-
mantics, we should see a degree of organisation to the lexicon, resulting in signifi-
cantly greater than chance co-occurrences of patterns for a given root where one
is derived from another. Furthermore, we would expect to be able both to identify
morphemes formally and to characterise them semantically on the basis of how
they derive verbs from the roots with which they combine.
24 The Arabic Verb

2.2.1 Data collection

In order to investigate the distribution of verbs within the lexicon according to

root and pattern, it was necessary to undertake an exhaustive survey of triliteral
and quadriliteral roots, not merely to record the number of verbs in each pattern
(these figures are available elsewhere: see Section 2.2.2), but to provide a compre-
hensive ‘map’ of how verbs in various patterns occur together for the same roots.

2.2.1.1  Method
Using Wehr’s Dictionary of Modern Written Arabic (1994), a comprehensive list-
ing of all triliteral roots which give rise to verbs in one or more patterns was com-
piled (including roots in which the second radical is doubled) in the form of a
spreadsheet. All attested verbal patterns in the range I to X (excluding IX) for each
root were tabulated, and instances of the other six possible morphological forms
noted where applicable. The exercise was repeated for quadriliteral roots giving
rise to one or more of patterns QI to QIV. A small extract from the table of triliteral
data is reproduced as Table 8. Note that root consonants appear in left-right order;
where C2 is doubled, the root appears in the table with C3 = C2 but conforms to the
dictionary ordering; under the respective patterns, ‘1’ indicates that the verb is at-
tested and a blank cell indicates not attested; multiple root entries are dealt with in
Section 2.2.1.2.2. Appendix I provides a full listing of data tables generated as part
of this research together with details of how they may be consulted.

2.2.1.2  Lexical issues

Modern Standard Arabic presents unique lexicological challenges. What does
‘modern’ mean for a language which relies for its grammatical forms and much of
its lexicon upon a classical language preserved with only minor changes since the
6th Century, due to its daily use by Muslims in the form of the Qur’an? What does
‘standard’ mean for a language spoken in a diverse range of dialects across a large
swathe of North Africa and the Middle East? If we define MSA as the contemporary
language of literature and official communication which is common to separate na-
tion states throughout the region, this at least presents a realistic target for the lexi-
cographer. Although it may only be practical to produce a dictionary of the written
language, this effectively defines what is ‘standard’ for the spoken language also.
Arabic, as a language with a dual system of communication (diglossia) confronts
the lexicographer with a variety of unexpected problems whose solution is not
always easy…. Before venturing any further step, the lexicographer had [sic] to
solve the cardinal question: What is [Modern Written Arabic] and what should
its lexicon look like? Or, in other words, the lexicographer has to define the iden-
tity of MWA in terms of its lexicon.  (Drozdík 1998: 211–212)
 Chapter 2.  Verbal morphology and the lexicon 25

Table 8.  Extract from data tabulated by root and pattern

Other
C1 C2 C3 I II III IV V VI VII VIII X Entry patterns
Ώ ϭ ϕ 1
Ώ ϭ ϝ 1 1 1 1
Ώ ϱ Ε 1 1 1
Ώ ϱ Ω 1 1
Ώ ϱ ν 1 1 1 IX
Ώ ϱ ω 1 1 1 1 1 1
Ώ ϱ ϥ 1 1 1 1 1 1 1
Ε ˯ έ 1
Ε Ώ Ώ 1 1
Ε Ώ έ 1
Ε Ώ ω 1 1 1 1 1 1 1
Ε Ώ ϝ 1 1 1
Ε Ν έ 1 1 1
Ε Ρ ϑ 1
Ε Υ Υ 1
Ε Υ ϡ 1 1 1 1
Ε Υ ϡ 1 1 2
Ε έ Ώ 1 1 1 1 1
Ε έ Ρ 1 1 1 1
Ε έ α 1
Ε έ ω 1 1
Ε έ ϑ 1 1 1

8: Extract from data tabulated by root and pattern

Insofar asTable
the current task of comprehensive data collection is concerned the res-
ervations which follow are noted, but for practical purposes cannot reasonably be
addressed. However, as closer examination of specific examples is undertaken, it
may be appropriate to revisit these issues.
2.2.1.2.1  Choice of dictionary. It is recognised that the method of data collection
employed here relies solely upon a single dictionary. However, Wehr’s work,
consulted here in translation from the original German, commands consider-
able respect:
26 The Arabic Verb

The excellence of Wehr’s Arabic-German dictionary was recognized in leading

centers of Arabic studies almost immediately after the appearance of the first edi-
tion (1952). The dictionary, compiled on sound lexicographical principles, pres-
ents the vocabulary and basic phraseology of Modern Written Arabic (MWA),
sometimes referred to as Modern Standard Arabic, the only codified variant of
present-day Arabic and the only one with an indisputable pan-Arab validity.
 (Drozdík 1998: 211)
The magnitude of the task of manually cross-checking every entry in a range of
other dictionaries would place it outside the scope of this research project. More-
over, Wehr has been the choice of others who have undertaken similar surveys
(McCarthy & Prince 1990a; Al-Qahtani 2003, 2005), which also has the merit of
allowing comparisons to be drawn.
2.2.1.2.2  Homomorphous (homonymous) roots. Where Wehr has listed more than
one entry for a particular root combination, and two or more of these give rise to
verbs, I have made separate, annotated entries in the data table e.g. ‫( تـخـم‬t‑kh‑m)
in Table 8. The rationale behind Wehr’s treatment of these roots does allow that
possible inconsistencies will result:
In the presentation of the entries in the dictionary, homonymous roots are given
separately in only a few especially clear instances. The arrangement of word en-
tries under a given root does not necessarily imply etymological relationship.
Consistent separation was dispensed with because the user … will not generally
be concerned with Semitic etymology.  (Wehr 1994: x)

Thus, Wehr’s decision to list separately or under a single entry may appear arbitrary.
For example, there are two entries for the root combination ‫( وزر‬w‑z‑r),
which I have tabulated accordingly (data extracted in Table 9). The verbs under
entry 1 share a common meaning involving bearing or taking on a burden or
sin (‫ – وزر‬wizr) and thus the decision to include them together seems justified.
Likewise, the verbs formed on patterns V (‘to become a minister’) and X (‘to
appoint as minister’) are classified under a separate entry, presumably identi-
fying a denominative derivation from ‫( وزيـر‬wazīr – ‘a minister’). However, a
complication arises in pattern VIII, where the verb can mean ‘to commit a sin’
or ‘to wear a loincloth’. We might speculate that Wehr’s decision to include ‫وزرة‬
(wizra – ‘a loincloth’) under entry 1 is motivated by a connection both in the
Qur’an (7: 22,26) and in the Hebrew scriptures which predate it (Gen. 3: 7,21),
where the sin of Adam and Eve results in their being clothed, initially with leaves
and subsequently with garments. In fact the word ‫( وزرة‬wizra) itself is to be
found nowhere in the Qur’an, though a later link cannot be discounted. Thus it
seems reasonable to assume that this is an example where Wehr has included the
­meaning together with others in the absence of clear etymological evidence to
 Chapter 2.  Verbal morphology and the lexicon 27

the contrary. We should be wary of making fanciful connections diachronically,

and in any case it is clear that this meaning of the pattern VIII verb is incongru-
ous synchronically, thus it appears that it deserves a separate entry. For reasons
of practicality Wehr’s classifications have been retained for the purposes of data
collection, though it should be considered in later chapters, where the focus is on
semantics, that there are often grounds for separating entries which have hereto-
fore been treated together.

Table 9.  Multiple entries for the root combination w-z-r

Other
C1 C 2 C3 I II III IV V VI VII VIII X Entry patterns
ϭ ί έ 1 1 1 1 1 1
ϭ ί έ 1 1 2

2.2.1.2.3  Colloquialisms. It must be accepted that there will be many instances

where dialectal words or colloquialisms are in regular use in the standard lan-
guage, often restricted by country or region. Again Wehr elaborates on this is-
sue in his introduction (1994: viii), and the decision was therefore taken that all
entries considered by Wehr to merit inclusion in a Dictionary of Modern Written
Arabic should also be included in the table here.
2.2.1.2.4  Synchronic gaps, classicisms and semantic change. Only a few verb forms
are precluded on purely morphological grounds: for example, roots beginning
with ‫( ن‬n) do not combine with pattern VII which utilises prefixed ‘n’. Thus non-
appearance of a given verb form in Wehr is not proof that it is not possible, has
never been in use or is not in current use somewhere within the Arabic-speaking
continuum. Conversely, attestation in the dictionary does not prove that a given
form is in true synchronic usage, as, concerning classicisms, Wehr (1994: ix) com-
ments that “Arab authors can and do frequently draw upon words which were al-
ready archaic in the Middle Ages” including “quotations from the Koran or from
classical literature” and therefore admits that “it is not possible to make a sharp
distinction between living and obsolete usage”. Moreover, there is the related
problem that where a pattern is attested for a root, there may be discrepancies
between its synchronic and classical semantic and syntactic usages.

2.2.2 Comparison of data with existing sources

I have already alluded to data collected and published by McCarthy & Prince
(1990a) and Al-Qahtani (2003, 2005). Summaries of their data, together with my
own totals, are presented in Table 10, which includes explanatory notes on some
of the totals calculated from Al-Qahtani’s figures.
28 The Arabic Verb

Table 10.  Lexical frequencies for triliteral and quadriliteral patterns

AL-QAHTANI McCARTHY & PRINCE DANKS
pattern ‘main entry’ ‘derived’ total biliteral triliteral total total
I 2512* 270 2299 2569 2523
II 290 1189 1479** 127 1271 1398 1416
III 27 428   455   18   445   463   465
IV 47 879   926   78   873   951   938
V 41 899   940   85   940 1025   953
VI 5 370   375   17   377   394   389
VII 7 244   251   31   229   260   267
VIII 12 578   590   68   553   621   606
IX 2 16    18    0    18    18    19
X 4 389   393   34   355   389   395
XI 0 2     2    0     2     2     2
XII 1 6     7 N/A     7     7     7
XIV 0 1     1 N/A     2     2     1
total total total
QI 275 296   294
QII   ** 111   131
QIII    †    1     1
QIV    3‡    8     8

Notes:
1. Al-Qahtani records his data in Semantic Valence of Arabic Verbs (2005: 54–90), referring to the list of
8327 verbs published as A Dictionary of Arabic Verbs (2003).
2. The counts by McCarthy & Prince are taken from their article Prosodic Morphology and Templatic
Morphology (1990a: 33–34).
3. No verbs in patterns XIII and XV are recorded in any of the three counts.
* This total was obtained by adding the figures for what Al-Qahtani terms biliteral and triliteral ‘ground’
verbs and subtracting 99 which he has listed separately due to medial vowelling variants (2005: 65).
** Al-Qahtani has included the ta- prefixed quadriliteral derivative verbs, elsewhere designated QII, in his
figures for pattern II. He does not, however, enumerate them separately.
† Al-Qahtani does not state where the total for this pattern has been included, if anywhere.
‡ This figure comprises verbs included by Al-Qahtani under the heading ‘quintiliteral’.

His methodology at times seems at odds with a synchronic, descriptivist per-

spective:
There are drawbacks in studying Arabic verbs in isolation from the numerous
studies conducted by Arab grammarians. This work has been done after careful
reading and consideration of their work, including contemporary works done by
those who were trained on the traditional framework of Arabic linguistics and
philology. What they have in common is that their studies on Arabic verbs are
based on logic and prescriptivism.  (Al-Qahtani 2005: 53)
 Chapter 2.  Verbal morphology and the lexicon 29

Although his system of classification reveals some unusual assignments and at

times cuts across well-established morphological categories, with the excep-
tion of the number for pattern II, where he strangely includes QII verbs, the
totals calculated from Al-Qahtani’s data are otherwise close to mine. A pos-
sible explanation for the minor discrepancies seen in the other totals is that Al-
Qahtani­ used as his source the 1974 reprint of the 3rd Edition of Hans Wehr’s
dictionary as opposed to my use of the 1994 reprint of the 4th Edition, which
includes additional material.6 This, and the fact that Al-Qahtani states that he
has excluded “verbs … coming from dialects such as Egyptian or Syrian, for
example” (2005: 55), explains why my totals are consistently somewhat higher.
It is less clear why there are discrepancies between Al-Qahtani and McCarthy
& Prince, since the latter cite a 1971 printing of Wehr, which is also the 3rd
edition. Particularly striking, and unexplained, is pattern V, where my total and
Al-Qahtani’s agree closely, while that of McCarthy & Prince is very different. In
the table of quadriliterals my significantly higher total for pattern QII may be
attributable to its productivity for borrowings, reflected in new entries in the
4th Edition of the dictionary.

2.2.3 Quantitative analysis of the data

There are a number of ways to analyse the raw data in order to gain insights into
how patterns occur by root.

2.2.3.1  Productivity by root

Table 11 shows the total numbers of triliteral and quadriliteral roots and the num-
bers of verbs derived from them, enabling the mean number of patterns repre-
sented per root to be calculated.

Table 11.  Mean patterns per triliteral and quadriliteral roots

No. of roots No. of verbs Mean patterns / root
Triliteral 2963 7981 2.69
Quadriliteral   350   434 1.24

6. For the complex publication history of Wehr’s dictionary in both German and English edi-
tions see Drozdík (1998).
30 The Arabic Verb

Tables 12 and 13 show frequencies for patterns per root for triliterals and quadri-
literals respectively. Roots which do not give rise to verbs have not been included
in the data, i.e. those with zero actual patterns per root, thus mean patterns per
root values must be interpreted accordingly. No triliteral roots were found to give
rise to verbs in more than nine patterns and no quadriliteral roots in more than
three. The distribution of roots in these tables shows that relatively few are highly
productive, while the category which is most numerous (more than one-third of
triliteral roots and over three-quarters of quadriliteral roots) is that which repre-
sents only one pattern per root. Thus lexical gaps, consisting of unused patterns
for given roots, are numerous.

Table 12.  Actual patterns per triliteral root

Actual patterns / root 1 2 3 4 5 6 7 8 9
No. of roots 905 696 532 377 246 124 62 17 4
Percentage 37.1 23.5 18.0 12.7 8.3 4.2 2.1 0.6 0.1

Table 13.  Actual patterns per quadriliteral root

Actual patterns / root 1 2 3
No. of roots 267 82 1
Percentage 76.3 23.4 0.3

2.2.3.2  Productivity by pattern

Of greater interest is productivity by pattern. Tables 14 and 15 reproduce my data
from Table 10, together with percentages of roots represented by each pattern. It
must be remembered that these figures can only reflect how productive a pattern
has been in the language up to this point in time. They cannot necessarily predict
whether a pattern will be more or less productive in the future for neologisms and
borrowings.

Table 14.  Percentage productivity by triliteral pattern

Pattern I II III IV V VI VII VIII IX X XI XII XIV
Freq. 2523 1416 465 938 953 389 267 606 19 395 2 7 1
% of roots 85.2 47.8 15.7 31.7 32.2 13.1 9.0 20.5 0.6 13.3 0.1 0.2 0.0

Table 15.  Percentage productivity by quadriliteral pattern

Pattern QI QII QIII QIV
Frequency 294 131 1 8
% of roots 84.0 37.4 0.3 2.3
 Chapter 2.  Verbal morphology and the lexicon 31

2.2.3.3  Co-occurrence of patterns

2.2.3.3.1  Overview. Clearly from the statistics presented above, some patterns oc-
cur more commonly in the lexicon than others. Since phonological constraints
are minimal, it is likely that the frequency with which a pattern occurs depends
on how semantically or syntactically ‘useful’ it is. Nevertheless, if the patterns are
all independent of one another, we would expect to see them distributed random-
ly amongst the roots. My data from Table 14 may be used to calculate predicted
frequencies for co-occurrence based on such a random distribution. For example,
if 31.7% of triliteral roots form a verb on pattern IV and 13.3% on pattern X, ran-
dom distribution predicts that 31.7% × 13.3% = 4.2% of roots will form verbs on
both patterns. By comparing actual co-occurrence figures with these predictions
it will be possible to identify which patterns have a tendency to occur together or
to be mutually exclusive.
Table 16 contains actual co-occurrence frequencies derived from my raw data
for all pairs of patterns from I to X, excluding pattern IX which, together with the
higher patterns, is excluded from further analysis as their frequencies are too low
to be reliably tested for statistical significance.

Table 16.  Actual co-occurrence frequencies (triliterals)

I II III IV V VI VII VIII X
I
II 1130
III 416 258
IV 853 558 216
V 816 661 196 379
VI 359 230 238 193 172
VII 256 136   54   98 116   41
VIII 567 318 183 245 274 160 75
X 365 265 128 267 211 130 32 128

In Table 17 predicted frequencies are presented as calculated above. There are

some striking differences between predicted and actual co-occurrences, for ex-
ample for the morphologically related patterns III and VI, but there appear to
be other more unexpected correlations, including those involving patterns VIII
and X with III and VI. However, it is not sufficiently rigorous to base conclusions
concerning co-occurrences on the raw data presented in these tables. In order to
assess the likelihood that co-occurrences are real phenomena and not mere coin-
cidences and to quantify the degree of correlation, the data must be examined for
statistical significance.
32 The Arabic Verb

Table 17.  Predicted co-occurrence frequencies (triliterals)

I II III IV V VI VII VIII X
I
II 1206
III   396 222
IV   799 448 147
V   811 455 150 302
VI   331 186   61 123 125
VII   227 128   42   85   86 35
VIII   516 290   95 192 195 80 55
X   336 189   62 125 127 52 36 81

2.2.3.3.2  Chi-square test methodology. The chi-square test is particularly suitable

for frequency data where the linguist wishes to compare actual frequencies with
those predicted by a particular hypothesis. For two independent variables A and
B, each of which can be present or absent (designated here as + or − respectively),
a 2 by 2 table is constructed, and data tested for significance according to the
method outlined here (Butler 1985: 118–121; Connor-Linton 2003). Actual fre-
quencies for each of the four logical possibilities are inserted into the table tem-
plate (Table 18), then predicted frequencies (assuming a null hypothesis of no
relationship between the variables) are calculated as in Table 19, where:
C = actual frequency of (A+ and B+)
D = actual frequency of (A+ and B−)
E = actual frequency of (A− and B+)
F = actual frequency of (A− and B−)

Table 18.  Chi-square testing: 2 by 2 table template for actual frequencies

Actual frequencies A+ A−
B+ c e
B− d f

Table 19.  Chi-square testing: 2 by 2 table template for calculating predicted frequencies

Predicted frequencies A+ A−
B+ (c+d)(c+e) / (c+d+e+f) (e+f)(c+e) / (c+d+e+f)
B− (c+d)(d+f) / (c+d+e+f) (e+f)(d+f) / (c+d+e+f)

The chi-square (χ2) value is calculated according to the following formula:

χ2 = ∑ ((actual-predicted)2/predicted) for each of the four cells in the table
 Chapter 2.  Verbal morphology and the lexicon 33

Significance can now be tested at various levels, by comparing the calculated chi-
square value with standard values. For 2 by 2 tables the parameter df (degrees of
freedom) = 1. Some of the relevant standard values are reproduced in Table 20,
in which significance at a given probability requires calculated chi-square to be
greater than or equal to the standard value. The probability level represents the
null hypothesis, namely that there is no relationship between the variables. Thus,
for example, if the calculated chi-square value lies between 10.83 and 15.14, the
probability that the variables are independent is somewhere between one chance
in 1000 and one in 10,000, or conversely there is a greater than 99.9% probability
that they are related.

Table 20.  Standard values of χ2 with df = 1 (Harter 1964: 234–239)

Probability level 0.05 0.01 0.001 0.0001
χ2 3.84 6.64 10.83 15.14

In addition to the chi-square value, a related parameter, the phi coefficient (Φ)
can be calculated (Butler 1985: 148–149):

Φ = (cf−de)
√((c+e)(d+f)(e+f)(c+d))

Whereas chi-square gives a measure of the probability that the variables are re-
lated, the phi coefficient measures to what degree the variables are related, either
positively or negatively: the greater the magnitude of Φ, the greater the degree of
correlation.
2.2.3.3.3  Chi-square and phi coefficient values for triliteral pattern co-occurrences.
Table 21 shows example data for the actual numbers of roots with patterns III and
VI present or absent and (in square brackets) the predicted values for the same
combinations of these variables, calculated according to Table 19. Chi-square
and phi coefficient values have also been calculated according to the formulae
given, and probability values obtained from standard chi-square tables (Harter
1964: 234–239).

Table 21.  Chi-square calculation for co-occurrence of patterns III and VI

+ pattern III − pattern III
+ pattern VI 238 [61.05]   151 [327.95]
− pattern VI 227 [403.95] 2347 [2170.05]
Chi-square (χ2) = 700 | Probability (p) < 0.0001 | Phi coefficient (Φ) = 0.49
34 The Arabic Verb

In Table 22 values of χ2, p and Φ have similarly been generated for all combina-
tions of patterns I–VIII and X. Some general observations concerning the inter-
pretation of these values will be helpful.

Table 22.  χ2, p and Φ values for co-occurrences of triliteral patterns

I II III IV V VI VII VIII
II (61.3) (chi-square χ2)
<.0001 probability p
[−0.14] [phi coefficient Φ]
III (8.11) (13.1)
<0.01 <0.001
[0.05] [0.07]
IV (36.4) (75.3) (55.2)
<.0001 <.0001 <.0001
[0.11] [0.16] [0.14]
V (0.25) (262) (25.2) (42.7)
>0.2 <.0001 <.0001 <.0001
[0.01] [0.30] [0.09] [0.12]
VI (18.0) (23.1) (700) (66.7) (29.8)
<.0001 <.0001 <.0001 <.0001 <.0001
[0.08] [0.09] [0.49] [0.15] [0.10]
VII (26.7) (1.17) (4.55) (3.45) (17.1) (1.28)
<.0001 >0.2 <0.05 <0.1 <.0001 >0.2
[0.09] [0.02] [0.04] [0.03] [0.08] [0.02]
VIII (42.7) (6.70) (121) (27.1) (59.5) (117) (10.5)
<.0001 <0.01 <.0001 <.0001 <.0001 <.0001 <0.005
[0.12] [0.05] [0.20] [0.10] [0.14] [0.20] [0.06]
X (19.0) (68.0) (96.2) (272) (94.4) (156) (0.46) (40.0)
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 >0.2 <0.0001
[0.08] [0.15] [0.18] [0.30] [0.18] [0.23] [−0.01] [0.12]

Firstly, the majority of cells in the table show extremely low values for p, even
though the associated phi coefficients may be relatively low. This is because the
data set is large, so even for weakly correlated patterns, the probability that they
are unrelated is small. In high-risk situations, such as medical trials for an ex-
pensive new drug or indeed in comparing texts in forensic linguistics, it is cus-
tomary to set a low value for p when looking for statistical significance. For our
purposes, where the consequences of misidentifying a correlation are relatively
minor, specifying a value of p < 0.01 (a 99% probability that a correlation is not
random) is highly conservative. Even on these stringent grounds, only six of the
36 combinations of pattern pairs fail the test for significance of co-occurrence. We
may therefore state with a considerable degree of certainty that the distribution
 Chapter 2.  Verbal morphology and the lexicon 35

of verbs according to morphological patterns in the lexicon is not random: there

must be morphosyntactic and/or morphosemantic factors which give rise to these
correlations. This may not be surprising news to scholars of Arabic, but it is objec-
tive evidence of form–meaning relationship in the verbal system of MSA and an
encouragement to pursue further investigation.
Secondly, care must be taken in interpreting the phi coefficient, which repre-
sents the degree to which variables are correlated but does not in itself establish
cause and effect. Butler (1985: 149–150) uses an example in which the height of
school pupils correlates highly with their scores in tests. Here there is a third vari-
able (age) which has a causal effect on both the other variables. The phi coeffi-
cients in Table 22, also represented graphically for ease of comparison in Figure 1,
confirm a high degree of correlation, as expected, between patterns which bear
close morphological relationships such as [III and VI] and [II and V]. The value
for [IV and X] lends support to the less obvious derivational relationship here

0.5
I
II
0.4 III
IV
V
VI
0.3
VII
VIII
X
Phi coefficient (Φ)

0.2

0.1

–0.1

–0.2
I II III IV V VI VII VIII X
Pattern

Figure 1.  Graphed phi coefficients for pattern co-occurrences (triliterals)

36 The Arabic Verb

also.7 These three pairs give rise to the largest phi coefficients highlighted in bold
type in the table. However other pairs also show a relatively high degree of cor-
relation, for example [VI and VIII]. This indicates that the presence of a pattern
VI verb for a given root is a relatively good predictor that the root will also have a
verb in pattern VIII and vice-versa.
However, it does not imply that a causal relationship exists between these
two patterns. In fact, both patterns independently show relatively high correla-
tion with pattern III, so it is possible for example that III and VI are related mor-
phologically while III and VIII are related syntactically or semantically. Quanti-
fied correlations allow claims which are not always backed by data to be assessed
for validity. For example, McCarthy & Prince (1990a: 34) state that “Form 5 …
tends to occur only together with roots having Form 2…, Form 6 with Form 3,
and Form Q2 with Q1”. Although this statement is perhaps phrased too strongly
and the use of the word ‘only’ is ambiguous, the tendency they identify is sup-
ported by my data for the triliterals.8 However, they also claim that “[t]his de-
pendency between different conjugations … is otherwise unknown in the Arabic
verb system”, though my figures demonstrate that the correlation between pat-
terns IV and X is as strong as that between II and V. What we can say is that the
correlations which exist between specific patterns certainly merit further inves-
tigation and explanation.
Lastly, a word on the phi coefficients for [I and II] and [VII and X], which
are the only two negative values in the table. The latter pair shows a barely nega-
tive correlation, which is not statistically significant and would therefore not be
worthy of further discussion, were it not that McCarthy & Prince (1990a: 38) have
claimed a negative correlation for this pair, stating that “[t]he scarcity of roots
that take both 7 and 10 is significant at the .05 level”. Similarly, they suggest that
roots avoid forming verbs in both patterns VII and VIII, whereas I demonstrate
a strong likelihood of a small positive correlation. Clearly they have taken the
trouble to analyse their data statistically, but since their methodology is not dis-
cussed it will not be possible to speculate as to the difference in their findings.
However, it is worth noting that patterns I and II are to some degree negatively
correlated according to my data, i.e. that there is some tendency for them not to
occur together. It is likely that this is largely due to the fact that these are the two
patterns most likely to occur in isolation: there are 636 roots which only exist in
pattern I and 160 only in pattern II.

7. See also Danks (2007).

8. See Section 2.2.3.3.4 for discussion of the quadriliterals.
 Chapter 2.  Verbal morphology and the lexicon 37

2.2.3.3.4  Quadriliteral pattern co-occurrences. Thus far we have excluded quadri-

literal pattern co-occurrences from the discussion. With only four patterns in to-
tal, two of which are rare in the lexicon, the only co-occurrence relationship which
is suitable for statistical treatment is [QI and QII], although this relationship is of
particular interest for its analogy to [II and V] and [III and VI] as discussed in
Section 2.1.2.5.1. Examination of the calculated values in Table 23 reveals that it is
almost certain that the two patterns are quite highly negatively correlated, i.e. that
QII has a tendency to occur independently of QI with significantly greater than
chance frequency. This is somewhat surprising and in direct contradiction of the
claim made by McCarthy & Prince (1990a: 34). The considerably higher diction-
ary count for QII in my data, based on a more recent edition of Wehr, may be
partially responsible for the discrepancy, particularly in terms of its productivity
for new borrowings, but further speculation is again impossible without access to
the methodology of McCarthy & Prince.

Table 23.  Chi-square calculation for co-occurrence of patterns QI and QII

+ pattern QI − pattern QI
+ pattern QII   82 [110.04] 49 [20.96]
− pattern QII 212 [183.96]   7 [35.04]
Chi-square (χ2) = 71.4 | Probability (p) < 0.0001 | Phi coefficient (Φ) = −0.45

2.3 Summary

The verbal morphology of MSA is highly systematic and lends itself to quantita-
tive analysis which reveals non-random distributions of verbal patterns by root
within the lexicon. Some of these distributions are attributable to morphological
dependencies between patterns whilst others suggest that semantic or syntactic
factors may be responsible. Several significant correlations have been identified
which merit further study.
 chapter 3

Alternative morphologies

It is clear from the introduction to Arabic verbal morphology in Chapter 2 that

it displays both a high level of systematicity and a complexity of interactions be-
tween morphological components. Thus conventional concatenative morphol-
ogy9 has traditionally been rejected in favour of a root-and-pattern model which
is distinctive of Semitic languages. Before we proceed to examine verbal seman-
tics in Chapter 4, some issues concerning the validity of the root-and-pattern
approach will be addressed. Thus alternative schemes which minimise the im-
portance of the root will be evaluated and the prosodic templatic model, which
will prove valuable in later chapters, will be introduced.

3.1 Alternatives to the Arabic root as the primary basis of derivation

The most basic assumption made in Chapter 2 is that the root sequence is the
foundation for Arabic word formation. On this basis, the system of verbal pat-
terns was introduced, in which it was shown that triliteral roots are most com-
monly encountered, with quadriliteral roots also occurring. We have also briefly
observed in 2.1.2.1 that what appear to be biliteral roots adopt the triliteral pat-
terns, where C2 also fills the C3 position and gemination occurs where phonologi-
cally appropriate.
Ussishkin (2006: 37) remarks that “[t]he consonantal root is not a construct of
modern, generative linguistics. Grammarians as far back as the Middle Ages, if not
earlier, had based their work on various Semitic languages on the consonantal root.”
As a native speaker of Modern Hebrew and psycholinguist, Shimron appeals not
only to centuries of scholarship but to intuition in support of the Semitic root:

9. Concatenative morphology is also described by Bauer (2003: 214) as the “linear approach

[in which] the morphs are accounted for in terms of their ordering in time (or on the page)”.
Sproat (1992: 44) explains and exemplifies concatenative morphology thus:
The simplest model of morphology that one can imagine is the situation where a morphologi-
cally complex word can be analyzed as a series of morphemes concatenated together. This situ-
ation obtains in … antidisestablishmentarianism, which we can analyze as being composed of
morphemes strung together like beads on a string.
40 The Arabic Verb

The root-template construct has traditionally appeared as a central morphologi-

cal structure of Semitic language because firstly … the root and the templates are
indeed there, in Semitic words, and cannot be simply overlooked. But it is also
the result of a long history of linguistic research… .  (Shimron 2003b: 4)

However, although the centrality of the root in word formation has formerly been
widely assumed by grammarians of Arabic, comparative Semiticists and most the-
oretical linguists, for example Greenberg (1950), its status and even the reality of
its existence have more recently been called into question. The debate is both lively
and current, largely between proponents of the root-based approach and those
who espouse a word- or stem-based approach to Arabic (and indeed Semitic) mor-
phology, with many significant papers appearing in the last decade. For example,
Shimron’s (2003a) volume opens with the editor’s own chapter (Shimron 2003b),
which not only may be profitably consulted as a summary of the traditional view
of Semitic morphology, but also introduces the debate which follows, as the vol-
ume includes contributions from notable opponents of the root such as Bat-El,
Heath and Benmamoun. However, adding to this already complex discussion, and
somewhat separate from it, is the ‘etymon-based’ approach to lexical organisation
proposed by Bohas (2006), which will be briefly discussed in Section 3.1.2.

3.1.1 Word- and stem-based approaches

The essential principle of word- and stem-based approaches is that word forma-
tion in Arabic is based upon fully vocalised words or stems,10 contrasting with the
root-and-pattern model in which the consonantal root, an abstract discontinuous
morpheme, is the basic morphological unit. Accompanying this, there is often an
underlying suggestion that the long-standing and once ubiquitous reliance upon
the consonantal root as the base morpheme may, at least in part, be an artefact
of traditional Arab lexicographic conventions. Bat-El (2003: 40–41), who claims
that “the consonantal root is a traditional notion” and that “tradition should be
respected by all means, but not at the cost of masking scientific inquiry”, also
remarks that Brockelmann explicitly expressed the view that the root is a mere
lexicographic convenience over a century ago. Larcher is among those who have
expressed similar views more recently:
La plupart des arabisants sont convaincus … que si la «racine» sert d’entrée aux
articles des grands dictionnaires arabes traditionnels et, à leur suite, arabisants,
c’est parce qu’elle sert pareillement d’entrée à la dérivation lexicale … De même,
un coup d’oeil dans les grands dictionnaires montre que la «racine» est une en-
trée purement formelle….  (Larcher 1999: 103)

10. With vowels inserted and thus pronounceable.

 Chapter 3.  Alternative morphologies 41

Whilst proponents of fully vocalised words or stems as the basis of derivation are
agreed that the Semitic root is not the fundamental morpheme, they diverge on
exactly what to replace it with. Thus while Heath (2003: 116) envisages “a core
of underived stems, e.g. the singular of simple nouns and the imperfective of
simple verbs [which] can be fed into derivational processes that produce derived
stems”, Benmamoun (1999, 2003b) emphasises the central role of the imperfective
(p-stem) verb alone, “enabling us to provide a unified account for aspects11 of nom-
inal and verbal morphology that have eluded previous treatments” (Benmamoun­
1999: 199). Ratcliffe (1997: 154) also writes in support of the p-stem verb as base
form in preference to the s-stem, which he views as derived despite its morpho-
logical simplicity.

3.1.1.1  Is the concept of root necessarily excluded?

Although Heath suggests that “[f]or the most part, the famous ‘consonantal
roots’ of Arabic are best consigned to oblivion”, Bat-El (2003: 41) remarks that
“arguments against the consonantal root are rarely made explicit”. For example,
in his treatment of doubled verbs, Gafos contributes evidence to the debate
without specifically rejecting a root-based analysis, claiming that “as soon as
doubled verbs are properly understood, they cease to provide crucial evidence
for root-based derivation” and that they “present an argument rather than a
problem for a stem-based view of Arabic morphology” (Gafos 2002: 82,84).
Moreover, it is uncommon for detractors of the root to dismiss it completely.
Although committed to the imperfective verb as the basis of derivation, Ben-
mamoun (1999: 199) cautions that “it is premature to argue at this point that all
productive Arabic morphology is word based rather than root based”. Similarly,
in his earlier work on denominal verbs in Modern Hebrew, Ussishkin is reluc-
tant to generalise both to other structures within that language and to other
Semitic languages:
In this analysis, I argue that the consonantal root plays no role in [Modern He-
brew] denominal verb formation. However, more work is necessary to determine
the status of roots in the language as a whole in order to verify if such entities
may be dispensed with entirely. Previous analyses have argued for the central
role of the consonantal root in all Semitic grammars. The large body of literature
on Arabic phonology and morphology, for example, takes the root as a neces-
sary element of the grammar. Further research is necessary in order to explore
whether my claims against the root result in superior accounts of word formation
throughout Semitic languages.  (Ussishkin 1999: 41)

11. Benmamoun uses ‘aspects’ here in the non-technical sense.

42 The Arabic Verb

It should be noted that some years later, however, the same author is prepared to
make the more sweeping claim that “[t]here is no need to refer to the consonantal
root; the word is the base of affixation” and hence “Semitic morphology resembles
more familiar morphology” (Ussishkin 2005: 172).
A strong conviction that a word-based analysis is correct does not necessarily
preclude a role for the root within it. Ratcliffe does not “deny the need for pro-
cesses operating on roots” (1997: 151) and he concludes that “[t]o the extent that
the consonantal root plays a role in the morphology it is as an intermediate form
extracted during a process of derivation” (1997: 169). Ratcliffe, in common with
other proponents of alternative approaches, focusses on specific shortcomings of
the traditional model and it may therefore be helpful to examine some of these in
the next section.

3.1.1.2  Specific arguments for a fully vocalised base

Ratcliffe (1997) begins by examining the vexing and frequently raised issue of
broken plurals, focussing on those exhibiting long ā in the second syllable and
analogous diminutives containing the ay diphthong, such as Examples (3a–c):

(3) a. ‫ كـلـب‬ ‫ كـالب‬ ‫ كـلـيـب‬

CvCC >> CvCvvC
kalb kilāb kulayb
dog:sg dog:pl dog:dim;sg
b. ‫ دقـتـر‬ ‫ دفـاتر‬ ‫دفـيتـر‬
CvCCvC >> CvCvvCvC
daftar dafātir dufaytir
notebook:sg notebook:pl notebook:dim;sg
c. ‫ سـلـطان‬ ‫ سالطيـن‬ ‫سـلـيــطيـن‬
CvCCvvC >> CvCvvCvvC
sulTān salāTīn sulayTīn
sultan:sg sultan:pl sultan:dim
 (after Ratcliffe 1997: 148)

Observing that it is unsatisfactory to consider the various templates which corre-

spond with these plurals and diminutives as separate morphemes as suggested by
the root-and-pattern model, he concludes that plural and diminutive formation
is not directly accessing the root, but rather modifying the fully-formed singular
noun in a largely consistent manner (Ratcliffe 1997: 147–148). Hence he charac-
terises the root-and-pattern model thus:
 Chapter 3.  Alternative morphologies 43

If we allow that some derivational rules must reference (phonologically-defined­

parts of) words rather than underlying consonantal roots, we are in effect
­admitting that the morpheme- (root-) based model of the Arabic lexicon argued
for in McCarthy (1979)12 is inadequate.  (Ratcliffe 1997: 149)

That a root-based model is therefore without merit is not, however, an inevi-

table conclusion. The work of McCarthy (1981, 1985), as further developed by
McCarthy­ & Prince, is foundational to the modern understanding of the root-
and-template model of Arabic morphology. However, applying and extending the
prosodic principles introduced in McCarthy & Prince (1990a), they allow that
[t]he broken plural … cannot be obtained with the ordinary resources of root-
and-template morphology. The category root is also morphologically inappropri-
ate as the basis of broken-plural formation, since some derivational affixes are
transferred intact….  (McCarthy & Prince 1990b: 219)

For them, there is no inconsistency in maintaining that the root is the morphemic
basis of derivation, but that other processes operate on words or stems derived
templatically from the root. Hammond (1988) also develops a model of broken
plural formation within root-and-template morphology. In contrast, although
admitting that it is possible that some derivational processes operate on the root
whilst others take words or stems as their base, Ratcliffe (1997: 150) considers that
“it is clearly preferable on grounds of simplicity to assume that they all [operate
on words or stems]”, citing Beard (1995) in support of his position that the word,
not the root, is the entry stored in the Arabic speaker’s memory:13
Bound grammatical morphemes cannot be defined other than as modifications
of major class lexical items. It follows from this undeniable fact that all major
class lexical items must have fully specified phonological representations.
(Beard 1995 in Ratcliffe 1997: 151)
However, reading further, it is clear that Beard (1995: 40) is using the term “bound
grammatical morphemes” exclusively in the narrowest sense of affixes. Given that
affixation is a concept applicable to concatenative rather than templatic morphol-
ogy, it is not immediately apparent that Beard’s constraint upon lexical items that
they be phonologically fully specified is necessarily relevant for the templatic
model of word formation which employs processes other than affixation and ap-
pears more adequate for Arabic.

12. McCarthy’s 1979 PhD dissertation was published as Formal Problems in Semitic Phonology
and Morphology in 1985.
13. Psycholinguistic evidence for the Arabic speaker’s mental lexicon is discussed in Section 3.1.1.3.
44 The Arabic Verb

A further problem for the root-based approach is that of the short vowels
in underived nouns and verbs. Using examples including the nouns in (4a–c),
Ratcliffe (1997: 151) argues that “if the three consonants of the stem are a separate
morpheme, then [the stem] vowel too must be separate [sic] morpheme”. How-
ever in these examples, as he continues, “the quality of the stem vowel is not pre-
dictable on semantic or grammatical grounds and may be any of the three short
vowels in the language”.

(4) a. ‫قـرد‬
qird
‘monkey’
b. ‫رمـح‬
rumH
‘spear’
c. ‫كـلـب‬
kalb
‘dog’  (after Ratcliffe 1997: 151)

It is clearly true that the short vowels here cannot be considered morphemic in
the sense that they carry any independent meaning when the root consonants are
subtracted from the words. However, the concept of empty morphs, having form
but not meaning, is not uncommon cross-linguistically. Some examples from
Indo-European languages are given in (5a–c):
(5) a. German compounds
Geburt + Jahr > Geburt-s-jahr
birth-?-year
‘year of birth’ (Bauer 2003: 30)
b. English neo-classical compounds
psych-o-logy
c. French adverbs
doux > douce > doucement
/du/ /dus/ /dusmã/
soft:msg soft:fsg softly
 (after Bauer 2003: 111)

In the German and English examples, the elements linking the compounds add
no meaning and appear merely to serve a phonological function, while in French,
the feminine inflection of the adjective is consistently realised in the derived ad-
verb, despite the category of gender being meaningless for this word class, ren-
dering this also an empty morph in this context. Bauer’s definition of an empty
 Chapter 3.  Alternative morphologies 45

morph as “a recurrent form in a language that does not appear to be related to

any element of meaning” (2003: 329) seems entirely applicable to the short vowels
in Examples (4a–c). Thus when Ratcliffe (1997: 151) argues that the short vowels
in the root-and-pattern model are “phonological elements … whose meaning or
function is empty”, he is justified regarding meaning, but it is entirely plausible
that they do fulfil a function which is purely phonological.
One further comment is necessary on the arbitrariness or otherwise of vo-
calisation in underived stems. Ratcliffe also uses three examples of pattern I verbs
(6a–c), in which he demonstrates that the vowel on the middle radical in the
p-stem can be any of the three available in the language, the choice of which he
claims is not predictable.

(6) a. ‫يـضـرب‬
yaDribu (corresponding s-stem Daraba)
‘he hits’
b. ‫يـكـتـب‬
yaktubu (corresponding s-stem kataba)
‘he writes’
c. ‫يـشـرب‬
yashrabu (corresponding s-stem shariba)
‘he drinks’  (after Ratcliffe 1997: 151)

This is also the position of Gafos (2002: 70), who furthermore specifies that “[n]
o phonological factors condition its choice”, although in the matter of phonol-
ogy Heath (2003: 151) differs, stating that ‘a’ is strongly favoured by the proxim-
ity of the pharyngeal consonants ‫( ح‬H) and ‫ع‬. However, as already discussed
in Section 3.1.2.2, there is evidence that there is some morphosemantic and/
or morphosyntactic significance to the pattern I s-stem medial vowel. Holes
(2004: 101) identifies alternations in this vowel as “broadly associated with dif-
ferent categories of transitivity and dynamic versus stative meaning”. Moreover,
it is undeniable that the quality of the s-stem vowel to some extent determines
the corresponding p-stem vowel, as in Table 2, where it may be observed that
only medial ‘a’ in the s-stem corresponds with unpredictable vowelling in the
p-stem. S-stem ‘i’ and ‘u’ are predictably ‘a’ and ‘u’ respectively in the corre-
sponding p-stems. It is therefore perhaps simplistic to dismiss the vowelling of
the underived p-stem as non-morphemic. Recall, however, that Ratcliffe does
not view the s-stem as basic, but favours the p-stem. His analysis requires that
p-stem vowelling be considered arbitrary and must therefore dismiss any cor-
relation of the vowelling of the s-stem (assumed to be derived) with transitivity
or stativity.
46 The Arabic Verb

3.1.1.3  External evidence

In recent years, much psycholinguistic evidence has been collected and analysed
in order to better characterise Semitic morphology. Although Ussishkin (1999;
2000; 2005) elsewhere argues in support of word-based models, he concedes that
“[p]sycholinguistic evidence regarding Semitic morphology tends to converge on
a root-based model, given results that indicate a role for lexical storage of the con-
sonantal root” (Ussishkin 2006: 38). A comprehensive overview of this evidence
is presented by Prunet (2006), largely comprising grammaticality judgements,
priming experiments and metathesis in various contexts. I will expand here only
upon the phenomenon of metathesis and will also briefly examine papers by
­Davis & Zawaydeh and Watson, the latter published simultaneously with Prunet’s
survey and thus not cited by him.
3.1.1.3.1  Aphasic metathesis. The case of an Arabic-French bilingual aphasic stroke
patient ‘ZT’ is documented by Prunet et al. (2000), providing a unique opportu-
nity to observe and contrast metathesis errors in Arabic with those in French, a
language which has no consonantal root of the kind attributed to Semitic lan-
guages. The following examples illustrate the metathesis errors made in Arabic by
ZT in a variety of oral and written tests:

(7) ‫ عـشـب‬ (target) → (output) ‫شـعـب‬

‫ع‬ushb shu‫ع‬b
‘grass’
(8) ‫ حـبـل‬ (target) → (output) ‫حـلـب‬
Hubul Hulub
‘ropes’

(9) ‫ مـجـهـود‬ (target) → (output) ‫مـجـدوه‬

majhūd majdūh
‘effort’
(10) ‫ تـفـحّـص‬ (target) → (output) ‫تـحـفّـص‬
tafaHHuS taHaffuS
‘scrutiny’
 (after Prunet et al. 2000: 613)

In Example (7), C1 and C2 of the root are transposed in the output, while in (8)
the metathesis involves C2 and C3. Significantly, consonants not belonging to the
root do not suffer metathesis, as in (9) where the ‘ma-’ prefix is retained. Similarly,
vowel melodies and templatic patterns are left unaltered: see Example (10) where
metathesis of the first two root consonants has occurred in the pattern V verbal
noun template ‘taC1aC2C2uC3’.
 Chapter 3.  Alternative morphologies 47

Of ZT’s 119 metathesis errors in Arabic in response to 1455 stimuli (8.18%),

none involved transposition of vowels and only one involved a consonant external
to the root. In contrast, far fewer metatheses were recorded for French: only 12 for
1498 stimuli (0.80%), of which only five (0.33%) are qualitatively equivalent to the
Arabic consonantal metatheses, the other seven involving transposition of vowels,
nasals, whole syllables or combinations of these. Thus equivalent metathesis is 25
times more common in ZT’s Arabic than in his French. Lest it should be assumed
that ZT’s overall competence in Arabic has been affected selectively, in tests of
his derivational and inflectional morphology error rates were almost identical for
Arabic (82%) and French (83%). Further metathesis evidence is presented in the
same paper from slips of the tongue and word games (see Section 3.1.1.3.2).
Prunet et al. (2000: 642–643) conclude “that the consonant metathesis errors
produced by ZT … provide external evidence for the existence of roots as lexical
units in the mental lexicon of speakers of Arabic”. They point to fundamental
differences between Semitic morphology and that of other languages, suggest-
ing that root consonants exist on a different representational tier from that of
the template pattern, unlike in French where consonants “are preanchored and
interspersed with vowels” (Prunet et al. 2000: 643), i.e. as an integral part of fully
vocalised stems. Data from an aphasic speaker of Modern Hebrew who makes
template rather than root errors is also presented (Prunet et al. 2000: 626–629) in
support of the existence of the template as a separate morphological tier, though
the authors note that there is no evidence thus far to support the notion of vocal-
ic melody as morphologically distinct from the template. The authors conclude
as follows:
The existence of roots as lexical units implies the existence of morphemes as lexi-
cal units and therefore favors morpheme-based models of morphology, or at least
models that can store morphemes on distinct tiers.  (Prunet et al. 2000: 643)

Discussion of the three-tier morphological model described by McCarthy &

Prince (1990a) will follow in Section 3.2.
3.1.1.3.2  Slip-of-the-tongue and word game metathesis. Metatheses occur not
only in aphasic patients but also in involuntary slips of the tongue. Prunet et al.
(2000: 625) and Mahfoudi (2007: 81) refer to the work of Berg & Abd-El-Jawad
(1996), which compares slips of the tongue in Jordanian Arabic with those in Ger-
man and English. They report significant differences in the types of metathesis
errors, with slips in Arabic mirroring the aphasic errors of ZT noted in the previ-
ous section. All root consonants were potential targets for involuntary metathesis,
regardless of their position with respect to the word or its syllable structure, with
80.5% of the errors occurring within words. In German and English however,
48 The Arabic Verb

errors tended to be far more common between words (93.6% and 86.3% respec-
tively) and were most often constrained by syllable structure on a like-for-like
basis, i.e. onset for onset or coda for coda. The following examples demonstrate
within-word (11) and between-word (12) metatheses in English, the latter popu-
larly recognised as a Spoonerism:
(11) remuneration (target) → (output) renumeration
(12) the dear old queen (target) → (output) the queer old dean

Note in each of these examples that it is syllable onsets which have been trans-
posed. The Arabic example in (13a), however, shows transposition of the con-
sonants in onset and coda positions in the first syllable and, moreover, similar
slips are observed in Arabic between root consonants across syllable boundar-
ies (13b).

(13) a. ‫ غـرفـة‬ (target) → (output) ‫رغـفـة‬

ghurfa rughfa
‘room’
b. ‫ حلم‬ (target) → (output) ‫ملح‬
Hilim miliH
‘dream’
 (after Berg & Abd-El-Jawad 1996: 303)

Berg & Abd-El-Jawad find the data to be statistically significant, thus providing
further evidence for the special status of the Semitic root.
Similarly, metathesis involving only the root consonants is reported in word
games (ludlings) in Moroccan and Bedouin Hijazi Arabic dialects, in which the
permutation process involved is believed to be unique (Bagemihl 1989: 539–
542).14 The assessment of Prunet et al. (2000: 625) is that such ludlings provide
additional evidence for the psychological reality of the root but they suggest that
further research is necessary to confirm that the facility with which native Arabic
(and indeed Semitic) speakers perform these consonantal permutations is not
matched by speakers of other languages.
3.1.1.3.3  Hypocoristics. Davis & Zawaydeh (2001) note that hypocoristics (pet
names) are widespread in colloquial Arabic and present data from Ammani-
Jordanian­speakers on one common hypocoristic pattern, exemplified in (14a–c):

14. See also Heath (1987), Berjaoui et al. (2007).

 Chapter 3.  Alternative morphologies 49

(14) a. ‫ خـالـد‬ (full name) → (hypocoristic) ‫خـلّـود‬

khaled khallūd (root kh-l-d )
b. ‫ سـلـيـم‬ (full name) → (hypocoristic) ‫سـلّـوم‬
salīm sallūm (root s-l-m)
c. ‫ هـنـد‬ (full name) → (hypocoristic) ‫هـنّـود‬
hind hannūd (root h-n-d )
 (after Davis & Zawaydeh 2001: 515)
Although the syllable structures and vowelling of the full names differ, the hypo-
coristics all take the pattern C1aC2C2ūC3, with consonants occurring in the same
sequence as in the original name and the second consonant doubled. The above
examples only contain root consonants and vowels. However, when hypocoristics
which are based on names in derived patterns containing additional non-root
consonants are examined, the same hypocoristic pattern is observed:

(15) a. ‫ مـحـ ّمـد‬ (full name) → (hypocoristic) ‫حـ ّمـود‬

muHammad Hammūd (root H-m-d )
b. ‫ أمـجـد‬ (full name) → (hypocoristic) ‫مـجّـود‬
’amjad majjūd (root m-j-d)
c. ‫ ابـتـسام‬ (full name) → (hypocoristic) ‫بـسّـوم‬
(i)btisām bassūm (root b-s-m)
 (after Davis & Zawaydeh 2001: 515)
Thus the consonants of the root and only the root are consistently selected for
inclusion in the hypocoristic template.
Although the conclusions reached by Davis & Zawaydeh differ somewhat
from those of Prunet et al. (2000) in terms of the implications for underlying
representations, they clearly view their work as supporting the existence of the
root and presenting “a challenge to strictly word-based approaches to morphol-
ogy”, stating that “the consonantal root is the object of a morphological strategy in
Arabic hypocoristic formation” (Davis & Zawaydeh 2001: 518–519).15
3.1.1.3.4  Diminutives in San’ani dialect. In a paper published in the same journal
issue as that of Prunet, Watson (2006) presents evidence relating to the formation
of diminutives in San’ani Arabic and other dialects. It is entirely appropriate that
the present debate with respect to MSA is informed by studies on Arabic dialects
as well as other Semitic languages, since the role of the consonantal root in mor-
phology is clearly a pan-Semitic issue. Thus Watson’s findings may justifiably be
extended to Arabic more generally.

15. See also Heath (1987).

50 The Arabic Verb

In interviews with San’ani native speaker informants, Watson elicited the mean-
ings of verbs on the productive diminutive pattern tCayCaC, reporting that the ex-
planations given most often employed basic triliteral nouns from the same root:
The explanatory use of a large number of different base forms, which share with
the diminutive verb only the consonantal root, suggests both that the basic con-
sonants are extractable from the tCayCaC form and that the triliteral consonantal
root is recognised by speakers as an independent morphological unit.
 (Watson 2006: 193)
However, Watson (2006: 195) also identifies counterexamples in the same pattern
such as the diminutive verb stem tmaydar, which on semantic grounds is clearly
derived from mudīr (‫‘ – مدير‬manager’) where the root is not m‑d‑r, but rather
d‑w‑r. On the basis of this and other examples, she concludes that the derivation
here is “from a fully vocalised nominal stem” (Watson 2006: 202). She also draws
on evidence from recent but restricted examples of San’ani diminutive noun for-
mation where she argues that “[t]he semantics indicates that the diminutive is
derived in these cases from the base noun – a vocalised stem – and not from the
more abstract root” (Watson 2006: 197).
Watson therefore argues “for neither an entirely root-based nor an entirely
stem-based approach, rather claiming, on the basis of data from a modern dialect,
that both types of word formation occur in Arabic” (Watson 2006: 190). Thus, in
conclusion, she writes:
Recognition of the root as an independent morphological unit at one level …
does not exclude the existence of constraints that transform one stem into an-
other stem without recourse to the root unit….  (Watson 2006: 202)

3.1.1.4 Conclusion
The consonantal root has both the weight of traditional interpretation and much
recent psycholinguistic and other external evidence to commend it.
While linguistics and psycholinguistics are different disciplines, they both deal
with language and we need principled reasons if we are asked to ignore the sys-
tematic convergences between psycholinguistic research and those linguistic
theories that posit consonantal roots. It seems implausible to me that speakers of
Semitic languages would find consonantal strings salient and pervasive enough
to systematically resort to them for either word-formation or processing, or both,
while assigning them no morphemic status in the construction of their mental
lexicons and grammars. (Prunet 2006: 62)

Nevertheless, Shimron (2003b: 1) counsels that “it may be useful to distinguish

sensitivity to the root and template structures from the actual involvement of the
root and the template in ongoing linguistic activity”. Thus, placing the evidence
 Chapter 3.  Alternative morphologies 51

presented to date by proponents of fully vocalised stems as basic to word formation

into the context of potentially contradictory evidence in favour of the root, the bal-
anced view is probably not too distant from that expressed by Shimron (2003b: 4),
“that although the root-template structure is obviously present in most Semitic
words, it may or may not play a role in all kinds of Semitic word formations”.
Thus, I am inclined to rephrase the statement made by Watson (2006: 202),
which was quoted in Section 3.1.1.3.4, to assert additionally, and perhaps more
pertinently for our present purposes, that the existence of constraints that trans-
form one stem into another stem without recourse to the root unit does not ex-
clude recognition of the root as an independent morphological unit.

3.1.2 Matrices, etymons and radicals

Contemporary with the emergence of word- and stem-based models in the last
decade or so is the even more radical approach proposed by Bohas, in which he
argues not for a basic unit larger than the Semitic root, but rather for one which is
smaller and even more abstract. This brief discussion of his proposal is based upon
Bohas (2006), which builds upon his 1997 and 2000 publications in French.

3.1.2.1 Bohas’s hypothesis

Bohas begins by identifying striking correspondences within the lexicon which
are not predicted according to the traditional root. So for example, he notes that
the following verbs which only share two root consonants (m and t) have closely
related meanings:
(16) a. ‫مـتى‬
matā (root m‑t‑y)
‘to stretch out (a rope)’
b. ‫مـتـأ‬
mata’a (root m‑t‑’ )
‘to pull out, stretch out (a rope)’
c. ‫مـتـع‬
mata‫ع‬a (root m‑t‑‫)ع‬
‘to lengthen, stretch out’
d. ‫مـتـن‬
matana (root m‑t‑n)
‘to pull out, stretch and lengthen something’  (after Bohas 2006: 4)16

16. I have been unable to verify contemporary usage of these and certain other examples using
Wehr (1994). Bohas appears to have obtained his data from the Arabic-French dictionary of de
Biberstein Kazimirski (1960), originally a 19th Century publication. See also Section 3.1.2.2.
52 The Arabic Verb

It is on the basis of examples such as these, which Bohas designates as a paradigm,

that he posits the existence of “the etymon, a binary composition of phonemes …
[extracted]… by establishing the common and constant phonetic and semantic
relationship between the words of the paradigm” (Bohas 2006: 4). Although the
identification of such biconsonantal ‘etymons’ within the lexicon is interesting in
itself, Bohas takes his argument further. Firstly, he claims that the elements of the
etymon are “not ordered in a linear fashion” (Bohas 2006: 17), i.e. that they do
not consistently occupy the same two positions in what has traditionally been re-
garded as the triconsonantal root. Thus he identifies the following examples with
similar meanings as sharing the etymon b‑j:

(17) a. ‫بـجـر‬
bajira (root b‑j‑r)
‘to have a large belly’
b. ‫حـبـج‬
Habija (root H‑b‑j)
‘to have a bloated or swollen belly’  (after Bohas 2006: 21)

Examples (18a–b) further demonstrate that the consonants of the etymon (b and

k) may not be consecutive and/or may occur in reverse sequence:

(18) a. ‫حـبـك‬
Habaka (root H‑b‑k)
‘to weave (also to braid, plait, knit)’
b. ‫كـرب‬
karaba (root k‑r‑b)
‘to twist, braid, make a rope’  (after Bohas 2006: 30)

However, Bohas’s hypothesis is more far-reaching than this. He argues for a more
basic level of lexical organisation defined as “a combination, not ordered in a lin-
ear fashion, of matrices of phonetic features linked to a semic nucleus” (Bohas
2006: 17). Thus, for example, he identifies the combination of a non-nasal, labial
consonant with an unvoiced, continuant consonant as relating to “movement of
air, wind … breathing” etc., citing the following examples, amongst others:

(19) a. ‫نـفـث‬
nafatha (etymon f - th)
‘to blow on something’
b. ‫فـ ّح‬
faHHa (etymon f - H)
‘to hiss (snake), to wheeze while sleeping’
 Chapter 3.  Alternative morphologies 53

c. ‫نـبـح‬
nabaHa (etymon b - H)
‘to hiss (snake)’
d. ‫بـ ّخ‬
bakhkha (etymon b - kh)
‘to snore while sleeping’  (after Bohas 2006: 18)

Inasmuch as these examples are to some extent onomatopoeic, it is not entirely

surprising that they share phonetic features which involve breathy exhalation. For
other examples of Bohas’s ‘matrices’, however, there is no onomatopoeic explana-
tion, though they are suggestive of the phenomenon of phonaesthemes (Bauer
2003: 119, 160; Bergen 2004; Bolinger 1968: 242), a concept which is widely recog-
nised, if considered controversial, cross-linguistically. Examples of suggested
phonaesthemes in English include words beginning with ‘gl-’ which frequently
have meanings relating to vision and light, such as glance, gleam, glow, glisten,
glimmer and glimpse. Bauer (2003: 160) states that the presence of phonaesthemes
is suggested “[w]here a lot of words share some phonological structure and also
some vague semantic structure”, whilst expressing the following note of caution:
Phonaesthemes can usually be seen as the result of post hoc analysis of existing
words, rather than the motivation for creating new words, and the boundaries of
the phonaestheme are typically rather fuzzy.  (Bauer 2003: 160)

It should be emphasised, however, that there is no inherent implication within

the concept of phonaesthemes that they are anything other than arbitrary: there
is no suggestion that the English phoneme combination ‘gl-’ stands in any di-
rect iconic relationship to the meaning with which it frequently appears to cor-
relate. Although they must necessarily be considered sub-morphemic units and
therefore, if valid, threaten the status of the morpheme as the smallest element of
meaning in language, phonaesthemes do not violate the Saussurean concept that
“the linguistic sign is arbitrary” (Saussure 1966: 67).
However, Bohas, who does not attempt to draw a parallel with phonaes-
themes, is not content with suggesting an arbitrary relationship between his
‘matrices’ and their meaning. Instead he describes them as “mimophonic … [by
which] we mean that there exists an analogy between the phonetic substance of
the matrix and its semic nucleus”, and elaborates concerning one specific conso-
nantal combination “that the mimophony of this matrix is due to the shape that
the tongue makes during the articulation” (Bohas 2006: 19). Thus Bohas, fully
recognising that his claims are contrary to Saussure’s arbitrariness of the sign, is
asserting a literal link between the articulatory shape adopted and the meaning of
the words containing the corresponding ‘matrix’ of phonetic features.
54 The Arabic Verb

More concerned with lexical organisation than morphology, Bohas is not

claiming that either the matrix or the etymon is morphemic. He introduces a
third level of lexical organisation, the ‘radical’, accounting for the extension of
the biconsonantal etymon to a largely triconsonantal morphology, defining the
radical as:
the etymon developed by the spreading of the last consonant or incrementation
of a sonorant, and including at least one vowel, and expanding the semic nucle-
us…. The radical is the domain in which various morphological and phonologi-
cal processes are carried out.  (Bohas 2006: 17)

3.1.2.2  Assessment of the matrix/etymon model

Leaving aside the ideological difficulties presented by Bohas’s literal interpreta-
tion of the phonetic shape of his matrices, there are a number of theoretical and
empirical problems for the concept of the biconsonantal matrix or etymon. Not
the least of these is that selection of the third consonant is not adequately dealt
with in Bohas’s scheme, nor is there a satisfactory explanation for why Arabic
morphology appears to be so dependent on a triconsonantal system if a bicon-
sonantal etymon is basic but not in itself morphemic. Six related objections are
discussed in more detail by Prunet (2006: 55–56), who allows that the etymon
may have diachronic value yet argues against the synchronic value attributed
by Bohas. Also, much as is argued concerning phonaesthemes, the domain of
­Bohas’s mimophones seems limited to certain special cases and they are therefore
not readily applicable across the language system as a whole, with many words
sharing features but no discernible semantic link. Mahfoudi (2007: 96) raises the
issues of there consequently being no clear way of identifying the etymon within
a word and that of the obscurity of its origins. Furthermore, it has already been
observed in footnote 16 on page 51 that many of the examples used by Bohas are
archaic, citing meanings which have not been verified synchronically. This casts
further doubt on whether the phenomena they are said to exemplify are justifiable
synchronically throughout the entire lexicon.
In support of the need for an alternative to the Semitic root as basic, ­Bohas
(2006: 7–16) produces experimental evidence that native speakers are largely
unable to consciously extract roots from a range of Arabic words, although the
methodology described intentionally draws heavily upon words in which the root
­contains one or both of the weak (glide) consonants ‫‘( و‬w’) and ‫‘( ي‬y’) or other
features which obscure the identity of the root. Whilst the performance of na-
tive speakers in these tests may nevertheless be disappointing for supporters of
root-based models, Bohas offers no comparable experimental data to suggest that
participants would have identified etymons or phonetic matrices more readily.
 Chapter 3.  Alternative morphologies 55

­ vidence in ­favour of the reality of the etymon in the mental lexicon is supplied by
E
Boudelaa & Marslen-Wilson (2001), who conducted masked and auditory-visual
cross-modal priming experiments which demonstrate faster reaction times when
consonant pairs recognised as etymons were present. However, as Prunet notes:
the argument in favour of etymons would benefit from showing that the priming
effects obtained when testing pairs of consonants in Arabic (identified as ety-
mons) are absent when comparable pairs of consonants are tested in non-Semitic
languages (since these supposedly have no etymons).  (Prunet 2006: 55)

Both Prunet (2006: 55) and Mahfoudi (2007: 91–92) observe that the distinction
between etymon and matrix is somewhat blurred by Boudelaa & Marslen-Wilson,
and Mahfoudi documents his own priming experiments, designed to test sepa-
rately these two levels of lexical organisation as defined by Bohas, also examining
the issue of whether the order of consonants in the etymon is relevant. Mahfoudi
summarises the results of his study as follows:
While there is psycholinguistic evidence for the etymon in its ordered version
that corroborates a previous study by Boudelaa and Marslen-Wilson (2001), the
non-ordered version of the etymon and its more abstract form (the phonetic ma-
trix) could not be supported by psycholinguistic data. (Mahfoudi 2007: 96)

3.1.2.3  Conclusion
Therefore, although the observations leading to Bohas’s concept of the etymon are
intriguing and will doubtless spawn further research, we must conclude for the
present, along with Mahfoudi (2007: 96), that “[t]he root remains a stronger no-
tion than the etymon to account for the organization of the Arabic mental lexicon”.
Moreover, in terms of morphological processes of derivation as distinct from lexical
organisation, there is little reason if any to reject the root in favour of the etymon.

3.2 Prosodic templatic morphology

As has already been discussed, there is a long tradition in Arabic of a Semitic

root-and-pattern morphology based on discontinuous morphemes. The most
fully developed expression of this non-concatenative morphology is presented in
the work of McCarthy (1981, 1985) and subsequently that of McCarthy & Prince
(1990a, 1990b), which “provides a comprehensive analysis of Arabic templatic
morphology within prosodic theory” (McCarthy & Prince 1990a: 49). Their mod-
el is also described by Watson (2002) and those points having particular relevance
to the present research are summarised here.
56 The Arabic Verb

3.2.1 Three morphemic tiers

As demonstrated in the following example of the pattern II s-stem passive of the

root ‫( كـتـب‬k‑t‑b), morphemes are arranged on three tiers:

(20) vowel melody u i s-stem (perfect), passive

| |
CV skeleton C v C C v C causative (pattern II)
| \/ |
root k t b ‘write’
 (after McCarthy & Prince 1990a: 5)

Thus each morphemic tier contributes both form and meaning to the resulting
derived stem. The vowel melody u‑i defines the passive perfect (s-stem), con-
trasting with both the active perfect (a‑a) and the passive imperfect or p-stem
(u‑a). The CV skeleton is that of derived pattern II, which has been somewhat
simplistically designated by McCarthy & Prince as causative (see Chapter 4 for
further discussion of pattern II meaning). The root supplies three consonants
k‑t‑b which broadly have the meaning ‘write’. Hence, combining the three tiers,
we have the (uninflected) stem kuttib ‘was caused to write’. Note that it is the
templatic tier, represented here as a CV skeleton, which supplies the character-
istic morphological shapes of the derived verbal patterns. It is this tier which
McCarthy & Prince have further described and indeed constrained on the basis
of prosodic theory.

3.2.2 Prosodic analysis

The Prosodic Morphology Hypothesis applies units of prosody to templatic mor-

phology. The phonological word (W) consists of one or more units designated as
a foot (F), which in turn consists minimally of a stressed syllable and optionally
also one or more unstressed syllables; each syllable (σ) contains one or more mo-
ras (μ), the units of syllable weight (Selkirk 1980 in Watson 2002: 129; McCarthy
& Prince 1990a: 3).

3.2.2.1  Syllable types in Arabic

Consider the following three syllable types found in Arabic: Cv, Cvv (i.e. con-
sonant followed by long vowel) and CvC. The first is a light syllable, considered
monomoraic, whilst the other two are heavy or bimoraic. Note that the syllable-
initial consonant is not considered to contribute a mora to syllable-weight:
 Chapter 3.  Alternative morphologies 57

(21) light (Cv) heavy (Cvv) heavy (CvC)

(after McCarthy & Prince 1990a: 6)

In fact, these are the only syllable types found medially in Arabic, i.e. when nei-
ther stem-initial nor stem-final. There are, however, special cases where syllables
are either initial or final.
The first is where the stem appears to begin with two consonants, as in the
derived verbal stems of patterns VII and upwards. When these patterns were in-
troduced in Chapter 2, we represented the s-stem templates as beginning with an
epenthetic vowel (i). This is the syllabic analysis for the paradigm pattern VIII
s-stem (i)fta‫ع‬al in different contexts:
(22) a. σ σ σ
/ | \ / | / | \
’ i f t a ‫ ع‬a l
b. σ σ σ σ
/ | / | \ / | / | \
q a d i f t a ‫ ع‬a l
c. σ σ σ
/ | \ / | / | \
w a f t a ‫ ع‬a l (after McCarthy & Prince 1990a: 11–12)

When the s-stem verb occurs utterance initially (post-pausally) as in (22a), the
first consonant of the stem (f ) is analysed as closing a syllable formed with an
epenthetic vowel and an initial glottal stop (hamza) which is supplied since all
Arabic syllables must have an initial consonant. This hamza which is external to
the pattern itself is known to Arab grammarians as ‫( هـمزة الوصـل‬hamzatu l‑waSl)
and is distinguished from a hamza integral to the root or pattern in fully marked
orthography. In (22b), which shows the verb in post-consonantal position, the
final consonant of the preceding word qad (‫ – قـد‬a past/perfective marker) forms
the onset of a syllable completed by the epenthetic vowel and the first stem con-
sonant. The final Example (22c), which shows the stem post-vocalically following
the word wa (‫‘ – َو‬and’) demonstrates that here no additional consonant or vowel
is supplied, the first stem consonant instead closing the final syllable of the pre-
ceding word. Thus in each case, McCarthy & Prince (1990a: 12) identify the initial
58 The Arabic Verb

stem consonant “as an extrametrical mora, one that is not linked to any syllable”
and consequently represent it in parentheses:
(23)

(after McCarthy & Prince 1990a: 12)

Syllables in stem-final position must also be treated separately, as all Arabic

stems must end in a consonant. Note that this is not true of words, which of-
ten end in vowels. The sequences which are available stem-finally are the three
medial syllable types followed by a further consonant: Cv+C, CvC+C and
Cvv+C, exemplified here in katab (‫‘ – كـتـب‬wrote’), baHr (‫‘ – بـحـر‬sea’) and qāmūs
(‫‘ – قامـوس‬dictionary’):
(24) a. Cv+C b. CvC+C c. Cvv+C

 (after McCarthy & Prince 1990a: 14)

The final consonant “is plausibly analyzed as extrametrical but not as moraic,
since it becomes an onset before vowel-initial suffixes or words” (McCarthy &
Prince 1990a: 14) for example qāmūs (24c) becomes qāmūsun with the nomina-
tive indefinite inflection.

3.2.2.2  Minimal stems

McCarthy & Prince develop an argument for a minimality constraint in Arabic,
also addressing the issue of words which appear to be counterexamples. They
conclude that the minimal stem in Arabic is a single quantitative trochaic foot,
which by definition has two moras: these moras (units of syllable weight) may
occur either in the same heavy syllable or in two light syllables (McCarthy &
Prince 1990a: 17–23). Example (24b) baHr shows a minimal stem with a bimoraic
(heavy) syllable on the pattern CvC, while (24a) katab is an example of a minimal
stem having two monomoraic (light) Cv syllables. The third possible minimal
stem, with a bimoraic Cvv syllable, is exemplified in bāb (‫‘ – بـاب‬door’):
 Chapter 3.  Alternative morphologies 59

(25)

3.2.2.3  Further noun stems

Examining basic nouns (those having no affixation and not derived from aug-
mented verbal patterns), in addition to nouns with minimal bimoraic stems the
following patterns containing three or four moras are encountered:
(26) a. light-heavy b. heavy-light c. heavy-light
CvCvv+C CvCCv+C CvvCv+C

d. heavy-heavy e. heavy-heavy
CvvCvv+C CvCCvv+C

(after McCarthy & Prince 1990a: 24)

The trimoraic stems may either consist of a light syllable followed by a heavy syl-
lable, as in (26a) wazīr (‫‘ – وزير‬minister’), or of a heavy syllable followed by a light
syllable, exemplified by (26b) funduq (‫‘ – فـنـدق‬hotel’) with first syllable CvC, and
(26c) kātib (‫‘ – كـاتـب‬writer’) with first syllable Cvv. In the examples with four mo-
ras, two patterns are observed, each having two heavy syllables: (26d) shows the
triliteral qāmūs (‫‘ – قامـوس‬dictionary’), which has two Cvv syllables, while in (26e)
finjān (‫‘ – فـنـجـان‬cup’) has CvC followed by Cvv. McCarthy & Prince (1990a: 25–33)
develop this analysis, introducing the Maximal Stem Constraint, which predicts
that the patterns thus far described are the only possible productive patterns for
underived nouns, and addressing the issues of diptotic broken plurals and rare
nouns with more than four consonants which are counterexamples.
60 The Arabic Verb

One matter raised which is of particular interest is that of the asymmetry in

the lexicon between nouns formed on the triconsonantal heavy-light and light-
heavy stems, i.e. (26c) and (26a) respectively. The data presented demonstrate that
light-heavy stems are both more diverse in vocalic melody and more common
than heavy-light stems, with “all CaaCiC nouns, constituting 97% of the CvvCvC
class [owing] their existence to a single morphological process, the formation of
the Form I active participle [fā‫ع‬il]” (McCarthy & Prince 1990a: 28). Prosodic the-
ory provides the explanation that the light-heavy combination is a recognisable
prosodic unit, the iambic foot, just as each of the bimoraic stem combinations
constitutes a single prosodic unit designated the trochaic foot and the heavy-
heavy combinations comprise two trochaic feet. In contrast, the CvvCv+C pat-
tern is not fully analysable in terms of prosodic feet, as it consists of a trochaic foot
plus a further syllable not constituting a foot.17 It is therefore concluded that this
pattern and hence nouns formed as the fā‫ع‬il active participle are derived, not un-
derived as heretofore assumed. The significance of this for the treatment of verbal
pattern III in subsequent chapters lies in the fact that this derived verbal stem has
precisely the same heavy-light syllable combination as the fā‫ع‬il active participle.
McCarthy & Prince (1990a) does not address the issue of syllables which appear
‘super-heavy’, such as in the CvvC+C pattern found in mādd (‫‘ – ما ّد‬extending’),
which is the active participle of a pattern I verb with C2 = C3. Arguably, however,
this retains both the trimoraic property of the regular fā‫ع‬il and its inability to be
analysed in terms of prosodic feet.

3.2.2.4  Verbal stems

Leaving aside patterns V, VI and QII, which have an obvious prefix, four prosodic
skeletons are proposed to account for all other verbal patterns:
(27) a. b. c. d.

 (McCarthy & Prince 1990a: 35)

17. The related matter of heavy-light quadriliterals (26b) is explained by McCarthy & Prince
(1990b: 31–32).
 Chapter 3.  Alternative morphologies 61

The four skeletons are characterised prosodically as either two light syllables
(27a) and (27c) or a heavy-light syllable sequence (27b) and (27d) followed by the
obligatory stem-final extrametrical consonant, with (27c) and (27d) additionally
prefixed by an extrasyllabic mora. The stems shown in Example (27) are patterns
I, II, VIII and X respectively, though Table 24 indicates the prosodic skeleton
applicable to each stem excluding the ta- prefixed patterns. Note that all verb
stems in this scheme are disyllabic and that the second syllable is always light
(monomoraic). McCarthy & Prince (1990a: 35–37) analyse the second syllable as
a morphemic suffix indicating the finite verb, contrasting it with a heavy second
syllable, which they designate a non-finite verb suffix morpheme, observed in the
majority of the corresponding verbal noun patterns. A consequence of the forego-
ing analysis is the conclusion that “none of the verb templates is basic … [r]ather,
all are derived” (McCarthy & Prince 1990a: 35).

Table 24.  Prosodic types by verbal stem

Pattern number Verbal stem Prosodic skeleton
I fa‫ع‬al, fa‫ع‬il, fa‫ع‬ul a
II fa‫ع ع‬al b
III fā‫ع‬al b
IV ’af‫ع‬al b
V tafa‫ع ع‬al (excluded)
VI tafā‫ع‬al (excluded)
VII nfa‫ع‬al c
VIII fta‫ع‬al c
IX f‫ع‬alal c
X staf‫ع‬al d
XI f‫ع‬ālal d
XII f‫ع‬aw‫ع‬al d
XIII f‫ع‬awwal d
XIV f‫ع‬anlal d
XV f‫ع‬anlay d
QI fa‫ع‬lal b
QII tafa‫ع‬lal (excluded)
QIII f‫ع‬anlal d
QIV f‫ع‬allal d

Morphemic status is also argued for the leading extrasyllabic mora analysed as
(σ) in patterns VII-XV, QIII and QIV. Although allowing that “a more precise
characterization of the semantics of the different [verbal patterns] could better
pin down the meaning of (σ)”, McCarthy & Prince (1990a: 39) consider that “it is
sufficient for [their] purposes to recognize that the different [verbal patterns] with
62 The Arabic Verb

initial (σ) have enough in common to warrant setting it up as a prefix”, proffering

a tentative identification of a detransitivising morpheme.
The final distinctive feature is the weight of the first syllable of the verbal stem.
McCarthy & Prince (1990a: 39) describe their analysis here as “somewhat subtle
and conjectural”, suggesting that the patterns with a light first syllable bear close
relationship to pattern I. They stop short of drawing parallels between the numer-
ous patterns sharing a heavy first syllable, preferring to analyse this as “a kind of
default base, appearing with all other [verbal patterns], a set of derivational pat-
terns that appear to have nothing in common” (McCarthy & Prince 1990a: 39).
Whether this is a potential weakness in their argument for morphemic status for
this prosodic feature will not be examined further here.
What may be of interest for our present study, however, is that patterns II, III,
IV and QI all begin with variations of the heavy (bimoraic) first syllable. This syl-
lable is realised differently from pattern to pattern, although II, IV and QI share
a CvC syllable and III alone has Cvv. Thus, in a study which examines form-
meaning­ relationships, there is both the potential for commonality of meaning
implied by the shared prosodic form of the syllable and differentiation of meaning
implied by the variant realisations as Cvv and CvC and, within the latter, further
variants according to consonant assignments. This will be seen to be significant as
we examine pattern III in more detail in later chapters.

3.3 Summary

Inasmuch as my research aims to identify and characterise morphemic compo-

nents of the Arabic verbal patterns which are additional to the consonants sup-
plied by the root (as understood in the root-and-pattern approach), it will be
necessary to assume that the root is not only a reality but also morphemic, i.e.
that it is the component bearing the basic meaning in any word derived from it.
Although it might be considered preferable to await a resolution of the issues con-
cerning the status of the root before attempting to analyse the semantic or syntac-
tic significance of patterns, it is unrealistic to believe that unanimity on the matter
is attainable. Since my approach is essentially descriptivist, however, I will not
attempt to propose derivational mechanisms. Such a pursuit is more applicable to
those who adopt a generative approach, for which the correct identification of the
base upon which these mechanisms operate assumes much greater importance.
Thus this present chapter has served largely to examine the claims of divergent
viewpoints and to introduce the prosodic templatic model which represents a
contemporary extension along the lines of traditional theory. For the present at
least, the question will remain open as to what implications my research may have
for the validity of these different approaches.
 chapter 4

Understanding Arabic verbal semantics

Form and meaning

In Chapter 2 it was established that the distribution of lexical items throughout

the verbal system of MSA is not random. Moreover, it has been demonstrated
that there are good reasons to believe that certain augmented patterns, princi-
pally II & V, III & VI and IV & X are morphologically related and indeed that the
second member of each pair is a derivative of the first. It has also been stated that
within the framework of Saussurean structuralism it is fundamental that form
and meaning are intimately linked within the linguistic sign, thus when there is a
formal realisation there must be an associated meaning within a given language
system. It would therefore be reasonable to expect that any description of the well-
characterised verbal patterns of MSA, with their distinctive morphological forms,
should be accompanied by a well-documented description of the distinctive se-
mantics specific to each pattern. Furthermore, given that it is basic to the study of
morphology that the morpheme is a meaning-bearing component of a language
system, one might expect that both descriptive and didactic grammarians would
be keen both to identify morphemes formally and to analyse and present the se-
mantic burden which each individual morpheme carries and contributes to each
verbal pattern. However, grammars of the language, though drawing on centuries
of tradition and observation, are woefully lacking in this respect, either aspiring
to a greater degree of specificity than they can deliver, or else admitting defeat
in this respect at the outset. This chapter will also examine the contributions of
Semiticists and general linguists to the form-meaning debate.

4.1 The grammarian’s dilemma

As explained in the previous chapter, the validity of the traditional analysis in

which meaning is attributed to triliteral and, less frequently, quadriliteral roots
will be assumed. Although some linguists have called this analysis into doubt, it is
the starting point of all the grammars consulted. Since they agree that the sequence
of root consonants carries meaning, whatever one’s preference may be regarding
morphological description, it is clear that the root sequence, “a semantic abstrac-
tion” (Holes 2004: 99), must be regarded as morphemic. In Examples (28a−c) the
root is k‑t‑b and the basic meaning ‘write’ can be seen in its derivatives:
64 The Arabic Verb

(28) a. ‫ كتب‬ [I]

kataba
‘to write’
b. ‫مكتب‬
maktab
‘office’
c. ‫ كاتب‬ [III]
kātaba
‘to correspond (with + d.o.)’

There is broad agreement that the unaugmented verb, in which “the root becomes
a Pattern I verb through the interdigitation of short vowels between its conso-
nants” (Holes 2004: 101) is basic and therefore carries an unaugmented meaning.
It is also uncontroversial to assert that the augmented verbal patterns build upon
the basic meaning of the root which they contain. The matter in question is to
what extent the semantics of the verbs thus derived can be predicted. In introduc-
ing derived forms Wright (1967: I.29) is non-committal, stating merely that they
“express various modifications of the idea conveyed by [pattern I]”. Writing in the
eighteenth century, Richardson (1969: 64), having elaborated upon the meanings
of the various triliteral verbal patterns, warns that “these derivative conjugations
are nevertheless frequently received in other senses” and allows that “many of
them [retain] the simple signification of their primitives”. Presumably in recog-
nition of the possible pitfalls involved in generalising the semantics of derived
patterns, Cowan (1958: 137) advises the student “to learn the meanings of the de-
rived verbs … without troubling himself unduly about the original or first form”.
Similarly, the view of Wickens is clear:
The Derived Forms are often spoken of as though they gave Arabic a sort of
mathematical exactness, that is as if the exact shade of meaning of every verb in
a Derived Form could immediately be recognised once one knew the relevant
formula; or as if every student of Arabic could “make up” his own Derived Forms
to suit his purpose. This is at the very least a gross exaggeration.
 (Wickens 1980: 64)

Badawi et al. (2004: 60) suggest that each augmented pattern “implies (though
not consistently) a specific semantic extension of the root meaning”. Meanwhile,
by choosing to call the augmented patterns “morphosemantic” Holes (2004: 100)
is making a clear statement that their morphology and semantics are intimately
linked, claiming that they “modify the root meaning semantically in (to a large
degree) predictable ways”. In her introduction to derived verbal patterns Ryding
elaborates further:
 Chapter 4.  Understanding Arabic verbal semantics 65

These variants all have a central related lexical meaning, but each verb form has
a different semantic slant on that meaning…. The interlocking of the lexical root
with the various verb form templates creates actual verbs whose meanings can
often be analyzed or deduced through the use of compositional semantics. That
is, the lexical meaning of the consonantal root plus the grammatical meaning of
the particular template combine to yield an actual word. This two-part formula
sometimes yields a very clear meaning derivable from the component parts, but
other times, the meaning is not as clear because of its evolution over time.
 (Ryding 2005: 434)

Schulz (2004: 28) asserts that “a functional-semantic description of the Forms is

possible”, but continues by admitting that “many verbs do not fit into a general
semantic system of the individual Forms”, thus highlighting the grammarian’s di-
lemma. On the one hand, either consciously aware of the claims of ­Saussurean
structuralism or perhaps unconsciously because of their own intuitive under-
standing of language systems in general, the grammarian seeks to assert the va-
lidity of the form-meaning relationship within the verbal patterns of MSA, whilst
on the other hand they feel obliged to weaken their claims by hedging their state-
ments with phrases such as “to a large degree” (Holes 2004: 100) or “not consis-
tently” (Badawi et al. 2004: 60). In essence, the dilemma is that when it comes to
descriptive examination of the individual augmented verbal patterns (II–XV), in
the absence of a clear morpheme-by-morpheme analysis of their semantics, and
desiring to present the reader or student of MSA with at least some framework
within which meaning may be recognised or attributed, grammarians appear to
resort to one or other of the following tactics: either explicit specification of all
possible semantic uses of the verbal pattern or reductionism, in which the se-
mantic significance of a given verbal pattern is simplified or generalised. It will be
shown that there are inherent drawbacks to each approach.

4.2 Specific approaches from the grammars

For ease of comparison, descriptions of the semantics of patterns II–X (exclud-

ing IX) from selected grammars spanning more than three centuries have been
summarised and presented in table form (Tables 25.1–25.2). This is not intended
to be an exhaustive survey of all Arabic grammars ever published, if that were
even possible, but rather an attempt to examine different approaches as a starting
point for comparison and to highlight some of the inconsistencies that emerge.18

18. Buckley (2004) was not available to me at the time of compiling this survey of grammars,
but also tends towards the simplified approach described in Section 4.2.1.
66 The Arabic Verb

Table 25.1  Meanings from grammars: patterns II to V

Table 25.1: Meanings from grammars: patterns II to V

 Chapter 4.  Understanding Arabic verbal semantics 67

Table 25.1 (continued)  Meanings from grammars: patterns II to V

Table 25.1 (cont.): Meanings from grammars: patterns II to V

68 The Arabic Verb

Table 25.2  Meanings from grammars: patterns VI to VIII and X

Table 25.2: Meanings from grammars: patterns VI to VIII and X

 Chapter 4.  Understanding Arabic verbal semantics 69

Table 25.2 (continued )  Meanings from grammars: patterns VI to VIII and X

Table 25.2 (cont.): Meanings from grammars: patterns VI to VIII and X

70 The Arabic Verb

The wording of the original sources is retained throughout, except that words
enclosed in brackets [ ] indicate my paraphrase, gloss or attempt to harmonise
terms. By way of illustrating the challenges involved in undertaking a semantic
description of the derived verbs, pattern II will be used as an example throughout
the discussions which follow.

4.2.1 The ‘reductionist’ approach

Among the modern grammars Badawi et al. (2004) represents an extreme, ap-
pearing to consistently strive to reduce the semantic categorisations of specific
patterns to a minimal set of possibilities. So, for example for pattern II they only
discuss factitive or causative and delocutive usages, to the exclusion of all others,
although in fairness in the preamble to the section they do recommend that their
“brief characterisations … should be reinforced by consultation of more detailed
sources” (Badawi et al. 2004: 60). This reductionism is somewhat curious given that
theirs is one of the weightier tomes and is subtitled A Comprehensive Grammar.
Doubtless there are good pedagogical reasons for not overburdening the student
beginning to study Arabic with too wide a range of semantic usages. However, the
authors claim that “it is not a pedagogical work in itself ” and that it is aimed at
(amongst others) “the student of Arabic at a relatively advanced level” and at “the
specialist in Arabic linguistics who needs data on which to base theories” (Badawi
et al. 2004: 1). Perhaps the clue to the authors’ reasons for their approach to verbal
semantics lies in the fact that they treat the verbal patterns in their foundational
opening chapter entitled ‘Forms’ and indeed the emphasis throughout the volume
is on the structure of the language, thus stressing its formal properties both mor-
phologically and syntactically over and above its semantic nuances.
However, even within this constraint, it is puzzling that this grammar is alone
within the ten surveyed in not mentioning the intensive use of pattern II, while
many also note the related extensive meaning. Likewise, Badawi et al. is alone in
drawing attention to delocutive alternates such as Example (29), whilst neglecting
to mention that in the modern language “Pattern II has been put to particularly
heavy use in the coining of denominative verbs” (Holes 2004: 101), i.e. in forming
verbs from nouns.
(29) ‫ كـبّـر‬ [II]
kabbara
‘to say allāhu ’akbar’  (Badawi et al. 2004: 60)
It is also perhaps noteworthy that while Badawi et al. employ a similar approach
for almost all the other patterns, when addressing the semantics of pattern VIII
they are forced to admit that “the meanings … are impossible to define with ac-
curacy” (Badawi et al. 2004: 61).
 Chapter 4.  Understanding Arabic verbal semantics 71

In the oldest grammar consulted, Richardson also tends towards a reduction-

ist strategy, though this is entirely consistent with the motivations explained in
his preface:
Among the many reasons which may be assigned for our limited knowledge of
this language, the first, and perhaps not the least, arises apparently from the per-
plexing obscurity and unengaging manner of grammarians; who, without suf-
ficiently attending to that simplicity and perspicuity which ought ever to be the
necessary guides to the uninformed minds of youth, bend their chief efforts to
the investigation of unuseful trifles, uninteresting definitions, and polemical sub-
tleties; involving the whole in such obscure terms, as to demand often as much
time, patience, and penetration to decypher [sic] the meaning of the teacher, as
to acquire the language itself through a less complicated medium.
 (Richardson 1969[1776]: vii)

4.2.2 The ‘explicit specification’ approach

The most striking example of this strategy is Wright, first published in 1859 and
last revised by Robertson Smith and de Goeje some forty years later, in which the
author has drawn both upon the works of Caspari and other Europeans and on
that of Arab grammarians, spanning the ancient and more contemporary (Wright
1967: v). It is significant that Wright is still held in high regard as a reference gram-
mar and a casual flick through its pages soon reveals a level of detail appropriate
to such a work. Indeed, in their respective prefaces, Thatcher and Cowan, whose
grammars are intentionally didactic, refer their students to Wright for further
grammatical study (Thatcher 1922: v; Cowan 1958: vi).
It will again be convenient to examine the treatment of pattern II semantics,
noting that Wright identifies and specifies six or even seven usages. Of these, the
causative or factitive usage is identified as the most common, and indeed is the
one example of usage mentioned by all the grammars surveyed. There is some
suggestion in his layout that Wright (1967: I.31–32) considers the declarative or
estimative usage, such as Example (30), and the denominative usage, as in Ex-
ample (31), as subsidiaries of the factitive:
ّ
(30) ‫ كـذب‬ [II]
kadhdhaba
‘to call [s.o.] a liar’  (c.f. pattern I: ‫ كـذب‬kadhaba – ‘to lie’)
(31) ‫ خـيّـم‬ [II]
khayyama
‘to pitch a tent’  (no pattern I verb from this root)

72 The Arabic Verb

Likewise, where some grammarians reduce the other primary meaning of pat-
tern II simply to ‘intensive’, Wright (1967: I.31) distinguishes three categories in
which “the signification agrees with the form”: intensive (“with great violence”);
extensive, with subdivision into temporal (“during a long time”) and numerical
(“to or by a number of individuals”); and iterative or frequentative (“repeatedly”).
Note that Wright is also claiming here that there exists a measure of iconicity in
the gemination of the middle radical. His examples include:

(32) ‫ ضـرّب‬ [II]

Darraba
‘to beat violently’ (intensive)
(33) ‫ قـتّـل‬ [II]
qattala
‘to massacre’ (numerically extensive)

It could be argued that Wright is describing the Classical language rather than
Modern Standard. Characterising Arabic in such terms implies that the former
variety is fixed and historically determinable, whereas the latter is necessarily un-
dergoing constant modification by those who use it and is thus subject to contem-
porary trends, with the concomitant possibility that older usages have been or are
being lost, modified or replaced. However, those modern grammarians whose
approach is similarly explicit continue to attest to the presence of all of Wright’s
categories in MSA. Thus, Holes (2004: 101–102) reiterates Wright’s designations,
with the exception of the iterative. In reference to current trends he places spe-
cial emphasis, as already noted, on modern denominative usage whilst observing
that there is a tendency for older extensive and intensive meanings to give way to
purely causative meanings. He gives Examples (34) and (35) respectively:

(34) ‫ مـوّت‬ [II]

mawwata
‘to cause to die’  (no longer ‘die in large numbers’)
(35) ‫ نـوّم‬ [II]
nawwama
‘to put to sleep’ (no longer ‘sleep deeply’)
(Holes 2004: 102)

What then of Wright’s iterative or frequentative category as a subdivision of the in-

tensive? Although this is not specified separately by most of the modern grammar-
ians, Ryding (2005: 491) does refer to “intensive or repeated action”, though without
giving an example to provide evidence that there is an iterative shade of meaning
distinct from the purely intensive. However, it can be seen in Examples (36a–b) that
intensive effect in pattern II can indeed be achieved by repeated or iterative action.
 Chapter 4.  Understanding Arabic verbal semantics 73

(36) a. ‫ قطع‬ [I]

qaTa‫ع‬a
‘to cut’
b. ‫ قطّع‬ [II]
qaTTa‫ع‬a
‘to cut to pieces’

4.2.3 The approaches compared

It is entirely appropriate for the various grammarians to tailor their treatment of

the verbal semantics of the derived patterns to their intended readership. Thus
the tendencies for didactic works to oversimplify and for reference works to ex-
amine detail should not be surprising. There is certainly merit for the student in
consulting a range of grammars, lest one fall into one or other of the twin traps of
either believing that Arabic verbal semantics is almost entirely predictable, within
a small and bounded set of meanings, or concluding that the system is more or
less chaotic, with consequently little reward of semantic predictability possible
from studying the formal realisations of each pattern. The student who errs in
the former direction will often tend to overgeneralise the meaning of a newly
encountered verb of a given pattern, whilst his counterpart who strays toward
the latter extreme will fail to avail himself of the benefits of noticing the semantic
commonalities which do exist between verbs employing the same pattern.

4.3 The linguist’s contribution

Linguists who choose to tackle the complexities of the Arabic verbal system do so
broadly from one of the following opposing perspectives: historical examination
of Arabic within the context of Semitic comparative linguistics and synchronic
description or modelling of the Arabic verbal system for theoretical purposes.

4.3.1 Semitic comparative linguistics

Although it is synchronic study of the verbal system of modern Arabic which

is central to our purposes, the possible benefits of examining Arabic within the
Semitic context, including historically, should not be neglected. In particular, it
may be valuable to ask whether there is evidence to support the notion that verbal
patterning in Semitic as a whole has a regular and defined semantic component.
However, preliminary examination reveals that there may be little cause for
optimism that a Semitic viewpoint will provide any more clarity. This should not
74 The Arabic Verb

be entirely surprising, since many of the grammarians of Classical Arabic, most

conspicuously Wright, have already referred extensively to the wider Semitic con-
text in drawing their conclusions. The following comment on Semitic as a whole
closely mirrors those found in Arabic grammars:
The Semitic verb has a set of themes or stems … in which formal changes cor-
respond to certain semantic variations and express different aspects of the action
connoted by the root. The semantic connexions may be somewhat fluctuating
and are not always readily identifiable.  (Moscati 1964: 122)

However, there is some prospect of semantic systematicity expressed here in

­Doron’s study of the Hebrew templatic system, the principles of which she ex-
tends more generally to Semitic:
Though the template system is on principle the same in all the Semitic languages,
the actual forms vary from language to language…. Since each and every active-
voice verb in Hebrew is derived by one of exactly three templates, it is natural
to suspect that the choice of template is not arbitrary, but that it indicates some
factor of the meaning of the derived verb. This indeed is the traditional view
concerning the templates, as is suggested, for example, by the term causative. Yet
modern linguists … have noted numerous examples where the semantic con-
tribution of the template is unpredictable … and have concluded that these ex-
amples doom to failure any attempt at a systematic analysis. Though I agree that
the semantic contribution of the templates is not transparent, I disagree that it is
not systematic.  (Doron 2003: 17)

Meanwhile, other comparative studies of the Semitic verb concentrate on its

distinctive morphology or on its tense-aspect properties. For example, Bennett
(1998) provides comprehensive verbal paradigms, but no analysis of pan-Semitic
semantics. Cohen introduces his work on the Semitic verbal system by declaring
that “[l]a principale richesse du verbe sémitique est une grande abondance de
thèmes pour chaque racine … [qui] peuvent exprimer le rôle du sujet par rapport
au procès … ou un mode du procès”. He proceeds to elaborate on the formal
realisations of these themes or patterns in Arabic. Disappointingly, however, he
concludes that “[i]l est hors du sujet du présent livre d’insister sur ces thèmes ver-
baux.” (Cohen 1924: 8–9). O’Leary (1969: 210) does venture further, contrasting
Semitic with Indo-European and tentatively offering his opinion that the limited
development of the tense system in Semitic might somehow be related to more
extensive development of themes or derived stems than in Indo-European lan-
guages, in which the tense system is more complex. This suggestion of typological
interdependency is potentially of interest, particularly in regard to the relatedness
 Chapter 4.  Understanding Arabic verbal semantics 75

of grammatical and lexical aspect in MSA,19 but O’Leary does not develop his
argument further, and it is of little use here in our search for semantic consistency
in the derived verb patterns.
If the prospect of an overarching description of Semitic verbal morphoseman-
tic relationships is unrealistic, insight might still be gained from comparison of
individual cognate patterns and their development. However, at the outset it must
be recognised that semantic diversification within certain individual Semitic pat-
terns has an extremely long history. In Danks (2007) I noted that the diversity and
unpredictability of MSA pattern X semantics is to be expected, given the pres-
ence of a morphological cognate with a broad range of semantic application in
­Akkadian, written evidence for which dates back over four millennia: specifically,
Akkadian of this period exhibits a “Št-lexical” stem with “a wide range of uses and
meanings … many of [which] are unpredictable” (Huehnergard­2000: 435).
Nevertheless, we will take pattern II, which with its gemination of the middle
radical is the most common type across the Semitic languages, as an example
of what can be gleaned from cross-linguistic comparison. O’Leary (1969: 210–
212), designating the pattern “intensitive”, begins by stating that intensive and
both temporally and numerically extensive usages are “basal”. He continues with
denominatives and extends usage of the pattern to “semi-causative”, highlight-
ing the commonality of its syntactico-semantic function across Semitic, render-
ing intransitives transitive and transitives doubly transitive. He fails, however,
to plausibly argue a link between intensitive and (semi-)causative: thus, why is
‫( فـرّح‬farraHa) ‘to gladden’ designated as intensitive, and why does deriving ‫قـتّـل‬
(qattala – ‘to slaughter’) from ‫( قـتـل‬qatala – ‘to kill’) not produce a verb which
is doubly transitive? For Doron (2003: 17–19) it is almost a matter of doctrine
that Semitic patterns involving middle radical gemination are intensive, drawing
a clear distinction between these and the causative alternations represented in
Arabic by pattern IV ‫’( أفـعل‬af‫ع‬ala). Wright (1967: I.31), similarly recognising the
historical Semitic evidence, holds that the pattern II intensive is original, whilst
the causative or factitive arises from it.20 Does Doron, then, allow the possibility
that what she terms the intensive template may be used causatively, even though
her terminology appears to preclude it? Her statement that “the intensive alterna-
tion is not [a valence-increasing alternation]” is subject to a caveat that this is a
generalisation which requires re-examination (Doron 2003: 19) and indeed she
does provide Hebrew examples

19. See Chapter 8.

20. See also Gray (1934: 79).
76 The Arabic Verb

where the intensive template involves a valence increase. This happens when the
simple verb is unaccusative. Unlike unergative and transitive simple verbs, where
the intensive template assigns the actor thematic role to one of the arguments of
the simple verb, if the simple verb is unaccusative, then the actor role is assigned
to an additional argument. The intensive verbs in [these examples] which cor-
respond to simple unaccusative verbs, are therefore just as transitive as the equi-
rooted causative verbs.  (Doron 2003: 26)

One such Hebrew example is given in (37).

(37) Root Simple verb Intensive Causative

(Unaccusative) (Transitive) (Transitive)
[t][b][’] [t]a[v]a[’] [t]i[b]ea[’] hi[t][b]ia[’]
‘drown’ ‘drown’ ‘drown’
 (Doron 2003: 26)
Now, if a verb in the “intensive alternation” is syntactically transitive to the same
degree as the corresponding verb in the “causative alternation” and if it involves
the participation of the additional argument in the “actor thematic role”, in what
way is it not causative? It must be recognised that the purpose of Doron’s paper is
to examine agency and indeed she does draw a distinction in degree of animacy
between the agents of these types of verbs. However, there is a danger of confus-
ing terminology and function here.
The practice of calling grammatical categories after their meanings is an invidi-
ous one and goes against the grain of structuralism, not to mention common
sense. The reason is that … linguistics is a search for meanings, and if we commit
ourselves to a meaning for a given form by naming the form after its supposed
meaning, it makes it difficult to change our mind about its meaning in the light
of further research.  (Beedham 2005: 19)

Thus to maintain that a verb is intensive rather than causative because it happens
to be homomorphic with a category we have already labelled ‘intensive’ and mor-
phologically distinct from one labelled ‘causative’ is to prejudge the issue.
Ryder devotes an entire monograph to what he prefers to call the “D-stem”
in Semitic, though he admits that “this study raises more questions than it will
answer” (1974: 9). His first chapter neatly summarises the aforementioned as-
sumptions in the historical approach and suggests alternative hypotheses. In par-
ticular, he cautions against a “pseudopsychological correlation of ‘strengthening’
between form and meaning (a confusion of linguistic forms with that which they
symbolize)”, later claiming that this is “an over-romanticizing of the stem’s func-
tion” (Ryder­1974: 11–12, 166). Clearly he does not share Doron’s conviction that
the pattern is fundamentally intensive, a meaning from which all others derive,
 Chapter 4.  Understanding Arabic verbal semantics 77

and goes as far as to suggest that this traditional interpretation is “a misconstruc-

tion of the reference which the Arab grammarians made to the D-stem, as denot-
ing [takthīr] or ‘increase’” (Ryder 1974: 166). It is also worthwhile to note that
­Ryder examines an alternative hypothesis applicable to form-meaning relation-
ships within this Semitic pattern, which perhaps appeals more to the Saussurean
view than the possibility of one form with two or more meanings. This hypoth-
esis allows that what we see in MSA could represent homomorphic forms with
“two or more phonological and/or morphological origins”, each of which has its
own function, or more succinctly, “the ‘Semitic D-stem’ may be a conglomerate of
forms similar only orthographically” (Ryder 1974: 11).
Goetze (1942: 2) also considers that “the meaning of the so-called intensive is
ill-defined” and shares Ryder’s view that middle radical gemination as symbolic
of intensification is a “romantic notion”. Attacking the primacy of intensitivity
among the meanings of the pattern as an unjustified assumption, and referring
here to the D-stem by the nomenclature used by scholars of Hebrew, Goetze as-
serts that:
[m]odern linguistics must reject such a prioris; it must demand that the meaning
of the picel is determined by a broad enough survey of actual occurrences and by
a definition of the conditions which govern them.  (Goetze 1942: 2)

Some insight as to why the traditional identification of intensitivity with gemina-

tion may have appealed to the Arab grammarians is provided here by Leemhuis,
who cautions that:
it should be remembered that they were not in the first place concerned with
a description of the language and how it was used; they were prescribing the
language for their contemporaries and how it should be ideally used. Moreover,
they became more and more convinced that the Arabic language, as a Godgiven
treasure, was a miracle of logic; the only task of the grammarian was to discover
and lay bare the bases of this absolute system and either to refer the facts of the
language to these bases or to reject them as improper. (Leemhuis 1977: 8)

We must as a minimum surely allow that the Semitic pattern characterised for-
mally by a geminated middle radical (Arabic pattern II) has both intensive and
causative meanings associated with it, diachronically and synchronically, and that
the prefixed pattern formally represented in Arabic by pattern IV is also capable
of causative meaning, albeit with the possibility that there may be different se-
mantic nuances between the causatives formally realised by these two distinct
patterns. It has already been noted that MSA pattern II verbs are increasingly
being used with causative meaning, a semantic domain previously more associ-
ated with pattern IV. Moreover it is recognised that “sometimes there is a subtle
78 The Arabic Verb

­ istinction between II and IV” (Wickens 1980: 67). Whether or not we accept a

d
multiple origin hypothesis for pattern II, which incidentally Ryder eventually re-
jects, it is important to emphasise that ‘intensive’ and ‘causative’ are semantic rath-
er than formal labels, and in keeping with Saussurean principles we should keep
form and meaning distinct, whilst recognising that they are inextricably linked.

4.3.2 The synchronic linguist’s perspective

The specialist linguist with a Saussurean perspective must necessarily be intrigued

by such a search for a coherent and consistent link between morphological forms
and meanings. We will therefore proceed to examine whether synchronic linguis-
tic studies have made relevant contributions towards solving the puzzle.
It must be recognised that although the Arabic verb is the subject of numer-
ous linguistic monographs and articles, those which treat the verbal system com-
prehensively are few. To reiterate, the linguist writes from completely different
motivations from the grammarian, and as such may be justified in only exploring
those components of the system which have direct bearing on the author’s thesis
or field of study. Thus, for example, while it might be supposed that a volume
entitled Structure and Function of the Arabic Verb (Bahloul 2008) would devote
space to the form-meaning relationships within the derived verbal patterns, this
topic is outwith the scope of his work, which concentrates on inflectional verbal
morphology and its relationship to the issues of tense, modality and aspect in
MSA. Meanwhile, Bakalla (1979) is concerned with a generative approach to ver-
bal morphology which concentrates on syntactic motivations.
Some do treat the derived verbal patterns, but only as a preliminary to devel-
oping their main argument. For example, Mohammad (1983) principally seeks
to pursue a generative semantic approach to the tense-aspect system and his
analysis of the semantics of the derived patterns, although present, is thus super-
ficial and incomplete. Similarly, Al-Qahtani (2005) sets out to apply Case Gram-
mar theory to MSA, and in passing contributes a considerable volume of data
on the verbal patterns and their lexical occurrence. His comment that “coun-
terexamples to any statement can be numerous and exceed the anticipation of
researchers” draws attention to the fact that the verbal system of MSA is rich in
lexical exceptions and prompts an admission that, inasmuch as he will attempt
to categorise the meanings of the various patterns, “statements here are approxi-
mate” (Al-Qahtani 2005: 102–103). There are also other works, such as The Pho-
nology and Morphology of Arabic (Watson 2002), which, while referencing the
verbal system of MSA, are principally concerned with dialects, in Watson’s case
San’ani and Cairene Arabic.
 Chapter 4.  Understanding Arabic verbal semantics 79

Thus, it is somewhat rare to discover evidence of any significant progress to-

wards tackling the intractable issue of the semantics of the Semitic verbal system
by examination of its synchronic manifestation in MSA. Fassi Fehri’s paper on
Arabic anti-causatives begins encouragingly, assuring the reader that “lexicologi-
cal research has shown that … lexical knowledge is predictable and derivable (to
a large extent) from the general principles of the grammar” and he therefore sets
out to tackle what he calls “puzzling descriptive problems that call for appropri-
ate solutions to make the lexical processes regular” (Fassi Fehri 1987: 1). Pattern
II causatives are among the verbs under scrutiny here, though the author reveals
traditionalist influences in referring to them as intensives (Fassi Fehri 1987: 7–8).
In fairness, he does make progress towards his goals inasmuch as he argues plau-
sibly for constraints on causativisation and anti-causativisation as processes, but
it is not within his remit at this point to investigate the underlying meaning of the
geminated pattern II or any other.
According to Cuvalay-Haak (1997: 89), “[s]peakers of MSA are able to extend
a root which is familiar in one pattern to others, and they usually agree on the
interpretation of the result.” She seems to imply that there is some prospect of
identifying a unifying semantic component to a given pattern. However, concern-
ing extension of the root to other patterns, El-Tikaina (1983) regards the ability of
traditional, structural grammar to produce non-existent verbal forms as evidence
that its very basis is flawed. Stressing the centrality of the lexicon, and using as
a foundation his lexical categorisation of Arabic verbs, he essentially sets out to
argue that form proceeds from meaning. Thus it is the meaning of the root which
imposes constraints upon the allowable patterns, rather than the semantics of
the formal pattern which combines with the root meaning to endow the derived
verb with a distinctive meaning of its own. Consequently he finds it necessary to
invoke two separate homomorphic derivations, both designated R2 to account for
the diversity of verbs formed on pattern II (El-Tikaina 1983: 35–36).
Despite Cuvalay-Haak’s initial optimism, she continues:
I will not try to produce a single unified meaning for each pattern. Although
some of the different uses of one pattern may be related to the same underlying
semantic principles, they must all be represented separately if we are to arrive at
an accurate description. The ascription of a single, all-encompassing meaning to
each pattern may contribute to our understanding of historical semantics, but
does not simplify the synchronic account of the functions involved.
 (Cuvalay-Haak 1997: 95)

Unusually, Cuvalay-Haak (1997: 97–101) classifies the various verbs of pattern

II according to their derivations as “deverbal”, “denominative” and “delocutive”,
rather than on the basis of their semantics, also helpfully surveying in brief the
80 The Arabic Verb

treatments of this pattern by a number of other authors including Chouémi (1966)

and Leemhuis (1977), both of whom are concerned with the Arabic of the Qur’an,
rather than MSA. Interestingly, although Chouémi (1966: 84, 108) maintains that
“La valeur fondamentale de la IIe forme est l’intensité qui provient du redouble-
ment de la 2e consonne”, his statistical analysis of the Qur’an reveals that 75%
of the verbs therein which are formed on this pattern are not strictly intensive.
With the exception of Greenberg (1991), however, Cuvalay-Haak’s sources pro-
vide little fresh insight, largely repeating material we have already encountered.
The contribution from Greenberg, eminent in the field of linguistic typol-
ogy and universals, properly belongs in the previous section, dealing as it does
with pattern II cognates cross-linguistically in Semitic. He notes from typological
study of the category of number in verbs that reduplication is “by no means the
only method, but a particularly prominent one in the expression of distributive
plurality in the verb” (Greenberg 1991: 584). This suggests that the notion that
middle radical gemination in Semitic may iconically find expression in the se-
mantics of the verb by means of some kind of sound symbolism may not merely
be romantic or motivated by a dogmatic elevation of Arabic to the status of some
ideal language, but that it fully deserves deeper analysis with regard to the prop-
erty of verbal number.
This challenge is taken up by Fassi Fehri, who, despite his earlier support for
the regularity of lexical processes, begins by admitting that “[t]he correspondence
between morphologically complex Arabic forms (or Semitic binyanim) and their
syntactico-semantic properties, alternations, and uses has until now resisted any
systematic and/or unifying treatment.” (Fassi Fehri 2003: 152). However, his paper
claims that the unifying property of pattern II gemination is indeed that of verbal
plurality as suggested by Greenberg: thus intensitivity is explained by “multiple or
repetitive action, interpreted as [event plurality]” while causativity involves “plu-
rality of participants” in which transitivity is attributable to “partitioned plurality
(alternating with assembled plurality in the repetitive)” (Fassi Fehri 2003: 162).
Although it may be argued that certain pattern II denominatives and delocutives
cannot readily be subsumed under this scheme, it is certainly a welcome step to-
wards a unified semantic description of at least one of the verbal patterns of MSA.
We shall return to the notion of verbal plurality in Chapter 7.

4.4 Summary

There is a widespread, if not universal, sense that the scheme of Arabic derived
verbal patterns ought to give rise to systematically predictable semantics. At
one extreme, overly simplistic claims of predictability have either been stated as
 Chapter 4.  Understanding Arabic verbal semantics 81

a­ ccepted fact, or argued on the basis of a priori reasoning, the origins of which
may have been influenced, whether consciously or not, by Islamic religious dog-
ma. At the other extreme, the search for systematic form-meaning correspon-
dence, whilst not intrinsically undesirable, has been almost stigmatised as vain
romanticising, with the result that many have simply cautioned that we must
accept that a coherent explanation is permanently beyond our grasp. Given the
diversity of the opinions expressed, and the evidence that this puzzle belongs as
much to Semitic as to MSA, it would indeed be tempting to abandon hope of
identifying any systematicity within Arabic verbal semantics, contenting one-
self with explanations that the answers lie safely, if frustratingly, buried in the
historical linguistics of Semitic and beyond the reach of synchronic study. How-
ever, we return to the Saussurean notion of a language as “un système où tout se
tient” (Meillet 1893: 318–319 in Koerner 1999: 26), and contend that Arabic, and
specifically MSA, is a system in the Saussurean sense. With that in mind, it may
not be possible to accurately reconstruct how the system of verbal derivations
arrived at the state we observe now, morphologically and semantically, but it
should be possible to analyse what its components mean synchronically in rela-
tion to each other and to the language as a whole.
 chapter 5

Evaluating the pattern III – pattern VI

semantic relationship

In the preceding three chapters, we have examined the verbal patterns of MSA
according to their lexical distribution, morphology and semantics. We are
now in a position to select a set of verbs for closer examination and thus this
chapter will examine the vowel-lengthening patterns III and VI according to
established semantic labels and with the emphasis on a detailed analysis of
dictionary-based data.

5.1 Selection of patterns III and VI

It will be shown here how the verbal patterns III and VI have been selected for
further study, having in mind the requirements of Beedham (2005) and consider-
ing their relationships to other patterns in the language system in recognition of
the principles of Saussurean structuralism more generally:
1. Patterns III and VI together show the highest degree of correlation of occur-
rence in the lexicon (Section 2.2.2.3.3).
2. There is a clear and relatively uncontroversial mechanism of derivation of
pattern VI from pattern III by ta- prefixation (Sections 2.1.2.5.1; 2.1.2.5.3;
3.2.2.4).
3. Patterns III and VI stand in the same morphological relationship to one an-
other as do patterns II and V and patterns QI and QII (Section 2.1.2.5.1).
4. Pattern III has been shown to share a prosodic skeleton with patterns II and
QI and their vowel melodies are identical; likewise pattern VI corresponds
with V and QII (Sections 3.2.2.4; 2.1.2.5.2).
5. There is broad agreement that pattern III most commonly involves mutual-
ity of action whilst pattern VI entails explicit reciprocity. However, there are
numerous exceptions to this generalisation (Tables 25.1–25.2, Chapter 4).
6. Both patterns III (465 verbs) and VI (389 verbs) occur commonly enough
in the lexicon for data to be statistically meaningful, whilst their numbers
are not so great that detailed semantic examination of lexical exceptions is
unmanageable (Section 2.2.2.2).
 84 order
In Theto Arabic Verb
assess to what degree the claims for pattern III mutuality and

pattern VI reciprocity are valid, and indeed to what extent pairs in both

patterns5.2 Mutuality
from the same root and
stand reciprocity
in direct relationship of meaning to one

another as derivational morphology would imply, a survey of all verbs

5.2.1 Data collection
formed according to these patterns was conducted using Wehr (1994).

In order
It was to assess
necessary to what degree
in a preliminary survey to themake
claims
manyfor pattern III mutuality and pattern
judgements
VI reciprocity are valid, and indeed to what extent pairs in both patterns from the
and assumptions based on available data. Ideally, each verb would be
same root stand in direct relationship of meaning to one another as derivational
observedmorphology
and assessedwould in the imply,
context aof
survey of allbutverbs
discourse, formedanaccording to these patterns
in practice
was search
exhaustive conducted usingexamining
of a corpus Wehr (1994).
each individual occurrence of
It was necessary in a preliminary survey to make many judgements and as-
hundreds of verbs is unworkable, and it must be assumed that the
sumptions based on available data. Ideally, each verb would be observed and as-
sessedhas
lexicographer in accurately
the context of discourse,
represented butuses.
all attested in practice
In the dataan exhaustive search of a corpus
tables
examining each individual occurrence of hundreds
listed in Appendix I, each verbal usage has been assigned a category label
of verbs is unworkable, and
it must be assumed that the lexicographer has accurately represented all attested
uses.
according In semantics
to its the dataand tables listed
syntax in Appendix
as listed by Wehr. A I, fulleach verbal
key to the usage has been assigned

categorya category
labels is alsolabel according
provided to its semantics
in Appendix and syntax
II. Table 26 shows exampleas listed by Wehr. A full key
to the category labels is also provided in Appendix II. Table 26 shows example
category data for two roots (Ϟ˰Θ˰ϗ – q-t-l and ωΰ˰ϧ – n-z-ω) which form verbs
category data for two roots (‫ – قـتـل‬q‑t‑l and ‫ – نـزع‬n‑z‑‫ )ع‬which form verbs in
patterns
in patterns III and III
VI. and VI.

Table 26.  Verb category data extract

C1 C2 C3 III VI
MUT REC
TRA/MUT/TRA-/MUT+/
REC/REC+
STA/TRA±

Table 26: Verb category data extract

For some roots the mutual and reciprocal meanings are clear, unambiguous and
in direct correspondence, for example:

(38) a. ‫ قـاتـل‬ [III] [MUT]

qātala (+ d.o.)
‘to fight (s.o.)’21
b. ‫ تـقـاتـل‬ [VI] [REC]
taqātala
‘to fight with one another’

In Examples (38a–b), only one meaning is listed for each pattern. The verb in (38a)
is transitive, taking a direct object which represents an implied mutual ­participant

21. See also Section 5.3 for discussion of the conative sense of this verb.
 Chapter 5.  Evaluating the pattern III – VI semantic relationship 85

in the action performed by the grammatical subject of the verb, whereas in (38b)
it is intransitive and reciprocal, necessarily requiring a subject which is grammati-
cally or functionally dual or plural. Thus sentences (39a−c) effectively differ only
in the thematic presentation of the roles of the participants:22

(39) a. ‫ قاتـل المـلـك الفـالح‬ [III]

qātal-a l-malik-u l-fallāH
fight;pst-3msg def-king-nom def-peasant
‘The king fought the peasant’
b. ‫ قاتـل الفـالح المـلـك‬ [III]
qātal-a l-fallāH-u l-malik
fight;pst-3msg def-peasant-nom def-king
‘The peasant fought the king’
c. ‫ تـقاتـل الفـالح والمـلـك‬ [VI]
taqātal-a l-fallāH wa-l-malik
fight_together;pst-3msg def-peasant and-def-king
‘The peasant and the king fought together’

Hence the pattern III and VI derivatives of the root q‑t‑l represent the ‘ideal’ types
of mutual and reciprocal verbs (monotransitive and intransitive respectively),23
which I have designated [MUT] and [REC] in my data. However, some verbs show
variations in transitivity whilst maintaining the sense of mutuality or reciprocity:

(40) a. ‫ تـنـازع‬ [VI] [REC/REC+]

tanāza‫ع‬a (optionally + d.o.)
‘to contend with one another / contest each other’s right to (s.th.)’
b. ‫[ تـنازع الفـالح والمـلـك الممـلـكـة‬VI]
tanāza‫ع‬-a l-fallāH
contest_right_together;pst-3msg def-peasant
wa-l-malik-u l-mamlaka
and-def-king-nom def-kingdom
‘The peasant and the king contested each other’s right to the kingdom’

The pattern VI verb in (40a–b) optionally takes a direct object, representing the
goal of the participants engaged in the reciprocal action of the verb. I record such

22. Fassi Fehri (2003: 160) views sentences similar to (39a–b) as not equivalent. See also Sec-
tion 7.2.2.2.
23. Although pattern III is not generally well represented in Arabic dialects, it may be signifi-
cant that in one dialect where it is common, Reinhardt (1894: 164, 171) reports only this type
of usage.
86 The Arabic Verb

a verbal usage as [REC+], though in fact the entry for this verb in the data appears
as [REC/REC+] since Wehr (1994) lists usages with and without a direct object.
Similarly, one of the entries for pattern III under the same root has the verb be-
ing used ditransitively with two direct objects. This usage of the verb in (41a–b)
is designated in the data as [MUT+], representing the ideal [MUT] verb with an
additional direct object.
(41) a. ‫ نـازع‬ [III] [MUT+]
nāza‫ع‬a (+ d.o.) (+ d.o.)
‘to contest the right of (s.o.) to (s.th.)’
b. ‫ نازع الفـالح المـلـك الممـلـكـة‬ [III]
nāza‫ع‬-a l-fallāH-u
contest_right;pst-3msg def-peasant-nom
l-malik-a l-mamlaka
def-king-acc def-kingdom
‘The peasant contested the right of the king to the kingdom’
However, alternative usages of this verb are listed by Wehr (1994) which are clear-
ly related in meaning though not necessarily implying mutuality, illustrating why
it is often not a simple matter of assigning a given verb to a single category such
as ‘mutual’:
(41) c. ‫ نـازع‬ [III] [MUT] / [TRA]
nāza‫ع‬a (+ d.o.)
‘to contend with (s.o.) / to combat (s.th.)’
d. ‫ نـازع في‬ [III] [TRA−]
nāza‫ع‬a fī (+ i.o.)
‘to contest (s.th.)’
e. ‫ نـازع‬ [III] [STA]
nāza‫ع‬a
‘to be in the throes of death’
f. ‫ نـازع الى‬ [III] [TRA±]
nāza‫ع‬a (+ d.o.) ilā (+ i.o)
‘to drive (s.o.) to do (s.th)’
In (41c), both mutual and non-mutual monotransitive meanings are shown, des-
ignated [MUT] and [TRA]. The same verb may also take an indirect object as in
(41d),24 whilst remaining functionally transitive, designated [TRA−], or a direct

24. The term ‘indirect object’ will be used to refer to any verbal argument in Arabic introduced
by a preposition and therefore in the genitive case, contrasting with the direct object in the ac-
cusative case.
 Chapter 5.  Evaluating the pattern III – VI semantic relationship 87

object and an indirect object as in (41f), designated [TRA±]. Without either di-
rect or indirect object in (41e), the intransitive usage is stative or [STA]. Further
complicating the entries for this particular verb is a ditransitive usage with two di-
rect objects which Wehr (1994) lists as ‘to attempt to wrest (from s.o. s.th.)’, which
appears superficially to be both conative and involving a directional transfer of
the inanimate direct object from the animate direct object to the subject, a direc-
tionality which argues against mutuality. However, I am reluctant to list the verb
separately as conative (in this case it would be [CON++], since it takes two direct
objects), as any distinction from the meaning elaborated in (41a–b) is probably a
mere artefact of translation into English, a phenomenon which we should be care-
ful to avoid. For further discussion of conativity, see Section 5.3.
Thus, in my data I list the pattern III verb from the root n‑z‑‫ ع‬and others
which display similar diversity as belonging to all the distinct semantic/syntactic
categories which I can identify and justify. Note that I have been guided by the
listings in Wehr (1994) in assessing when direct and indirect objects are optional
or obligatory for each verb. Interpretation of the data must take account of the
extent to which subjective decisions have had to be made in order to arrive at a
working set of data. I have therefore tried whenever interpreting the data to err
on the side of caution.

5.2.2 Data interpreted

5.2.2.1  Pattern III mutuality

Totals are presented in Table 27 for the presence or absence of mutuality of action
in pattern III according to whether or not the root also forms a verb in pattern VI.
In order to arrive at a binary classification of verbs as either mutual (+[MUT]) or
non-mutual (−[MUT]), it has been necessary to simplify the data in two ways.
Firstly, verbs such as nāza‫ع‬a (‫ )نـازع‬in Section 5.2.1, which have entries attributed
to several categories one or more of which is mutual, are assumed to be +[MUT].

Table 27.  Occurrence of mutual meaning [MUT] in pattern III

+ [pattern VI] − [pattern VI] TOTAL
+ [MUT] 196 165 361 (77.6%)
− [MUT]   42   62 104 (22.4%)
TOTAL 238 227 465

Secondly, all verb categories as defined in Appendix II which exhibit implied mu-
tuality of action are included as +[MUT], whether or not the verb takes an ad-
ditional direct or indirect object. Examples (42) and (43) show verbs in categories
with additional arguments:
88 The Arabic Verb

(42) ‫ شـاطـر‬ [III] [MUT+]

shāTara (+ d.o.) (+ d.o.)
‘to halve with (s.o.) (s.th.)’
(43) ‫ واطـأ على‬ [III] [MUT±]
wāTa’a (+ d.o.) ‫ع‬alā (+ i.o.)
‘to be in agreement with (s.o.) on (s.th.)’

Note that there are also verbs with mutual meaning where the implied partici-
pant is present as the indirect object, hence giving rise to two other categories
with and without a second indirect object ([MUT-±] and [MUT−]) within my
scheme of classification, though significantly I have not found these categories
represented in pattern III, which is always directly transitive with respect to the
implied participant.
In addition, I have tended towards a liberal interpretation of mutuality in
cases where it is not clear. Compare Examples (44) and (45):

(44) ‫ راكـض‬ [III]

rākaDa (+ d.o.)
‘to race (s.o.)’
(45) ‫ زايـد‬ [III]
zāyada (+ d.o.)
‘to outbid / make a higher bid than (s.o.)’

It is clear in (44) that the participants represented by both the grammatical sub-
ject and the direct object of the verb must necessarily be equal partners in the ver-
bal event ‘racing’: it is impossible for the subject to race unless he is racing against
someone else who is also racing. There are many pattern III verbs which repre-
sent this kind of equal partnership in a mutual action, a large number of which
involve competition or conflict.25 However, in (45) the relationship between the
participants is different: there is a sense in which the subject is in a position of
superiority with respect to the direct object. Nevertheless, the participation of
the second party in the verbal event ‘bidding’ is necessary to be able to say that
the subject of the verb has outbid him. I therefore contend that this still implies
mutuality of action between the participants represented by the subject and the
direct object, though with an asymmetric relationship between them. Hence un-
like the symmetrically mutual verb in (39a–b), equivalence in meaning does not
exist between (46a–b), though the fully reciprocal pattern VI verb in (46c) does
describe the overall ‘bidding’ event in terms of involvement of both participants
in the process, lending weight to the argument that mutuality is implicit in the
pattern III verb from which it is derived.

25. See also the verbs in Table 26.

 Chapter 5.  Evaluating the pattern III – VI semantic relationship 89

(46) a. ‫ زايـد المـلـك الفـالح‬ [III]

zāyad-a l-malik-u l-fallāH
outbid;pst-3msg def-king-nom def-peasant
‘The king outbid the peasant’
b. ‫ زايـد الفـالح المـلـك‬ [III]
zāyad-a l-fallāH-u l-malik
outbid;pst-3msg def-peasant-nom def-king
‘The peasant outbid the king’
c. ‫ تـزايـد الفـالح والمـلـك‬ [VI]
tazāyad-a l-fallāH
outbid_each_other;pst-3msg def-peasant-nom
wa-l-malik
and-def-king
‘The peasant and the king outbid one another’

This type of asymmetric mutuality is evident in many pattern III verbs, including
a number with meanings involving cooperation, such as (47):

(47) ‫ عـاون في‬ [III]

‫ع‬āwana (+ d.o.) fī (+ i.o.)
‘to help, assist (s.o.) in/with (s.th.)’

The possibility of asymmetry in pattern III mutuality is recognised by Fleisch

(1944: 62–73), who distinguishes “participation: action avec un autre (un réci-
proque implicite)” from “action afficiente: action sur un autre (sans réaction ou
contrepartie de celui-ci)”.
Even though the simplifications and assignments I have made in the data
have been biased in favour of mutual interpretation, it may be seen in Table 27
that only around three-quarters of pattern III verbs represent mutuality of action.
Thus there remain over one hundred pattern III verbs for which no mutual inter-
pretation is possible. Example (48) shows a verb which is transitive but does not
involve a mutual participant, while (49) and (50) are both intransitive.

(48) ‫ واثـب‬ [III] [TRA]

wāthaba (+ d.o.)
‘to pounce / fall upon (s.th.)’
(49) ‫ نـاور‬ [III] [INT]
nāwara
‘to manoeuvre’
(50) ‫ سافـر‬ [III] [INT]
sāfara
‘to travel / go on a journey’
90 The Arabic Verb

Closer examination of (49) reveals that the verb is not derived from a Semitic
root, but rather the result of back-formation from the noun ‫( مـناورة‬munāwara), it-
self a borrowing from English or French manoeuvre, which rather fortuitously re-
sembles a pattern III verbal noun. There are isolated examples of this kind which
need not trouble us unduly as they are not true Arabic exceptions, but rather
serve to show how the root-and-pattern morphology is pressed into service to as-
similate borrowings. In contrast, (50) is a very common verb which is not a recent
borrowing.26 Whilst it may be denominative, having a meaning not obviously re-
lated to the pattern I verb from the same root sequence s‑f‑r, this is not a sufficient
explanation for why the verb appears to synchronically violate the form-meaning
relationship for pattern III. Thus (48) and (50) exemplify true lexical exceptions
to pattern III mutual meaning.

5.2.2.2  Pattern VI reciprocity

Table 28 gives data for the presence or absence of reciprocity [REC] in pattern VI
according to whether or not the root also forms a verb in pattern III.

Table 28.  Occurrence of reciprocal meaning [REC] in pattern VI

+ [pattern III] − [pattern III] TOTAL
+ [REC] 181   68 249 (64.0%)
− [REC]   57   83 140 (37.0%)
TOTAL 238 151 389

Verbs with meanings attributable to several categories have been counted as posi-
tive for reciprocity as long as at least one meaning is reciprocal and, as before,
categories differing in transitivity have been included, exemplified by (51) to (53).
Note that no ditransitive reciprocals were found.

(51) ‫ تـهـامـس‬ [VI] [REC]

tahāmasa
‘to whisper together’
(52) ‫ تـنـاقـل‬ [VI] [REC+]
tanāqala (+ d.o.)
‘to tell each other (s.th.)’
(53) ‫ تـظـافـرعلى‬ [VI] [REC±]
taDHāfara ‫ع‬alā (+ i.o.)
‘to join forces against (s.o./s.th.)’

26. 206 instances found in 1001 Nights, including verb forms and active participle (arabiCorpus).
 Chapter 5.  Evaluating the pattern III – VI semantic relationship 91

Although explicit reciprocity is the dominant meaning for pattern VI, this only
accounts for somewhat less than two-thirds of all verbs in this pattern, as shown
by the data in Table 28. Holes (2004: 103) and others27 report simulative meanings
for some roots in pattern VI, involving pretence as in (54):

(54) ‫ تـمـارض‬ [VI] [SIM]

tamāraDa
‘to feign illness’

Table 29.  Occurrence of simulative meaning [SIM] in pattern VI

+ [pattern III] − [pattern III] TOTAL
+ [SIM]    8 (3.6%)   29 (19.2%)   37 (9.5%)
− [SIM] 230 (96.4%) 122 (80.8%) 352 (90.5%)
TOTAL 238 151 389

The data for the simulative verb category [SIM] as recorded in Table 29 interest-
ingly shows that this pattern VI meaning is five times more common for roots
which do not also form a verb in pattern III. This will be examined further in the
following section, together with other pattern III – pattern VI correlations. As in
the case of reciprocal verbs, simulative verbs in pattern VI occur with varying
transitivity, for example:

(55) ‫ تـنـاكـر‬ [VI] [SIM+]

tanākara (+ d.o.)
‘to pretend not to know (s.th.)’
(56) ‫ تـعـامى عن‬ [VI] [SIM±]
ta‫ع‬āmā ‫ع‬an (+ i.o.)
‘to pretend not to see (s.th.)’

It is noteworthy that only one verb was found to be a possible candidate for both
reciprocal and simulative meanings:

(57) a. ‫ تـراءى‬ [VI] [REC]

tarā’ā
‘to see one another’
b. ‫ تـراءى بـ‬ [VI] [SIM±]
tarā’ā bi (+ i.o.)
‘to feign/simulate (s.th.)’

27. Including Buckley (2004: 403), not represented in Table 25.2.

92 The Arabic Verb

However, other verbs in pattern VI cannot be categorised as either reciprocal or

simulative, for example:

(58) ‫ تـآكـل‬ [VI] [PAS]

ta’ākala bi (+ i.o.)
‘to be consumed / corrode / become worn’
(59) ‫ تـغـايـد‬ [VI] [INT]
taghāyada
‘to walk with a graceful, swinging gait’

Whilst (59) is most likely denominative, with no verbs in other patterns from the
root gh‑y‑d, (58), which I have categorised as intransitive with a passive meaning,
is representative of a small number of verbs which Ryding (2005: 543) describes
as denoting gradual change and in this case is derived from a very productive root
which includes a pattern III mutual verb (60):

(60) ‫ آكـل‬ [III] [MUT]

’ākala (+d.o.)
‘to eat/dine with (s.o.)’

Verbs such as these, together with those identified as simulative, have been classi-
fied as exceptions to pattern VI reciprocal meaning.

5.2.2.3  Pattern III – pattern VI correlations

The data presented in Tables 27–29 have been tested for statistical significance
using the chi-square test described in Section 2.2.2.3.2 and the results shown in
Tables 30 and 31. Although there is most likely a correlation between a root giv-
ing rise to mutual meaning in pattern III and its also forming a verb in pattern
VI, there is almost a three times higher degree of correlation between reciprocal
meaning in pattern VI and presence of a pattern III verb from the same root. To
rephrase this, the presence of pattern III for a pattern VI verb is a relatively good
predictor that it has a reciprocal meaning.

Table 30.  Significance of [MUT] and [REC] co-occurrences with patterns III and VI
Chi-square (χ2) Probability (p) Phi coefficient (Φ)
Pattern III [MUT] according to   6.25 <0.025 0.12
presence of pattern VI (Table 27) (significant)
Pattern VI [REC] according to 38.6 <0.0001 0.31
presence of pattern III (Table 28) (very significant)
 Chapter 5.  Evaluating the pattern III – VI semantic relationship 93

Table 31.  Significance of pattern VI [SIM] co-occurrences with pattern III

Chi-square (χ2) Probability (p) Phi coefficient (Φ)
Pattern VI [SIM] according to 27.0 <0.0001 −0.26
presence of pattern III (Table 29) (very significant)

Conversely, simulative meaning in pattern VI is negatively correlated with the

presence of a pattern III verb for the same root (Table 31), i.e. it is significantly
more likely when pattern III is absent. However, the above correlations do not
in themselves prove that pairs of pattern III – pattern VI verbs tend to have cor-
responding mutual-reciprocal meanings. Thus Table 32 contains data for all pat-
tern III – pattern VI pairs, categorised independently for mutuality in pattern III
and reciprocity in pattern VI, demonstrating that there is a highly significant and
relatively strong correlation between these two properties.

Table 32.  Occurrence and significance of [MUT]/[REC] correlations in patterns III and VI

VI = + [REC] VI = − [REC] TOTAL
III = + [MUT] 166 (69.7%) 30 (12.6%) 196
III = − [MUT]   15 (6.3%) 27 (11.3%)   42
TOTAL 181 57 238
Chi-square (χ2) = 45.6 | Prob. (p) < 0.0001 (very significant) | Phi coeff. (Φ) = 0.44

Note, however, that there are 72 instances (over 30%) of co-occurrent pattern III
and VI verbs where the mutual-reciprocal relationship does not hold, including
27 where neither paradigmatic meaning applies.

5.2.2.4  Correlations with pattern I

The construction of the database allows correlations with other properties to be
investigated. Thus, Table 33 presents correlation data for the presence or other-
wise of a pattern I verb for a given root with corresponding pattern III mutuality
and pattern VI reciprocity. Although there is weak correlation in each case with
the presence of pattern I verbs, the chi-square values are not high enough to es-
tablish statistical significance at the level of p = 0.05. However, if we examine the
correlations on the basis of the medial vowelling of the pattern I verb (Table 34),
the results are more promising, with the most significant chi-square values indi-
cating that pattern VI reciprocal meanings are more likely for a root with a pat-
tern I medial vowel ‘a’ (fa‫ع‬ala) verb and less likely if the corresponding vowel is
‘u’ (fa‫ع‬ula).
94 The Arabic Verb

Table 33.  Occurrence of pattern III [MUT] and pattern VI [REC] with pattern I verb
a. + [pattern I] − [pattern I]
III = +[MUT] 327 34
III = −[MUT]   89 15
Chi-square = 2.15 p > 0.1 Phi = 0.07
NOT SIGNIFICANT

b. + [pattern I] − [pattern I]
VI = + [REC] 234 15
VI = − [REC] 125 15
Chi-square = 2.77 p > 0.05 Phi = 0.08
NOT SIGNIFICANT

Table 34.  Correlations for pattern III [MUT] and pattern VI [REC] with patterns Ia, Ii, Iu
a. + [pattern Ia] − [pattern Ia]
III = +[MUT] 279 82
III = −[MUT]   83 21
Chi-square = 0.30 p > 0.5 Phi = −0.03
NOT SIGNIFICANT

b. + [pattern Ia] − [pattern Ia]

VI = + [REC] 197 52
VI = − [REC]   90 50
Chi-square = 10.2 p < 0.005 Phi = 0.16
VERY SIGNIFICANT

c. + [pattern Ii] − [pattern Ii]

III = +[MUT] 80 281
III = −[MUT] 15   89
Chi-square = 2.97 p > 0.05 Phi = 0. 08
NOT SIGNIFICANT

d. + [pattern Ii] − [pattern Ii]

VI = + [REC] 53 196
VI = − [REC] 40 100
Chi-square = 2.62 p > 0.1 Phi = −0.08
NOT SIGNIFICANT
 Chapter 5.  Evaluating the pattern III – VI semantic relationship 95

e. + [pattern Iu] − [pattern Iu]

III = +[MUT] 41 320
III = −[MUT] 10 94
Chi-square = 0.25 p > 0.6 Phi = 0.02
NOT SIGNIFICANT

f. + [pattern Iu] − [pattern Iu]

VI = + [REC] 24 225
VI = − [REC] 30 110
Chi-square = 10.4 p < 0.005 Phi = −0.16
VERY SIGNIFICANT

The pattern I variants, which I have designated Ia, Ii and Iu according to their
s-stem medial vowel, were introduced in Section 2.1.2.2, while in Section 3.1.1.2
it was noted that Holes (2004: 101) views them as “broadly associated with differ-
ent categories of transitivity and dynamic versus stative meaning”. Specifically, he
states that Ia “generally denotes an action, transitive or intransitive, performed
by an agent”, Ii “also frequently denotes actions, transitive and intransitive, but
ones in which … the agent is agent moyen:28 not an agent pure and simple but
one that affects itself in some way by the performance of its action”, while Iu “is
always intransitive and denotes the possession or acquisition of a quality that is
permanent” (Holes 2004: 101). Before proceeding further, we may also examine
the correlations between the presence of pattern I verbs and the distribution of
mutual/reciprocal verb pairs in patterns III and VI.

Table 35.  Correlations for [MUT/REC] pairs in patterns III and VI with pattern I
a. + [pattern I] − [pattern I]
+[MUT/REC] 154 12
−[MUT/REC]   64  8
Chi-square = 0.98 p > 0.3 Phi = 0.06
NOT SIGNIFICANT

b. + [pattern Ia] − [pattern Ia]

+[MUT/REC] 131 35
−[MUT/REC]   58 14
Chi-square = 0.08 p > 0.7 Phi = −0.02
NOT SIGNIFICANT

28. In the words of Fleisch (1979: 231).

96 The Arabic Verb

c. + [pattern Ii] − [pattern Ii]

+[MUT/REC] 35 131
−[MUT/REC] 10   62
Chi-square = 1.70 p > 0.1 Phi = 0.08
NOT SIGNIFICANT

d. + [pattern Iu] − [pattern Iu]

+[MUT/REC] 22 144
−[MUT/REC]  8   64
Chi-square = 0.21 p > 0.5 Phi = 0.03
NOT SIGNIFICANT

However, no statistically significant correlations are found (see Table 35), even

when medial vowelling variants are specified. We have already noted from Ta-
ble 30 that presence of a pattern III verb for a given root is highly correlated at a
significant probability level with reciprocal meaning in pattern VI. It is possible,
therefore, that for roots with pattern III present, pattern VI meaning is indepen-
dent of the presence and vowelling of pattern I, whereas correlations may still be
found for pattern VI verbs where pattern III is absent.

Table 36.  Correlations for pattern VI [REC] with pattern I when pattern III present
a. + [pattern I] − [pattern I]
+[REC] 168 13
−[REC]   50  7
Chi-square = 1.46 p > 0.2 Phi = 0.08
NOT SIGNIFICANT

b. + [pattern Ia] − [pattern Ia]

+[REC] 144 37
−[REC]   45 12
Chi-square = 0.01 p > 0.9 Phi = 0.01
NOT SIGNIFICANT

c. + [pattern Ii] − [pattern Ii]

+[REC] 36 145
−[REC]  9   48
Chi-square = 0.48 p > 0.4 Phi = 0.04
NOT SIGNIFICANT
 Chapter 5.  Evaluating the pattern III – VI semantic relationship 97

d. + [pattern Iu] − [pattern Iu]

+[REC] 23 158
−[REC]  7   50
Chi-square = 0.01 p > 0.9 Phi = 0.01
NOT SIGNIFICANT

Table 37.  Correlations for pattern VI [REC] with pattern I when pattern III absent
a. + [pattern I] − [pattern I]
+[REC] 66  2
−[REC] 75  8
Chi-square = 2.71 p > 0.05 Phi = 0.13
NOT SIGNIFICANT

b. + [pattern Ia] − [pattern Ia]

+[REC] 53 15
−[REC] 45 38
Chi-square = 9.24 p < 0.005 Phi = 0.25
VERY SIGNIFICANT

c. + [pattern Ii] − [pattern Ii]

+[REC] 17 51
−[REC] 31 52
Chi-square = 2.63 p > 0.1 Phi = −0.13
NOT SIGNIFICANT

d. + [pattern Iu] − [pattern Iu]

+[REC]  1 67
−[REC] 23 60
Chi-square = 19.3 p < 0.0001 Phi = −0.36
VERY SIGNIFICANT

The data presented in Tables 36 and 37 confirm that there is good evidence to
suggest that the medial vowelling of the pattern I verb (specifically patterns Ia
and Iu) is strongly correlated with pattern VI reciprocal meaning only if pattern
III is absent. It must be noted that although the data in Table 37d are highly in-
dicative of a significant correlation, care should be taken in interpreting the abso-
lute values here and in Table 37a, as accuracy of the chi-square test can be com-
promised when individual cell values are small. For this methodological reason,
98 The Arabic Verb

e­ xamination of the distribution of the relatively rare pattern VI simulative verbs

has been restricted to correlations with pattern I medial vowelling in the absence
of pattern III (see Table 38).

Table 38.  Correlations for pattern VI [SIM] with pattern I when pattern III absent
a. + [pattern Ia] − [pattern Ia]
+[SIM] 11 18
−[SIM] 87 35
Chi-square = 11.5 p < 0.001 Phi = −0.28
VERY SIGNIFICANT

b. + [pattern Ii] − [pattern Ii]

+[SIM] 17 12
−[SIM] 31 91
Chi-square = 11.9 p < 0.001 Phi = 0.28
VERY SIGNIFICANT

c. + [pattern Iu] − [pattern Iu]

+[SIM] 10   19
−[SIM] 14 108
Chi-square = 9.28 p < 0.005 Phi = 0.25
VERY SIGNIFICANT

It is not surprising that pattern Ia shows significant negative correlation with sim-
ulative meaning, while pattern Iu is positively correlated, since [REC] and [SIM]
meanings are not independent but rather in complementary distribution and it
has already been demonstrated that the reverse correlations are found for pat-
tern VI reciprocals (Table 37). What is perhaps more surprising is that pattern
Ii, which usually correlates only weakly with other properties, shows significant
co-occurrence with pattern VI simulative meaning.
Finally, for the sake of completeness, Tables 39 and 40 record data for pattern
III mutual meaning correlated with pattern I according to the presence or other-
wise of pattern VI. Only pattern Ii shows a significant correlation in the case that
pattern VI is also present and indeed there is a corresponding weak correlation
with the presence of any pattern I verb. Thus these data indicate that the presence
and vowelling of the pattern I verb have little or no effect on whether pattern III
will assume a mutual meaning.
 Chapter 5.  Evaluating the pattern III – VI semantic relationship 99

Table 39.  Correlations for pattern III [MUT] with pattern I when pattern VI present
a. + [pattern I] − [pattern I]
+[MUT] 183 13
−[MUT]   35  7
Chi-square = 4.52 p < 0.05 Phi = 0.14
SIGNIFICANT

b. + [pattern Ia] − [pattern Ia]

+[MUT] 157 39
−[MUT]   32 10
Chi-square = 0.32 p > 0.5 Phi = 0.04
NOT SIGNIFICANT

c. + [pattern Ii] − [pattern Ii]

+[MUT] 43 153
−[MUT]  2   40
Chi-square = 6.66 p < 0.01 Phi = 0.17
SIGNIFICANT

d. + [pattern Iu] − [pattern Iu]

+[MUT] 27 169
−[MUT]  3   39
Chi-square = 1.38 p > 0.2 Phi = 0.08
NOT SIGNIFICANT

Table 40.  Correlations for pattern III [MUT] with pattern I when pattern VI absent
a. + [pattern I] − [pattern I]
+[MUT] 142 21
−[MUT]   56  8
Chi-square = 0.01 p > 0.05 Phi = −0.01
NOT SIGNIFICANT

b. + [pattern Ia] − [pattern Ia]

+[MUT] 120 43
−[MUT]   53 11
Chi-square = 2.14 p > 0.1 Phi = −0.10
NOT SIGNIFICANT
100 The Arabic Verb

c. + [pattern Ii] − [pattern Ii]

+[MUT] 38 124
−[MUT] 12   53
Chi-square = 0.67 p < 0.025 Phi = 0.05
NOT SIGNIFICANT

d. + [pattern Iu] − [pattern Iu]

+[MUT] 14 148
−[MUT]  7   58
Chi-square = 0.25 p > 0.6 Phi = −0.03
NOT SIGNIFICANT

In interpreting the data presented above for correlations with pattern I, it must be
recognised that my method of data collection has limitations. In addition to those
reservations noted in Section 5.2.1, my raw data has not been corrected for roots
where pattern I meanings do not correspond obviously with derived patterns III
and VI. A case might be made for considering these as deserving the status of sep-
arate homomorphous roots, though I have consistently followed the assignments
made by Wehr (1994).29 The raw data are also subject to the phenomenon of roots
forming more than one pattern I variant, which are sometimes synonymous but
frequently unrelated in meaning. For example, in (61) the root forms both pattern
Ia and Iu variants with different basic meanings, though it is clearly the Ia verb
which is related in meaning to the pattern VI [REC] derivative, as the correlations
recorded in Table 37 predict. However, my data records this as the only example
of pattern VI [REC] corresponding with pattern Iu in the absence of pattern III
(see Table 37d), thus effectively reporting an anomaly in error.

(61) a. ‫ رشَـق بـ‬ [Ia]

rashaqa (+ d.o.) bi (+ i.o.)
‘to pelt/hurt (s.o.) with (s.th)’
b. ‫ ر ُشـق‬ [Iu]
rashuqa
‘to be shapely, elegant’
c. ‫ تـراشـق‬ [VI] [REC]
tarāshaqa
‘to pelt/hurt one another’

29. See Section 2.2.1.2.2.

 Chapter 5.  Evaluating the pattern III – VI semantic relationship 101

While re-examination of the data to minimise these inconsistencies would be

possible, this is both time-consuming and highly dependent on subjective judge-
ments. It is my contention that the limitations of the data will cause a tendency
towards understatement of the magnitude and significance of the correlations.
Thus the fact that significant correlations have nevertheless been identified sug-
gests that they are of sufficient interest to warrant further investigation.

5.3 Pattern III conativity

It was noted in the previous chapter that some grammarians30 report conative
usage for pattern III verbs. For example, Holes (2004: 102) considers that “[t]he
basic meaning of Pattern III is conative, that is, it denotes the making of an effort
to achieve the Pattern I root meaning”. Certainly, this is a possible interpretation
for the example he gives, the verb ‫( قـاتـل‬qātala) previously seen in Example (38a)
above, where the pattern I verb means ‘to kill’ and thus pattern III which is usually
translated as ‘to fight (with s.o.)’ may be interpreted as ‘to try to kill (s.o.)’. This is
certainly the position expressed by Brockelmann:

Der III. Stamm ‫ فاعل‬drückt das Streben oder den Versuch aus, die Handlung an
einer Person auszuüben, auf eine Person oder Sache einzuwirken, z.B. ‫ قـتـل‬töten,
‫ قاتـل‬zu töten suchen = jemand bekämpfen.  (Brockelmann 1904: 27)

Fischer (2002: 88), using the same example, analyses it as having “eine Handlung
zum Ziel”. However, the mutuality of action expressed by this verb is undeniable
and thus we must allow that it is capable of both mutual and conative interpre-
tations. Similarly, Gaudefroy-Demombynes & Blachère (1952: 54) describe pat-
tern III as having “essentiellemnt le sens de but”, although they allow that “cette
notion … est insuffisante; il faut mêler celle de «se rapprocher de, se joindre à
quelqu’un en accomplissant l’acte»”, adding that “on tend ainsi vers la réciprocité
exprimée par la 6e forme”. Whilst maintaining that pattern III expresses cona-
tivity in Classical Arabic, Cantineau (1934: 149–150) adds that in the dialect of
Palmyre (Syria­), it rarely coexists with pattern I, and where it does “dans un cas
elle exprime une nuance de réciprocité”. Whereas there are numerous pattern III
verbs which express mutuality but no conativity, such as Example (60) above and
others cited by Holes (2004: 102), I have identified only one verb in this pattern as
possibly conative but not mutual:

30. These also include Buckley (2004: 398), not represented in Table 25.1.

102 The Arabic Verb

(62) a. ‫ وارى‬ [III] [TRA] / [CON+]

wārā (+ d.o.)
‘to hide / conceal / try to keep secret (s.th.)’
b. ‫ ورّى‬ [II] [TRA]
warrā (+ d.o.)
‘to hide / conceal / keep secret (s.th.)’

In the case of Example (62a), the corresponding pattern I verb does not share the
basic meaning and comparison is therefore only possible with the verb in pattern II
(62b). Although I have followed Wehr (1994) and classified (62a) as either a simple
transitive [TRA] or conative with respect to a direct object [CON+], since the action
of hiding or concealing something is arguably an attempt to keep it secret anyway, it
is unclear whether the conative distinction between pattern III and pattern II is real
or not. Native speaker evidence might help to clarify this, but since this verb is the
only candidate I have identified for a non-mutual conative category my conclusion
is that it is unnecessary to treat conative meanings for pattern III separately, merely
to note that pattern III mutual meaning also often implies conativity.
Thus, on the basis of data which show that mutuality without conativity is
encountered frequently but not conativity without mutuality, I reach the opposite
conclusion from Holes and contend that the dominant meaning of pattern III is
mutual.

5.4 Summary

A summary of the conclusions reached in this chapter will be helpful before pro-
ceeding to investigate patterns III and VI further.
1. It has been established that implied mutuality is the dominant meaning for
pattern III and that reciprocity is the dominant meaning for pattern VI.
2. However, approximately one-quarter of pattern III and one-third of pattern
VI verbs do not conform to these dominant meanings.
3. The mutual-reciprocal relationship between the patterns has been validated
statistically as a real phenomenon, but it is still inadequate to explain the
meanings of over 30% of the pattern III–pattern VI verb pairs.
4. The medial vowelling of pattern I verbs from the corresponding root is cor-
related with pattern VI meaning when no pattern III verb is present. Since
any systematicity in pattern I vowelling is attributable to differences in tran-
sitivity and the nature of the agent, the data suggest that it may be fruitful to
investigate these properties with respect to the derived forms. This will be the
subject of the following chapter.
 chapter 6

Transitivity and valency

In Chapter 5, it was established that pattern I medial s-stem vowelling, a property

linked with transitivity, shows some correlation with pattern VI semantics for
verbs from the same root. It was also noted that variations in degrees of transitiv-
ity exist for both patterns III and VI, even when they carry their respective mutual
and reciprocal dominant meanings. This chapter will explore issues of transitivity
and the related verbal property of valency with particular reference to patterns III
and VI, their relationship with one another and with pattern I. Extensive data
will again be presented and analysed and a new model for evaluating valency and
hence valency change in MSA will be posited.

6.1 Defining transitivity in Arabic

6.1.1 Transitive verbs

A transitive verb in Arabic, as in English, is most often defined as one which takes
a direct object (Saad 1982: 1; Badawi et al. 2004: 778; Ryding 2005: 64). Arabic is
typologically a nominative-accusative alignment language, in which both the sin-
gle argument (S) of an intransitive verb and the agent (A) of a transitive verb are
in the nominative case, while the patient (P) of a transitive verb is distinguished
by the accusative case.31 Although the obligatory case inflections of Classical Ara-
bic are not always pronounced in MSA and only a few of these inflections are evi-
dent in unvowelled text, Example (63) shows a nominal direct object with overt
accusative marking, while in (64) the verb has a suffixed direct object pronoun.

(63) ً ‫ضـرب المـلـك فـالحـا‬

Darab-a l-malik fallaH-an
hit;pst-3msg def-king peasant-acc.indf
‘The king hit a peasant’

31. These labels for verbal arguments follow Comrie (1989: 110–111).

104 The Arabic Verb

(64) ‫ضـربـه المـلـك‬

Darab-a-hu l-malik
hit;pst-3msg-obj.3msg def-king
‘The king hit him’

6.1.2 Intransitive verbs

Although the tendency for medial vowelling ‘u’ and to some extent ‘i’ to be iden-
tified somewhat with intransitivity has already been noted (Section 6.2.2.4), the
formal test of intransitivity in Arabic, as in English, is the lack of a direct object
(Saad 1982: 1; Badawi et al. 2004: 774; Ryding 2005: 64), as in (65):

(65) ‫ضـحـك الـمـلـك‬

DaHik-a l-malik
laugh;pst-3msg def-king
‘The king laughed’
However, Example (66) shows that caution must be exercised in the special case
of the verbal noun used emphatically (‫ – المفعول المطلق‬al-maf‫ع‬ūlu l-muTlaq), also
known as a cognate accusative, internal or absolute object, or objective comple-
ment (Wright 1967: I.53–57; Cachia 1973: 14; Ryding 2005: 174; Badawi et al.
2004: 145–146).
(66) ً ‫ ضـحـك الـمـلـك ضـحـكـا‬
DaHik-a l-malik DaHk-an
laugh;pst-3msg def-king laugh-acc.indf
‘The king laughed heartily’
A similar, though much more restricted phenomenon is observed with a small
number of English verbs such as ‘dream’ and ‘laugh’ which are properly desig-
nated intransitive but are capable of taking an object-like cognate noun phrase
(NP), although
[a] cognate NP is always likely to include some modifiers (*He laughed a laugh
sounds infelicitous) and is likely to be used because there are much greater pos-
sibilities for adjectival etc. modification of a noun than there are for adverbial
modification of a verb. (Dixon 2005: 305)

Levin examines the meaning of the Arabic term for this phenomenon in some
detail, challenging the use of the terminology employed by Wright and others and
stating that “[t]he [Arab] grammarians emphasize that the [verbal noun] called
[al-maf‫ع‬ūlu l‑muTlaq] is not an object” (Levin 1991: 924). Thus although the
­verbal noun in the accusative in Example (66) is formally identical to the direct
object of a transitive verb, the verb must be treated as intransitive.
 Chapter 6.  Transitivity and valency 105

6.1.3 Ambitransitivity

MSA also allows ambitransitivity: where consistent with the semantics, verbs
which are otherwise transitive may also be used intransitively, i.e. the direct object
is optional. Examples (67a–b) show transitive and intransitive usages respectively
for the same verb. Note in these examples that, as in English, the presence of an
optional prepositional phrase (‘about the war’) does not affect the assessment of
the verb as behaving transitively or intransitively.

(67) a. ‫كـتـب المـلـك كـتـابا ً عن الحـرب‬

katab-a l-malik kitāb-an ‫ع‬an al-Harb
write;pst-3msg def-king book-acc.indf about def-war
‘The king wrote a book about the war’
b. ‫كـتـب المـلـك عن الحـرب‬
katab-a l-malik ‫ع‬an al-Harb
write;pst-3msg def-king about def-war
‘The king wrote about the war’

6.1.4 Transitivity through a preposition

The Arabic term for transitive verbs, ‫( األفعال المتعدية بأنفسها‬al-’af‫ع‬ālu l-muta‫ع‬addiya
bi-’anfusihā), may literally be rendered “Verbs that pass on … through them-
selves” (Cachia 1973: ٦٣). However, grammarians also recognise verbs which are
semantically transitive but govern their object indirectly, thus ‫األفعال المتعدية بـغـيـرها‬
(al-’af‫ع‬ālu l‑muta‫ع‬addiya bi-ghayrihā) or “Verbs that pass on … through some-
thing other than themselves (viz. through a preposition)” (Cachia 1973: ٦٣). Ex-
ample (68a) shows this kind of indirectly transitive verb, whilst (68b) is an alter-
nate usage of the same verb with direct transitivity: hence the object in the two
examples appears in the genitive and the accusative respectively.

(68) a. ‫أعـلـن بـاسـتـعـداده‬

’a‫ع‬lan-a bi-sti‫ع‬dād-i-hi
announce;pst-3msg with-readiness-gen-poss.3msg
‘He announced his readiness’
b. ‫أعـلـن اسـتـعـداده‬
’a‫ع‬lan-a sti‫ع‬dād-a-hu
announce;pst-3msg readiness-acc-poss.3msg
‘He announced his readiness’
(after Badawi et al. 2004: 381)
106 The Arabic Verb

There is clearly a semantic requirement in Arabic for the verb in these examples to
have an obligatory object/patient argument. Thus the verbal syntax, including the
identity of the preposition preceding the indirect object in the genitive in (68a),
is lexically specified. Hence this type of lexically specified prepositional phrase as
indirect object should not be confused with optional prepositional phrases used
adjectivally or adverbially as in (67a–b).32

6.1.5 Multi-transitive verbs

Many verbs may be doubly transitive in the sense of taking two direct objects in
the accusative, for example:

(69) ً ‫أعـطـى المـلـك خـدامه كـتـابـا‬

’a‫ع‬Tā l-malik khaddām-a-hu kitāb-an
give;pst.3msg def-king servant-acc-poss.3msg book-acc.indf
‘The king gave his servant a book’

However, other verbs are directly transitive with respect to one object and indirect-
ly transitive with respect to a second object in the genitive through a preposition:

(70) ‫عـاهـد المـلـك شـعـب المـغـرب على السالم‬

‫ع‬āhad-a l-malik sha‫ع‬b-a l-maghrib
promise;pst-3msg def-king people-acc def-Morocco
‫ع‬alā s-salām
on def-peace
‘The king promised the people of Morocco peace’

Although for many such verbs the second object may be semantically and syn-
tactically optional, the preposition governing it is lexically determined and thus
a distinction may still be drawn between this type of prepositional object phrase
and prepositional phrases used adverbially.
Triple direct transitivity is considered possible by Badawi et al. (2004: 380),
though corpus analysis yielded no actual examples. However, El-Kassas (2007: 2)
discusses “verbs of communication and speech” which take three direct objects
in the accusative and refers to Wright’s documentation of “the fourth form of the
‫[ أفـعـال القـلـب‬verbs of the heart]” in the nineteenth century:

32. However, Dickins & Watson (2009: 531) regard these indirect objects as “adverbial com-
plements”.
 Chapter 6.  Transitivity and valency 107

(71) ً‫أرى هللا الـنـاس أيـوب صـابـرا‬

’arā llāh-u n-nās ’ayūb
make_think;pst-3msg God-nom def-people Job
Sābir-an
patient-acc.indf
‘God has made people think Job patient’ (after Wright 1967: II.51)

Note in this example that the predicative adjective ‫( صـابـر‬Sābir) takes the accusa-
tive case, indicating its role as one of three direct objects.33 The recent encyclo-
paedia article on transitivity in Arabic by Dickins & Watson includes examples of
triply transitive verbs in both Standard Arabic and Khartoum dialect (2009: 530),
although the article by Soltan, which follows in the same volume, denies the
­validity of triple transitivity not only in Arabic but cross-linguistically (2009: 539),
dismissing examples of “believe-class” verbs such as ‫( ظن‬DHanna) and its deriva-
tives as not properly multi-transitive. My examination of entries in Wehr (1994)
has yielded a small number of verbs which are capable of taking three objects, up
to two of which may be direct. For example, the verb in (72) has two direct objects
and one indirect object:

(72) ‫أبـدـل المـلـك خـدامه ذهـبا ً بـبيتـه‬

’abdal-a l-malik
give_in_exchange;pst-3msg def-king
khaddām-a-hu dhahab-an
servant-acc-poss.3msg gold-acc.indf
bi-bayt-i-hi
with-house-gen-poss.3msg
‘The king gave his servant gold in exchange for his house’

Thus it is evident that transitivity in Arabic is more complex than a simple binary
opposition between transitive and intransitive. Even strictly in terms of lexically
specified verbal syntax, transitivity must instead be viewed as a continuum be-
tween the least transitive verbs, i.e. those which only allow cognate verbal noun
complements, and the most transitive which take three objects.34

33. For further discussion on the ambiguity of the categories of noun and adjective in Arabic
see 7.3.1.
34. Dickins & Watson allow for “degrees of transitivity” even when “[t]aking transitivity in the
narrow sense” of “a state of direct dependence of a nominal phrase … on a verbal element”
(2009: 528–530).
108 The Arabic Verb

6.2 Valency

Related to transitivity, linguistic valency is a concept borrowed from chemistry

by Tesnière (1959: 238ff.) and is analogous to the number of bonding sites on an
atom of a given element. Thus, just as it is an inherent property of an oxygen atom,
for example, to have two bonding sites or a nitrogen atom to have three, each of
which must be filled to constitute a complete molecule, the valency of a verb is
also considered to be lexically specified according to the number of elements it
must be accompanied by in order for a sentence to be grammatically complete.
These elements may include the subject and both direct and indirect objects. Tes-
nière recognised verbal valencies ranging from zero (avalent) to three (trivalent).
Verbs with zero valency are found in English with an ‘empty’ subject (Allerton
1982: 5) for which no other noun phrase may be substituted as in (73):
(73) It snowed.

The following examples, in which the elements contributing to valency are indi-
cated in italics, show valencies ranging between one and three:
(74) a. Jonny slept.
{1}
[monovalent]
b. Deborah hit the ball.
{1} {2}
[divalent]
c. Will considers housework a chore.
{1} {2} {3}
[trivalent]

In each of these examples, deletion of any one of the elements renders the sen-
tence ungrammatical,35 thus Herbst (1999) calls these elements obligatory com-
plements.36 Example (75) demonstrates that a prepositional phrase may also be
an obligatory complement:
(75) Newlyn lies at the western end of Mount’s Bay  (Herbst 1999)
{1} {2}
[divalent]

35. The verb ‘consider’ may also be divalent, but with a different meaning. Compare: ‘He con-
sidered the situation.’
36. For clarity, the terminology throughout is that of Herbst (1999), though other authors pre-
fer different nomenclature.
 Chapter 6.  Transitivity and valency 109

In contrast, elements which can be freely inserted or deleted, typically adverbials,

are considered adjuncts and do not contribute to the valency count. These ele-
ments are underlined in (76a−c)
(76) a. Jonny slept for hours.
{1}
[monovalent]
b. Deborah hit the ball with the bat.
{1} {2}
[divalent]
c. Under normal circumstances, Will considers housework a chore.
{1} {2} {3}
[trivalent]
Whilst it is clear in the above examples which elements are obligatory complements
contributing towards valency and which are adjuncts and hence not contributing,
an intermediate class of optional complements is possible for some verbs. These
are elements which do contribute to valency but may optionally be deleted without
violating grammaticality, such as the underlined elements in (77) and (78):
(77) Emily was reading a book.
{1} {2}
[divalent]
(78) Most creative people object to the notion that the work
{1} {2}
they do comes easily
[divalent]  (Herbst 1999)
Because the optional complements in both (77) and (78) are lexically specified
as to their form for the verb in question (respectively a direct object noun phrase
and a prepositional phrase with ‘to’), they must be considered part of the valency
structure of the verb. These complements, although deletable, are thus distinctly
different from adjuncts, which are typically adverbial and may take a wide range
of forms. Furthermore Herbst (1999) draws attention to what Allerton (1982: 68–
69) calls “contextual deletion”: for example in (78) the underlined complement of
the verb ‘object’ is only contextually optional, i.e. rendering the usage monovalent
by deleting the complement is only possible if context makes it clear what is being
objected to. Note that optionality of complements provides an explanation for the
phenomenon of ambitransitivity discussed in Section 6.1.3.
The concept of valency is only outlined above in its simplest form. Classifica-
tion of verbs according to different types of complements allows the identification
of a range of possible structures for each numerical value of valency. The valency
110 The Arabic Verb

structures of English are discussed comprehensively by Allerton (1982). However,

for our present purposes we are concerned with the application of the concept of
valency to the Arabic verb in order to arrive at a more adequate classification than
that provided by transitivity alone, especially in regard to multi-transitive verbs
(6.1.5) and those which are transitive through a preposition (6.1.4).
Examples (63) to (70) in Section 6.1 have already introduced monovalent, diva-
lent and trivalent Arabic verbs. It will be noted, however, that the verbs in (71) and
(72) each have four complements. Tesnière (1959: 258) states that “il semble bien
qu’il n’existe dans aucune langue de formes verbales simples comportant plus de
trois valences”. Although Allerton (1982: 116) writes concerning tetravalent struc-
tures for simple verbs that “it is usually assumed that there are none in English”,
he proceeds to argue a case for their existence. There is therefore no reason not
to recognise the existence of tetravalent verbs in Arabic. As for zero valency, the
weather verbs in Arabic, unlike English, are capable of taking an overt subject and
while verbs such as ‫( وجب‬wajaba – ‘to be necessary’) are commonly impersonal,
they are not exclusively so and will normally have an indirect object and/or subordi-
nate clause complement. There may thus be no properly zero-valent verbs in MSA.
Thus we will proceed to categorise the pattern III and pattern VI verbs and
pattern I verbs from the same roots on a valency scale ranging from one to four
and hence investigate how valency is distributed within the various patterns and
whether morphological derivations are linked with valency change.

6.3 Data collection

Semantic category labels with accompanying information on direct and indirect

objects were already assigned as described in Section 6.2.1. Each label has been
converted to an alphanumeric code consisting of the numeric valency followed
by a letter indicating the combination of direct and indirect objects (if present).
Table 41 presents the codes (from high to low) for all valency structures so far
encountered or reported for MSA, ranked according to subject, then number of
direct objects and finally number of indirect objects.
As with data collection in previous chapters, the assumption has been made that
the lexicographic information provided by Wehr (1994) is accurate, complete and
consistent. Thus all usages recorded in the dictionary which differ in any or all of the
three parameters of subject, direct object and indirect object give rise to a separate
code for a given verb. However, inasmuch as the dictionary entries may be assumed
to be reliable and since the codes are allocated on purely formal grounds, the data
recorded and analysed in this chapter are objective and not as in the previous chapter
where allocations to categories were decided subjectively on the basis of semantics.
 Chapter 6.  Transitivity and valency 111

Table 41.  Valency structure codes for MSA

Valency Subject D.O. I.O Code
4 1 3 0 4D
4 1 2 1 4C
4 1 1 2 4B
4 1 0 3 4A
3 1 2 0 3C
3 1 1 1 3B
3 1 0 2 3A
2 1 1 0 2B
2 1 0 1 2A
1 1 0 0 1B
1 0 0 1 1A

Examples of alternative valency structures may be seen in the extract from the
data
from the data tables 37
tables 37 presented
presented in where
in Table 42, Table 42, where
the root έΪ˰Α
the root
combination combination ‫( بـدر‬b‑d‑r) has
(b-d-r) hasthree different
three different structure
structure codes:
codes: two twoand
divalent divalent and one trivalent.
one trivalent.

Table 42.  Data extract from valency structure codes for patterns III and VI
C1 C2 C3 III VI
3B 1B/3A

3B/2B/2A 1A

3C 2B

2A 1B

2B 1B

2B/2A 1B

2B 1B/2A

2A 1B/2A

2B 1B

3B/2A 1B/2A

2B 1B

2B 1B/2A

Table 42: Data extract from valency structure codes for patterns III and VI

Table 43 is a summary of the distribution of numerical valency across all

pattern III 37. Data tables

and pattern areforms.
VI verb listedNote
in Appendix I. for each pattern in
that the totals

this table are greater than the actual totals for the corresponding verb forms

in the dictionary, since a verb which displays two or more valency structures

contributes to data in more than one column. However, we may observe that
112 The Arabic Verb

Table 43 is a summary of the distribution of numerical valency across all pat-

tern III and pattern VI verb forms. Note that the totals for each pattern in this
table are greater than the actual totals for the corresponding verb forms in the
dictionary, since a verb which displays two or more valency structures contributes
to data in more than one column. However, we may observe that 88% of the 465
pattern III verb forms have at least one usage with a divalent structure and 86%
of the 389 pattern VI verb forms have at least one usage with a monovalent struc-
ture. The ‘ideal’ mutual and reciprocal categories [MUT] and [REC] introduced
in Section 5.2.1 correspond with these numeric valencies. However, it was also
noted in that discussion that mutual and reciprocal meanings occur with varying
transitivity and thus these and other semantic labels will be represented in the
valency data by different values. Although it may readily be observed that there
is a distinct difference in distribution of valencies between patterns III and VI, it
is also clear that a given pattern does not reliably specify valency. Although only
approximately 8% of pattern III verbs have monovalent usages listed, as many as
24% show valencies greater than two, while nearly 44% of pattern VI verbs have
usages with valencies greater than one.

Table 43.  Valency distribution in patterns III and VI

Valency 1 2 3 4
Pattern III   36 411 109 2
Pattern VI 333 155   15 0

6.4 Data analysis

6.4.1 Valency and the ta- prefix

Watson (2002: 139–142) argues that the ta- prefix in pattern VI is “detransitiviz-

ing” and that patterns V–X are characterised by “a reduction or minimization of
the valence [sic] of the underlying verb”. Although Watson quotes from McCarthy
& Prince (1990a: 38), strictly these authors only refer to patterns VII–X in this
context, having excluded patterns V, VI and QII from their analysis. Nevertheless,
Watson’s hypothesis may now be tested using the data obtained for pattern VI and
for pattern III which is its corresponding ‘underlying’ verb.

Table 44.  Numerical valency reduction and minimisation in pattern VI

Total Reduction Minimisation % reduced/minimised
Roots with patterns III and VI 238 134 10 60.5
Roots with pattern VI only 151 – 91 60.3
 Chapter 6.  Transitivity and valency 113

The property of valency reduction is clearly only applicable to pattern VI verbs

which also have a pattern III verb from the same root, whereas a minimum va-
lency state of 1 is theoretically possible for all pattern VI verbs. Data for valency
reduction in Table 44 only represent unambiguous instances where all pattern VI
usages show reduction relative to all pattern III usages. Likewise, minimisation is
defined as when all usages are monovalent and is only recorded when unambigu-
ous reduction is not present or is inapplicable. With multiple valencies possible
for many verbs, however, these data are somewhat crude. Nevertheless, if valency
reduction and minimisation are viewed as a whole, there is remarkable consis-
tency across pattern VI, whether or not pattern III is present for the root.

6.4.1.1  A hierarchical approach to valency structures

At the conclusion of Section 6.1, we noted that transitivity in Arabic must be
treated as more than a binary opposition, i.e. transitive versus intransitive, lead-
ing to an examination of valency as a more adequate model. Although different
valency structures have been introduced, there has thus far been no attempt
to define degrees of transitivity within the same numerical valency. We may
hypothesise, however, that a verb which is transitive through a direct object
is treated in Arabic as more transitive than one which is transitive through a
preposition and indirect object. Thus Table 41 may be viewed as one possible
hierarchical ordering of valency structures from high to low transitivity and
hence detransitivisation involves moving from any given valency structure code
to any code lower in the table: we will call this ‘hierarchical valency reduction’.
For consistency, valency minimisation in the data will be recorded as before:
thus codes [1A] and [1B] both represent minimisation, since [1A] is restricted
to impersonal subjects and is not available semantically to most verbs.38 Thus
Table 45 is analogous to the Table 44 data for pattern III – pattern VI pairs, but
allowing for hierarchical reduction.

Table 45.  Hierarchical valency reduction and minimisation in pattern VI

Total Reduction Minimisation % reduced/minimised
Roots with patterns III and VI 238 188 10 83.2

It may be seen from the above data that the hypothesis which redefines transitiv-
ity and hence detransitivisation on a hierarchical basis leads to a much higher
percentage of pattern III – pattern VI verb pairs conforming to the proposal that
the ta- prefix is a detransitivising morpheme. The hierarchy defined by Table 41

38. The pattern VI verb from the root ‫( بدر‬b-d-r) is the only instance of code [1A] in the data.
114 The Arabic Verb

assumes a ranking based upon numerical valency, number of direct objects

and number of indirect objects in that order. Clearly this is not the only or-
der in which these parameters and hence the valency structure codes may be
ranked, but it is likely to be the most successful since there are numerous pat-
tern III – pattern­ VI pairs in which a complement encoded as direct object in
pattern III is encoded as indirect object in pattern VI, often with no substantial
change in meaning. For example, in (79a) the pattern III verb puts ‫رئـيـس الوزراء‬
(‘the prime minister’) in the accusative (ra’īsa l‑wuzarā’), hence is designated
[2B] for valency, whilst in (79b) the corresponding pattern VI verb is coded as
[2A] as it takes the preposition ‫( مـع‬ma‫ع‬a) followed by the same noun phrase in
the genitive (ra’īsi l‑wuzarā’).

(79) a. ‫ حـادث المـلـك رئـيـس الوزراء‬ [III] [MUT] [2B]

Hādath-a l-malik ra’īs-a l-wuzarā’
talk;pst-3msg def-king president-acc def-minister;pl
‘The king talked to the prime minister’
b. ‫ تـحـادث المـلـك مـع رئـيـس الوزراء‬ [VI] [MUT−] [2A]
taHādath-a l-malik ma‫ع‬a ra’īs-i
talk;pst-3msg def-king with president-gen
l-wuzarā’
def-minister;pl
‘The king talked with the prime minister’

6.4.1.2  Hierarchical exceptions

We are now left with 40 verb pairs which appear to constitute a set of lexical
exceptions to the hierarchical valency reduction rule. There are various reasons
why these pattern III – pattern VI pairs do not show hierarchical reduction or
valency minimisation in the crude data in Table 45 and they are categorised in
Table 46 according to whether I have been able to postulate an explanation for
their exceptionality.

Table 46.  Exceptional pattern III – pattern VI pairs by explanation

Potentially explicable Non-explicable
Unrelated Compliant Hierarchical Synonyms Other
meanings in at least challenges
one usage
Number of 5 23 4 6 2
verb pairs 32 8
 Chapter 6.  Transitivity and valency 115

6.4.1.2.1  Unrelated meanings. Five pairs, including (80a–b), have meanings

which appear to be unrelated in any of their usages. In this example the verbs
have a reversed hierarchical relationship, with pattern VI directly transitive
and pattern III transitive through a preposition. However since they appear to
be unrelated in meaning they might more properly be classified under separate
dictionary entries, in which case they would not be relevant to an examination of
pattern III – pattern VI pairs.

(80) a. ‫ حـامى عـن‬ [III] [2A]

Hāmā ‫ع‬an (+ i.o.)
‘to defend/protect (s.o./s.th.)’
b. ‫ تـحـامى‬ [VI] [2B]
taHāmā (+ d.o.)
‘to shun/avoid (s.o./s.th.)’

Particularly perplexing is the verb pair in (81a–b), which I have classified as unrelat-
ed, although they have usages which are effectively antonyms in patterns III and VI.

(81) a. ‫ نـاول لـ‬ [III] [3B]

nāwala (+ d.o) li (+ i.o.)
‘to give (s.th) to (s.o.)’
b. ‫ تـنـاول مـن‬ [VI] [3B]
tanāwala (+ d.o.) min (+ i.o.)
‘to receive (s.th) from (s.o.)’

6.4.1.2.2  Compliance with the hierarchy in at least one usage. A further 23 verb
pairs were found to comply with hierarchical valency reduction in at least one
corresponding usage. Examples (82a–b) form a mutual-reciprocal pair, where the
verb in each pattern additionally takes an indirect object which is functionally
identical. These verbs thus show simple numerical valency reduction, though the
dictionary entry is complicated by another indirectly transitive usage of the pat-
tern III verb with a related meaning and a different preposition (82c).

(82) a. ‫ خـاطـر عـلـى‬ [III] [MUT±] [3B]

khāTara (+ d.o.) ‫ع‬alā (+ i.o.)
‘to bet (s.o.) (s.th. [a stake])’
b. ‫ تـخـاطـر عـلـى‬ [VI] [REC±] [2A]
takhāTara ‫ع‬alā (+ i.o.)
‘to make a bet [together] against (s.th. [a stake])’
c. ‫ خـاطـر بـ‬ [III] [TRA−] [2A]
khāTara bi (+ i.o.)
‘to risk/stake (s.th.)’
116 The Arabic Verb

Examples such as this, in which the phenomenon of hierarchical reduction is cer-

tainly being observed for all pattern VI usages and their pattern III equivalents,
may arguably be reclassified as complying with detransitivisation. However, for
several roots a problematic usage is encountered due to the pattern III verb op-
tionally taking either a direct or an indirect object without a change of meaning,
for example in (83):

(83) a. ‫ شـاجـر‬ [III] [MUT] [2B]

shājara (+ d.o.)
‘to quarrel with (s.o.)
b. ‫ تـشـاجـر‬ [VI] [REC] [1B]
tashājara
‘to quarrel with one another’
c. ‫ شـاجـرمـع‬ [III] [MUT−] [2A]
shājara ma‫ع‬a (+ i.o.)
‘to quarrel with (s.o.)’
d. ‫ تـشـاجـرمـع‬ [VI] [MUT−][2A]
tashājara ma‫ع‬a (+ i.o.)
‘to quarrel with (s.o.)’

When the paradigm mutual [MUT] usage of pattern III (83a) is compared with
either of the pattern VI usages in (83b) and (83d), the hierarchical relationship is
preserved: (83b) is a paradigm reciprocal [REC] and (83d) is an indirectly transi-
tive mutual [MUT−] standing in the same relationship to its pattern III equivalent
as in (79b). The difficulty arises with the indirectly transitive mutual [MUT−] in
(83c), which is rare amongst pattern III verbs. Although Badawi et al. (2004: 381)
present no concrete diachronic data, they allude to a change in MSA in which “a
verb which originally had no preposition may now be seen with one” and also
remark that “[a] noticeable tendency is the occurrence of [ma‫ع‬a] ‫“ مـع‬with” …
with verbs of reciprocity, i.e. stems III, VI and VIII”. It is possible that a change
which may obscure the detransitivisation effect commonly observed between pat-
terns III and VI and has perhaps been induced or encouraged by contact with
European languages, especially English, is underway with some verbs. If this is
so, the modern language might be expected to adapt in one of a number of ways
in order to preserve a Saussurean system, either eliminating one of the verbs al-
together or restricting it to usages which conform to a hierarchical relationship
within the pair. For the verbs in (83), corpus data (arabiCorpus) reveals that the
pattern III verb is actually extremely uncommon, while pattern VI is used exten-
sively both intransitively as the reciprocal and transitively with the preposition.
 Chapter 6.  Transitivity and valency 117

6.4.1.2.3  Hierarchical challenges: Other potentially explicable exceptions. The re-

maining four verb pairs which I have designated ‘potentially explicable’ represent
related challenges to the adequacy of the valency structure hierarchy I have pro-
posed in Table 41. The first challenge concerns the hierarchical position of verbs
with two or more indirect object complements. Since I have assumed that overall
numerical valency is the most highly weighted parameter, a trivalent verb with
two indirect objects [3A] is ranked higher than a divalent verb with one direct
object [2B]. Thus the usages of the verbs in (84a–b) do not conform to the hier-
archy, even though the participant in the mutual action represented by the direct
object in (84a) is represented by the indirect object in (84b) in conformity with
the hierarchical reduction exemplified in (79). It is the addition of a third comple-
ment, i.e. the matter being consulted about, which has raised the valency in (84b)
and violated the hierarchy.

(84) a. ‫ شـاور‬ [III] [MUT] [2B]

shāwara (+ d.o.)
‘to consult with (s.o.)’
b. ‫ تـشـاور مـع فـي‬ [VI] [MUT-±] [3A]
tashāwara ma‫ع‬a (+ i.o.) fī (+ i.o.)
‘to consult with (s.o.) about (s.th.)’

It seems, however, that this situation may simply be an artefact of Wehr’s lexicog-
raphy and the texts on which it is based. The arabiCorpus does indeed contain
numerous examples of the pattern VI verb with two prepositional complements
(84b). Although the less common pattern III verb, when it is used, most often ap-
pears with only the direct object as in (84a), the following example demonstrates
that it may appear in the [3B] valency structure with both direct and indirect
object complements:

(85) ‫ أو يشاوره في األمر‬،‫بعث رسالته… دون أن يخبر صاحبه‬

ba‫ع‬ath-a risālat-a-hu … dūna
send;pst-3msg letter-acc-poss.3msg without
’an yukhbir-a SāHib-a-hu,
that inform;npst.3msg-sbjv friend-acc-poss.3msg
’aw yushāwir-a-hu fī l-’amr
or consult;npst.3msg-sbjv-obj.3msg in def-matter
‘He sent his letter … without informing his friend or
consulting him about the matter’
(arabiCorpus: 309dinehayate13.txt)
118 The Arabic Verb

In the only other pair where the same hierarchical conflict occurs, the pattern III
verb ‫( خـابـر‬khābara) is found only rarely in arabiCorpus and never in the sense
of ‘to negotiate’, which is the usage which gives rise to the difficulty. Thus on the
available evidence, it seems unnecessary to consider a change to the hierarchy as
already presented in order to accommodate these two verb pairs.
A similar difficulty is encountered with the pair in (86a–b), in which an ad-
ditional complement for the pattern VI verb complicates what would otherwise
be a paradigm mutual-reciprocal relationship:

(86) a. ‫ دارس‬ [III] [MUT] [2B]

dārasa (+ d.o.)
‘to study with (s.o.)’
b. ‫ تـدارس‬ [VI] [REC+] [2B]
tadārasa (+ d.o.)
‘to study (s.th.) carefully together’

Both verbs are found in the arabiCorpus newspaper corpus, though the pattern
VI verb predominates. Within the purely formal definitions we have adopted in
this chapter, we must accept that they are equal in transitivity and valency and
therefore constitute an exception to the hierarchy.
A related challenge is how to incorporate verbs which are transitive through
the preposition ‫( بـيـن‬bayna – ‘between’). In the data, I have treated this preposi-
tion in the same manner as any other, such that it contributes one to the intransi-
tive object count for the verb. Although other verbs, for example ‫( داول‬dāwala – in
the sense ‘to alternate between’), may take complements introduced by this prep-
osition, the only such verb involved in a pair which is not compliant in at least
one usage and therefore has been classified as challenging the hierarchical order
is shown together with its pattern VI counterpart in (87a−c) :

(87) a. ‫ بـاعـد بـيـن‬ [III] [CAU−] [2A]

bā‫ع‬ada bayna (+ i.o.)
‘to cause a separation between / separate (s.o./s.th.)’
b. ‫ تـبـاعـد‬ [VI] [REC] [1B]
tabā‫ع‬ada
‘to separate / be separated (from one another)’
c. ‫ تـبـاعـد عـن‬ [VI] [MUT−] [2A]
tabā‫ع‬ada ‫ع‬an (+ i.o.)
‘to move away from (s.o./s.th.)’

The verb in (87b–c) is typical of pattern VI usages: either indicating explicitly re-
ciprocal separation [REC] or mutual separation with respect to an indirect object
 Chapter 6.  Transitivity and valency 119

[MUT−]. However, the causative usage in (87a), which gives rise to the problem-
atic [2A] classification, is unusual for pattern III in that the action of the agent
affects only the patient represented by the indirect object and not the agent itself.
Where the patient comprises a plural or dual pronoun or noun, the analysis of a
[2A] valency structure appears sound:

(88) ‫وما الذي يجمع ويباعد بينهما؟‬

wa-mā lladhī yajma‫ع‬-u
and-what rel;msg bring_together;npst.3msg-ind
wa-yubā‫ع‬id-u bayna-humā?
and-separate;npst.3msg-ind between-obj.3du
‘And what is it which brings them [both] together and separates them [both]?
(arabiCorpus: GEN1997: 1364)

However, when the objects or persons to be separated are represented by two dif-
ferent nouns or object pronouns, Arabic allows two syntactic variations: the first
is used most often when both are nouns, the second being preferred when one is
a pronoun and obligatory when both are pronouns.

(89) ‫ما الذي يباعد بين العرب وأوروبا؟‬

mā lladhī yubā‫ع‬id-u bayna
what rel;msg separate;npst.3msg-ind between
l-‫ع‬arab wa-’ūrūbā?
def-Arabs and-Europe
‘What is it which separates the Arabs from Europe?’
(arabiCorpus: 052799OPIN01)
(90) ‫فكل يوم يمر يباعد بين آل جور وبين منصب الرئيس االمريكي‬
fa-kull yawm yamurr-u
and_so-every day pass;npst.3msg-ind
yubā‫ع‬id-u bayna ’āl gūr
separate;npst.3msg-ind between Al Gore
wa-bayna manSib-i r-ra’īs-i l-’amrīkī
and-between office-gen def-president-gen def-American
‘So every day which passes separates Al Gore from the office of
American President.’  (arabiCorpus: 102699REPO03)

Whether the preposition ‫( بـيـن‬bayna) takes two indirect object arguments with
a conjunction introducing the second (89) or is repeated with each instance of
it taking a single indirect object argument (90), it is clear that both object argu-
ments are fully necessary to the sense of the verb in any conceivable context in
which there is no dual or plural object as in (88) or indeed when the action of the
verb takes place between such an object and any other. There is therefore adequate
120 The Arabic Verb

justification for treating both objects as obligatory complements, raising the va-
lency of the verb to three. With two indirect object complements, the valency
structure code then becomes [3A] and hierarchical reduction relative to pattern
VI usages is established.
6.4.1.2.4  Synonyms. Six verb pairs have been classified as synonymous in their
corresponding usages. The pattern III verb in Examples (91a–b) has alternative
usages: it may be either directly or indirectly transitive with meanings which are
distinct yet related. However, there is no hierarchical reduction in pattern VI
(91c–d) since in each instance it is both synonymous with and syntactically iden-
tical to the corresponding usage of its pattern III counterpart (91a–b).

(91) a. ‫ جاوز‬ [III] [2B]

jāwaza (+ d.o.)
‘to pass / go beyond (s.th.)’
b. ‫ جاوز عـن‬ [III] [2A]
jāwaza ‫ع‬an (+ i.o.)
‘to pass over / disregard (s.th.)’
c. ‫ تـجـاوز‬ [VI] [2B]
tajāwaza (+ d.o.)
‘to pass / go beyond (s.th.)’
d. ‫ تـجـاوز عـن‬ [VI] [2A]
tajāwaza ‫ع‬an (+ i.o.)
‘to pass over / disregard (s.th.)’

This phenomenon is contrary to the Saussurean understanding of form and

meaning, since it appears to render the ta- prefix morph meaningless. Thus these
verb pairs remain inexplicable in terms of the valency structure hierarchy or any
other formal model.
6.4.1.2.5  Other exceptions. The two remaining verb pairs show meanings which
are related but not quite synonymous: the problematic usages are shown in
(92a–b) and (93a–b).

(92) a. ‫ نـاكـر‬ [III] [2B]

nākara (+ d.o.)
‘to disapprove of / reject (s.o)’
b. ‫ تـنـاكـر‬ [VI] [2B]
tanākara (+ d.o.)
‘to be ignorant/feign ignorance of (s.th)/ to snub,
pretend not to know (s.o.)’
 Chapter 6.  Transitivity and valency 121

(93) a. ‫ غـافـل‬ [III] [2B]

ghāfala (+ d.o.)
‘to take advantage of (s.o.’s) negligence’
b. ‫ تـغـافـل‬ [VI] [2B]
taghāfala (+ d.o.)
‘to neglect (s.o./s.th.)’

Both roots have negative connotations (‘not know’ and ‘not heed’ respectively)
and arguably the derived patterns are all capable of a simulative interpretation, i.e.
to affect a behaviour. However, together with the synonymous pairs in the previ-
ous section, these remain exceptions to the valency structure hierarchy.
6.4.1.2.6  List of exceptions. Having examined the data and investigated in more
depth the 40 pattern III – pattern VI pairs which appear exceptional, the nine verb
pairs listed in Table 47 are those which I have been unable to reconcile in any
manner with the valency structure hierarchy.

Table 47.  Pattern III – pattern VI exceptions to hierarchical

valency reduction or minimisation39
Root Classification
‫جـنـب‬ j-n-b synonymous39
‫جـوز‬ j-w-z synonymous
‫دخـل‬ d-kh-l synonymous
‫درس‬ d-r-s additional argument
‫روح‬ r-w-H synonymous
‫سـرع‬ s-r-‫ع‬ synonymous
‫غـفـل‬ gh-f-l near-synonyms
‫غـلـو‬ gh-l-w synonymous
‫نـكـر‬ n-k-r near-synonyms

6.4.1.3  Summary
The ta- prefix constitutes the distinctive morphological difference between pat-
terns III and VI. In the previous chapter, we examined the claim that it derives a
reciprocal pattern VI verb from a mutual pattern III verb. Data presented in Sec-
tion 5.2.2.3 revealed that this interpretation is only possible for 69.7% of the 238
verb pairs, i.e. there are 72 exceptions.

39. For this and other examples, patterns III and VI are synonymous for usages where the root
meaning is shared.
122 The Arabic Verb

In this chapter it was suggested that ta- is a morphosyntactic device involved

in detransitivisation. However, when numerical valency reduction and minimi-
sation were taken together, only 60.5% of the verb pairs were found to comply.
When the hierarchy of valency structures in Table 41 was substituted for simple
numerical valency, the percentage of compliant pairs rose to 83.2%, leaving 40
exceptions. Closer examination of these exceptions, and applying a degree of lib-
erality of interpretation where at least one corresponding usage complies, reduced
the number of pairs which are clear exceptions to only nine.
Thus characterising the ta- prefix as a detransitivising morpheme has a suc-
cess rate of up to 96.2% in pattern III – pattern VI verb pairs, provided that detran-
sitivisation is defined as a reduction in the valency structure hierarchy involving
either reduction or minimisation of the number of complements or replacement
of one or more direct object complements with the same number of indirect ob-
ject complements.40

6.4.2 Valency and relationships with pattern I

Having established the usefulness of the valency structure hierarchy of Table 41

for characterising detransitivisation in the pattern III – pattern VI relationship, we
may use it as a tool to investigate whether quantifiable differences in transitivity
also exist between each of these two patterns and pattern I.
It is a basic tenet of the root-and-pattern approach to Arabic morphology
discussed in Chapter 3 that verb patterns such as III are not derived from fully vo-
calised stems such as the pattern I verb stem but formed from tiered morphemes
of which the root provides the base meaning (Section 3.2.1). However, since it is
also normally held that the pattern I verb represents the base meaning, it is ap-
propriate to use pattern I for comparison.

6.4.2.1  Pattern I and pattern III valency

The investigation of the pattern III – pattern VI valency relationship began in 6.4.1
with the hypothesis that the prefix which derives VI from III is detransitivising. It
will be helpful to approach the pattern I – pattern III relationship in the same way.
Thus we begin by asking what difference in valency, if any, we might expect.
We have already noted that for some grammarians pattern III, rare across the
breadth of Semitic languages, is a specialised variant of pattern II (see 2.1.2.5.3) in
which “duplication of the medial has taken place but has then failed, the loss of the
consonant producing compensatory lengthening in the preceding vowel” (O’Leary
1969: 217). MacDonald (1963: 102) comments that on this basis “we shall expect

40. A comparable study of patterns II & V and QI & QII is suggested, but outwith the scope of
the present research.
 Chapter 6.  Transitivity and valency 123

the meaning of [pattern III] to be closely associated with that of [pattern  II]”.41
According to the prosodic templatic analysis of McCarthy & Prince (1990a) as
noted in 3.2.2.4, patterns II and III are indeed prosodically identical, both begin-
ning with a heavy (bimoraic) syllable. The same prosodic template applies to pat-
tern IV, which like II is CvC while III is Cvv. Patterns II and IV are often held to
involve causativisation of the base meaning and “the status of causativisation as a
valency-increasing operation is beyond doubt” (Fassi-Fehri 1987: 23; also Hallman
2005: 6). This is illustrated by the following examples, in which an extra direct ob-
ject complement is added in the augmented patterns thus increasing the valency:
(94) a. ‫ كـتـب‬ [I] [TRA] [2B]
kataba (+ d.o.)
‘to write (s.th.)’
b. ‫ كـتّـب‬ [II] [CAU+] [3C]
kattaba (+ d.o.) (+ d.o.)
‘to make (s.o) write (s.th.)’
c. ‫ أكـتـب‬ [IV] [CAU+] [3C]
’aktaba (+ d.o.) (+ d.o.)
‘to make (s.o.) write (s.th.) / dictate to (s.o.) (s.th.)’
If shared prosody between patterns II, III and IV is a valid predictor of meaning
and/or pattern III is a variant of pattern II, we would expect that pattern III would
also be valency-increasing or transitivising with respect to pattern I.
Valency increase is implied by Wright’s (1967: I.33) statement that “when
[pattern I] denotes a quality or state, [pattern III] indicates that one person makes
use of that quality towards another and affects him thereby, or brings him into that
state”. Furthermore, just as ta- prefixation may be associated with rendering a
direct object complement indirect, as in (79), it is suggested that in like manner
pattern III may convert an indirect object complement of the pattern I verb into
a direct object (Mullins 2005: 17; Wright 1967:I.33), which is also a process of
transitivisation as defined by the hierarchical valency structure in this chapter.
Thus as a starting point we shall assume the hypothesis that pattern III is valency-
increasing with respect to pattern I.
Data recorded in Table 48 represent instances of unambiguous valency in-
crease according to the established valency structure hierarchy. Thus for a root to
be included in the data in the first column, all recorded pattern III usages must
rank higher than all pattern I usages in the hierarchy. In the subsequent columns,
the requirement has been limited to pattern I forms with the specified medial

41. Whilst noting some functional similarities between patterns II and III and concurring with
the theory that they are phonologically conditioned variants, Zaborski (1997: 257–260) estab-
lishes that the vowel-lengthening variant can be traced back to Proto-Hamitosemitic.
124 The Arabic Verb

Table 48.  Roots showing hierarchical valency increase from pattern I to pattern III
Valency increase Valency increase Valency increase Valency increase
patterns I–III pattern Ia–III pattern Ii–III pattern Iu–III
No. with   55 (27.8%) 45 (26.0%) 30 (60.0%) 19 (90.5%)
pattern VI absent
No. with   56 (25.7%) 47 (24.6%) 24 (55.8%) 24 (77.4%)
pattern VI present
TOTAL 111 (26.7%) 92 (25.3%) 54 (58.1%) 43 (82.7%)

vowel. It may be observed that data for roots having a pattern III verb occurring
with and without a corresponding pattern VI verb show no significant difference.
It is also clear that these data on the whole do not support the hypothesis that the
pattern III derivation is valency-increasing relative to the base form. However,
data for valency increase relative to pattern Iu look more promising and a more
thorough examination is warranted.
Regarding the valency of pattern Iu, recall that Holes (2004: 101) states that it
“is always intransitive and denotes the possession or acquisition of a quality that is
permanent”. It is unclear from the context whether Holes intends ‘intransitive’ to
be understood to mean not taking a direct object, thus allowing that verbs which
are transitive through a preposition (see 6.1.4) are in this sense intransitive. Of the
52 pattern Iu verbs in my present data set (those occurring alongside pattern III,
9 of which do not show valency increase), 16 were found to have a usage classified
as [2A], i.e. they may be monotransitive through a preposition, for example:

(95) ‫ ضـرُع الـى‬ [Iu] [2A]

Daru‫ع‬a ‘ilā (+ i.o.)
‘to implore / be submissive before (s.o.)’

However, of these 16, three pattern Iu forms were found in Wehr (1994) which
appear to allow a direct object, including two which adopt a valency of three with
an additional indirect object [3B].

(96) ‫كان في حاجة إلي أن يأنس في نفسه الحزم والشجاعة‬

kān-a fī Hāja ’ilay-ya ‘an
be;pst-3msg in need to-obj.1sg that
ya’nus-a fī nafs-i-hi
sense;npst.3msg-sbjv in self-gen-obj.3msg
l-Hazm wa-sh-shajā‫ع‬a
def-determination and-def-courage
‘He needed me in order to sense determination and courage in himself ’
(arabiCorpus: 042499WRIT02)
 Chapter 6.  Transitivity and valency 125

An example of one of these [3B] verbs from the corpus is shown in (96), though
it is the nature of modern written Arabic that it is impossible to be sure of the
vowelling of most pattern I verbs where alternatives exist. According to Wehr, this
transliteration with p-stem vowelling corresponding to s-stem medial ‘u’ is valid
but the alternate reading as ya’nasa is also possible with p-stem vowelling cor-
responding to s-stem medial ‘i’. Similar ambiguities exist for the other examples,
though contemporary data from native speakers might be sought to corroborate
Wehr’s analysis.
More demonstrably problematic for any hypothesis of pattern Iu – pattern III
transitivisation is that the number reported as complying in the crude data in
Table 48 contains a large number of ‘false positives’. Many of these result from
what are probably best analysed as homomorphic root combinations which give
rise to verbs in patterns Iu and III with apparently unrelated meanings. In (97a−c)
the pattern III verb is clearly related semantically to pattern Ia but not to Iu:

(97) a. ‫ سـ ُمـر‬ [Iu] [1B]

samura
‘to be brown / turn brown’
b. ‫ سـ َمـر‬ [Ia] [1B]
samara
‘to chat (in the evening)’
c. ‫ سـامـر‬ [III] [2B]
sāmara (+ d.o.)
‘to chat (in the evening) with (s.o.)’

I estimate from an examination of the lexical entries that around 30% of the pat-
tern Iu verbs apparently undergoing transitivisation are semantically unrelated to
their pattern III counterparts. Thus it is likely that we are not observing a transi-
tivisation process deriving pattern III from the base form, rather simply that Iu
tends to intransitivity whilst III is most often transitive, i.e. that the data prove no
causal relationship.
Further evidence against pattern III derivation being considered transitivis-
ing comes from examples such as (98a–b), in which the pattern I valency struc-
ture ranks level with that for the corresponding pattern III usage,42 and (99a–b),
in which it ranks higher:

42. The lower valency structure [3B] with one direct and one direct object is given as an alter-
native for (98a) with the same meaning but not for (98b).
126 The Arabic Verb

(98) a. ‫ مـنَـع‬ [Ia] [3C]

mana‫ع‬a (+ d.o.) (+ d.o.)
‘to deny (s.o.) (s.th.)’
b. ‫ مـانـع‬ [III] [3C]
māna‫ع‬a (+ d.o.) (+ d.o.)
‘to deny (s.o.) (s.th.)’
(99) a. ‫ مـنَـح‬ [Ia] [3C]
manaHa (+ d.o.) (+ d.o.)
‘to bestow (s.th.) on (s.o.)’
b. ‫ مـانـح‬ [III] [2B]
mānaHa (+ d.o.)
‘to bestow favours on (s.o.)’
Thus there is little evidence in favour of characterising pattern III derivation as a
process of transitivisation relative to the base meaning and indeed some evidence
to the contrary. It is difficult to devise a chi-square test for statistical significance
regarding valency increase since the nature of the test requires that the variables
are not dependent. However it is possible to test whether pattern I transitivity is
relevant to transitivity in pattern III. Thus in Table 49 I have tested for any correla-
tion between pattern I intransitives and pattern III transitives43 for corresponding
roots, any alternates leading to ambiguity being treated as negatives. The chi-square
value is not quite high enough to establish significance at the p = 0.05 level, but
note that any correlation which may exist is negative, suggesting that if anything
pattern III transitivity is weakly associated with corresponding pattern I transitiv-
ity. This also explains why no significant correlations were found between pattern I
medial vowelling and either pattern III mutuality or the pattern III – pattern VI
mutual – reciprocal relationship in Section 5.2.2.4, since pattern I transitivity is
largely irrelevant to pattern III transitivity, which is an indispensable prerequisite
for mutual meaning and for forming a reciprocal in pattern VI.

Table 49.  Correlation between pattern I intransitivity and pattern III transitivity

Pattern III +[trans] Pattern III −[trans] TOTAL
pattern I +[intrans]   84 34 118
pattern I −[intrans] 135 63 198
TOTAL 219 97 316
Chi-square (χ2) = 2.78 | Prob. (p) > 0.05 (not significant) | Phi coeff. (Φ) = −0.08

43. The criterion used is the presence or absence of one or more direct objects, thus valency
structures [1B], [2A], [3A] and [4A] have been treated as intransitive and all others as transitive,
to allow that transitivisation may involve replacement of an indirect object with a direct object.
 Chapter 6.  Transitivity and valency 127

Although I am cautious about drawing firm conclusions on the basis of the

data I have assembled, which as previously discussed are subject to a num-
ber of limitations, it is evident that transitivisation as an explanation for pat-
tern III meaning is not worth pursuing further at this point, given the results
of the preliminary investigation detailed here and given that the exceptions
are in the majority.

6.4.2.2  Pattern I and pattern VI valency

In the previous chapter (Section 5.2.2.4), correlations were established between
pattern I medial vowelling, often associated with transitivity, and pattern VI
semantics when pattern III is absent. We may use the valency structure data
obtained in this chapter to investigate this further.
There is little to be gained from direct comparison of valency between pat-
terns Iu and VI, since we have already established that intransitive usages pre-
dominate in each of these patterns. However, what may be of interest is whether it
is intransitivity or transitivity in all pattern I variants, rather than specifically the
medial vowelling, which has a bearing on the semantic and syntactic outcome for
pattern VI. In all the tables below, the criteria for transitivity and intransitivity are
as in footnote 43, with ambiguous instances again treated as negatives.
The data in Tables 50–52 give rise to some observations. Intransitivity in pat-
tern VI correlates very significantly with intransitivity in pattern I and is inde-
pendent of whether the root also gives rise to pattern III. This phenomenon is
probably attributable to the semantics of the base meaning. However, pattern I
transitivity is not significantly correlated with transitivity in pattern VI. Simply
put, intransitive base forms frequently result in pattern VI intransitives, but there
is no preference for derivation of pattern VI transitives.

Table 50.  Correlations of intransitivity and transitivity in patterns I and VI

a. VI +[intrans] VI −[intrans]
I +[intrans] 123  4
I −[intrans] 195 37
Chi-square = 13.3 p < 0.0005 Phi = 0.19
VERY SIGNIFICANT

b. VI +[trans] VI −[trans]
I +[trans] 7   89
I −[trans] 8 255
Chi-square = 3.17 p > 0.05 Phi = 0.09
NOT SIGNIFICANT
128 The Arabic Verb

Table 51.  Intransitivity in patterns I and VI according to presence or absence of pattern III

a. Pattern III present VI +[intrans] VI −[intrans]
I +[intrans]   62  2
I −[intrans] 130 24
Chi-square = 6.68 p < 0.01 Phi = 0.18
SIGNIFICANT

b. Pattern III absent VI +[intrans] VI −[intrans]

I +[intrans]   61  2
I −[intrans]   65 13
Chi-square = 6.67 p > 0.05 Phi = 0.22
SIGNIFICANT

Table 52.  Pattern I intransitives and transitives with pattern VI semantics: pattern III absent
a. VI +[SIM] VI −[SIM]
I +[intrans] 20 43
I −[intrans]  8 70
Chi-square = 10.1 p < 0.005 Phi = 0.27
VERY SIGNIFICANT

b. VI +[SIM] VI −[SIM]
I +[trans]  3 29
I −[trans] 25 84
Chi-square = 2.86 p > 0.05 Phi = −0.14
NOT SIGNIFICANT

c. VI +[REC] VI −[REC]
I +[intrans] 16 47
I −[intrans] 50 28
Chi-square = 21.0 p < 0.0001 Phi = −0.39
VERY SIGNIFICANT

d. VI +[REC] VI −[REC]
I +[trans] 23  9
I −[trans] 43 66
Chi-square = 10.4 p < 0.005 Phi = 0.27
VERY SIGNIFICANT
 Chapter 6.  Transitivity and valency 129

What is more interesting is that the data confirm that semantic correlations for
pattern VI in the absence of pattern III are due to transitivity and intransitivity
in pattern I.
The most striking tendency is for pattern VI reciprocals with no correspond-
ing pattern III verb to occur much more readily with pattern I transitives. A par-
tial explanation, which would require further research, is that when pattern I is
capable of being interpreted as mutual [MUT], which is necessarily transitive,
pattern VI may be derived directly without the need for a pattern III intermediate.
Inasmuch as pattern III would thus involve redundant morphology, i.e. form with
no meaning, its absence has a certain Saussurean elegance. Examples (100) and
(101) suggest this possibility:

(100) a. ‫ صـفَـع‬ [Ia] [MUT] [2B]

Safa‫ع‬a (+ d.o.)
‘to slap (s.o.)’
b. ‫ تـصـافـع‬ [VI] [REC] [1B]
taSāfa‫ع‬a
‘to slap one another’
(101) a. ‫ رفَـس‬ [Ia] [MUT] [2B]
rafasa (+ d.o.)
‘to kick (s.o.)’
b. ‫ تـرافـس‬ [VI] [REC] [1B]
tarāfasa (+ d.o.)
‘to kick one another’

6.5 Summary

Application of a valency structure hierarchy to the traditional understanding of

transitivity has enabled characterisation of the function of the ta- prefix in pat-
tern VI as detransitivising with few true exceptions. However, the evidence is that
the pattern III template is not transitivising relative to the base form.
Thus far, reference has not been made to the influential paper on transitivity
by Hopper & Thompson (1980), though in some respects my approach to tran-
sitivity as a hierarchy rather than a binary opposition is inspired by their work.
Although I have limited my data collection here to that which is formal and may
thus be objectively measured, i.e. the subject and object complements specified in
the lexicon, I am conscious that transitivity cross-linguistically is demonstrably
more complex than this. In particular, I am drawn to their view that transitivity
130 The Arabic Verb

“involves a number of components, only one of which is the presence of an ob-

ject of the verb” (Hopper & Thompson 1980: 251). Three of the properties which
­Hopper & Thompson relate to transitivity appear particularly pertinent to an ex-
amination of patterns III and VI: the concepts of affectedness and individuation
are relevant to the discussion of verbal plurality in Chapter 7, while aspect will be
the subject of Chapters 8 to 10.
 chapter 7

The pattern III template

From form to meaning

In the previous chapter, the ta- prefix morpheme was shown to be essentially de-
transitivising in deriving pattern VI from pattern III. However, thus far we have
been largely unsuccessful in establishing a unified description for the meaning of
pattern III itself. Since pattern VI is uncontroversially derived from pattern III by
simple concatenative morphology, i.e. the addition of a prefix, we note that the
remaining morphological form of pattern III is shared by its derived counterpart,
at least in the s-stem. Thus there is a formal basis for assuming that the remaining
morphemic content of the derived pattern is identical with that of its source stem,
pattern III, and we will therefore concentrate in this chapter on the characteristics
of that pattern, though conscious that any findings are applicable to its ta- deriva-
tion also. On the basis that the defining formal feature of the pattern III and its de-
rivative is vowel lengthening, the link between this and verbal plurality will first be
investigated, whilst Section 7.3 will explore other instances of vowel lengthening,
introducing the possibility that it may be associated with aspectual properties.
Returning to the prosodic template approach of McCarthy & Prince (1990a)
introduced in Chapter 3, the component parts of a pattern III verb form may be
analysed as in Example (102) ‫( كـاتـب‬kātaba – ‘to correspond with’):
(102) vowel melody a a s-stem (perfect), active
/\ |
CV skeleton C v v C v C pattern III
| | |
root k t b ‘write’

Recall also that the prosodic analysis of patterns III and II represented in (103)
leads to identification of the first syllable in both as heavy or bimoraic: the pro-
sodic ‘quantity’ of each pattern is the same.
(103) CvvCv [III] CvCCv [II]

However, although they are quantitatively identical they are qualitatively differ-
ent (namely in vowel lengthening as opposed to consonant gemination), despite
exhibiting the same vowel melodies in both s- and p-stem forms (see 2.1.2.5.2).
132 The Arabic Verb

The motivation for attempting to characterise the form-meaning relationship

as realised in Arabic verbal pattern III was originally to evaluate the application of
Beedham’s (2005) method of lexical exceptions to MSA. The research which has
been documented in previous chapters has established a formal morphological
foundation on which such an investigation might be expected to yield a positive
outcome and has also examined pattern III in the context of the verbal system as
a whole. However, as we observed in Chapter 1, Beedham (2005: 153) describes
his method “in terms of the Hegelian triad of thesis, antithesis, and synthesis”
and thus far we have concentrated on thesis and antithesis, i.e. on establishing
what the basic meaning associated with the formal realisation of pattern III is not.
Thus, in search of a synthesis capable of describing pattern III, I am led to apply
Beedham’s methodology afresh and ask what it is which is most characteristic of
pattern III formally.

7.1 Formal characterisation of pattern III

When the verbal patterns were introduced in Chapter 2, both s- and p-stem conju-
gations were elaborated. These stems, together with other pattern III verbal deriva-
tives are presented in Tables 53 and 54 according to whether or not they exhibit the
C1āC2 sequence. Four alternative verbal noun forms are included in these tables,
which I have numbered in accordance with Wright (1967: I.116–117), who notes
that type 1 with the C1āC2 sequence is most common and claims that type 3 is origi-
nal, with 2 and 4 derived from it by phonological change. This exhaustive listing of
pattern III verbal and nominal forms thus leads to the following observations:
1. all pattern III word forms contain at least one Cvv syllable;44
2. in all but two less common verbal noun forms (which may be phonologically
derived) there exists a Cvv syllable with onset at the first root consonant C1;
3. in the majority of forms the Cvv syllable gives rise to a C1āC2 sequence;
4. the only form which contains neither C1āC2 nor C2āC3 is the s-stem passive
verb.
It may therefore be argued that the most characteristic feature of pattern III mor-
phology is vowel lengthening45 and we will thus direct our attention to examining
where vowel lengthening is employed elsewhere in the language system of MSA
and with what meaning.

44. See Section 3.2.2.1 for discussion of the extrametricality of the final consonant.
45. This property is the basis of the L classification of pattern III in Table 7 (Section 2.1.2.5.3).
 Chapter 7.  The pattern III template: From form to meaning 133

Table 53.  Pattern III verbal stems and derivatives with C1āC2 sequence
Description Paradigm form CV template
s-stem (active) verb ‫فـاعَـل‬ C1āC2aC3
p-stem (active) verb ‫يُـفـا ِعـل‬ yuC1āC2iC3
active participle ‫ُمـفـا ِعـل‬ muC1āC2iC3
p-stem (passive) verb ‫يُـفـاعَـل‬ yuC1āC2aC3
passive participle ‫ُمـفـاعَـل‬ muC1āC2aC3
verbal noun (1) ‫ُمـفـاعَـلـة‬ muC1āC2aC3a

Table 54.  Pattern III verbal stems and derivatives without C1āC2 sequence
Description Paradigm form CV template
s-stem (passive) verb ‫فُـو ِعـل‬ C1ūC2iC3
verbal noun (2) ‫فِـعـال‬ C1iC2āC3
verbal noun (3) ‫فِـيعـال‬ C1īC2āC3
verbal noun (4) ‫فِـعّـال‬ C1i C2C2āC3

7.2 Vowel lengthening and plurality

The morphology of MSA employs two distinct means of forming nominal (and ad-
jectival) plurals. Some nouns take regular masculine or feminine inflectional suf-
fixes according to conventional concatenative morphology, as in (104) and (105):

(104) ‫ مـسـلـم‬ ‫مـسـلـمـون‬

muslim muslim-ūn
Muslim Muslim-mpl.nom
(105) ‫ سـيـاسـة‬ ‫سـيـاسـات‬
siyāsa siyās-āt
policy policy-fpl

However a more common strategy is the formation of broken plurals, so called

because morphemic material is inserted between the root consonants according
to a number of different patterns, many of which involve vowel lengthening or
insertion of long vowels. Some productive broken plural patterns are illustrated
in (106) to (108):

(106) ‫ كـتـب‬ ‫مـكـاتـب‬

aktab makātib
office office;pl
(107) ‫ درس‬ ‫دروس‬
dars durūs
lesson lesson;pl
134 The Arabic Verb

(108) ‫ كـلـب‬ ‫كـالب‬

kalb kilāb
dog dog;pl (after Benmamoun 2003a: 57)

McCarthy & Prince (1990b) provide an account of broken plural formation within
the scheme of prosodic morphology and detailed discussion of the mechanisms
involved in forming these plurals will not be undertaken here. What is pertinent
is the preponderance of long vowels in the broken plural patterns, as observed
by Benmamoun (2003a: 56), who maintains “that vowel length encodes plurality
… within the nominal system where the bulk of the so-called broken plurals dif-
fer from the corresponding singular form by having a long vowel”. Benmamoun
(1999) argues for the central role of the p-stem in word formation46 and, taking
this as the basis for derivation of pattern III, he concludes that a unified analysis
is possible in which “the phenomenon of broken plurals is present in both the
nominal and verbal systems” (Benmamoun 2003a: 61). Before examining Ben-
mamoun’s analysis in more detail it will be helpful to investigate the notion of
plurality as it applies to verbal systems.

7.2.1 An introduction to verbal plurality

We are accustomed to number being a category applicable to nominal expres-

sions: plurality is thus a means of indicating reference to multiple persons or
objects. It is common cross-linguistically to encounter verbal inflections which
represent the corresponding nominal number on the verb according to its gram-
matical subject or less commonly its object. This kind of plural inflection is not
what we are concerned with here, but rather a semantic plurality which is integral
to the lexical entry and may be encoded in derivational morphology.
Noting that verbal plurality is particularly common in Amerindian lan-
guages, Greenberg (1991) begins his seminal examination of the phenomenon
in Semitic by referring to the work of Swadesh on Chitimacha (Louisiana), in
which he identifies verbal plurality as potentially being manifested as “temporal
repetition”, “spatial dispersion”, “action by many” or “action on many” (Swadesh
1946 in Greenberg 1991: 577). Greenberg proposes that Arabic pattern II and its
Semitic cognates display many characteristics associated with verbal plurality in
other languages and the argument with respect to Arabic specifically is further
developed by Fassi Fehri (2003).47

46. See also Section 3.1.1.

47. See also Section 4.3.2.
 Chapter 7.  The pattern III template: From form to meaning 135

In cross-linguistic studies, verbal plurality is typologically marked like nomi-

nal plurality and “often shown by partial or complete reduplication or an affix,
both contrasted with the zero of the singular” (Greenberg 1991: 577). Consistent
with this observation, gemination of the medial root consonant in Arabic pat-
tern II constitutes a partial reduplication with respect to the base pattern I and
frequency data presented in Chapter 2 are also consistent with pattern II being
marked relative to pattern I. The following examples illustrate plurality of mean-
ing in pattern II, recalling the semantic categories for that pattern according to
Wright (1967: I.31) referred to in Section 4.2.2:

(109) a. ‫ قـطـع‬ [I]

qaTa‫ع‬a (+ d.o.)
‘to cut (s.th.)’
b. ‫ قـطّـع‬ [II] [temporal repetition]
qaTTa‫ع‬a (+ d.o.)
‘to cut into pieces (s.th.)’
(110) a. ‫ قـتـل‬ [I]
qatala (+ d.o.)
‘to kill (s.o.)’
b. ‫ قـتّـل‬ [II] [action on many]
qattala (+ plural d.o.)
‘to massacre (many persons)’

It is self-evident that the verbs in both (109b) and (110b) involve repeated ac-
tions: cutting and killing respectively. Wehr’s entry for the latter specifies a plural
object, emphasising that the action must necessarily be performed upon multiple
persons, i.e. its plurality is lexically specified. The following examples from Fassi
Fehri illustrate that some verbs may be used either for repeated action on multiple
objects (interpreted distributively) or on the same object (interpreted intensively):

(111) a. ‫[  جـرّح الـجـنـود‬II] [distributive: action on many]

jarraH-a l-junūd
wound_intensively;pst-3msg def-soldier;pl
‘He wounded many [of the] soldiers’
b. ‫ جـرّح الـرجـل‬ [II] [intensive: temporal repetition]
jarraH-a r-rajul
wound_intensively;pst-3msg def-man
‘He inflicted many wounds on the man’
(after Fassi Fehri 2003: 155)
136 The Arabic Verb

Furthermore, it is suggested that pattern II when intransitive is also capable of

interpretation as ‘action by many’ and that the pattern therefore demonstrates
‘natural ergativity’, an alignment analogous to ergative-absolutive case-marking
languages in which the grammatical subject of an intransitive and the patient
of a transitive share the property of plurality (Greenberg 1991: 577–578; Fassi
Fehri­ 2003: 155). However evidence for ‘action by many’ in MSA is doubtful:
Fassi Fehri (2003: 179) considers it insufficiently documented, though he refers to
­Greenberg’s archaic example from Wright (112b), which is not attested by Wehr
(1994) with this meaning.

(112) a. ‫ بـرك الجـمـل‬ [I]

barak-a l-jamal
kneel;pst-3msg def-camel
‘The camel knelt down’
b. ‫ بـرّك الـنـعـم‬ [II]
barrak-a n-na‫ع‬am
kneel;pst-3msg def-camel_drove
‘The [whole drove of] camels knelt down’
(after Wright 1967: I.31)48

A wide-ranging definition of verbal plurality is favoured by Cusic, who suggests

that:
[it] should be broadly construed to include the multiplicity of actions, events, oc-
currences, occasions and so on; but in addition whatever indicates extension or
increase, whether in time or space, of actions or states of affairs.
 (Cusic 1981: 64)

Dixon (2000: 34) recognises both reduplication and consonant repetition as mor-

phological realisations of causativity, citing Gulf Arabic as an example of the latter
process according to Holes (1990: 185). Fassi Fehri (2003: 158) poses this ques-
tion: “If reduplication is a mark of verbal plurality, how can it be the source of
both the “intensive” … and the “causative”…?” However, the argument may be
advanced that rendering a verb causative represents an increase in the complex-
ity of the situation, specifically adding a participant – the causer – to an action or
state of which the ‘causee’ is subject, hence increasing the valency of the verb by
one (Dixon 2000: 30). Therefore causativity is consistent with a broad definition of
verbal plurality. Fassi Fehri (2003: 158) thus concludes that pattern II “is a plural
form … [which] can be read ambiguously”, arguing for a collective/distributive

48. Wright also cites ‫ موّت‬with an intransitive plural meaning which is not attested by Wehr.
 Chapter 7.  The pattern III template: From form to meaning 137

distinction between intensitivity and causativity. However, recall that pattern IV

is often considered the canonical causative form and that it shares a CvC bimo-
raic first syllable with pattern II according to prosodic morphology. Specifically,
Fassi Fehri (2003: 159) draws attention to pattern IV glottal prefixation, which
is exemplified in (113) and compares it with broken plurals which also exhibit
glottal prefixation such as (114), arguing that the glottal prefix is “the essential
morphological segment for marking plurality”.

(113) ‫ أكـتـب‬ [IV]

’aktaba (+ d.o.) (+ d.o.)
‘to make (s.o.) write (s.th.)’
(114) ‫ كـرش‬ ‫أكـراش‬
kirsh ’akrāsh
‘stomach’ ‘stomachs’

Thus Fassi Fehri relates this parallel morphological feature of the glottal prefix to
plurality in nouns and pattern IV verbs in much the same way as ­Benmamoun
analyses vowel lengthening in pattern III as indicative of plurality as observed
above.

7.2.2 Mutuality and reciprocity as verbal plurality

If we allow that there are reasonable arguments in favour of recognising the phe-
nomenon of verbal plurality in MSA pattern II and a possibility that it is also ap-
plicable to pattern IV, we must still examine critically Benmamoun’s (2003a: 53)
claim “that what has been labelled a reciprocal verb such as [kātab] in Arabic is
in fact a plural form of the verb katab (plurality of events, each involving at least
one agent)”. Note that Benmamoun uses the term ‘reciprocal’ here of a verb for
which I prefer the designation ‘mutual’, since I reserve ‘reciprocal’ for the explicit
reciprocity most frequently found in pattern VI, which Benmamoun describes as
reflexive and inchoative.
Two questions must be addressed: is Benmamoun’s analysis of vowel length-
ening as a formal characteristic which is common to pattern III verbs and broken
nominal plurals valid, and is verbal plurality a viable interpretation of shared or
mutual action?

7.2.2.1 Formal comparison of vowel lengthening in pattern III

and broken nominal plurals
In drawing a comparison between the vowel lengthening found in the pattern III
template and that in broken nominal plurals, two striking differences emerge.
138 The Arabic Verb

As already exemplified in (107) and (108), it is only the length and not the
identity of the vowel which appears to be common to broken nominal plurals.
Although we have thus far supposed that it is vowel length which is characteristic
of pattern III also, it is evident from the forms listed in Tables 53 and 54 that it is
specifically the long ā which is basic to the pattern.49 It was already noted that the
only form which does not contain ā is the s-stem passive. Cross-linguistically we
expect the passive to be marked relative to the active and therefore it is justifiable
to consider the vowelling of the active as the default. Furthermore, I will present
evidence in Chapter 11 that use of the passive vowel melody which gives rise to ū
in the s-stem is largely incompatible with pattern III (and pattern VI). Thus it is
entirely possible that the identity of the vowel subject to lengthening, i.e. ā rather
than ū, is important for the formal realisation of pattern III, casting doubt upon
the parallel being drawn.
The other discrepancy is that of the position of the long vowel relative to the
root consonants. Again, in Tables 53 and 54 it is shown that the characteristic
position for the long vowel in pattern III is between the first and second con-
sonants, while Examples (107) and (108) demonstrate that in broken plurals of
minimal stem nouns the long vowel occurs in the other available intra-root posi-
tion between C2 and C3. Significantly, Benmamoun (2003a: 57) finds that “[a]n
extension of [McCarthy’s] analysis of the plural to the reciprocal would yield the
wrong results”, i.e. the verbal stem *katāb instead of kātab. However, by substitut-
ing the pattern I p-stem as the input for this analysis of pattern III derivation,
Benmamoun achieves the desired output (115), which he compares with the bro-
ken nominal plural derivation in (116):

(115) ‫( يـكـتـب‬input form) → ‫( يـكـاتـب‬output form)

yaktubu yukātibu
p-stem [I] ‘he writes’ p-stem [III] ‘he corresponds’
(116) ‫( مـكـتـب‬input form) → ‫( مـكـاتـب‬output form)
maktab makātib
office offices
(after Benmamoun 2003a: 59)

Benmamoun (2003a: 59) thus claims that “taking the imperfective [p-stem] tem-
plate as input to the derivation of the reciprocal allows for a unified analysis of
the verbal and nominal “broken” plural formation”. Although his examples appear
superficially convincing, I believe there are causes for concern. I do not find it en-
tirely satisfactory that the analysis depends upon an input form which is already

49. This is the position of Buckley (2004: 397).

 Chapter 7.  The pattern III template: From form to meaning 139

inflected for person and gender (3rd person masculine)50 and hence derives a
similarly inflected output. The implication is that other persons in the pattern
III p-stem paradigm are derived separately from corresponding inflected inputs,
thus 1st person singular ‫’( أكـاتب‬ukātib) from ‫’( أكـتـب‬aktub) etc., since there is no a
priori reason to prefer 3msg as input form over any other minimally morphologi-
cally complex form in the paradigm. Moreover, in order to then obtain the s-stem
paradigm one must presumably delete the inflectional prefix before inflecting for
person and number. The other issue with the example in (116) is that the noun is
already derived: it is a noun of place prefixed with ma-, as is the example ‫مـسـجـد‬
(masjid – ‘mosque’) which Benmamoun (2003a: 55) uses to demonstrate the pro-
sodic derivational mechanism. He highlights the observation that “in both cases,
the output of the derivation is a word with a second syllable whose onset is the
first radical of the root and whose vowel is long” (Benmamoun 2003a: 58). How-
ever his own examples demonstrate that this outcome is only obtained when the
input form is already prefixed, whereas the simple underived noun stems in (107)
and (108) produce plural forms with the syllable containing the long vowel having
the middle radical of the root as onset. What is consistent, even for quadriliteral
nouns as in (117), is that nominal inputs yield plural forms with a long vowel in
the second syllable.

(117) ‫( قـنـديـل‬input form) → ‫( قـنـاديـل‬output form)

qindīl qanādīl
‘lamp’ ‘lamps’
(after Benmamoun 2003a: 54)

In fact (117) is directly analogous to both (115) and (116) in that all three ex-
amples have four consonants, whether the first is supplied by the root or by pre-
fixation.
Before drawing conclusions we should further note that Benmamoun is not
suggesting that the iambic plural pattern which he has based his analysis upon
is the only pattern on which broken plurals are formed. In Section 7.2.1 we met
the glottal prefix broken plural pattern which Fassi Fehri (2003) uses to argue for
plurality in verbal pattern IV, exemplified in (114) with both glottal prefix and
second syllable long vowel. There are many other patterns which have neither
vowel lengthening nor the glottal prefix, such as (118) in which the long vowel of
the singular is absent in the plural:

50. The category of number in the third person of the p-stem only being represented in the
suffix.
140 The Arabic Verb

(118) ‫ كـتـاب‬ ‫كـتـب‬

kitāb kutub
‘book’ books’

Ultimately the mechanisms of derivation proposed by both McCarthy & Prince

and Benmamoun are only theoretical and are not in themselves formal evidence.
Hence from a purely descriptivist formal perspective, we can state only that an
unspecified long vowel in second syllable position is a frequent characteristic of
nominal broken plurals and that long vowel ā with onset C1 is a basic character-
istic of all pattern III active verbal paradigm forms. Thus it is my contention that
it is too great a leap to draw a parallel, for example, between the long vowels in
‫( دروس‬durūs – ‘lessons’) and ‫( دارس‬dārasa – ‘to study with (s.o.)’) and on this basis
to argue that they are both formal realisations of plurality.

7.2.2.2  Is plurality a valid interpretation of mutual action?

Benmamoun’s argument for verbal plurality is focussed on the formal and deriva-
tional parallels he observes between pattern III verbs and nominal broken plurals:
as such, his treatment of semantics is brief, though somewhat curiously, given his
use of the designation ‘reciprocal’ for pattern III, he maintains that:
support for the characterization of the reciprocals as plural verbs comes from
the fact that the verbs in question do not always have the meaning of reciproc-
ity (where for example one is both an agent and patient) but may also have the
meaning of sharing in the same activity.  (Benmamoun 2003a: 56)

Thus the only semantic grounds which he provides for an interpretation of

plurality are that pattern III is transitive, “whereas reciprocal formation cross-
linguistically­usually yields intransitive forms” (Benmamoun 2003a: 56). Howev-
er, I maintain that pattern VI is the true reciprocal and is usually intransitive (or at
least, less transitive than the corresponding pattern III verb), while still possessing
the long vowel which Benmamoun would characterise as a marker of plurality.
It is clear from evidence presented in previous chapters that pattern VI usages
are distinct from the base meanings of corresponding pattern I verbs which have
neither the long vowel nor the ta- prefix. It has also been demonstrated that pat-
tern III is not predominantly transitivising relative to pattern I,51 thus we cannot
suppose that pattern VI is simply a form transitivised by vowel lengthening and

51. See Section 6.4.2.1

 Chapter 7.  The pattern III template: From form to meaning 141

simultaneously detransitivised by ta- prefixation. Moreover, I would suggest that

it is the often intransitive pattern VI true reciprocal, whose subject must be either
grammatically or logically dual or plural, which most closely meets one of the
criteria for verbal plurality, i.e. action by many.
However, the semantic case for pattern III mutuality as an expression of ver-
bal plurality is made more convincingly by Fassi Fehri (2003: 160), who begins by
arguing that in the following examples “the subject is more active than the object,
and the two sentences are not equivalent”:

(119) a. ً‫ سـابـق زيد عـمـرا‬ [III]

sābaq-a zayd ‫ع‬amr-an
race;pst-3msg Zayd Amr-acc
‘Zayd raced Amr’
b. ً‫ سـابـق عـمـر زيدا‬ [III]
sābaq-a ‫ع‬amr zayd-an
race;pst-3msg Amr Zayd-acc
‘Amr raced Zayd’  (after Fassi Fehri 2003: 160)

Whilst I do not accept that the subject here is more active than the object in
the literal sense of the action being performed, I have previously suggested that
the difference in similar examples is thematic52 and furthermore the choice of
grammatical roles may be indicative of which participant is the initiator of the
action. This is perhaps what Fassi Fehri means by “not equivalent”, and I have
already noted that there are many verbs in pattern III which I have described
as asymmetrically mutual,53 where the non-equivalence of subject and object is
more pronounced. Thus it may be necessary to recognise asymmetry as inherent
in pattern III, though consisting of a continuum ranging from equal partners in
the action differentiated only thematically through emphasis on initiator to dif-
ferentiation of active and passive participants. Fassi Fehri concludes that pattern
III vowel lengthening:
expresses plurality of participants. But participants are not treated as ‘equal’.
[Plurality­] is partitioned, hence the transitivity. In the reciprocal [pattern VI],
the participants are assembled, hence the intransitivity, even though the [event]
is plural. (Fassi Fehri 2003: 162)

52. See Section 5.2.1

53. See Section 5.2.2.1
142 The Arabic Verb

Thus vowel lengthening represents semantic plurality in both patterns III and
VI; partitioning of participants in a transitive construction allows asymmetric
differentiation in pattern III; and detransitivising by ta- prefixation removes the
­asymmetry and assembles the participants in one grammatical subject. Hence
there appears to be a good semantic case for accepting verbal plurality as a prop-
erty of both pattern III mutual verbs and pattern VI reciprocal verbs.

7.2.3 Summary – Plurality

The evidence discussed in the previous two sections leads us to conclude that
verbal plurality is a viable semantic explanation of pattern III mutuality and pat-
tern VI reciprocity, though the formal equivalence of vowel lengthening in these
patterns with that in certain nominal broken plurals is doubtful.
However, a major difficulty for establishing the form–meaning relationship
in these verbal patterns remains. It appears that Benmamoun (2003a) is mis-
guidedly using ‘reciprocal’, which should be reserved as a semantic label, for
all formal realisations identified as pattern III, a practice which as previously
discussed54 can lead to prejudging the meaning of a given morphological form.
In fact, it has already been demonstrated in Chapter 5 that sharing in an activ-
ity, or mutuality of action, is the dominant meaning for pattern III, but that over
one-quarter of all verbs in this pattern do not express this dominant meaning.
This is an issue which Benmamoun does not address. Thus even if his case for
pattern III plurality is valid, his analysis of the whole set of 465 verbs which
share this common template is flawed: he has simply chosen to ignore the sub-
stantial subset of the verbs in this pattern which are not capable of mutual and
hence plural interpretation.
This continued inability to account for the lexical exceptions to mutual-
ity in pattern III leads me to investigate whether the characteristic long ā of
pattern III (and hence pattern VI), if not equivalent to vowel lengthening in
nominal broken plurals, is nevertheless found elsewhere in the morphological
system of MSA.

54. See Sections 2.1.2.5.3 and 4.3.1

 Chapter 7.  The pattern III template: From form to meaning 143

7.3 The long ā in Arabic morphology

Rather than proceeding from meaning to form, as Benmamoun (2003a) appears

to do for plurality, I consider it preferable to start with form, since this evidence
is objective. Thus Tables 55 and 56, largely based on Wright (1967) and Schulz
(2004), bring together an exhaustive list of forms (broadly arranged in order of
morphological complexity) which exhibit long ā in C1āC2 and C2āC3 sequences
respectively, repeating some of the data for pattern III verbal forms and deriva-
tives from Tables 53 and 54 for ease of comparison. We may readily make the
following observations on the basis of the data in these tables:
1. whilst long ā is present in a number of broken nominal plurals, it is by no
means restricted to these forms;
2. all verbal patterns from I to XII (except XI) have at least one verbal noun form
with long ā;
3. both C1āC2 and C2āC3 sequences are commonly encountered, though only
the former is found in verbal patterns.

Table 55.  All forms with C1āC2 sequence

Description Paradigm form CV template
active participle [I] ‫فا ِعل‬ C1āC2iC3
noun of instrument ‫فا ِعلة‬ C1āC2iC3a
noun of instrument ‫فاعول‬ C1āC2ūC3
s-stem (active) verb [III] ‫فـاعَـل‬ C1āC2aC3
p-stem (active) verb [III] ‫يُـفـا ِعـل‬ yuC1āC2iC3
p-stem (passive) verb [III] ‫يُـفـاعَـل‬ yuC1āC2aC3
active participle [III] ‫ُمـفـا ِعـل‬ muC1āC2iC3
passive participle [III] ‫ُمـفـاعَـل‬ muC1āC2aC3
verbal noun [III] ‫ُمـفـاعَـلـة‬ muC1āC2aC3a
broken plural of verbal noun [IV] ‫أفـاعـيـل‬ ’aC1āC2īC3
verbal noun [II]; broken plural ‫تَـفـاعـيـل‬ taC1āC2īC3
s-stem (active) verb [VI] ‫تـفـاعَـل‬ taC1āC2aC3
verbal noun [VI] ‫تـفـاعَـل‬ taC1āC2aC3
verbal noun [VI] ‫تـفـا ِعـل‬ taC1āC2iC3
verbal noun [VI] ‫تـفـاعُـل‬ taC1āC2uC3
p-stem (active) verb [VI] ‫يَـتـفـاعَـل‬ yataC1āC2aC3
p-stem (passive) verb [VI] ‫يُـتـفـاعَـل‬ yutaC1āC2aC3
active participle [VI] ‫ُمـتـفـا ِعـل‬ mutaC1āC2iC3
passive participle [VI] ‫ُمـتـفـاعَـل‬ mutaC1āC2aC3
144 The Arabic Verb

Table 56.  All forms with C2āC3 sequence5556

Description Paradigm form CV template
verbal noun [I]; broken plural56 ‫فَـعال‬ C1aC2āC3
verbal noun [I]; broken plural ‫فَـعالة‬ C1aC2āC3a
broken plural ‫فَـعـالـى‬ C1aC2āC3ā
verbal noun [I]; verbal noun [III]; noun ‫فِـعال‬ C1iC2āC3
of instrument etc.; broken plural
verbal noun [I]; broken plural; noun of occupation ‫فِـعالة‬ C1iC2āC3a
verbal noun [I]35 ‫فُـعال‬ C1uC2āC3
verbal noun [I]; noun of fragmentation ‫فُـعالة‬ C1uC2āC3a
broken plural ‫فُـعـالـى‬ C1uC2āC3ā
verbal noun [I] ‫فَـعالـية‬ C1aC2āC3iya
verbal noun [III] ‫فِـيعـال‬ C1īC2āC3
noun of occupation/intensity ‫فَـعَّـال‬ C1aC2C2āC3
noun of instrument ‫فَـعَّـالة‬ C1aC2C2āC3a
verbal noun [II]; verbal noun [III]; ‫فِـعَّـال‬ C1iC2C2āC3
verbal noun [VIII]
broken plural ‫فُـعَّـال‬ C1uC2C2āC3
broken plural ‫أَفـعـال‬ ’aC1C2āC3
verbal noun [IV] ‫إفـعـال‬ ’iC1C2āC3
noun of instrument etc. ‫ِمـفـعـال‬ miC1C2āC3
active participle [XI] ‫ُمـفـعـا ّل‬ muC1C2āC3C3
verbal noun [II] ‫تَـفـعـال‬ taC1C2āC3
verbal noun [II] ‫تِـفـعـال‬ tiC1C2āC3
verbal noun [V] ‫تِـفِـعَّـال‬ tiC1iC2C2āC3
verbal noun [VII] ‫ا ْنـفِـعـال‬ (i)nC1iC2āC3
verbal noun [VIII] ‫ا ْفـتِعـال‬ (i)C1tiC2āC3
verbal noun [X] ‫اسـتِـ ْفـعـال‬ (i)stiC1C2āC3
verbal noun [XII] ‫ا ْفـ ِعـيعـال‬ (i)C1C2īC2āC3

Some further clarification of both the terminology and usage of the main gram-
matical categories presented in Tables 55 and 56 is required.

55. There are additionally three quadriliteral broken plural forms with this sequence not re-
corded here.
56. Wright (1967: I.133) also recognises “adjectives … assimilated to the participles” with
these forms.
 Chapter 7.  The pattern III template: From form to meaning 145

7.3.1 Nominal templates with long ā

7.3.1.1  Participles
Following Holes (2004), for example, I have used the terms ‘active participle’ and
‘passive participle’ since the concept of participle is familiar from English grammar.
However, once again we risk prejudging our analysis of elements of the language by
naming them according to our perception of their meaning rather than their form.
Wright (1967) frequently, though not entirely consistently, uses Latin termi-
nology, thus nomen agentis and nomen patientis or agent and patient nouns re-
spectively. If we instead turn for insight to the terminology of Arab grammarians,
they are presented as ‫( اسـم الفـاعـل‬ismu l-fā‫ع‬il – ‘noun of doing’) and ‫اسـم المـفـعـول‬
(ismu l-maf‫ع‬ūl – ‘noun of done’). In Arabic these terms are essentially the formal
designations we seek, since the noun in construct in each of them is simply the
corresponding pattern I participle of the paradigm verb ‫( فـعـل‬fa‫ع‬ala – ‘to do’),
though translation necessarily involves sacrificing the formal link, although my
rendition into English of the second term, if ungrammatical, is both appropriate
and revealing. Since our main focus is the verbal system, there has thus far been
no need to discuss the relationship between the categories of noun and adjec-
tive in Arabic, but it will suffice to note that the distinction between nouns and
adjectives is fuzzy, an understanding of which makes sense of this explanation by
Wright, which might at first appear self-contradictory:
The nouns which Arab grammarians call … nomina agentis [agent nouns], and
… nomina patientis [patient nouns], are verbal adjectives, i.e. adjectives derived
from verbs, and nearly correspond in nature and signification to what we call
participles. (Wright 1967: I.131)

This should not be too surprising: we are familiar with the ambiguity which ex-
ists concerning the English gerund, which may often be used nominally (120a) or
adjectivally (120b), as well as verbally for the continuous aspect. However English
also permits this ambiguity of category for the second participle57 for some verbs
(121a–b), though others like ‘to do’ are lexically excluded from nominal usage,
which is what makes my translation of the Arabic term as ‘noun of done’ ungram-
matical, although it renders faithfully the concept of the original.
(120) a. Smoking is not permitted. (noun)
b. The smoking barbecue was extinguished. (adjective)
(121) a. The deceased were buried on the battlefield. (noun)
b. The deceased dictator was buried today. (adjective)

57. See footnote 1.

146 The Arabic Verb

We shall return in Chapter 11 to consider whether there is any evidence for

comparable lexical exclusion of some Arabic verbs from usage in the passive
participle form.
Regarding the active participle, when used adjectivally it broadly means ‘per-
forming the action designated by the verb’ and nominally it designates the ‘per-
former of the action designated by the verb’, i.e. the agent, whether animate or
inanimate. Schulz (2004: 70) notes that adjectival and nominal usages are often
lexicalised as in Examples (122) and (123), with some verbal nouns used in either
category (124).

(122) ‫ بـارد‬ [APT:I]

bārid
‘cold’
(123) ‫ طـابـع‬ [APT:I]
Tābi‫ع‬
‘stamp [for making an imprint]’
(124) ‫ عـامـل‬ [APT:I]
‫ع‬āmil
‘active’/‘worker’ (after Schulz 2004: 70)

The following examples of the wāw al-Hāl (‫‘“ – واو الحـال‬and’ of circumstance”)
construction (Cachia 1973: ٣٤) employing a p-stem verb and an active participle
respectively demonstrate their interchangeability:

(125) a. ‫وهو يـنـظـر إل ّي‬

wa-huwa yanDHur-u ’ilay-ya
and-he look;npst.3msg-ind to-obj.1sg
‘while [he was] looking at me’
b. ‫ وهو جـالـس مـع يـسـيـرو‬
wa-huwa jālis ma‫ع‬a yasīr
and-he sit;apt with Yasir
‘while [he was] sitting with Yasir’ (after Al-Tarouti 1991: 131–133)

Indeed, Al-Tarouti (1991: 134–135) states that “[t]he above examples illustrate the
compatibility of the [p-stem] verb and the active participle to the extent that they
can be substituted with each other”. Al-Saqi writes extensively on the nature of
the active participle, describing it in the title of his work as “between nominality
and verbality”, and concurs that it is interchangeable with its p-stem verb (Al-Saqi
1970: 41). Thus, note that the active participle represents ongoing, ­incomplete
­action in the same manner as the p-stem verb, which is usually designated imper-
fect or imperfective.
 Chapter 7.  The pattern III template: From form to meaning 147

Recall also at this point (see Section 3.2.2.3) that the prosodic analysis of
McCarthy­ & Prince (1990a: 28–29) reveals that the nominal CvvCvC prosodic
template, on which the pattern I active participle stem is based and which ac-
counts for 97% of nouns in this class, has “no role … as a primitive, underived
template”. Indeed they conclude that it is derived by lengthening the initial vowel
of the verbal pattern in the same manner as the pattern III verbal stem. According
to this analysis, it would be entirely consistent to expect a commonality of mean-
ing between the active participle and the pattern III verb.
The passive participle is also used adjectivally in the sense of ‘state resulting
from performance of the action designated by the verb’ or nominally to mean
‘entity which has had the action designated by the verb performed on it’, i.e. the
patient, as in Examples (126) and (127) respectively.

(126) ‫ مـفـتـوح‬ [PPT:I]

maftūH
‘open’ (i.e. ‘having been opened’)
(127) ‫ مـوظّـف‬ [PPT:II]
muwaDHDHaf
‘employee’ (i.e. ‘one who has been employed’)

It is important to be aware that the Arabic passive participle is not directly

equivalent to the English second participle. Beedham’s (2005: 33–46) analysis
of the passive as aspect in English draws the distinction between actional and
statal passives, which both make use of what he characterises as the “be + V‑ed”
construction involving the second participle. The passive sentence using this
construction in (128a) is ambiguous as to actional or statal interpretation, as
(128b–c) demonstrate:
(128) a. The door was closed.
b. The door was closed by the janitor at 6.00. [actional]
c. The door was closed when I walked past it at 6.00 [statal]
(after Beedham 2005: 34)

However, MSA does make a distinction between actional and statal passive, re-
serving the passive participle for statal use, consistent with its adjectival quality,
as in (129):

(129) ً ‫كـان البـاب مغـلـقـا‬

kān-a l-bāb mughlaq-an
be;pst-3msg def-door close;ppt-acc.indf
‘The door was closed’ [statal]
148 The Arabic Verb

Emphasis on state is demonstrated by the following corpus Example (130) of the

same verbal participle in context, in which the implication is that the doors (to
peace) were closed throughout the former Prime Minister’s term of office:

(130) ‫بهدف أن يفتح أمامه أبوابا كانت مغلقة في عهد رئيس الوزراء السابق بنيامين نيتانياهو‬
bi-hadaf ’an yaftaH-a ’amāma-hu
with-goal that open;npst.3msg-sbjv before-obj.3msg
’abwāb kān-at mughlaq-a fī ‫ع‬ahd
door;pl be;pst-3fsg close;ppt-f in time
ra’īs-i l-wuzarā’-i s-sābiq binyāmīn
president-gen def-minister;pl-gen def-former Benjamin
nītānyāhū
Netanyahu
‘… with the goal of opening doors before him which were closed
in the time of former Prime Minister, Benjamin Netanyahu’
(arabiCorpus: 121499FILE01)

In contrast, the Arabic actional passive is more naturally rendered using one or
other of the constructions shown in (131) and (132), employing vowel melody
change or a pattern VII inchoative or ‘middle’ verb respectively.

(131) ‫أُغـلِـق البـاب‬

’ughliq-a l-bāb
close;pst.pass-3msg def-door
‘The door was closed’ [actional]
(132) ‫انـغـلـق الـبـاب‬
inghalaq-a l-bāb
be_closed;pst-3msg def-door
‘The door was closed’ (or ‘The door closed’)58 [actional]

The absence in Arabic of the equivalent of the agentive by-phrase usually associ-
ated with the English passive59 does not compromise actional interpretation, and
moreover:

58. Any doubt as to the equivalence of the passive and the inchoative is removed by the under-
standing that external agency of some kind is necessarily implied, since a door is incapable of
closing itself.
59. Holes (2004: 319–320) reports that constructions representing overt expression of the agent
of a passive verb are now becoming commonplace in journalistic Arabic, however, due to trans-
fer from European languages. He also notes a further construction employing a grammatically
active verb with passive meaning, which is irrelevant to the present discussion.
 Chapter 7.  The pattern III template: From form to meaning 149

[t]he agentive by-phrase [in English] is optional. In fact, it is rare. According

to Quirk et al. (1985: 164–165) approximately four out of five English passive
sentences have no expressed agent, usually because the agent is irrelevant or un-
known, as in My house was burgled last night.  (Beedham 2005: 35)

Thus the Arabic passive participle emphasises result not process, i.e. the state
which results rather than the action which produced that state. In (133), the same
construction as (125a–b), the passive participle describes the result of the action
of the corresponding verb upon what would be the patient of the verb (‘her eyes’)
as opposed to the ongoing performance of the verb:

(133) ‫وعـيـنـيـها مـفـتـوحـة‬

wa-‫ع‬ayn-ay-hā maftūH-a
and-eye-du-poss.3fsg open;ppt-f
‘[while] her eyes [were] open[ed] / with her eyes open’
(after Al-Tarouti 1991: 135)

For our present purposes, it may be helpful to think of the active and passive par-
ticiples as ‘doing’ and ‘done’ forms and to note that the paradigmatic ‘doing’ form
(that of pattern I) and both the ‘doing’ and ‘done’ forms of patterns III and VI all
contain the C1āC2 sequence.

7.3.1.2  Verbal nouns

Known in Arabic as the ‫( مـصـدر‬maSdar – ‘source’), the verbal noun is the closest
counterpart of the English infinitive and indeed it is described as an infinitive
by Schulz (2004: 58). Although “systematically related to specific verb [patterns]
… from triliteral or quadriliteral roots” (Ryding 2005: 75), the verbal noun ex-
ists in numerous templates, especially in the lower numbered patterns. Among
these templates, as already observed, all triliteral verb patterns from I to XII
(excluding XI) have at least one verbal noun template with long ā, usually in a
C2āC3 sequence, although II, III and VI all have verbal nouns with C1āC2. ­Ryding
(2005: 75) suggests that the verbal noun is best understood as “[naming] the ac-
tion denoted by its corresponding verb”. Thus it also frequently fulfils a role simi-
lar to the nominal use of the English gerund and there is considerable overlap in
usage with the active participle. Wright both draws a distinction and acknowl-
edges the similarities:
The [verbal nouns] are by their very nature substantives, but have come to be
used also as adjectives; the [active and passive participles] are by their very
nature adjectives, but have come to be used also as substantives.
 (Wright 1967: I.109)
150 The Arabic Verb

Although the verbal noun may be lexicalised as in Examples (134a–b), it is fre-

quently employed to express the action of the verb, especially when the agent
is not relevant or non-specific and thus “like passivization, allows “unattribut-
able” claims to be made” (Holes 2004: 320). Thus the finite subjunctive verb ‫يـصـلـح‬
(yuSliHa) in (135a), which necessarily possesses a subject, may be replaced by its
(non-finite) verbal noun ‫’( إصـالح‬iSlāH [vn:iv]), allowing the agent of reform to
remain unattributed as in (135b).
(134) a. ‫كـذب‬
kidhb
‘lie’
b. ‫سؤال‬
su’āl
‘question’
(135) a. ‫طـلـب زعـيم المـعـارضة أن يـصـلـح رئـيـس الوزراء االقـتـصـاد‬
Talab-a za‫ع‬īm-u l-mu‫ع‬āraDa
demand;pst-3msg leader-nom def-opposition
’an yuSliH-a ra’īs-u
that reform;npst.3msg-sbjv president-nom
l-wuzarā’-i l-iqtiSād
def-minister;pl-gen def-economy
‘The leader of the opposition demanded that the prime minister
reform the economy’
b. ‫طـلـب زعـيم المـعـارضة إصـالح االقـتـصـاد‬
Talab-a za‫ع‬īm-u l-mu‫ع‬āraDa
demand;pst-3msg leader-nom def-opposition
’iSlāH-a l-iqtiSād
reform[vn:iv]-acc def-economy
‘The leader of the opposition demanded reform of the economy
[demanded that the economy be reformed]’

When used verbally in this way the verbal noun retains the lexical parameters
specified for its corresponding verb, in particular the valency of the verb and any
prepositions introducing indirect objects. The following Examples (136a−c) show
that the first specified argument, if either subject or direct object, goes into con-
struct with the verbal noun and the remaining verbal arguments are represented
as direct or indirect objects as specified by the verb’s lexical parameters.
(136) a. ‫قـبـل مـغـادرته الـعـاصـمة‬
qabla mughādarat-i-hi l-‫ع‬āSima[t-a]
before leaving[vn:iii]-gen-poss.3msg def-capital[-acc]
‘before his leaving the capital’
 Chapter 7.  The pattern III template: From form to meaning 151

b. ً‫ تـعـيـيـن الـلواء مـديـرا‬

ta‫ع‬yīn-u l-liwā’-i mudīr-an
appointing-nom def-general-gen director-acc.indf
‘appointing the general as director’
c. ‫تـحـويـل الحـلـم إلى حـقـيـقـة‬
taHwīl-u l-Hulm ’ilā Haqīqa
transforming-nom def-dream to reality
‘transforming the dream into reality’ (after Ryding 2005: 81–83)

That the verbal noun retains the lexically specified parameters of its correspond-
ing verb will be important in the consideration of aspect which will be developed
further in the following chapter.

7.3.1.3  Nouns of instrument

A further subset of nominal templates comprises nouns of instrument and certain
related and specialised usages for vehicles and vessels. The templates included
in Tables 55 and 56 are not the only ones available for nouns of instrument and,
significantly, Schulz (2004: 78) notes that “[i]n Modern Standard Arabic the ac-
tive participle is far more widespread to denote instruments”, a usage which we
have already noted. Examples of the five templates containing long ā are shown
in (137a–e) alongside the corresponding verbs, the first three having a C2āC3
sequence and the other two having C1āC2:
­
(137) a. ‫ مـفـتـاح‬ ‫فـتـح‬
miftāH fataHa
‘key’ ‘to open’
b. ‫ حـجاب‬ ‫حـجـب‬
Hijāb Hajaba
‘veil/covering’ ‘to veil/cover’
c. ‫ غـسّـالـة‬ ‫غـسـل‬
ghassāla ghasala
‘washing machine’ ‘to wash’
d. ‫ قاطـرة‬ ‫قـطـر‬
qāTira qaTara
‘locomotive’ ‘to tow’
e. ‫ حاسـوب‬ ‫حـسـب‬
Hāsūb Hasaba
‘computer’ ‘to calculate’

As the terminology implies, nouns of instrument are agent nouns, being the (in-
animate) entity which performs the action of the verb. Concerning (137e), this is
152 The Arabic Verb

clearly a recent coinage and the other three examples given by Schulz (2004: 78)
on this template appear not to be deverbatives. Watson (2002: 131) remarks that
“[a] large number of nouns on [this] template are originally loan words”.

7.3.1.4  Nouns of occupation and intensity

Related to the C1aC2C2āC3a template exemplified in (137c), which has the femi-
nine morphological suffix, is the C1aC2C2āC3 template commonly used for oc-
cupations. Both these templates include not only long ā, but also reduplication
of the medial consonant, a phenomenon already observed to be associated with
verbal plurality.60 Intensitivity has also been established as a kind of plurality of
action, and indeed the same template which serves to represent occupations may
also be used to designate or describe an entity, whether animate or inanimate,
which performs the verb iteratively or intensively. The following examples dem-
onstrate both nominal and adjectival uses:

(138) a. ‫ أ ّكال‬ ‫أكـل‬

’akkāl ’akala
‘glutton’ ‘to eat’
b. ‫ كذاب‬ ّ ‫كـذب‬
kadhdhāb kadhaba
‘[habitual] liar’ ‘to lie’
c. ‫ عالّم‬ ‫عـلـم‬
‫ع‬allām ‫ع‬alima
‘knowing thoroughly’ ‘to know’
d. ‫ ب ّكاء‬ ‫بـكـى‬
bakkā’ 61 bakā
‘given to weeping frequently’ ‘to weep’

From the iterative and therefore habitual meanings of (138a–b), it is not difficult
to see that the application of this template to occupations or professions is merely
an extension of the concept of ‘one who performs the verb iteratively or habitu-
ally’, hence (139a–b):

(139) a. ‫ خـ ّدام‬ ‫خـدم‬

khaddām khadama
‘servant’ ‘to serve’

60. See Section 7.2.1.

61. The form of this noun is modified from the standard template to accommodate the weak
final root consonant.
 Chapter 7.  The pattern III template: From form to meaning 153

b. ‫ خـبّـاز‬ ‫جـبـز‬
khabbāz khabaza
‘baker’ ‘to bake’

Similarly, the C1iC2āC3a template may also represent an occupational meaning,

though an abstract noun referring to performance of a trade or profession as op-
posed to an agent noun referring to the performer, hence ‘performing the verb
habitually by way of an occupation’ (140):

(140) ‫ نـجـارة‬ ‫نـجـر‬

nijāra najara
‘carpentry’ ‘to hew/carve/plane [wood]’

Words formed on these templates are not always deverbal, such as those in (141)
which are most likely derived from nouns; this emphasises that the templates are
productive for occupations and their practitioners.

(141) a. ‫ بـحّـار‬ ‫بـحـر‬

baHHār baHr
‘sailor’ ‘sea’
b. ‫ نـحـالـة‬ ‫نـحـلة‬
niHāla naHla
‘bee-keeping’ ‘bee’

7.3.2 Commonalities of meaning

Drawing together our analyses of the nominal forms described in the previous
sections, if we maintain that the formal representation of long ā bears meaning,
then what might that meaning be? The following observations will help to clarify
some shared properties of the long ā forms:
1. they are almost always associated with activity as opposed to passivity:
describing or designating the agent not the patient;
2. they generally focus on process as opposed to result: ‘doing’ not ‘done’;
3. they frequently represent situations as possessing internal temporal
complexity: durative as opposed to punctual; habitual or iterative
as opposed to a single action.
In the light of these observations it may be helpful to reconsider the pattern III
and pattern VI verbs and what distinguishes them semantically from their pattern
I counterparts where these exist.
154 The Arabic Verb

7.3.2.1  Agency and patiency

Consider the following examples:
(142) a. ‫ قـتـل‬ [I]
qatala (+ d.o.)
‘to kill (s.o.)’
b. ‫ قـاتـل‬ [III]
qātala (+ d.o.)
‘to fight [with] (s.o.)’
c. ‫ تـقـاتـل‬ [VI]
taqātala
‘to fight each other’
In the pattern I example, the direct object is patient, i.e. it receives the action of
the verb, whereas for the corresponding pattern III verb, which we have previ-
ously designated mutual, the direct object is a participant in the action rather than
passively receiving the action, i.e. not patient, whilst in pattern VI the participants
are co-agents.
Compare the following examples in English, which demonstrate that murder
and fight behave differently:
(143) a. The terrorists murdered a man
b. A man was murdered by the terrorists.
c. I saw a murdered man.
d. John fought a man.
e. ?A man was fought by John.

f. *I saw a fought man.

g. John and the man fought [together].
h. *John and the man were fought.
Example (143e) is at best odd as a passive, though may be marginally accept-
able with suitable compositional context, while the attributive use of the sec-
ond participle in (143f) is much less acceptable than (143c). Furthermore, the
reciprocal action in (143g) is not passivisable as the verb is intransitive. If the
corresponding Arabic verbs in (142a–c) behave similarly with regard to the
passive and are representative of their patterns, we would expect both patterns
III and VI to resist passivisation by the vowel melody mechanism. Moreover, I
would contend that in (143d) and (143e) both participants may be designated
fighters, i.e. the English agent noun, whilst *fightee to describe the direct ob-
ject on the paradigm ­employer–employee would be both contrived and redun-
dant. Since the corresponding Arabic verbs in the above examples either have
no ­direct object (pattern VI) or it is not patient (pattern III), we would expect
them generally to have no use for a passive participle, since this is a nomen
 Chapter 7.  The pattern III template: From form to meaning 155

patientis or ‘noun of done’ and there is no patient to specify and no result to

describe. Hence we could describe both participants in (142b–c) individually as
‫( مـقـا تِـل‬muqātil – ‘fighter/fighting’) but neither as *‫*( مـقـا تَـل‬muqātal – ‘fought’).
Is it possible that all Arabic pattern III and pattern VI verbs behave similarly?
I will present evidence in Chapter 11 in support of these related hypotheses of
non-passivisability, also explaining the apparent anomaly of passive participle
templates containing the C1āC2 sequence, which I have noted above as being
associated with activity or agency.

7.3.2.2  Process and result

Regarding the English examples in the previous section, Beedham (2005: 23)
notes that “[t]he possibility or impossibility of the perfect [second] participle ap-
pearing in attributive position, i.e. before the noun, also indicates lexical aspect
of the verb”. Analysing the English passive as “action + state”, he finds that it is
the verb’s telicity, i.e. whether it has an inherent endpoint, which affects passivis-
ability (Beedham 2005: 43–44). Building upon the examples above, consider the
following verbs:

(144) a. ‫ كـتـب‬ [I]

kataba (+ d.o.)
‘to write (s.th.)’
b. ‫ كـاتـب‬ [III]
kātaba (+ d.o.)
‘to correspond with (s.o.)’
c. ‫ تـكـاتـب‬ [VI]
takātaba
‘to keep up a correspondence [with each other]’

The direct object of the pattern I verb in (144a), be it a letter, a book etc., is affected
by the action of the verb such that at the termination of the action it has entered
into a new state: it may be described by the passive participle ‫( مكتوب‬maktūb – ‘writ-
ten’), the endpoint of the action. Both the action itself leading up to the endpoint
and the agent of that action may be designated ‫( كـاتـب‬kātib – ‘writing/writer’), the
active participle with long ā focusing on the process which leads up to the end-
point. However, for the pattern III and pattern VI verbs, while there is no problem
conceptualising either the process or the participants (‫ – مـكاتب‬mukātib – ‘corre-
sponding/correspondent’), what can we plausibly identify as result? Whilst long-
term correspondence may result in a pile of letters, it implies no resulting change
of state upon either or any of the two or more participants, whether they are gram-
matically subject or direct object. Thus we have preliminary evidence for atelicity
in patterns III and VI compared with their telic pattern I counterpart.
156 The Arabic Verb

7.3.2.3  Temporal complexity

There is a strong sense of temporal complexity among pattern III and pattern VI
verbs. Thus (145) most naturally suggests an interpretation with both participants
active more-or-less continuously throughout the specified time period:

(145) ‫ كانا يتـقـاتالن طوال ساعـتين‬ [VI]

kān-ā yataqātal-ā-n Tiwāla sā‫ع‬at-ayn
be;pst-3md fight_together;npst.3m-du-ind during hour-du.gen
‘They were fighting [together] for two hours’

Example (146) represents repeated individual activity on the part of each par-

ticipant, i.e. an iterative interpretation, each taking turns at being the writer and
the recipient of the writing, and the situation described as a whole must neces-
sarily therefore be durative. In this example the durative adverbial phrase implies
sustained activity, but even if only one exchange of letters takes place the event
is temporally complex, since more than one act of letter-writing must have taken
place to justify correspondence.

(146) ‫ كانا يـتـكـاتبان طـوال سـنـتـين‬ [VI]

kān-ā yatakātab-ā-n Tiwāla sanat-ayn
be;pst-3mdu correspond_together;npst.3m-du-ind during year-du.gen
‘They were corresponding [together] for two years’

Clearly durativity and iterativity are aspectual properties, relating to the way in
which events pass through time. The examples presented above represent tem-
porally complex situations, i.e. situations which may be viewed from within and
perceived as having an internal texture, thus they are both durative and dynam-
ic, involving change. Although Al-Tarouti (1991: 139) observes that “reciprocal
events are internally complex and thus viewed as imperfective”, I will proceed
to demonstrate in the following chapter that terms such as imperfective are best
reserved for grammatical aspect, while the aspectual properties in these examples
belong to the category of lexical aspect.

7.3.3 Summary – Hypothesis of long ā as an aspectual marker

On the basis of the evidence presented above, I am led to hypothesise that the
formal expression of long ā in the C1āC2 sequence in both verbs and nouns has
a consistent aspectual significance and that this may also extend to the other
­inter-radical position C2āC3. The following chapter will establish the framework
within which the specific aspectual properties of the vowel-lengthening patterns
will subsequently be examined.
 chapter 8

An aspectual model for

Modern Standard Arabic

The complexities surrounding the characterisation of verbal aspect are such that
different scholars have proposed various models which recognise some or all of
a raft of properties, among them telicity, durativity, dynamicity and iterativity, as
aspectual or not. Thus this chapter will assess an aspectual model by Olsen (1997),
drawing in turn extensively on Vendler (1967), for its applicability to Arabic, in
order for us to address in the following chapter the question of which specific
aspectual property is associated with long ā verbal and nominal templates.

8.1 Defining aspect

Recognition of the category of verbal aspect has its origins in the study of Slavic
linguistics. Consequently, the particular realisation of aspect in Slavic, i.e. the
morphological opposition of perfective and imperfective verbs, has been held by
some to be the defining standard for the category of aspect cross-linguistically­,
such that any potentially aspectual expression in a given language which does not
equate to the Slavic criteria is not aspect. This is the position adopted by ­Zandvoort
(1962), though others have sought to redefine and extend the category of aspect
in recognition of the cross-linguistic complexity of representing the relationship
of the verb to the time line of the event it describes in a manner which is not
simply a matter of tense. Much of the debate centres around whether “aspect as
grammaticalisation of the relevant semantic distinctions” must be held as distinct
from Aktionsart, a German term applied to “lexicalisation of the distinctions”,
whether or not “lexicalisation is by means of derivational morphology” (Comrie
1976: 6–7). In fact Comrie’s approach dispenses with the term ­Aktionsart and,
although Bache recognises the significance of Comrie’s contribution, he con-
cludes that both Comrie (1976) and Lyons (1977) “fail to come to grips with one
crucial problem: the distinction between aspect and Aktionsart” and proceeds to
“attempt to show that a strict distinction between aspect and Aktionsart must be
insisted on” (Bache 1982: 59).
158 The Arabic Verb

It is not my intention here to enter more than absolutely necessary into the
theoretical debate concerning the scope of the category of aspect. I am convinced
of two things: firstly, that the category of tense is insufficient as the only temporal
descriptor grammatically realised in the verb in Arabic, therefore the language
must possess an aspectual system. Secondly, that the aspect of the verb realised
in the inflectional morphology of MSA (see Section 8.2) must interact with ver-
bal temporal semantics, realised in derivational morphology and hence lexically
specified. Thus our present requirement is a workable model of aspect for MSA
which may be used diagnostically to determine the precise aspectual property or
properties the existence of which I have hypothesised for pattern III and VI verbs
and the nominal forms whose templates share the C1āC2 sequence.
In pursuit of such a model I will assume the definition that “aspects are dif-
ferent ways of viewing the internal temporal constituency of a situation” (Comrie
1976: 3), and thus non-deictic, contrasted with tense which is deictic and “relates
the time of the situation to some other time, usually to the moment of speaking”
(Comrie 1976: 1–2). The following definition may also be helpful:
[A]spect is an expression of the way in which an action/event passes through
time, e.g. as a continuous/extended activity, as an event with a final result, as the
beginning of an action, with emphasis on the intensity of an action, etc.
 (Beedham 2005: 19)

I concur with Beedham (2005: 21) that “[a]spect is formally realized in three

different ways in the world’s languages”, though his characterisation of the first
as “Auxiliary + Participle”, while applicable to the Indo-European languages he
has worked with, is too narrow for our purposes. I will therefore use the term
‘grammatical aspect’ to encompass this and other morphosyntactic realisations,
thus including inflectional (but not derivational) morphology. The realisation
of aspect often described as Aktionsart, which is what is relevant for attribut-
ing meaning to verbal patterns III and VI and the formally similar nominal
templates, we will call ‘lexical aspect’, which may or may not have overt formal
expression in derivational morphology. Lastly, we must recognise the existence
of ‘compositional aspect’, noting that verbs in context respond differently with
respect to the temporal constituency of the situation depicted according to their
compositional environment, including, but not limited to, the categories of num-
ber and definiteness in their agents and patients (Verkuyl 1993). We will also
observe that compositional context leads to pragmatically determined aspect in
certain situations.
 Chapter 8.  An aspectual model for Modern Standard Arabic 159

8.2 The Arabic s-stem and p-stem verb forms:

Tense or grammatical aspect?

The two stems of the Arabic verb were introduced in Chapter 2, where we referred
to them as the s- and p-stems following Holes (2004), whilst noting that they are
most often referred to as the perfect and imperfect respectively, for example by
Wright, whose grammar dates from the mid-nineteenth century. He refers to the
verbal stems as either “states” or “tenses”, but writes:
The temporal forms of the Arabic verb are but two in number, the one express-
ing a finished act, one that is done and completed in relation to other acts (the
Perfect); the other an unfinished act, one that is just commencing or in progress
(the Imperfect). (Wright 1967: I.51)

From this characterisation, one might speculate that had Wright been aware of
modern linguistic theory and terminology, he would instead have described the
two stems not as tenses but as aspects.62 The non-deictic and hence aspectual
nature of the Arabic stems leads Wright (1967: I.51) to observe that “[a] Semitic
Perfect or Imperfect has, in and of itself, no reference to the temporal relations
of the speaker … and of other actions which are brought into juxtaposition with
it.” Are we therefore justified in categorising the Arabic stems as aspect and dis-
pensing with the terminology of tense altogether? Wright (1967: I.51) acknowl-
edges that the nomenclature used by Arab grammarians refers to deictic time, i.e.
tense, arguing that they “have given an undue importance to the idea of time, in
connection with the verbal forms”. Use of the term ‫( الـماضي‬al‑māDī – ‘the past’)
for the s-stem is near universal in Arabic linguistic metalanguage, although the
term referring to the p-stem, ‫( الـمـضـارع‬al‑muDāri‫)ع‬, has no explicit time con-
notation, rather meaning ‘alike’ or ‘similar’. Wright observes, however, that Arab
grammarians also assign the terms denoting present (‫ – الـحال‬al-Hāl or ‫–  الـحاضـر‬
al-HāDir) and the future (‫ – الـمسـتـقـبـل‬al-mustaqbal) to the p-stem.
Pragmatically, we may suggest that completed events are most often located
in the speaker’s past, while events represented as incomplete may be located any-
where on the time line. This simplified view is broadly aligned with the usage
of the s- and p-stems respectively. Thus the p-stem functions for all events not
yet completed at the time referred to, including the present, the future, the ne-
gated past (with ‫ – لم‬lam – and the subjunctive)63 and the imperative. However,
under certain circumstances it is clear that the notion of completeness, which is

62. As do Buckley (2004: 537ff.) and Fleisch (1974).

63. The most common realisation of past negation in MSA (Bahloul 2008: 45).
160 The Arabic Verb

non-deictic and aspectual, appears to override deictic temporal considerations.

Thus Example (147), also quoted by Comrie (1976: 79), is notable both for the
absence of the future marker ‫( سوف‬sawfa) with the first (p-stem) verb and the use
of the s-stem with future sense for the second:

(147) ‫أجـيـئـك إذا احـمـ ّر الـبـسـر‬

’ajī’-u-ka ’idhā Hmarr-a
1sg-come;npst-ind-obj.2msg when redden;pst-3msg
l-busr
def-unripe_date[collective]
‘I shall come to you when the [unripe] date ripens’
(after Wright 1967: I.9)

The ripening of the dates is thus presented aspectually as complete at the time
of an as-yet hypothetical visit by use of the s-stem, normally associated with the
past. Although such usage may be somewhat archaic, the p-stem being preferred
by many in MSA, similar conditional usages of the English progressive past and
simple past in (148a–b) to refer to possible future events demonstrate that choice
of tense is often not obvious and that aspectual considerations are important:
(148) a. I would buy a new car if I was earning enough money.
b. I would buy a new car if I won the lottery.

Thus we must be aware that:

[w]hat is classified as a tense, a mood or an aspect in any particular language may
have a range of functions, some or all of which may fall outside the scope of the
general definition of the grammatical category in question…. Furthermore, it is
undoubtedly the case that the terms conventionally used to describe the func-
tions of the tenses, moods and aspects in certain languages are very misleading.
This point must be borne constantly in mind.  (Lyons 1977: 682)

There is no a priori reason in language why the existence of a form with perfective
aspect implies a corresponding imperfective form, nor why a past tense is neces-
sarily juxtaposed with a contrasting non-past tense or tenses. As Lyons (1977: 688)
continues, “although tense and aspect may be found in the same language, it is not
uncommon for there to be gaps and asymmetries”.
Comrie’s (1976: 78–81) position is that Arabic has combined tense-aspect
oppositions, a view supported by Al-Tarouti (2001: 206–207), and Mohammad
(1983: 122) elaborates a full tense-aspect system. Meanwhile Bahloul (2008: 71),
in recognition that for the Arabic perfect “anteriority is not always deictically
defined”, argues for “a taxis-aspect” system following Jakobson (1971). Holes
(2004: 217), who is particularly concerned with continuing developments in
 Chapter 8.  An aspectual model for Modern Standard Arabic 161

modern Arabic and also draws upon Eisele (1999), writes that “[t]he s-stem/
p-stem distinction was historically not one of tense but of verbal aspect – al-
though synchronically … it is evolving in both MSA and the dialects towards
a tense system.” Indeed Fassi Fehri (2004: 236) argues that “Arabic is more of a
‘tense language’ than an ‘aspect language’”.
Thus the MSA tense-aspect debate is complex and may not be capable of sat-
isfactory resolution. I must reiterate that my interest in it for present purposes is
limited to the interaction of grammatical aspect and tense with lexical aspect. In
glossing examples, I have chosen to represent the s-stem as ‘pst’, in recognition of
its tense function as past, or at least relative past, in modern Arabic. This is not in-
tended to exclude its perfective function, however. Similarly the p-stem is glossed
as ‘npst’, which is consistent with an analysis of the p-stem as the default or tem-
porally and aspectually unmarked stem, much as argued by Benmamoun (1999).
However, since the s- and p-stems are frequently not used in isolation but in
combination with each other and with the particles ‫( قـد‬qad) and ‫( سـوف‬sawfa),
it is possible to construct or identify examples of language in which aspect and
tense are clearly specified. The particle ‫( سـوف‬sawfa), also grammaticalised as pre-
fixed ‫( سـ‬sa-), combines with the p-stem, which serves all non-past functions, to
represent future time. However, the uses and meanings of ‫( قـد‬qad), “sometimes
augmented to laqad” (Dahl & Talmoudi 1979), are more complex.64 Combined
with the p-stem, it expresses uncertainty regarding events not yet completed, much
like the English modal auxiliary ‘may’. With the s-stem, however, the sense is one
of certainty and completeness, thus “[i]f one wishes to emphasize that the action is
complete one may prefix the particle ‫ قـد‬qad or ‫( لـقـد‬laqad) to the perfect” (Cowan­
1958: 56), although Wright (1967: I.286) allows that “[i]t also serves to mark the
position of a past act or event as prior to the present time or to another past act or
event, and consequently expresses merely [the English] Perf. or Pluperf.”. Although
Dahl & Talmoudi (1979) argue that qad has a pragmatic evidential function and
regard Comrie’s representation of it as an explicit perfect marker as misleading, it
undoubtedly has aspectual function also and I will proceed along with Badawi et
al. (2004: 366–367), who claim that “qad reinforces the perfective aspect of [the s-
stem] verb form”. Thus I will treat the qad + s-stem construction as both explicitly
past and explicitly perfective, in the sense in which perfectivity views the situation
from outside as a whole, that is it “involves lack of explicit reference to the inter-
nal temporal constituency of a situation, rather than explicitly implying the lack
of such internal temporal constituency” (Comrie 1976: 21). In Example (149), the
particle qad followed by an s-stem verb has thus been glossed as perfective:

64. Ryding (2005: 450) notes that other augmentations of qad are also found.
162 The Arabic Verb

(149) ‫قـد كـتـبـت الرسـالة‬

qad katab-tu r-risāla
pfv write;pst-1sg def-letter
‘I have written the letter’

In similar fashion, although I regard the p-stem verb when used by itself as not
specifically marked for imperfectivity, when used together with the s-stem of the
copula verb ‫( كـان‬kāna) a past situation is represented and marked as imperfec-
tive, i.e. there is “explicit reference to the internal temporal structure of a situa-
tion, viewing a situation from within” (Comrie 1976: 24). Thus this construction,
much like the French imparfait, expresses past continuous/progressive and past
habitual/iterative situations:

(150) ‫كـانـت تـرتـدي قـمـيـصـا ً أزرق‬

kān-at tartadī qamīS-an ’azraq
be;pst-3fsg wear;npst.3fsg.ind shirt-acc.indf blue;m
‘She was wearing a blue shirt’
(151) ‫كـان يـعـمـل في الـمـطـبـخ‬
kān-a ya‫ع‬mal-u fī l-maTbakh
be;pst-3msg work;npst.3msg-ind in def-kitchen
‘He used to work in the kitchen’
(after Ryding 2005: 447)

Implicit and explicit realisations of grammatical aspect will be encountered

when examples demonstrating the interaction with lexical aspect are examined
in Section 8.4.

8.3 Vendler’s aspectual categories

The most influential work in the categorisation of verbs according to lexical aspect
is a chapter in Vendler (1967: 97–121),65 which establishes four “time schemata”
for English verbs on the basis of how they respond to the progressive and whether
the events they describe are bounded. In stating that “the use of a verb may …
suggest the particular way in which that verb presupposes and involves the no-
tion of time” (Vendler 1967: 97), he is, without pre-empting a later terminology,66
presenting the existence and relevance of aspect as integral to a verb’s entry in the

65. First published in a substantially similar article as Vendler (1957).

66. Vendler (1967: 97) refers to “finer aspects” of time reference, but is not using ‘aspect’ in the
technical sense.
 Chapter 8.  An aspectual model for Modern Standard Arabic 163

lexicon and implying that all verbs in the language may be attributed to a small
and finite set of aspectual categories. Olsen (1997: 11) claims that “all languages
have lexical aspect” and I will hold as a priori that it is therefore possible to con-
struct a model along the lines of Vendler’s schemata which is applicable to MSA.
Verkuyl (1993) examines Vendler’s scheme in some detail, referring to
other philosophical works by Ryle (1949) and Kenny (1963), who also draw
on ­Aristotle, and to later adaptations of Vendler by Mourelatos (1978), Dowty
(1979), Hoeksema (1984)67 and others. As a linguist rather than a philosopher,
Verkuyl (1993: 33–34) argues that Vendler “did not distinguish well between cri-
teria based on (some sort of) agentivity and criteria based on purely temporal
properties of situations”. Nevertheless, Vendler’s categories have been and con-
tinue to be foundational to this field, with subsequent work variously extending,
reducing or modifying Vendler’s scheme and either redefining or renaming the
parameters upon which distinctions are made. The following English examples
illustrate Vendler’s categories:
(152) Tom loves football. (STATE)
(153) Bruce is running. (ACTIVITY)
(154) Harry constructed a model aircraft. (ACCOMPLISHMENT)
(155) Emily recognised James. (ACHIEVEMENT)

Verkuyl (1993: 65) identifies Vendler’s underlying parameters as “process” and

“definiteness”, representing them as equipollent features [± definite] and [± pro-
cess], i.e. they are either present or absent (Table 57).

Table 57.  Verkuyl’s analysis of Vendler’s categories

[− process] [+ process]
[− definite] STATE ACTIVITY
[+ definite] ACHIEVEMENT ACCOMPLISHMENT

However, there are limitations to Vendler’s scheme, among them the failure to take
account of compositionality. Thus the verb in (153), which is ambitransitive, can
be rendered an ACCOMPLISHMENT68 by the addition of a direct object (156):
(156) Bruce is running the Edinburgh marathon.  (ACCOMPLISHMENT)

67. Published as Hoeksema (1985).

68. From this point, terms denoting Vendler’s categories will be rendered in capitals to indicate
this usage.
164 The Arabic Verb

The difference between (153) and (156) is one of telicity: the latter has a built-in
endpoint, whereas the former does not. We might therefore allow that the English
verb ‘to run’ has two lexical entries, one an intransitive ACTIVITY and the other
a transitive ACCOMPLISHMENT. However, Example (157) demonstrates that it
is not the verb’s transitivity which determines telicity:
(157) Bruce is running marathons regularly. (ACTIVITY)

With an indefinite, unspecific direct object, the situation is once again atelic, thus
an ACTIVITY. If this is unclear, it will be helpful to imagine each situation inter-
rupted by some event. Thus we can imagine the following scenarios:
(158) a. Bruce was running marathons regularly when injury forced him to
retire from the sport.
b. Bruce ran marathons regularly.
(159) a. Bruce was running the Edinburgh marathon when he retired injured
half-way.
b. Bruce ran the Edinburgh marathon.

Clearly, if (158a) is true then (158b) is also true: (158a) entails (158b). However
the equivalent entailment does not apply to (159a–b), as the expected completion
or telic endpoint was never reached due to the situation terminating early. Thus,
compositional aspect interacts with lexical aspect for this and other verbs, appar-
ently cutting across categorisations on Vendler’s scheme.
Is an analysis therefore possible which accounts for compositional aspect and
yet recognises inherent aspect in a verb’s lexical specification? This was one of the
issues confronting me in my search for a suitable aspectual model for MSA, along
with the challenge presented by a plethora of proposed aspectual properties, their
oppositions and interactions.

Table 58.  Some aspectual oppositions (Comrie 1976; Bache 1982)

imperfective perfective
continuous habitual
progressive non-progressive
dynamic stative
durative punctual (momentaneous)
telic atelic
semelfactive iterative (repetitive)
resultative non-resultative
 Chapter 8.  An aspectual model for Modern Standard Arabic 165

In Table 58 is given a selection of the most used terminology, presented as op-

posing pairs. However, many monographs and papers have been devoted to the
subject of aspect, thus other scholars mention further properties or rename them
according to some nuance. Thus, for example, punctual is alternatively referred to
as momentaneous or iterative as repetitive. Furthermore, many of these aspectual
properties interact or are sub-divisions of another category. Complicating the is-
sue further is that instead of simple oppositions, related aspect features may be
represented, for example, as stative against non-stative and dynamic against non-
dynamic, instead of merely in opposition to one another.
Thus a suitable model with which to examine derivationally realised lexical
aspect in MSA must explain how lexical aspect features interact with grammatical
aspects and tenses, possess a logical structure capable of comprehension, applica-
tion and if necessary extension, and have no major inconsistencies. With this in
mind I have examined in some detail the scheme proposed by Olsen (1997)69,
which builds upon Vendler, and have found it to meet these criteria.

8.4 Olsen’s scheme

8.4.1 Overview

In her introduction Olsen writes:

Given the variety of lexical, grammatical, semantic, and pragmatic elements con-
tributing to the interpretation of aspect, it is not surprising that analyses … differ
widely. The variation between these studies masks the relations among lexical
and grammatical aspect phenomena. Furthermore, few (if any) theories provide
an explanatory account of the relation between aspect and tense and the tendency
for each, in the absence of the other, to acquire default temporal interpretations
resembling those generally assigned to the other.  (Olsen 1997: 3)

Thus Olsen’s work addresses these issues, adopting a scheme in which “the lexi-
cal aspect classes are not primitives, but are generally assumed to represent clus-
ters of values for lexical aspect features” (Olsen 1997: 11). Moreover, she presents
these features together with those relating to grammatical aspect and tense as
privative: instead of equipollent features, such as [+dynamic] in opposition with
[− dynamic], Olsen’s marked features contrast with absence of marking for that
feature. Thus, for example, a verb is either marked [+dynamic] or is unmarked

69. Also employed by Mughazy (2005) in his examination of lexical aspect in Egyptian Arabic.
166 The Arabic Verb

for ­dynamicity, which we may represent as [Ødynamic]. This distinction between

­being marked as [−dynamic] and being designated as unmarked [Ødynamic] al-
lows a verb to be interpreted as dynamic even if not marked [+dynamic]. Thus the
concept of privative features permits a verb to function flexibly in those proper-
ties for which it is not marked, so allowing for the compositional aspect we have
already encountered and enabling pragmatic interpretations in context. Thus Ta-
ble 59 indicates the seven privative features recognised by Olsen, though it should
be noted that not all languages employ all these marked features and they may
place restrictions on which ones occur together.

Table 59.  Olsen’s privative features

LEXICAL ASPECT
[+durative] [+dynamic] [+telic]
GRAMMATICAL ASPECT
[+perfective] [+imperfective]
TENSE
[+past] [+future]

Olsen also describes situations as having temporal structure and she uses the
concepts of ‘nucleus’ and ‘coda’ borrowed from phonological analysis of syllable
structure. Thus the situation described by a verb may have two distinct phases,
such as in Example (160):
(160) Matthew built a house
|-----------------------------------------|----------------------->
nucleus coda
‘building the house ‘building the
is in progress’’ house is complete’

The horizontal line in the event diagram is a time line representing the situa-
tion, running from past to future, left to right. If we ‘drop in’ at any point during
the nucleus phase, we can say that building the house is in progress. However,
there comes a distinct point in time at which completion occurs, and for any
time after that if we ‘drop in’ in the same way it is no longer true to say that
building the house is in progress, but it is true to say that building the house
is complete. Regarding grammatical aspect, Olsen’s scheme maintains that if a
verb is marked as [+perfective] it makes reference to the CODA phase, whereas
if it is marked [+imperfective] it makes reference to the NUCLEUS phase and
if unmarked for both features it may refer to either. There are two constructions
in English which mark [+perfective] and [+imperfective]: the perfect aspect,
 Chapter 8.  An aspectual model for Modern Standard Arabic 167

perhaps most concisely defined as a past event with present relevance (Comrie
1976: 52),70 and the progressive aspect. Note that both constructions are realised
in English by an auxiliary verb plus a participle: have + V-ed for the perfect and
be + V-ing for the progressive. We may compare the interactions of grammatical
aspect with the time line in (161a–b):
(161) a. Matthew has built a house’ [+perfective]
RT/ST
↓
|------------------------------------------------|----------------->
nucleus coda

b. Matthew was building a house [+imperfective] [+past]

RT ST
↓ ↓
|------------------------------------------------ - - - - - - - - - ->
nucleus

In the first time line (161a), the [+perfective] marked verb points at the coda. ST
represents the speech time, i.e. the time at which the utterance is made. For the
English present perfect, this coincides with RT, the reference time. At this time the
coda situation holds, that is it is true to say ‘building the house is complete’ and
the nucleus situation ‘building the house is in process’ no longer holds: it is over,
though by implication it must have preceded the coda phase. Thus (162) is not
a valid utterance, unless we are using the dissonance for some sort of pragmatic
effect like irony.
(162) *Matthew has built a house and he is still in the process of building it.

The second time line (161b) employs the past progressive, marked [+imperfec-
tive] and [+past]. Note that RT (reference time) now precedes ST (speech time),
as the situation is located by tense in the past. However, RT now points at the
nucleus phase and it is impossible to judge from the sentence whether or not the
coda phase (in which building is complete) has been entered by ST or not. Thus
we could continue the sentence in any of the following ways:

70. Strictly this is the present perfect, although the auxiliary in the perfect construction may
also have past or future marking, so ‘relative present relevance’ would be a more accurate de-
scription.
168 The Arabic Verb

(163) a. Matthew was building a house when we met last year but he hasn’t
finished building it yet.
b. Matthew was building a house when we met last year and he finished
building it today.
c. Matthew was building a house when we met last year but I don’t know
whether he has finished building it yet.
d. Matthew was building a house when we met last year but he will never
finish building it.

It is as if the coda, although normally expected for a verb of this lexical category,
is completely invisible to the imperfective verb, since it hasn’t happened yet. In
fact, the verb ‘to build’ is one of Vendler’s ACCOMPLISHMENTS: it has a defi-
nite or bounded endpoint. Hence in Olsen’s description it manifests a meaningful
coda phase, which she associates with a marked [+telic] feature. ACCOMPLISH-
MENTS in Olsen’s scheme are also marked with the features [+dynamic] and
[+durative], that is ‘building’ involves a changing situation which happens over a
period of time. Olsen views these as nucleus features: they affect only that phase
of event time and how it proceeds.
Table 60 shows how Olsen characterises Vendler’s categories in terms of the
three privative features relevant to lexical aspect. She maintains that a verb must
be minimally [+dynamic] or [+durative], i.e. it must have a nucleus marked with
one or both of these features, but that the [+telic] feature and therefore the coda
are optional. This leaves two other allowable combinations, and Olsen extends
Vendler’s classes to include SEMELFACTIVES,71 which are single actions pre-
sented as punctual, hence [+dynamic] but not [+durative] and with no built-in
resulting endpoint, for example ‘sneeze’ or ‘hit’. She also explores what some have
called ‘stage-level states’ which would be marked [+durative] and [+telic], but she
remains doubtful whether they exist in English (Olsen 1997: 48–49).72
Having understood the basic principles of Olsen’s scheme, which she devel-
ops with particular reference to English and Koine Greek, we may now investigate
its applicability to MSA.

71. Heard (1827: 142) introduces the term ‘semelfactive’ as expressing “the sudden and single
occurrence of an action”, although it is used by Comrie (1976: 42) to contrast specifically with
‘iterative’. Comrie also considers ‘semelfactive’ as a telic aspectual category with an endpoint. I
will henceforth adopt Olsen’s use of the term throughout.
72. See Section 8.4.2.6 for arguments against recognising the validity of the category of STAGE-
LEVEL STATE.
 Chapter 8.  An aspectual model for Modern Standard Arabic 169

Table 60.  Olsen’s extended characterisation of Vendler’s categories

CATEGORY NUCLEUS CODA
STATE [+durative]
ACTIVITY [+durative] [+dynamic]
ACHIEVEMENT [+dynamic] [+telic]
ACCOMPLISHMENT [+durative] [+dynamic] [+telic]
SEMELFACTIVE [+dynamic]
STAGE-LEVEL STATE? [+durative] [+telic]

8.4.2 Application to MSA

I will proceed to examine each of the categories from Table 60 and their in-
teractions with perfective and imperfective grammatical aspects in order to
demonstrate the validity of Olsen’s analysis to MSA. I will adhere to Olsen’s
terminology as far as possible, though while conscious of her reasons for doing
so, I have not found it necessary to distinguish deictic centre (C) from speech
time (ST) in the event time diagrams here. Many examples of actual language
have been obtained from arabiCorpus or the World Wide Web, while others
were tested on three native Arabic speaker informants by questionnaire.73 A
guide to the abbreviations and conventions used in the examples and event
diagrams is found on page xvii.

8.4.2.1  States
A STATE verb is marked as follows:

Nucleus: [+durative] Coda: [Øtelic], i.e. unmarked.

We will take as an example the following pattern III verb:

(164) ‫ جاور‬ [III]

jāwara
‘to adjoin’ (+ d.o.)

All native speaker informants found this verb to be compatible with the imperfec-
tive past construction:

73. Full questionnaire data and methodology are contained in Appendix III.
170 The Arabic Verb

(165) a. [+imperfective] [+past]

‫ كان بيتنا يجاور المكتبة‬ {YYY}
kān-a bayt-u-nā yujāwir-u
be;pst-3msg house-nom-poss.1pl adjoin;npst.3msg-ind
l-maktaba
def-library
‘Our house adjoined/used to adjoin the library’ /
*‘Our house was adjoining the library’  (Questionnaire: 1)
Recall that a verb marked as [+imperfective] references the nucleus phase, thus
in the event diagram (165b), the reference time RT intersects with the nucleus
marked as [+durative], so the situation described holds over a period of past time
and is continuous though not progressive, as it is [Ødynamic], which explains
why the English be + V‑ing construction, although often an appropriate transla-
tion of the Arabic imperfective past, is not valid here.
(165) b. RT ST
↓ ↓
ET: <------------------------------------------>
NUCLEUS
Two out of three informants found Example (165c) less acceptable:
(165) c. [+past]
‫ جـاور بيتنا المكتبة‬ {N?Y}
jāwar-a bayt-u-nā l-maktaba
adjoin;pst-3msg house-nom-poss.1pl def-library
‘Our house adjoined the library’  (Questionnaire: 49)

In my analysis, the s-stem verb is marked [+past] but is not marked [+perfective]
when it stands alone, thus it is capable of the same imperfective interpretation
according to the event diagram in (165b), though it is not surprising that two
informants preferred the explicitly marked [+imperfective] of (165a).
The following verb, also in pattern III, is interesting in that it might be sup-
posed to imply change and hence be [+dynamic] and possibly also [+telic].
Equivalent verbs in English were designated “continuatively durative” verbs by
Poutsma (1926: 289), who noted that they have the sense of “continuing beyond
a certain point of time”.
(166) ‫جاوز‬
jāwaza
‘to exceed, surpass’ (+ d.o.)
This verb is capable of being interpreted dynamically in (167), whilst in (168)
the situation is clearly static (non-dynamic), thus it is consistent with Olsen’s
 Chapter 8.  An aspectual model for Modern Standard Arabic 171

­ esignation of [+durative] [Ødynamic] for verbs of STATE, i.e. absence of the

d
privative feature [+dynamic] allows both dynamic and non-dynamic readings.
(167) [+imperfective] [+past]
‫ مليون شخص‬21 ‫كان عدد سكان القاهرة يجاوز اآلن‬
kān-a ‫ع‬adad sukkān-i l-qāhira
be;pst-3msg number inhabitant;pl-gen def-Cairo
yujāwiz-u l-’ān 21 milyūn shakhS
surpass;npst.3msg-ind def-time 21 million person
‘The population of Cairo was now exceeding / growing beyond
21 million people’  (arabiCorpus: GEN1997: 17420)
(168) [+imperfective] [+past]
‫ما كان يجاوز السبعة والثالثين من الـعـمـر‬
mā kān-a yujāwiz-u s-saba‫ع‬a
neg be;pst-3msg surpass;npst.3msg-ind def-seven
wa-th-thalāth-īn min al-‫ع‬umr
and-def-thirty-acc from def-age
‘He was not more than thirty-seven years of age’
(arabiCorpus: GEN1997: 33758)

Note that (167) also demonstrates that the verb is [Øtelic] as it does not reach an
endpoint in coda. Rather, the sentence includes a goal which is external to the
verb itself, namely the figure of ‘21 million’, and growth can continue indefinitely
past this external goal, which is achieved at an unspecified point during event
time nucleus and is thus located in a range which includes the reference time
‘now’, but is not ordered with respect to RT. Thus any of the three following event
diagrams is possible for (167), although it is pragmatically likely that RT and goal
are close in time, though not coincident.
(169) a. goal about to be reached at RT
RT ST
↓ ↓
ET: <----------------------------------------------->
NUCLEUS ↑
X: goal

b. goal has just been reached at RT

RT ST
↓ ↓
ET: <----------------------------------------------->
NUCLEUS ↑
X: goal
172 The Arabic Verb

c. goal coincidental with RT

RT ST
↓ ↓
ET: <----------------------------------------------->
NUCLEUS ↑
X: goal

8.4.2.2  Activities
An ACTIVITY verb is marked as follows:

Nucleus: [+durative] [+dynamic] Coda: [Øtelic], i.e. unmarked.

A great many verbs in pattern III are in this category, for example:

(170) ‫ قاتل‬
qātala
‘to fight (with)’ (+ d.o.)

The s-stem verb in (171a) is unmarked for grammatical aspect as already noted,
allowing flexibility of interpretation. However, the verb’s nucleus features [+du-
rative] and [+dynamic] require that the situation described held over a period
of time (continuous) and was progressive, i.e. (171a) entails that (171b) is also a
valid statement:
(171) a. [+past]
‫قاتلوا اسرائيل في جنوب لبنان والضفة الغربية‬
qātal-ū ’isrā’īl fī janūb lubnān
fight;pst-3mpl Israel in south Lebanon
wa-D-Diffa al-gharbīy-a
and-def-bank def-west-f
‘They fought Israel in Southern Lebanon and the West Bank’
(arabiCorpus: GEN1996: 6441)
b. [+imperfective] [+past]
‫كانوا يقاتلون اسرائيل في جنوب لبنان والضفة الغربية‬
kān-ū yuqātil-ū-n ’isrā’īl fī janūb
be;pst.3mpl fight;npst.3m-pl-ind Israel in south
lubnān wa-D-Diffa al-gharbīy-a
Lebanon and-def-bank def-west-f
‘They were fighting Israel in Southern Lebanon and the West Bank’

Pragmatically, the situation described by the ACTIVITY verb must have had a be-
ginning, i.e. it was entered into at some unspecified point in time, as represented
in the following event diagram:
 Chapter 8.  An aspectual model for Modern Standard Arabic 173

(171) c. RT ST
↓ ↓
ET: 0------------------------------------------>
nucleus

In (172a) the beginning of the situation is anchored externally by a temporal ad-

verbial within the time-frame of a specific month:
(172) a. [+imperfective] [+past]
1947 ‫كانوا يقاتلون اليهود منذ نونبر‬
kān-ū yuqātil-ū-n al-yahūd mundhu
be;pst-3mpl fight;npst.3m-pl-ind def-Jews since
nuwanbir 1947
November 1947
‘They were fighting/had been fighting the Jews since November 1947’.
 (arabiCorpus: 514isslamic3503.txt)
b. RT ST
↓ ↓
ET: |------------------------------------------>
↓ NUCLEUS
X: |---|
‘November 1947’

Note also that the Arabic s-stem [+past] is not incompatible with durative ad-
verbials, all informants agreeing on the grammaticality of (173a), in which the
beginning and end of the situation are temporally defined relative to one another,
as illustrated in the event diagram in (173b):
(173) a. [+past]
‫ قاتـل أحمد صديقه طوال عشرين دقيقة‬ {YYY}
qātal-a ’aHmad Sadīq-a-hu Tiwāla
fight;pst-3msg Ahmad friend-acc-poss.3msg during
‫ع‬ashrīn daqīqa
twenty-acc minute
‘Ahmad fought his friend for twenty minutes’
(Questionnaire: 6)
b. RT ST
↓ ↓
ET: |----------------------------------------------- |
↓ NUCLEUS ↓
X: |- - - - - - - - - - - - - - - - - - - - - - - - - - - -|
‘twenty minutes’
174 The Arabic Verb

However, although the end of the situation may be externally bounded by such an
adverbial, it should be noted that the verb itself has no built-in endpoint, i.e. the
situation might have continued longer. In (174a) the verb is explicitly marked as
[+perfective] by the particle ‫( قد‬qad), though I take the view that the s-stem in this
perfective role is no longer explicitly marked as [+past], since a present perfect
interpretation is possible:
(174) a. [+perfective] [Øpast]
‫ قد قاتلوا تاريخيا ً من أجل عدم الربط بين التأ ييد إلسرائيل وبين الهجرة إلسرائيل‬
qad qātal-ū tārīkhīy-an min_’ajl ‫ع‬adam-i
pfv fight;pst-3mpl historical-acc.indf because_of lack-gen
r-rabT bayna t-ta’yīd li-’isrā’īl
def-connection between def-support for-Israel
wa-bayna l-hijra li-’isrā’īl
and-between def-emigration to-Israel
‘Historically they have fought on account of74 the lack of connection
between support for Israel and emigration to Israel’
 (arabiCorpus: archive22843)
The situation is represented in event diagram (174b) as ongoing at the present
time where RT and ST coincide:
(174) b. RT/ST
↓
ET: 0----------------------------------------------->
NUCLEUS
While it is certain that the [+durative] [+dynamic] nucleus features held for all
points in past time since the implied commencement of the situation, hence
(174a) entails (174c), the truth of (174d) is also a possibility. The verb is unspeci-
fied for telicity [Øtelic], so there is no coda phase, no natural intersection with the
perfective and hence no inherent endpoint. An arbitrary endpoint is possible, i.e.
they may have finished fighting before RT is reached (174c), but not implicit, as
they may still be fighting (174d):
(174) c. … ‫كانوا يقاتلون من أجل عدم الربط‬
kān-ū yuqātil-ū-n min _ajl ‫ع‬adam-i
be;pst-3mpl fight;npst.3m-pl-ind because_of lack-gen
r-rabT
def-connection
‘They were fighting because of the lack of connection …’

74. An alternative translation has been suggested here for ‫من أجل‬, but this does not impact on
the aspectual argument.
 Chapter 8.  An aspectual model for Modern Standard Arabic 175

d. … ‫ما زالوا يقاتلون من أجل عدم الربط‬

mā zāl-ū yuqātil-ū-n min_’ajl
neg cease;pst-3mpl fight;npst.3m-pl-ind because_of
‫ع‬adam-i r-rabT
lack-gen def-connection
‘They are still fighting because of the lack of connection …’

8.4.2.3  Accomplishments
An ACCOMPLISHMENT verb is marked as follows:
Nucleus: [+durative] [+dynamic] Coda: [+telic]

The following verb will serve as an illustration:

(175) ‫بنى‬
banā
‘to build’ (+ d.o.)

An ACCOMPLISHMENT verb, like an ACTIVITY, has both [+durative] and

[+dynamic] features in the nucleus, thus (176a) implies that (176b) is also valid
for the nucleus phase, cf. (171a–b):
(176) a. [+past]
‫بنى لنفسه بيتا‬
banā li-nafs-i-hi bayt-an
build;pst;3msg for-self-gen-poss.3msg house-acc.indf
‘He built himself a house’  (arabiCorpus: GEN1996: 11400)
b. [+imperfective] [+past]
‫كان يبني بيتا‬
kān-a yabnī bayt-an
be;pst-3msg build;npst.3msg house-acc.indf
‘He was building a house’

Unlike ACTIVITY verbs, ACCOMPLISHMENT verbs have the [+telic] coda fea-
ture also, such that there is a built-in endpoint. Recalling the definition of perfec-
tivity as an outside view of the situation as a whole, the perfective interpretation
of (176a) is that it entails (176c), i.e. in (176a) the endpoint is a completed house.
(176) c. ‫بيته مبني‬
bayt-u-hu mabnī
house-nom-poss.3msg build;ppt
‘His house is built’

Whether or not the endpoint is reached is irrelevant for the validity of the
[+telic] feature, although the perfective reading is often supplied as a matter of
176 The Arabic Verb

­ ragmatics, the assumption being that an imperfective verb would have been
p
employed had the process still been ongoing, hence the optionality of the per-
fectivising particle ‫( قد‬qad ). However, since the s-stem verb itself is not marked
for grammatical aspect, either the nucleus or the coda phase can be the locus
of the intersection of RT with ET, thus (176d) represents the imperfective view
obligatory for the marked [+imperfective] verb in (176b) and optional for the
[Øimperfective] verb in (176a), although the latter is pragmatically more likely
to be understood perfectively as (176e) where RT intersects the coda for which
the result (176c) is true.
(176) d. RT ST
↓ ↓
ET: 0-------------------------------|------------------->
NUCLEUS CODA
e. RT ST
↓ ↓
ET: 0-------------------------------|---------------->
NUCLEUS CODA

Adverbial phrases are often useful diagnostic tools for establishing aspectual
categories. For example, the process adverbial ‘slowly’ is only compatible with
situations displaying both durativity and dynamicity, thus it is incompatible with
STATE verbs (177):75

(177) *‫جـاور بيتنا المكتبة ببطء‬

jāwar-a bayt-u-nā l-maktaba bi-buT’
adjoin;pst-3msg house-nom-poss.1pl def-library with-slowness
*‘Our house adjoined the library slowly’

It is precisely because this type of process adverbial requires durativity and dy-
namicity that it draws attention to the nucleus phase of an ACCOMPLISHMENT
which is marked for these features, effectively locating RT in the nucleus as in (178),
which must therefore correspond to the imperfective event diagram (176d).

(178) ‫بنى بيته ببطء‬

banā bayt-a-hu bi-buT’
build;pst;3msg house-acc-poss.3msg with-slowness
‘He built his house slowly’

75. This is not meant to imply that ‘slowly’ is compatible with all [+dynamic] verbs: testing with
a range of diagnostic adverbials may be required in order to arrive at a conclusive categorisation.
 Chapter 8.  An aspectual model for Modern Standard Arabic 177

In contrast, an adverbial phrase may compositionally supply perfectivity, em-

phasising completion, i.e. the [+telic] result, and locating RT in the coda as in
(179a), which presents an ordered timeline (179b) consisting of building process
(nucleus) followed in order by completion (coda), earthquake and utterance (ST):

(179) a. ‫بنى بيته قبل الزلزال‬

banā bayt-a-hu qabla z-zilzāl
build;pst;3msg house-acc-poss.3msg before def-earthquake
‘He built his house before the earthquake’
b. RT ST
↓ ↓
ET: 0-------------------------------|------------->
NUCLEUS CODA ↑
X: ‘the earthquake’

Of course, specific [+perfective] verbal marking may also be employed to place

the focus not upon the process of building but on the result (180a):
(180) a. [+perfective]
‫قد بنى فيها البريطانيون إبان االستعمار منازل لمديري الشركات الكبيرة‬
qad banā fī-hā l-barīTānīy-ūn
pfv build;pst;3msg in-obj.3fsg def-British-pl.nom
’ibāna l-’isti‫ع‬mār manāzil
during def-colonial_rule home;pl
li-mudīr-ī sh-sharik-āt-i l-kabīr-a
for-director-pl.gen def-company-pl-gen def-big-f
‘During the colonial rule, the British had built homes there
for the directors of big companies’ (arabiCorpus: NEW1996: 32749)
b. RT (ST)
↓ ↓
ET: 0--------------------|---------------->
NUCLEUS CODA
X: |------------------------------------------------------|
‘colonial rule’

Thus in (180a–b) the emphasis is on the outcome that homes were completed at
some point prior to ST, since [+perfective] marking ensures that RT intersects
with ET at the coda. The temporally extensive adverbial ‘during the colonial rule’,
in combination with [+perfective] grammatical aspect and [+telic] lexical aspect,
places the entire event within an extended and externally defined period of time
X, ensuring the interpretation that the situation commenced and ran to comple-
tion during that period, i.e. that nucleus and coda are both located within it. Thus
178 The Arabic Verb

at some otherwise undefined time RT during the external situation (‘colonial

rule’), the nucleus (the process of building) has been completed and the coda
holds and we can confidently assert both (180c) and (180d):
(180) c. [+imperfective] [+past]
‫كان البريطانيون يبنيون منازل إبان االستعمار‬
kān-a l-barīTānīy-ūn yabniy-ū-n
be;pst-3msg def-British-pl.nom build;npst.3m-pl-ind
manāzil ’ibāna l-’isti‫ع‬mār
home;pl during def-colonial_rule
‘The British were building homes during the colonial rule’
d. ‫انـتـهـى البريطانيون من بناء منازل إبان االستعمار‬
intahā l-barīTānīy-ūn min binā’
finish;pst.3msg def-British-pl.nom from building[vn:i]
manāzil ’ibāna l-’isti‫ع‬mār
home;pl during def-colonial_rule
‘The British finished building homes during the colonial rule’

In my translation of (180a), the choice of the past perfect illustrates that pragmat-
ic interpretation of the Arabic perfective construction with ‫( قد‬qad ) is required in
order to place RT with respect to ST, due to the lack of marking for tense [Øpast].
I could instead have translated with the English present perfect (ST coincident
with RT), though in this example extra-linguistic knowledge is brought to bear,
namely that British colonial rule in Aden had ended when the text was written,
hence ST is at a time following the end of the situation in X, and since RT is con-
tained within X, it must also be anterior with respect to ST. Because of this anteri-
ority the coda has no explicit relevance at ST (the homes could have subsequently
been demolished), only at RT and implicitly for an undefined time beyond it.
Before proceeding, we should re-examine the event diagram presented in
(176d), in which an ACCOMPLISHMENT verb marked [+imperfective] for
grammatical aspect is seen to direct RT to reference the nucleus phase. Let us
use a fresh Example (181a) in order not to prejudice the analysis with the kind of
extra-linguistic knowledge which was employed in interpreting (180a–d):
(181) a. [+imperfective] [+past]
‫كان [مجلس اإلعمار] يبني منازل ومساكن‬
kān-a [majlis-u l-’i‫ع‬mār]
be;pst-3msg council-nom def-development[vn:iv]
yabnī manāzil
build;npst.3msg;ind home;pl
‘[The development council] was building homes’
(arabiCorpus: BUS1996: 36787)
 Chapter 8.  An aspectual model for Modern Standard Arabic 179

A better representation of the event diagram for this situation is given here:
(181) b. RT ST
↓ ↓
ET: 0--------------------> |
NUCLEUS CODA

The imperfective intersects with ET during the nucleus, marked as [+durative]

[+dynamic], representing the situation as a process, comparable with ACTIVITY
verbs which have no coda. Effectively the imperfective viewpoint has no access to
the [+telic] coda, which is posterior (future) relative to RT, beyond the implication
that this is the logical endpoint if the process is not interrupted before completion.
Thus the situation is presented as incomplete at RT: it is still in process, and may or
may not reach its logical endpoint in coda – there is simply no way of telling in the
absence of evidence external to the utterance, whether linguistic or extra-linguis-
tic. In view of the indefinite plural direct object, pragmatic interpretation allows
for three possible outcomes subsequent to RT in (181a): (a) none of the homes
were completed; (b) all of the homes were completed; (c) only some of the homes
were completed. If (a) is the actual outcome, the statement in (181c) is false.
(181) c. [+perfective]
‫قد بنى [مجلس اإلعمار] منازل ومساكن‬
qad banā [majlis-u l-’i‫ع‬mār]
pfv build;pst;3msg council-nom def-development[vn:iv]
manāzil
home;pl
‘[The development council] has built homes’

Hence, we can deduce that for an ACCOMPLISHMENT the imperfective (181a)

does not entail the perfective (181c). This lack of entailment is characteristic of
telicity.

8.4.2.4  Achievements
An ACHIEVEMENT verb is marked as follows:

Nucleus: [+dynamic] Coda: [+telic]

The following verb in pattern VIII will be used as an example:

(182) ‫انتصر على‬

[i]ntaSara (‫ع‬alā)
‘to win’ / ‘to triumph over’ (+ i.o.)
180 The Arabic Verb

Firstly, it will be helpful to demonstrate that this verb can be used in the imperfec-
tive past:
(183) a. [+imperfective] [+past]
‫الرئيس … كان ينتصر… في معاركة غير متكافئة‬
ar-ra’īs kān-a yantaSir-u fī
def-president be;pst-3msg win;npst.3msg-ind in
mu‫ع‬āraka ghayr mutakāfi’-a
fight other_than equal-f
‘The President … was winning … in an unequal fight’
(arabiCorpus: GEN1996: 12617)
b. RT ST
↓ ↓
ET: 0--------------------> |
NUCLEUS CODA

The event time (183b) has a structure identical to (181b), in which the verb was
marked as [+durative], but the fact that an ACHIEVEMENT verb is unmarked
for this privative feature does not prevent a durative reading, hence intersection
of RT with a durative nucleus is indicated, preceding an expected endpoint in
coda which may or may not be reached. In (184a) however, with the verb formally
specified only as [+past], since the [Ødurative] lexical aspect status of the verb
permits non-durative or punctual interpretation, RT intersects with a nucleus
which appears to have no extension in time. In other words, the dynamic transi-
tion to coda is presented as instantaneous.
(184) a. [+past]
‫انتصر المجاهدون المسلمون على القوات الصينية‬
intaSar-a l-mujāhid-ūna
triumph;pst-3msg def-warrior-mpl.nom
l-muslim-ūn ‫ع‬alā l-quww-āt-i S-Sīnīy-a
def-Muslim-mpl.nom over def-force-pl-gen def-Chinese-f
‘The Muslim warriors triumphed over the Chinese forces’
(arabiCorpus: GEN1997: 5202)
b. RT ST
↓ ↓
ET: ||----------------------------------->
NUCLEUS CODA
 Chapter 8.  An aspectual model for Modern Standard Arabic 181

This instantaneous understanding of the dynamic phase can be further demon-

strated by adding a punctual adverbial phrase (184c), where RT is now anchored
to an external time:
(184) c. [+past]
‫انتصر المجاهدون المسلمون على القوات الصينية في الساعة الثانية‬
intaSar-a l-mujāhid-ūna
triumph;pst-3msg def-warrior-mpl.nom
l-muslim-ūn ‫ع‬alā l-quww-āt-i
def-Muslim-mpl.nom over def-force-pl-gen
S-Sīnīy-a fī s-sā‫ع‬a ath-thāniy-a
def-Chinese-f in def-hour def-second-f
‘The Muslim warriors triumphed over the Chinese forces at two o’clock’

Pragmatically, we recognise that such a triumph is unlikely to have been instanta-

neous: there must have existed a durative situation prior to the coda, though this
pre-existing situation is not necessarily that described by the event ‘triumph’, only
a situation implied by it. This is shown more clearly by changing the adverbial
phrase (184d–e):
(184) d. [+past]
‫انتصر المجاهدون المسلمون على القوات الصينية في خالل أسبوعين‬
intaSar-a l-mujāhid-ūna
triumph;pst-3msg def-warrior-mpl.nom
l-muslim-ūn ‫ع‬alā l-quww-āt-i
def-Muslim-mpl.nom over def-force-pl-gen
S-Sīnīy-a fī khilāl ’usbū‫ع‬-ayn
def-Chinese-f in interval week-du.gen
‘The Muslim warriors triumphed over the Chinese forces in[side] two
weeks’
e. RT ST
↓ ↓
ET: ||------------------------------->
NUCLEUS CODA
X: |--------------------------|
‘two weeks [of fighting]’

A period of fighting lasting two weeks is now implied (the pre-existing situa-
tion, represented here as X, an event external to the ET), which culminates in the
triumph of the Muslim warriors (the situation pertinent to the ET described).
However, (184e) does not entail (184f):
182 The Arabic Verb

(184) f. [+imperfective] [+past]

‫كان ينتصر المجاهدون المسلمون في ك ّل وقت في خالل األسبوعين‬
kān-a yantaSir-u
be;pst-3msg triumph;npst.3msg-ind
l-mujāhidūn al-muslim-ūn
def-warrior-mpl.nom def-Muslim-mpl.nom
fī kull waqt fī khilāl ’usbū‫ع‬-ayn
in every time in interval week-du.gen
‘The Muslim warriors were triumphing at all times during the two weeks’

Thus, the verb in (184a) is most naturally interpreted as representing a punctual

event, entirely compatible with its [Ødurative] lexical aspect. If the verb had the
[+durative] feature, a punctual interpretation would be impossible. Hence, in
Vendler’s scheme, it is an ACHIEVEMENT rather than an ACCOMPLISHMENT.
Let us now examine the same verb but marked for perfectivity (185a):
(185) a. [+perfective]
‫الجزائر قد انتصرت على اإلرهاب‬
al-jazā’ir qad intaSar-at ‫ع‬alā l-’irhāb
def-Algeria pfv triumph;pst-3fsg over def-terrorism[vn:iv]
‘Algeria has triumphed over terrorism’ (arabiCorpus: GEN1997: 2975)
b. RT/ST
↓
ET: ||-------------------------------->
NUCLEUS CODA
The [+perfective] aspect ensures that RT intersects with ET at the coda, placing
the emphasis on the resulting endpoint rather than the transition to the endpoint,
in this example focusing not upon the triumph itself but upon its continuing rel-
evance at RT. It is irrelevant in this example whether the nucleus is durative or
punctual, which is to be expected since presence or absence of [+durative] is a
nucleus feature with no bearing upon the coda. It was noted in (181a−c) that for
a [+durative] ACCOMPLISHMENT, the imperfective does not entail the perfec-
tive, since the coda, although the natural endpoint, may never actually be reached.
Likewise, (185c) does not entail (185a):
(185) c. [+imperfective] [+past]
‫كانت الجزائر تنتصر على اإلرهاب‬
kān-at al-jazā’ir tantaSir-u
be;pst-3fsg def-Algeria triumph;npst.3fsg-ind
‫ع‬alā l-’irhāb
over def-terrorism[vn:iv]
‘Algeria was triumphing over terrorism’
 Chapter 8.  An aspectual model for Modern Standard Arabic 183

In contrast with an ACCOMPLISHMENT, however, since a [Ødurative] verb is

uncertain as to durativity, it is also true that for an ACHIEVEMENT the perfec-
tive (185a) does not entail the imperfective (185c). To clarify, consider an Olym-
pic track race: the athlete who takes the gold medal may have been ahead of the
field for some time before reaching the finishing line, in which case the athlete
would have been said to have been winning the race approaching the line; how-
ever, if the gold medal winner comes from behind to take the race on the line in a
photo-finish, the act of winning is a punctual event and it would not be true to say
that the athlete had at any point been winning prior to the actual victory.

8.4.2.5  Semelfactives76
A SEMELFACTIVE verb is marked as follows:

Nucleus: [+dynamic] Coda: Ø

The following pattern I verbs are typical of SEMELFACTIVES:

(186) ‫عطس‬
‫ع‬aTasa
‘to sneeze’
(187) ‫سعل‬
sa‫ع‬ala
‘to cough’

Semelfactives, being [Ødurative], are most naturally interpreted as punc-

tual, i.e. the nucleus is presented as having no duration (188a):
(188) a. [+past]
‫عطس الرجال دفعة واحدة عطسة‬
‫ع‬aTas-a r-rijāl duf‫ع‬at-an
sneeze;pst-3msg def-man;pl instance-acc.indf
wāhid-a ‫ع‬aTsa
one-f sneeze
‘The men sneezed all at once’ (arabiCorpus: FisherCastle12: 20)
b. RT ST
↓ ↓
ET: ||
NUCLEUS

76. This category is absent from Mughazy’s (2005) account of Olsen’s scheme and he appar-
ently discounts its existence in Egyptian Arabic.
184 The Arabic Verb

As for the ACHIEVEMENT verb in (184a−c), it is clear that a sneeze is not lit-
erally instantaneous. Nevertheless, it is of sufficiently short duration that when
combined with a temporally extensive adverbial phrase, the meaning is most
naturally pragmatically determined as iterative rather than durative, i.e. multiple
sneezing events (188c):
(188) c. [+past]
‫عطس الرجل في خالل دقيـقـتـيـن‬
‫ع‬aTas-a r-rajul fī khilāl daqīqat-ayn
sneeze;pst-3msg def-man in interval minute-du.gen
‘The man sneezed for two minutes’
d. RT ST
↓ ↓
ET: || || || || || ||
nucleus
X: |--------------|
’2 minutes’

However, it may be possible to imagine an alternative reading such as (188e),

given a suitable causing event in the context, suggesting an event diagram (188f)
which is essentially the same as (188b).
(188) e. ‘The man sneezed within two minutes’
f. RT ST
↓ ↓
ET: ||
NUCLEUS
X: |-------------|
↑’2 minutes’
causing event

Note that the English adverbial phrases in the translations do not permit such am-
biguity. In contrast, use of the imperfective past (189a) with a SEMELFACTIVE
forces a non-punctual, hence iterative (or possibly habitual), interpretation of the
nucleus, with which RT must intersect. Thus, the event diagram (189b) is equiva-
lent to (188d), except that the duration of the iterative situation is not bounded by
an external adverbial phrase.
(189) a. [+imperfective] [+past]
‫ذات يوم كان يسعل بعنف‬
dhāt_yawm kān-a yas‫ع‬ul-u bi-‫ع‬anf
one_day be;pst-3msg cough;npst.3msg-ind with-violence
‘One day, he was coughing violently’ (arabiCorpus: 061999WRIT03)
 Chapter 8.  An aspectual model for Modern Standard Arabic 185

b. RT ST
↓ ↓
ET: < || || || || || || >
NUCLEUS

It may be significant that no examples were found in the corpus for either of these
semelfactive verbs with the perfective ‫( قد‬qad ) construction. There is no
[+telic] feature for the perfective to intersect with in coda and it is therefore likely
that it is rendered superfluous. However, if we form a sentence using the con-
struction, we can use this diagnostically.
(190) [+perfective]
‫قد سعل بعنف‬
qad sa‫ع‬al-a bi-‫ع‬anf
pfv cough;pst-3msg with-violence
‘He [has] coughed violently’

It is clear that the imperfective (189a) does entail the perfective (190), which
is confirmation that the verb is not marked as [+telic], cf. activities. Con-
versely, to assert that (192) is true on the basis of (191) is doubtful at best. This
is because semelfactive perfectives do not entail the corresponding imper-
fective, due to the punctual readings permitted by their [Ødurative] status, cf.
achievements.
(191) [+past]
ً‫سعل الرجل مرة‬
sa‫ع‬al-a r-rajul marrat-an
cough;pst-3msg def-man time-acc.indf
‘The man coughed once’
(192) [+imperfective] [+past]
‫كان الرجل يسعل‬
kān-a r-rajul yas‫ع‬ul-u
be;pst-3msg def-man cough;npst.3msg-ind
‘The man was coughing’

8.4.2.6  Stage-level states

Olsen (1997: 48–50) appears less than convinced that the category of STAGE-
LEVEL STATE exists, although the combination is allowable in her scheme,
which requires only that “[a] VERB must be minimally [+dynamic] or [+dura-
tive]” (Olsen 1997: 51), hence the feature marking as follows:
Nucleus: [+durative] Coda: [+telic]
186 The Arabic Verb

Passonneau’s (1988: 47) analysis distinguishes “transition events”, conflating

Vendler’s ACHIEVEMENTS and ACCOMPLISHMENTS, in which a “new state
or process comes into being as a result of the initial process”. Processes require
dynamicity and states are therefore excluded from transition events as they “have
no kinesis, they cannot culminate in new situations”. The argument that without
change there can be no inherent endpoint appears thoroughly sound, therefore
it is my contention that [+telic] situations must necessarily be [+dynamic] and
hence states marked only as [+durative] in the nucleus phase are excluded from
being marked [+telic] in the coda.
It is clear, however, that some states are more temporary than others. Com-
pare the following English examples:
(193) a. Emma has been pregnant for three months.
b. *Emma has always been pregnant.
(194) a. Craig has been sick for three months.
b. Craig has always been sick.
(195) a. Water supplies have sufficed for three months.
b. Water supplies have always sufficed.
(196) a. ?Daniel has been intelligent for three months.
b. Daniel has always been intelligent.

According to Olsen (1997: 48–49), the verbal expressions ‘be pregnant’, ‘be sick’
and ‘suffice’ have been variously proposed as examples of STAGE-LEVEL STATES
in English, hence marked as [+telic]. However, as the following examples demon-
strate, it is precisely the verb categories marked [+telic] which are incompatible
with the English perfective have + V-ed construction together with a temporally
extensive adverbial:
(197) *Andy has built himself a house for three years.  (ACCOMPLISHMENT)
(198) *Hannah has won the race for twenty minutes. (ACHIEVEMENT)
(199) Alastair has lived in Scotland for three years. (STATE)
(200) Louise has cycled for twenty minutes. (ACTIVITY)
(201) Henry has coughed for twenty minutes. (SEMELFACTIVE)

Since the [+telic] feature is present in (197)–(198), the intersection of RT with

ET at the coda specified by the [+perfective] verb requires the interpretation that
there was a point in time prior to RT at which the nucleus phase was completed,
thus a temporally extensive adverbial can only refer back to a nucleus which is
not being referenced by the perfective. The problem does not exist in (199)–(201),
as there is no coda feature to be referred to, hence RT intersects with ET at the
 Chapter 8.  An aspectual model for Modern Standard Arabic 187

nucleus, to which the durative adverbial applies, forcing an iterative interpreta-

tion in the case of (201). This is also true of (193)–(195), providing clear evidence
that these are STATES like any other. Somewhat paradoxically, although uncon-
troversially recognised as a STATE, it is ‘be intelligent’ (196) for which the use of
a temporally extensive adverbial is questionable. This leads me to the conclusion
that whether any of the above states is more or less temporary is determined prag-
matically and not by any lexical feature such as telicity.
No Arabic examples for STAGE-LEVEL STATES will be proposed,77 as I have
encountered no data to support the existence of verbs in this category and I am in
any case convinced by Passonneau’s argument on the basis of dynamicity that the
category itself has no place in Olsen’s scheme. Olsen does, however, observe that
Smith’s (1991: 110) data from Mandarin show “that “certain stage-level stative
predicates” may have ingressive interpretations, implicating a dynamic entrance
into the state named by the verb” (Olsen 1997: 50). This raises an interesting point,
which Olsen does not attempt to pursue here, though she later returns to the
subject of ingressives in her chapter on grammatical aspect. Her conclusion is
that “ingressive interpretation (also known as inceptive or inchoative) seems to
be based on properties orthogonal to [+imperfective] and [+perfective] aspect”
(Olsen­1997: 108). However in Chapter 10, using examples from Arabic and Eng-
lish, I will argue that a separate category of verbs with ingressive or inceptive
interpretation must be established, as a further extension to Vendler’s system as
explained and expanded by Olsen, and that Olsen’s scheme is capable of further
expansion to accommodate this new category.

8.5 Summary

We have briefly examined aspectual theory in general and established that MSA
exhibits realisations of both morphosyntactic grammatical aspect and lexical
­aspect. Vendler’s (1967) verbal categories have been shown to be applicable to
­Arabic, through the application of Olsen’s (1997) scheme, which extends Vendler’s
categories and provides a consistent and logical framework within which to ex-
amine lexical aspect. With this theoretical model in place we may now examine in
more detail the lexical aspect properties of verbs in patterns III and VI, i.e. those
with the C1āC2 template sequence.

77. Mughazy (2005: 138) claims the existence of stage-level statives in Egyptian Arabic, also
calling them “recurrent statives” (2005: 151). However, I am not convinced that the examples
he gives are [+telic].
 chapter 9

Aspectual categorisation of patterns III and VI

Having established that lexical aspect is a valid property in the verbal system of
MSA and is manifested in categories of the type proposed by Vendler (1967) and
explained in Olsen’s (1997) model, we will need to examine whether there is any
aspectual consistency within patterns III and VI to establish whether the C1āC2
sequence is indeed an aspectual marker, either for a verbal category or for one or
more privative features as defined by Olsen. Dictionary and corpus data will be
analysed, leading to a preliminary hypothesis that the sequence marks atelicity.
Many apparent exceptions will be eliminated, though it will also be demonstrated
that there is a significant number of verbs which share inceptive semantic proper-
ties and need closer inspection in the following chapter.

9.1 Data

A cursory examination of the verbs in patterns III and VI reveals that ACTIVITY
verbs are very common, as exemplified in (202) and (203):

(202) ‫[ كان أغلبهم يشاهد التلفزيون‬III]

kān-a ’aghlab-u-hum
be;pst-3msg majority-nom-poss.3mpl
yushāhid-u t-tilifizyūn
watch;npst.3msg-ind def-television
‘Most of them were watching television’ (arabiCorpus: 031399WRIT01)
(203) ‫[ ما زال أعضاء مجلس األمن يتحاورون‬VI]
mā zāl-a ’a‫ع‬Dā’ majlis-i
neg cease;pst-3msg member;pl council-gen
l-’amn yataHāwar-ū-n
def-security discuss_together;npst.3m-pl-ind
‘The members of the security council are still carrying on a discussion’
 (arabiCorpus: 110999OPIN07)

However, there are significant numbers of verbs which appear more properly to
be classified as STATES, such as the verb in (204), which we used as an example
of a STATE in the previous chapter:
190 The Arabic Verb

(204) ‫جـاور‬
jāwara (+ d.o.)
‘to adjoin (s.th.) / be the neighbour of (s.o.)’

Since the imperfective past in Arabic, like the French imparfait, cannot make the
distinction between the progressive and the habitual (unlike the English con-
structions was + V-ing and used to + V), dynamicity cannot be tested on this
basis. We may therefore justifiably argue that the ACTIVITY/STATE distinction
in MSA is relatively subtle, as any formal testing of this distinction must rely upon
assessing compositional compatibility of each verb with certain adverbial phrases.

9.1.1 Verbs of ACTIVITY and STATE

In gathering preliminary lexical aspect data on patterns III and VI from Hans
Wehr (1994), I have elected to conflate the ACTIVITY and STATE categories on
the basis of their shared privative feature marking, i.e. that they are marked [+du-
rative] and unmarked for telicity [Øtelic]. In order to improve the quality of the
data, I have split entries for root consonant combinations where it appears that
homomorphic roots give rise to verbs with unrelated meanings, allowing for as-
pectual variation between homonymous verbs with different root meanings to be
recorded. This explains why the total numbers of verbs attributed to patterns III
and VI in Tables 61 and 62 are somewhat greater than the raw verb counts quoted
in earlier chapters.

Table 61.  Pattern III verbs by aspectual category and co-occurrence

+ [pattern VI] − [pattern VI] TOTAL
III = ACTIVITY/STATE 228 (83.2%) 192 (81.0%) 420 (82.2%)
III = OTHER   46 (16.8%)   45 (19.0%)   91 (17.8%)
TOTAL 274 237 511
Chi-square (χ2) = 0.42 | Prob. (p) > 0.5 (not significant) | Phi coeff. (Φ) = 0.03

Table 62.  Pattern VI verbs by aspectual category and co-occurrence

+ [pattern III] − [pattern III] TOTAL
VI = ACTIVITY/STATE 236 (84.6%) 137 (84.0%) 373 (84.4%)
VI = OTHER   43 (15.4%)   26 (16.0%)   69 (15.6%)
TOTAL 279 163 442
Chi-square (χ2) = 0.02 | Prob. (p) > 0.8 (not significant) | Phi coeff. (Φ) = 0.01
 Chapter 9.  Aspectual categorisation of patterns III and VI 191

As previously, the data here are presented according to whether or not patterns III
and VI co-occur for the root, and testing for statistical significance has been ap-
plied. The figures reveal remarkable consistency across the two patterns, with the
percentages of verbs of ACTIVITY/STATE clearly independent of whether or
not the patterns co-occur and averaging 83.2%. That aspect is independent of
co-occurrence contrasts with the findings for mutual and reciprocal meanings
presented in Section 5.2.2.3 and provides a clear indication that lexical aspect, at
least for the patterns examined here, is a direct function of derivational morphol-
ogy. It will also be noted that aspectual categorisation as ACTIVITY/STATE is a
more successful description than mutuality is of pattern III (77.6%)78 or reciproc-
ity of pattern VI (64.0%),79 all the more so because the data collected for seman-
tic categories were treated conservatively, as discussed in Section 5.2.2.1, favour-
ing the traditional hypothesis of mutual-reciprocal meanings for these patterns.
However, although the preliminary findings for categorising verbs with the C1āC2
sequence as ACTIVITY/STATE are promising, 16.8% of the verbs in Tables 61
and 62 appear to not be readily attributable to this combined aspectual category.
One question which arises is whether the percentages of verbs obtained from
a dictionary survey accurately reflect the extent to which the ACTIVITY/STATE
lexical aspect is associated with patterns III and VI in contemporary language use.
This is pertinent to the relationship between form and meaning from a psycho-
linguistic viewpoint, since regular usage will reinforce a speaker’s perception of
a meaning being associated with the corresponding form. Thus instead of solely
relying on a lexical survey, where each verb listed has equal weight in the data no
matter whether it is in contemporary use, a corpus survey of all pattern III and
pattern VI verbs was undertaken using the online arabiCorpus in order to obtain
data more representative of actual MSA usage.

9.1.1.1  Corpus data counts – Methodology

By searching arabiCorpus for pattern III and pattern VI verb forms, two different
figures may be obtained: a ‘type’ count records the different verb forms only once
each, no matter how frequently they occur in the corpus, while a ‘token’ count
enumerates every instance of a form in the corpus.80 Thus a type count includes
all verbs which are in common usage and gives them equal weighting, while

78. See Section 5.2.2.1

79. See Section 5.2.2.2
80. For an early use of this distinction, see Thomson & Thompson (1915). For the terminology,
see McEnery & Wilson (2001: 82).
192 The Arabic Verb

e­ xcluding any verb which is too archaic or otherwise uncommon to be found in

the corpus. A token count goes further in that the contribution of any given verb
towards the totals, and hence the percentages, is proportional to how commonly
it is used. For example, no instances of (205) were found in arabiCorpus and it
therefore does not contribute to either the type count or the token count. How-
ever both (206) and (207) do occur in the corpus, though the latter is 12.5 times
more common than the former, thus both contribute one type to the count while
the latter contributes 12.5 times as many tokens.

(205) ‫ شـابـه‬ [III]

shābaha (+ d.o.) (token count: 0)
‘to resemble (s.o./s.th.)’
(206) ‫ شـاكـس‬ [III]
shākasa (+.d.o.) (token count: 24)
‘to quarrel with (s.o.)’
(207) ‫ تـسـابـق‬ [VI]
tasābaqa (token count: 300)
‘to compete with one another’

Clarification is required of the methodology used to obtain the counts quoted in

the above examples and contributing to the data presented in the following sec-
tion. In order to obtain meaningful comparable data from arabiCorpus, I elected
to interrogate the database specifically for the p-stem 3rd person masculine sin-
gular form of each verb, since it was necessary to identify a string to search for
which was unique to the verb in question. Thus for example, had I searched for
the pattern III verb kātaba (’to correspond [with s.o.]’) using the string kātb
(‫)كـاتـب‬, the search results would not only have returned all s-stem and p-stem
instances of the verb in question, but would also have yielded all instances of
the pattern VI verb takātaba (‫‘ – تـكـاتـب‬to correspond [together]’) as well as the
nouns kātib (‫‘ – كـاتـب‬writing/writer’), mukātaba (‫‘ – مـكـاتـبـة‬correspondence’) and
the broken plural makātib (‫‘ – مـكـاتـب‬libraries/offices’). However by searching for
the string ykātb (‫ )يـكـاتـب‬only pattern III verb instances are returned, including
all p-stem 3rd person masculine paradigm forms with and without suffixed ob-
ject pronouns. This is not a perfect solution as there are some anomalies, such as
verbs which may pragmatically only occur with feminine subjects81 or most of-
ten with plural or dual inanimate subjects (and hence in the 3rd person feminine
singular paradigm form). Also, the string search specified will under-represent
pattern VI verbs with final radical ‫( ي‬yā) where there is an orthographic change

81. For example ‫( حـبلت‬Habilat – ‘she became pregnant’).

 Chapter 9.  Aspectual categorisation of patterns III and VI 193

when an object pronoun is suffixed, though these are sufficiently rare as to not
affect the data significantly. Furthermore it is conceivable that some verbs are
over- or under-represented in the p-form compared with the s-form, though I
have no data to support such an objection.
In the relatively rare cases where homonymous verbs exist with different aspec-
tual category properties, I have attempted to apportion the token count according
to the meanings apparent from the corpus context. To do so with complete accuracy
would be a prohibitively time consuming task and thus where instances are numer-
ous I have made judgements based on examining a selection of examples.
Finally, some notes on arabiCorpus itself. The data were obtained using the
‘newspaper’ corpus subset82 of arabiCorpus, interrogated online over a period83
during which no additions to the corpus were made. The newspaper corpus was
chosen as relatively homogeneous, representative of contemporary usage and less
susceptible to (though clearly not entirely free from) archaism, poeticism and
colloquialism than literature. By restricting my searches to uniquely specified
strings, and with the minor provisos outlined above, I believe that the data gener-
ated and presented in the following section are comparable across the entire set of
pattern III and pattern VI verbs.

9.1.1.2  Results
Tables 63 and 64 present the counts and percentages for patterns III and VI re-
spectively by type and token. The data reveal that although the range of pattern VI
verbs in contemporary usage is comparable to that of pattern III, they are en-
countered only one-third as often in the corpus texts. Nevertheless there is still
remarkable consistency between the percentages of verbs in each pattern which
are attributable to the ACTIVITY/STATE category. The percentages for the type
counts are very close to those obtained from the dictionary search, demonstrat-
ing that there is no preferential absence of verbs from the corpus according to the
ACTIVITY/STATE lexical aspect category.

Table 63.  Type and token counts for pattern III verbs by aspectual category
Type Token
ACTIVITY/STATE 291 (82.2%) 125,314 (91.0%)
OTHER   63 (17.8%)   12,425 (9.0%)
TOTAL 354 137,739

82. At the time of writing (18 June 2009) approximately 65 million words.
83. February 2008.
194 The Arabic Verb

Table 64.  Type and token counts for pattern VI verbs by aspectual category
Type Token
ACTIVITY/STATE 254 (82.7%) 43,723 (92.4%)
OTHER   53 (17.3%)   3,578 (7.6%)
TOTAL 307 47,301

However, the token counts reveal that much higher percentages of actual in-
stances of pattern III and pattern VI verbs are of those in the ACTIVITY/STATE
category, therefore the association of this lexical aspect meaning with the formal
C1āC2 sequence is enhanced. Thus by examining corpus data we have established
that over 91% of actual pattern III and pattern VI usage conforms to [+durative]
and [Øtelic] aspectual feature marking.

9.1.2 Other categories represented

Whilst the number of exceptions to the ACTIVITY/STATE categorisation seems

large at 160 and consequently the task of examining them in more detail appears
daunting, many of these apparently exceptional verbs are attributable to groups
whose members share a basic meaning and/or aspectual properties.

9.1.2.1  Verbs of surprise

Three of the more common verbs in this category are given in (208) to (210):

(208) ‫ فـاجأ‬ [III]

fāja’a (+ d.o.)
‘to come suddenly upon (s.o.) / to take (s.o.) by surprise’
(209) ‫ صادف‬ [III]
Sādafa (+ d.o.)
‘to meet (s.o.) unexpectedly’
(210) ‫ تـكـالـب على‬ [VI]
takālaba ‫ع‬alā (+ i.o.)
‘to fall, pounce upon / assail (s.o.)’

This set of verbs, having a shared meaning involving sudden action with explicit
or implicit surprise, represents 20 exceptions in total (Table 65), accounting for
approximately 2% of all verbs in the lexicon with the C1āC2 sequence, though
representing just over 1% of corpus tokens.
 Chapter 9.  Aspectual categorisation of patterns III and VI 195

Table 65.  Frequency of verbs of surprise by pattern

III = ‘SURPRISE’ VI = ‘SURPRISE’ TOTAL
dictionary count ,   12 (2.3%)    8 (1.6%)    ,20 (2.1%)
corpus type count ,    8 (2.3%)    7 (2.2%)    ,15 (2.3%)
corpus token count 1,784 (1.3%) 339 (0.7%) 2,123 (1.1%)

How then should this small but significant group of verbs be categorised aspectu-
ally? We will take as our example the verb in (208) and test it firstly for durativity
using examples assessed by native speakers. Example (211a) is uncontroversial, all
three native speaker informants finding it acceptable:

(211) a. ‫ فاجأ أحمد صديقه‬ {YYY}

fāja’-a ’aHmad Sadīq-a-hu
surprise;pst-3msg Ahmad friend-acc-poss.3msg
‘Ahmad surprised his friend’

However, when a temporally extensive adverbial phrase was appended (211b),

only one of the three informants judged it good Arabic, one marking it as margin-
ally acceptable and the other as unacceptable.

(211) b. ‫ فاجأ أحمد صديقه طوال عشرين دقيقة‬ {Y?N}

fāja’-a ’aHmad Sadīq-a-hu
surprise;pst-3msg Ahmad friend-acc-poss.3msg
Tiwāla ‫ع‬ashr-īn daqīqa
during twenty-gen minute
? ‘Ahmad surprised his friend for twenty minutes’  (Questionnaire: 12)

It seems likely that any possible interpretation with a temporally extensive ad-
verbial is iterative rather than durative, as with the perfectly acceptable explicitly
iterative adverbial in (211c):

(211) c. ‫ فاجأ أحمد صديقه مرات كثيرة‬ {YYY}

fāja’-a ’aHmad Sadīq-a-hu
surprise;pst-3msg Ahmad friend-acc-poss.3msg
marr-āt kathir-a
occasion-pl many-f
‘Ahmad surprised his friend on many occasions’ (Questionnaire: 8)

Further evidence is provided in (211d), where the [+past] [+imperfective] verb

which might otherwise have been interpreted as habitual was judged as incom-
patible or only marginally compatible with the temporally extensive adverbial.
196 The Arabic Verb

(211) d. ‫ كان أحمد يفاجئ صديقه طوال عشرين دقيقة‬ {??N}

kān-a yufāji’-u ’aHmad
be;pst-3msg surprise;npst.3msg-ind Ahmad
Sadīq-a-hu Tiwāla ‫ع‬ashr-īn daqīqa
friend-acc-poss.3msg during twenty-gen minute
? ‘Ahmad was surprising / used to surprise his friend for twenty
minutes’  (Questionnaire: 32)

Finally, compatibility with a punctual adverbial confirms the non-durative nature

of this verb:

(211) e. ‫فاجأ أحمد صديقه في الساعة الثانية‬

fāja’-a ’aHmad Sadīq-a-hu
surprise;pst-3msg Ahmad friend-acc-poss.3msg
fī s-sā‫ع‬a ath-thāniy-a
in def-hour def-second-f
‘Ahmad surprised his friend at two o’clock’

We may therefore conclude, since this verb resists durative interpretation, that
its nucleus is unmarked for durativity, i.e. [Ødurative]. Thus it must either be
a SEMELFACTIVE or an ACHIEVEMENT, depending on whether or not it is
marked [+telic] in coda.
The test of telicity which we applied in the previous chapter was that for a
[+telic] verb the imperfective does not entail the perfective. Certainly (212a) does
entail (212b), suggesting atelicity and therefore a SEMELFACTIVE categorisa-
tion, cf. Section 8.4.2.5.
(212) a. [+imperfective] [+past]
‫كان يفاجئنا هو دائما بآرائه‬
kān-a yufāji’-u-nā huwa
be;pst-3msg surprise;npst.3msg-ind-obj.1pl sbj.3msg
dā’im-an bi-’ārā’-i-hi
lasting-acc.indf with-opinion;pl-gen-poss.3msg
‘He was always surprising us / always used to surprise us with his
opinions’ (arabiCorpus: 020799ARTS09)
b. [+perfective]
‫قد فاجأنا هو بآرائه‬
qad fāja’-a-nā huwa
pfv surprise;pst-3msg-obj.1pl sbj.3msg
bi-’ārā’-i-hi
with-opinion;pl-gen-poss.3msg
‘He [has] surprised us with his opinions’
 Chapter 9.  Aspectual categorisation of patterns III and VI 197

However, the extreme resistance of the example verb to temporally extensive

adverbials is unlike the SEMELFACTIVE examples used in the previous chap-
ter, thus leaving open the possibility that the situation described consists of
discrete, iterative telic events. It would therefore seem expedient to employ a
further test for telicity and, since this verb is transitive, testing for passivisability
is a possible solution.
In Section 7.3.1.1 the distinction was drawn between actional and statal pas-
sives and it was noted that MSA reserves the passive participle or ‘done’ form,
which implies result, for statal passives. Recall also that verbs marked [+telic]
exhibit a coda phase in which the emphasis is upon the resulting endpoint of the
action they describe. Beedham (2005: 44) states explicitly that telicity is required
for passivisability, i.e. that a verbal expression must have a built in endpoint in
order to passivise, or more rigorously to form a statal passive. Thus we shall return
to Example (211a) and examine whether it may be passivised.
The responses of native speakers indicated that they have little or no prob-
lem with the verb being used as an actional passive (213a), but they unanimously
judged the statal passive construction with the passive participle (213b) to be un-
acceptable:84

(213) a. ‫ فوجئ صديقه‬ {?YY}

fūji’-a Sadīq-u-hu
surprise;pst.pass-3msg friend-nom-poss.3msg
‘His friend was surprised’ (Questionnaire: 21)
b. *ً ‫ كان صديقه مفا َجأ‬ {NNN}
kān-a Sadīq-u-hu mufāja’-an
be;pst-3msg friend-nom-poss.3msg surprise;ppt-acc.indf
‘His friend was surprised’  (Questionnaire: 46)

This is a clear indication of atelicity. However, my attention was initially drawn

to the ‘verbs of surprise’ during my preliminary categorisation of lexical aspect
precisely because I was not convinced that these verbs are atelic. The following
Example (214) from English gives insight into the problem:
(214) As long as we remain surprised by our economic success … we may be all right.
(http://www.independent.co.uk/opinion/commentators/hamish-mcrae/
a-healthy-economy-is-a-worrying-one-567405.html, 15 July 2009)

84. For further discussion of the compatibility of the passive participle with patterns III and VI
see Chapter 11.
198 The Arabic Verb

Clearly English allows statal passive use of the second participle of the verb ‘to sur-
prise’. It is therefore most likely marked [+telic] and hence is a verb of ACHIEVE-
MENT. However, the evidence from native Arabic speakers is that in MSA the
verb ‫( فاجأ‬fāja’a) is [Øtelic] and therefore SEMELFACTIVE. In fact my initial re-
luctance to accept its atelicity is possibly an artefact of translation. For readability
I have been translating the verb as ‘to surprise’, whereas properly it should be
translated as ‘to take by surprise’, these two verbal expressions being aspectually
distinct in English. We will not examine each ‘verb of surprise’ individually, but it
is clear that example verbs (209) and (210) are more obviously SEMELFACTIVE
when translated than is Example (208).
Thus to summarise the findings of this section, the ‘verbs of surprise’, which
form a small but significant subset of C1āC2 sequence verbs, are SEMELFACTIVE
and therefore share with ACTIVITY and STATE verbs the property of being un-
marked for telicity, i.e. [Øtelic].

9.1.2.2  Verbs of giving

A small group of C1āC2 sequence verbs which are not classifiable as ACTIVITY/
STATE share a common meaning of ‘giving’. However, the data in Table 66 show
that these constitute less than 1% of the pattern III and pattern VI verbs in the lex-
icon and that there are only three such verbs, shown in Examples (215) to (217),
which are actually attested in the corpus. Moreover, they are used infrequently,
accounting for only 0.05% of total attested instances in these patterns.

Table 66.  Frequency of verbs of giving by pattern

III = ‘GIVING’ VI = ‘GIVING’ TOTAL
dictionary count   6 (1.2%) 1 (0.2%)   7 (0.7%)
corpus type count   3 (0.9%) 0 (0.0%)   3 (0.5%)
corpus token count 95 (0.07%) 0 (0.0%) 95 (0.05%)

(215) ‫قـايـض بـ‬

qāyaDa (+ d.o.) (+ d.o.) bi (+ i.o.)
‘to give (s.o.) (s.th.) in exchange for (s.th)’
(216) ‫نـاول‬
nāwala (+ d.o.) [or optionally li (+ i.o.)] (+ d.o.)
‘to give [to] (s.o.) (s.th.)’
(217) ‫وافـى بـ‬
wāfā (+ d.o.) bi (+ i.o.)
‘to deliver to (s.o.) (s.th.) / supply (s.o) with (s.th)’
 Chapter 9.  Aspectual categorisation of patterns III and VI 199

It might be expected that ‘verbs of giving’ are categorised as either ACHIEVE-

MENTS or ACCOMPLISHMENTS, since the situation described by the verb ap-
pears to involves a change of state for the object as a result of the action, namely
a change in who has it in their possession, whether this transition is punctual or
durative, i.e. we would expect [+telic] marking. However, consider the following
English examples:
(218) a. Christopher gave his books to a charity shop.
b. Christopher donated his books to a charity shop.
c. *A charity shop sells given goods.
d. A charity shop sells donated goods.

Although the verbs in (218a–b) appear synonymous, they behave differently in

the use of their second participles in (218c–d), indicative of an aspectual differ-
ence. We must therefore be careful when comparing apparently similar examples
and furthermore avoid transferring expectations from English to the Arabic ex-
amples. In fact although the act of giving is most readily conceptualised as the
transfer of possession of a physical object from giver to receiver, where the (gram-
matical) object transferred is abstract it does not undergo the same change of
state. Thus in (219), the receiver is now in possession of the transferred abstract
object, although the giver has not actually parted with it.
(219) The doctor gave Ewen advice about his condition.

The following example from Arabic (220a) suggests the time line in (220b); fur-
thermore, since the indirect object (’his criticisms’) is unbounded, for any point
RT within the nucleus phase, it would be true to assert (220c) according to the
time line in (220d):
(220) a. [+imperfective] [+past]
‫كان يوافيه بنقداته لنظرات المنفلوطي‬
kān-a yuwāfī-hi
be;pst-3msg supply;npst.3msg-obj.3msg
bi-naqd-āt-i-hi li-naDHar-āt-i l-manfalūTī
with-criticism-pl-gen-poss.3msg of-view-pl-gen def-Manfaluti
‘He was giving him [supplying him with] his criticisms of Manfaluti’s
views’ (arabiCorpus: GEN1996: 5209)
b. RT ST
↓ ↓
ET: 0----------------------------------------->
nucleus
200 The Arabic Verb

c. [+perfective]
‫قد وافاه بنقداته‬
qad wāfā-hu bi-naqd-āt-i-hi
pfv supply;pst;3msg-obj.3msg with-criticism-pl-gen-poss.3msg
‘He has given him [supplied him with] his criticisms’
d. RT/ST
↓
ET: 0----------------------------------------------->
nucleus

This behaviour is typical of an ACTIVITY and thus, as the verb is capable of

atelic interpretation, we must attribute to it null marking for telicity, i.e. [Øtelic].
The following examples from each of the other two verbs in this group which
are actually attested in the corpus make it clear that they too are capable of atelic
interpretation, though the scarcity of examples clearly marked for grammatical
aspect makes further investigation from the corpus difficult.
(221) [+imperfective] [+past]
‫العالم االسالمي كان يقايض التوابل بالذهب‬
al-‫ع‬ālam-u l-’islāmī kān-a
def-world-nom def-Islamic be;pst-3msg
yuqāyiD-u t-tawābil bi-dh-dhahab
exchange;npst.3msg-ind def-spice;pl with-def-gold
‘The Islamic world was exchanging [used to exchange] spices for gold’
 (arabiCorpus: GEN1997: 35361)
(222) ‫كان الكاهن يناول القربان الى ذوي الشهداء وسائر المصلّين‬
kān-a l-kāhin yunāwil-u
be;pst-3msg def-priest give;npst.3msg-ind
l-qurbān ’ilā dhaw-ī sh-shuhadā’
def-Eucharist to possessor-gen.pl def-martyr
wa-sā’ir-i l-muSall-īn
and-remainder-gen def-worshipper-gen.pl
‘The priest was giving the Eucharist to the families of the martyrs and the
rest of the worshippers’ (arabiCorpus: NEW1996: 30463)

It is also significant that, unlike other verbs with meanings of ‘to give’ in other
patterns such as ‫’( أعـطى‬a‫ع‬Tā – pattern IV) and ‫( وهـب‬wahaba – pattern I), none of
the three verbs examined here has a passive participle listed in Wehr (1994). If in-
deed this is an accurate indication that they may not be used in the statal passive,
it is further evidence for their atelicity. Certainly I have found no evidence in the
corpus to contradict Wehr’s omission of the passive participles from the lexicon.
 Chapter 9.  Aspectual categorisation of patterns III and VI 201

To summarise, this group contributes little numerically to the data, but pre-
liminary findings based on limited data suggest that they too are unmarked for
telicity, i.e. [Øtelic].

9.1.2.3  Verbs of inception

Unlike the previous two groupings, the verbs which I have gathered together un-
der this heading do not share a basic meaning, but rather they appear to be aspec-
tually similar, describing a situation which involves entry into a state. Thus at first
sight they appear to display telicity and are not capable of inclusion in the catego-
ry of ACTIVITY/STATE. Table 67 shows the counts for the verbs I have grouped
according to inceptivity. Although their representation in the corpus by token is
only around half that by type, these verbs constitute a relatively large grouping
and account for the majority of exceptions to the ACTIVITY/STATE category
amongst verbs with the C1āC2 sequence. Clearly any characterisation of the form-
meaning relationship for patterns III and VI which involves lexical aspect must
provide a satisfactory account of the role of these verbs within the system.

Table 67.  Frequency of verbs of inception by pattern

III = ‘INCEPTION’ VI = ‘INCEPTION’ TOTAL
dictionary count    63 (12.3%)    55 (12.4%) ,   118 (12.4%)
corpus type count    46 (13.0%)    42 (13.7%) ,    88 (13.3%)
corpus token count 9726 (7.1%) 2881 (6.1%) 12,607 (6.8%)

Many of the verbs in this grouping have previously been identified as mutual or
reciprocal, but rather than representing a static mutual relationship or participa-
tion together in an activity, they describe entry by two or more entities into a
relationship which then persists. Thus they may be said to involve a transition
(necessarily dynamic) into a durative state. Examples from patterns III and VI are
given in (223) and (224):

(223) ‫ حـالـف‬ [III]

Hālafa (+ d.o.)
‘to make an alliance with (s.o.)’
(224) ‫ تـعـارف‬ [VI]
ta‫ع‬ārafa
‘to become mutually acquainted’

However, other verbs included in the grouping do not conform to a mutual inter-
pretation in pattern III or reciprocal interpretation in pattern VI, and many of these,
like Example (225), may represent either a state or the attainment of that state.
202 The Arabic Verb

(225) ‫ تـكامـل‬ [VI]

takāmala
‘to be/become complete’
Verbs in Arabic which exhibit this state/entry into state dichotomy of meaning are
common, such as Examples (226) to (228) drawn from other patterns:
(226) ‫ حـبـل‬ [I]
Habila85
‘to be/become pregnant’
(227) ‫ اسو ّد‬ [IX]
(i)swadda
‘to be/become black’
(228) ‫ كـبـر‬ [I]
kabura
‘to be/become large/great’
As the translations suggest, English does not have equivalent verbs capable of inter-
pretation as either state or entry into state and it makes this distinction by means of
the aspectual difference between ‘be’ and ‘become’. However it appears that verbs of
this type in MSA do not conform to the categories defined by Vendler (1967) and
extended by Olsen (1997). In the following chapter we will proceed to investigate
whether further extension of Olsen’s scheme is required in order to incorporate
these verbs, before examining the pattern III and pattern VI verbs which I have
identified as carrying an inceptive meaning. At present, however, we will treat this
group of verbs as potential exceptions to the [Øtelic] feature marking observed for
the other C1āC2 sequence verbs examined in Sections 9.1.1, 9.1.2.1 and 9.1.2.2.

9.1.2.4  Other verbs

There remains a small number of verbs, contributing less than 1% to the token
count, which must be examined on a case-by-case basis, as the preliminary dic-
tionary search suggests that they may not belong in any of the groups identified
above. The counts for these verbs are given in Table 68.

9.1.2.4.1  Denominative and delocutive verbs. There are five verbs which are clearly
derived from nouns or sayings. Examples (229) and (230) are delocutive, i.e. de-
rivatives designating utterance of the source word or phrase, and are therefore
SEMELFACTIVE and hence [Øtelic].

85. The verb here is given in the standard citation form, though note that for pragmatic rea-
sons it would not normally be found with masculine inflection.
 Chapter 9.  Aspectual categorisation of patterns III and VI 203

Table 68.  Frequency of other anomalous verbs by pattern

III = ‘OTHER’ VI = ‘OTHER’ TOTAL
dictionary count   10 (2.0%)    5 (1.1%)    15 (1.6%)
corpus type count    6 (1.7%)    4 (1.3%)    10 (1.5%)
corpus token count 820 (0.6%) 358 (0.6%) 1178 (0.6%)

(229) ‫ العـن‬ [III]

lā‫ع‬ana (token count: 1)
‘to utter the li‫ع‬ān [oath of condemnation]’
(230) ‫ عـايـد عـلـى‬ [III]
‫ع‬āyada ‫ع‬alā (+ i.o.) (token count: 3)
‘to wish (s.o.) a merry feast’

Example (231), although most likely an ACCOMPLISHMENT, appears to be un-

related in meaning to the root combination t-b-l, but is rather a denominative
derivative of ‫( تـابـل‬tābal/tābil – ‘coriander/spice’).

(231) ‫ تـابـل‬ [III]

tābala (+ d.o.) (token count: 0)
‘to spice, season (s.th.)’

It is also listed in the dictionary with a QI variant ‫( تـوبـل‬tawbala), in which C2 is

the weak consonant wāw (‫ )و‬forming the first syllable taw, possibly a reanalysis of
the stressed syllable tā which is integral to the source noun. This suggests that the
resemblance to pattern III is incidental and the verb does not strictly contain the
C1āC2 sequence in which the long vowel is part of the derivational morphology.
It should also be noted that the token count for this verb in its pattern III form in
the arabiCorpus newspaper corpus was zero.
A similar analysis is suggested for Examples (232a–b), which are listed as pat-
tern III and pattern VI derivatives respectively of the phrase ‫( ال شـيء‬lā shay’ –
‘nothing’). Again, the ā is integral to the source phrase and as such should not be
analysed as belonging to the derivational morphology.

(232) a. ‫ الشى‬ [III]

lāshā (+ d.o.) (token count: 1)
‘to suppress/crush/destroy (s.th.)’
b. ‫ تـالشى‬ [VI]
talāshā  (token count: 192)
‘to be suppressed/crushed/destroyed’, ‘to dwindle/disappear’
204 The Arabic Verb

9.1.2.4.2  Potentially [+telic] verbs. The ten verbs listed in Examples (233)–(242)

were identified according to the meanings given below86 as either ACCOM-
PLISHMENTS or ACHIEVEMENTS, i.e. belonging to one of the categories with
[+telic] feature marking in Olsen’s scheme.

(233) ‫ دارك‬ [III]

dāraka (+ d.o.) (token count: 0)
‘to reach (s.o./s.th.)’
(234) ‫ تـسـافـك‬ [VI]
tasāfaka (token count: 0)
‘to murder each other’
(235) ‫ طـابـق‬ [III]
Tābaqa (+ d.o.) (token count: 0)
‘to adapt/adjust/trim (s.th.)’
(236) ‫ طـاهـر‬ [III]
Tāhara (+ d.o.) (token count: 0)
‘to circumcise (s.o.)’
(237) ‫ عـايـر‬ [III]
‫ع‬āyara (+ d.o.) (token count: 9)
‘to gauge / test the accuracy of (s.th.)’
(238) ‫ تـفـانـى‬ [VI]
tafānā (token count: 17)
‘to annihilate each other’
(239) ‫ نـاهـز‬ [III]
nāhaza (+ d.o.) (token count: 580)
‘to attain, reach, seize (s.th.)’
(240) ]‫ تـنـاهـى [ إلى‬ [VI]
tanāhā [’ilā (+ i.o)] (token count: 41)
‘to come to an end, to reach (s.o.)’
(241) ‫ بـارك‬ [III]
bāraka (+ d.o.) (token count: 226)
‘to bless (s.o.)’
(242) ‫ تـعـافـى‬ [VI]
ta‫ع‬āfā (token count: 108)
‘to recuperate, recover, regain health’

86. Note that there are also alternative meanings which are unambiguously atelic for some of
these verbs.
 Chapter 9.  Aspectual categorisation of patterns III and VI 205

It will be noted that four of the verbs listed scored zero in the corpus token count:
thus the data suggest that these verbs may have little relevance for contemporary
Arabic and any assessment of alternate meanings where these are suggested by
Wehr (1994) will be difficult. It is therefore proposed to concentrate on the re-
maining six verbs.
Examination of corpus data for the verb in (237) reveals that rather than the
literal meaning of testing weights and measures, contemporary usage suggests a
figurative meaning, of which (243) is an example:

(243) ‫كان أصحابه هناك يعايرونه بما كتبه عن النساء‬

kān-a ’aSHāb-u-hu hunāk
be;pst-3msg friend;pl-nom-poss.3msg there
yu‫ع‬āyir-ū-na-hu bi-mā
measure;npst.3m-pl-ind-obj.3msg with-what
katab-a-hu ‫ع‬an an-nisā’
write;pst- 3msg-obj.3msg about def-women
‘His friends there used to measure him by what he wrote about women’
 (arabiCorpus: 050899WRIT01)

This figurative usage is consistent with an ACTIVITY rather than an ACCOM-

PLISHMENT, although it should be noted that the verbal noun ‫( مـعايـرة‬mu‫ع‬āyara)
is more common than the verb itself and retains literal usage in MSA, as demon-
strated in (244):

(244) ‫ جمعية الحرفيين … ال تعترف بأي طريقة لمعايرة الذهب إال بالطريقة التي يستخدمونها‬
jama‫ع‬īat-u l-Hirafīy-īn … lā
society-nom def-craftsman-pl.gen neg
ta‫ع‬tarif-u bi-’ayy Tarīqa
recognise;npst.3fsg-ind with-any method
li-mu‫ع‬āyarat-i dh-dhahab ’ilā bi-T-Tarīqa
of-measurement-gen def-gold except with-def-method
allatī yastakhdim-ū-na-hā
rel;3fsg employ;npst.3m-pl-ind-obj.3fsg
‘The craftsmen’s guild … does not recognise any method of assaying gold
other than the method which they employ’
(arabiCorpus: archive64102)

Similarly, contemporary usage of the verb from Example (238), of which the cor-
pus example in (245) is typical and represents a STATE, appears to exclude the
[+telic] meaning of ‘to annihilate each other’.
206 The Arabic Verb

(245) ‫جميل جدا ان يتفانى المعلم في عمله‬

jamīl jiddan ’an yatafānā
beautiful very that be_consumed;npst.3msg
l-mu‫ع‬allim fī ‫ع‬amal-i-hi
def-teacher in work-gen-poss.3msg
‘It is very pleasing that the teacher is consumed in his work’
(arabiCorpus: 011223t45037MQAL)

For (239), I have not identified any corpus examples supporting the meaning of
‘to seize’ in MSA and it appears that identification of the meaning ‘to reach’ for
this verb with an ACHIEVEMENT may simply be an artefact of translation. As
Example (246) demonstrates, typical usage of the verb closely resembles the ‘con-
tinuatively durative’ Example (166) in Section 8.4.2.1, which was classified as a
STATE verb.

(246) ‫عدد الطالب … كان يناهز المئتين عندما درست في الجامعة‬

‫ع‬adad-u T-Tullāb … kān-a
number-nom def-student;pl be;pst-3msg
yunāhiz-u l-mi’at-ayn ‫ع‬indamā
approach;npst.3msg-ind def-hundred-du.acc when
daras-tu fī l-jāmi‫ع‬a
study;pst-1sg in def-university
‘The number of students … was approaching two hundred when I studied
at the university’  (arabiCorpus: GEN1997: 41591)

One of the meanings of Example (240) is also ‘to reach’, though closer investiga-
tion of corpus data reveals that it is almost exclusively used figuratively of sounds,
news etc. reaching the attention of a hearer, as in (247):

(247) … ‫لقد تناهى الى علم الملك حسين ان‬

laqad tanāhā ’ilā ‫ع‬ilm-i
pfv reach;pst;3msg to knowledge-gen
l-malik Hussayn ’an …
def-king Hussein that …
‘It had reached King Hussein’s ears that …’
(arabiCorpus: NEW1996: 17682)
However, when we conceptualise sound or news reaching a hearer, we must note
that its receipt does not terminate the propagation of that sound or the dissemina-
tion of that news in the way that reaching the summit of a mountain terminates
the process of climbing. Example (248), in which the verb has the same meaning
but no target recipient is specified, clarifies that no endpoint is implied by the verb
in this usage.
 Chapter 9.  Aspectual categorisation of patterns III and VI 207

(248) ‫تناهى صوتها من المطبخ‬

tanāhā Sawt-u-hā min al-maTbakh
come;pst;3msg sound-nom-obj.3fsg from def-kitchen
‘The sound of it came from the kitchen’ (arabiCorpus: GEN1996: 5540)

Thus although other derivatives of the root of ‫( تـنـاهـى‬tanāhā) have a meaning

involving ‘ending’, the corpus evidence is that the usage of the pattern VI verb is
overwhelmingly atelic.
At first sight, Example (241) is an ACCOMPLISHMENT verb, involving a
process culminating in a resultant state upon the direct object, i.e. a state of ‘being
blessed’. However, the following example sentences using this verb (249a–f) were
submitted to native speakers for assessment:

(249) a. ‫ بارك هللا ابراهيم طوال عشرين سنة‬ {YYY}

bārak-a llah ’ibrāhīm Tiwāla ‫ع‬ashr-īn sana
bless;pst-3msg God Abraham during twenty-gen year
‘God blessed Abraham for twenty years’ (Questionnaire: 45)
b. ‫ كان هللا يبارك ابراهيم طوال عشرين سنة‬ {NYY}
kān-a llah yubārik-u ’ibrāhīm
be;pst-3msg God bless;npst.3msg-ind Abraham
Tiwāla ‫ع‬ashr-īn sana
during twenty-gen year
‘God was blessing Abraham for twenty years’ (Questionnaire: 41)
c. ‫ بارك هللا ابراهيم مـرّات كـثيرة‬ {YY?}
bārak-a llah ’ibrāhīm marr-āt kathir-a
bless;pst-3msg God Abraham occasion-pl many-f
‘God blessed Abraham many times’ (Questionnaire: 53)
d. ‫ كان هللا يبارك ابراهيم حـتى أواخر أيّـامه‬ {?YY}
kān-a llah yubārik-u ’ibrāhīm
be;pst-3msg God bless;npst.3msg-ind Abraham
hattā ’awākhir ’ayyām-i-hi
until end;pl day;pl-gen-poss.3msg
‘God was blessing Abraham unto the end of his days’ (Questionnaire: 59)
e. ‫ بورك ابراهيم‬ {NYY}
būrik-a ’ibrāhīm
bless;pst.pass-3msg Abraham
‘Abraham was blessed’ (actional passive)  (Questionnaire: 44)
f. ً ‫ كـان ابراهيم مـبـاركـا‬ {Y??}
kān-a ’ibrāhīm mubārak-an
be;pst-3msg Abraham bless;ppt-acc.indf
‘Abraham was blessed’ (statal passive)  (Questionnaire: 36)
208 The Arabic Verb

The use of a temporally extensive adverbial with the s-stem in (249a) was ac-
ceptable to all three informants, providing evidence that the verb represents a
durative process. There was some disagreement regarding the imperfective past
in (249b), though the majority acceptance of (249c) demonstrates that the action
of the verb may also be performed iteratively upon an individual, and this is also
the most likely interpretation of (249d). However, this usage is not consistent with
­ACCOMPLISHMENT verbs. Recall the [+telic] example verb ‫( بـنـى‬banā – ‘to
build’) in Section 8.4.2.3: it is not possible to perform this verb iteratively upon
the same (i.e. definite and specific) entity, due to its telicity. Once the endpoint is
reached, the action represented by the verb cannot continue with respect to that
patient. Hence (250) is not well-formed either in Arabic or in English:

(250) *‫ بنى مكتبة الجامعة مرّات كثيرة‬

*banā maktabat-a l-jāmi‫ع‬a
build;pst;3msg library-acc def-university
marr-āt kathīr-a
occasion-pl many-f
*‘He built the university library many times’

Nevertheless, (249e) and (249f) demonstrate that not only may ‫( بـارك‬bāraka) be
used in the actional passive, but for at least some speakers the statal passive is
also acceptable, which is normally indicative of telicity, and indeed Wehr (1994)
includes the passive participle in his entry for this verb. Therefore we must ask
whether it is possible for a verb to be both resultative in that it brings about a new
state upon its patient and yet atelic in that it involves no built-in endpoint.
In Section 8.4.2.1, we discussed the verb ‫( جـاوز‬jāwaza – ‘to exceed/surpass’)
which corresponded with Poutsma’s (1926: 289) designation ‘continuatively du-
rative’. Although the situation represented by the verb involves an external goal,
the action of the verb may continue indefinitely beyond the goal: it was therefore
designated [Øtelic] as no endpoint is inherent. It is possible that a similar ex-
planation is pertinent to verbs like ‫( بـارك‬bāraka). Consider the following event
diagram for (249d):
(251) RT ST
↓ ↓
ET: <-----------------------------------------------|
NUCLEUS
↑
X: Abraham’s death
 Chapter 9.  Aspectual categorisation of patterns III and VI 209

The beginning of the situation in (251) is unspecified and its end is bounded by
the external event of Abraham’s death. However, at any RT within that period it is
true to assert both (252a) and (252b):
(252) a. [+perfective]
‫ قد بارك هللا ابراهيم‬
qad bārak-a llah ’ibrāhīm
pfv bless;pst-3msg God Abraham
‘God has blessed Abraham’
b. ‫ابراهيم مـبـارك‬
’ibrāhīm mubārak
Abraham bless;ppt-acc.indf
‘Abraham is blessed’

That the imperfective (249d) entails the perfective (252a) is indicative of atelicity
and thus the verb would be analysed as an ACTIVITY. However, it is undeniable
that the state of ‘being blessed’ represented in (252b) is directly resultative from
the performance of the action of the verb. Nevertheless, it is vital to realise that
the state of ‘being blessed’ is fundamentally different to the state of ‘being built’ or
‘being killed’. Thus the statement in (253) is possible, while (254) is not:

(253) ‫قد بارك هللا ابراهيم وما زال يـبـاركـه‬

qad bārak-a llah ’ibrāhīm wa-mā
pfv bless;pst-3msg God Abraham and-neg
zāl-a yubārik-u-hu
cease;pst-3msg bless;npst.3msg-ind-obj.3msg
‘God has blessed Abraham and he is still blessing him’
(254) *‫قد قتل أحمد صديقه وما زال يـقـتـله‬
qad qatal-a ’aHmad Sadīq-a-hu
pfv kill;pst-3msg Ahmad friend-acc-poss.3msg
wa-mā zāl-a yaqtul-u-hu
and-neg cease;pst-3msg kill;npst.3msg-ind-obj.3msg
*‘Ahmad has killed his friend and he is still killing him’

The difference is that ‘being killed’, like ‘being built’, is a final and absolute state,
thus telic, whilst ‘being blessed’ is transitional and relative and thus atelic. One
person may be more blessed than another, whilst one may not be more dead than
another, or one house be more built than the next. Thus in the same way that
the situation described by a ‘continuatively durative’ verb may proceed past an
external goal, the situation described by ‫( بارك‬bāraka) may proceed beyond an
intermediate state of ‘being blessed’ to subsequent states of ‘being more blessed’.
210 The Arabic Verb

This is only possible because the [+perfective] verb in (253) does not point to the
coda, because there is none for a [Øtelic] verb. Before proceeding, it is also worth
noting that the implication which this verb carries of a cumulative effect rules out
the possibility of discrete, telic, iterative events, since there is no sense that the
state brought about by the initial performance of the action of the verb is anything
other than persistent. In conclusion, despite the compatibility of this verb with the
statal passive, the state described is not a final result and the evidence is therefore
that it should be categorised as an ACTIVITY.
Finally, I am forced to conclude that Example (242) is indeed anomalous and
therefore constitutes a true exception. The evidence is that it is [+telic], i.e. it has
a built-in endpoint, as it is clearly incorrect to assert the perfective (255b) on the
basis of the imperfective (255a). Since durativity is also inherent to its meaning, it
is properly categorised as an ACCOMPLISHMENT.
(255) a. [+imperfective]
‫ال يزال الرئيس الكوبي فيدل كاسترو يتعافى من عملية جراحية‬
lā yazāl-u r-ra’īs-u
neg cease;npst.3msg-ind def-president-nom
l-kūbī fīdal kāstrū yata‫ع‬āfā min
def-Cuban Fidel Castro recover;npst.3msg from
‫ع‬amalīya jirāHīya
operation surgical
‘The Cuban President Fidel Castro is still recovering from surgery’
 (arabiCorpus: archive61186)
b. [+perfective]
‫قد تعافى الرئيس الكوبي فيدل كاسترو من عملية جراحية‬
qad ta‫ع‬āfā r-ra’īs-u l-kūbī
pfv recover;pst;3msg def-president-nom def-Cuban
fīdl kāstrū min ‫ع‬amalīya jirāHīya
Fidel Castro from operation surgical
‘The Cuban President Fidel Castro has recovered from surgery’

9.1.2.4.3  Section summary. To summarise the findings of this section (see also

Table 69): the majority of the verbs presented here for which we have corpus data
have been found to represent STATES, ACTIVITIES or SEMELFACTIVES and
hence are [Øtelic]. Of the remainder, only one verb present in the corpus and
with long ā properly belonging to its derivational morphology has been identi-
fied as [+telic].
 Chapter 9.  Aspectual categorisation of patterns III and VI 211

Table 69.  Categorisation of verbs provisionally thought to be anomalous

[Øtelic] non-derivational ā [+telic]87
dictionary count    7    3    5
corpus type count    7    2    1
corpus token count 877 193 108

9.2 Summary

Based on corpus data, there is good evidence that over 93% of actual usage of
pattern III and pattern VI verbal forms is attributable to verbs in categories un-
marked for telicity, i.e. [Øtelic]. Only one verb attested in the corpus was posi-
tively identified as [+telic], contributing less than 0.1% to actual usage. With three
verbs reanalysed as properly quadriliteral, i.e. with ā not integral to derivational
morphology, only the verbs classed in my survey as inceptives and resistant to
classification under Olsen’s scheme remain as potential exceptions to an analysis
of the C1āC2 sequence being identified as a formal morphological marker of atel-
icity (Table 70).

Table 70.  Categorisation of pattern III and pattern VI verbs by telicity87

[Øtelic] ? inceptives [+telic] excluded
dictionary count ,    827 (86.8%)   ,  118 (12.4%)    5 (0.5%)    3 (0.3%)
corpus type count ,    570 (86.2%) ,    88 (13.3%)    1 (0.2%)    2 (0.3%)
corpus token count 172,132 (93.0%) 12,607 (6.8%) 108 (0.1%) 193 (0.1%)

Thus in Chapter 10 we will endeavour to explain these verbs of inception in terms

of Olsen’s scheme and in particular to identify whether or not they exhibit a
[+telic] feature-marked coda phase. Chapter 11 will present additional evidence
regarding pattern III and pattern VI passive participle formation and the limited
compatibility of these verbs with the statal passive.

87. Included here are those verbs absent from the corpus which cannot be fully assessed for
telicity.
 chapter 10

Inceptive aspect

This chapter will examine inceptivity as defining a verbal class, specifically in

MSA, but with the implication also that it has cross-linguistic application. An
extension of Olsen’s (1997) scheme will therefore be argued and then applied
specifically to the group of verbs identified as inceptive in the previous chapter.
Henceforth, when referring to verbs which describe dynamic entry into a dura-
tive state, the term INCEPTIVE will be used, which the OED defines as “express-
ing the beginning of action”. This term is preferred here over ‘ingressive’, as the
latter, sharing a morpheme with words such as ‘regress’ and ‘progress’, may imply
durativity or process which is not necessarily a feature of the onset of INCEP-
TIVES. Whereas ‘punctual inception’ and ‘durative inception’ are both possible,
it is likely that ‘punctual ingression’ is an oxymoron and ‘durative ingression’ is
tautological. Indeed, one might categorise ingression as [+durative] and incep-
tion as [Ødurative] and thus we will refer to INCEPTIVES without prejudicing
the possibility that the initial phase of the situation represented by the verb may
be durative. Similarly, INCEPTIVE is preferred over ‘inchoative’, since the latter
term is often associated with entry into a state without agency88 and there is no
such lack of agency in some of the examples already cited in Section 9.1.2.3.

10.1 The case for a lexical aspect category of INCEPTIVE

We will begin by using the verb ‫( حبل‬Habila – ‘to be/become pregnant’) as our
example, since the inception and subsequent state represented by the verb are
readily conceptualised. In (256a), although the verbs are marked only as [+past],
the adverbial phrase favours a perfective reading and thus an inceptive interpreta-
tion of Habila:

88. E.g. ‘to sink’ used intransitively, meaning ‘to become sunk’.
214 The Arabic Verb

(256) a. ً ‫وفي غضون سنة حبلت حنة وأنجبت ابنا‬

wa-fī_ ghuDūn sana Habil-at Hanna
and-within year [become]_pregnant;pst-3fsg Hannah
wa-’anjab-at ’ibn-a-n
and-give_birth_to;pst-3fsg son-acc-indf
‘And within a year Hannah became pregnant and gave birth to a son’
 (ALAB: 1 Samuel 1: 6)
However, in (256b) with an overtly marked [+future] verb, it is no longer clear
that an inceptive interpretation is required:

(256) b. ‫ ستحبلين وتلدين ابنا‬

sa-taHbal-īn
fut-[be/become]_pregnant;npst.2fsg-ind
wa-talid-īn ’ibn-a-n
and-give_birth_to;npst.2fsg-ind son-acc-indf
‘You will be [become] pregnant and give birth to a son’
 (ALAB: Judges 13: 3)

It is obvious pragmatically that in order to be pregnant one must first become

pregnant. Thus we must allow the possibility that inceptivity in Arabic is merely
pragmatically determined, rather than a lexical aspect feature, and that any sug-
gestion that it is otherwise is simply an artefact of translation into English, which
formally distinguishes inceptive situations periphrastically with ‘become’.
It should be noted at this point that inception of a state may be either a punc-
tual event, as in (256a–b), or a durative process. The following Example (257)
which employs the verb ‫(( اسو ّد‬i)swadda – ‘to be/become black’) demonstrates du-
rative inception:

(257) ‫قد اسو ّدت وجوههم وتقلّصت عضالتهم‬

qad iswadd-at wujūh-u-hum
pfv [become]_black;pst-3fsg face;pl-nom-poss.3mpl
wa-taqallaS-at ‫ع‬aDal-āt-u-hum
and-shrink;pst-3fsg muscle-pl-nom-poss.3mpl
‘Their faces had blackened and their muscles had shrunk’
(arabiCorpus: GEN1996: 7768)
Conveniently in (257) the phrase containing the verb of interest is followed by a
phrase containing the verb ‫( تقـلص‬taqallaSa – ‘to shrink’), which is unambiguously
durative and dynamic and is therefore an ACTIVITY, introduced by the con-
junction ‫( و‬wa – ‘and’), thus suggesting two processes occurring simultaneously.
Thus (i)swadda is non-punctual and hence we must consider that it is potentially
marked [+durative].
 Chapter 10.  Inceptive aspect 215

However, verbs of punctual inception present two related problems for

­ lsen’s (1997) analysis. Firstly, a verb cannot be both a STATE and [Ødurative].
O
Consider these English sentences:
(258) a. Charlotte was pregnant for nine months
b. Charlotte became pregnant * for twenty minutes / ?for nine months

Example (258a) is a straightforward STATE, marked [+durative] and therefore

compatible with an appropriate durative adverbial ‘for nine months’. However,
the event of becoming pregnant is perceived as punctual, whatever the actual time
needed for the biological process involved. This is demonstrated in (258b), where
the relatively short durative ‘for twenty minutes’ is incompatible with the verb,
although the longer durative ‘for nine months’ is rendered marginally acceptable
by the interpretation that it is the STATE of ‘being pregnant’ ensuing from the
inceptive event ‘become pregnant’ which is durative. This is not a problem for
English, since we can designate expressions formed with ‘be’ as STATES [+dura-
tive] (Olsen 1997: 155) and inceptives with ‘become’ as [Ødurative]. However, in
Arabic it would appear that many verbs such as ‫( حبل‬Habila) have both stative
and punctually inceptive interpretations. The most obvious solution is to simply
designate them as [Ødurative], allowing both possibilities.
However, this leads to the second problem. Again, consider some English ex-
amples:
(259) a. What Charlotte did was become pregnant
b. *What Charlotte did was be pregnant
c. Charlotte became pregnant and so did Susan
d. *Charlotte was pregnant and so did Susan

Pseudo-cleft constructions (259a–b), and indeed other constructions with ‘do’

(259c–d), are incompatible with states as they require dynamicity (Binnick
1991: 174; Olsen 1997: 35). Again, for English this is not problematic, as we des-
ignate the stative copula expressions as [+durative][Ødynamic] and inceptives
with ‘become’ as [Ødurative][+dynamic]. Now, since we have been forced to des-
ignate Habila as [Ødurative] to allow for the punctually inceptive interpretation
‘become pregnant’, we must also designate it [+dynamic], as [Ødurative][Ødy-
namic] is disallowed: “A VERB must be minimally [+dynamic] or [+durative]”
(Olsen 1997: 51). However, verbs designated [Ødurative] [+dynamic] are either
SEMELFACTIVES if [Øtelic] or ACHIEVEMENTS if [+telic]. If Habila is either
of these, it must be a [+telic] ACHIEVEMENT, since there is a built in endpoint
to becoming pregnant, a persistent resulting state (i.e. pregnancy) which is not
true of SEMELFACTIVES. It will be noted that the [Ødurative][+dynamic] nu-
cleus designation is the opposite of a STATE, which is [+durative] [Ødynamic].
216 The Arabic Verb

Thus we are caught in a paradox, since the objective in designating Habila as

[Ødurative] was precisely in order to allow it to be interpreted as either stative or
punctually inceptive. Under Olsen’s scheme, we must either designate the verb as
[Ødurative][+dynamic] for the nucleus and accept that the verb is incapable of
stative interpretation, or designate it as [+durative] [Ødynamic] and accept that it
is incapable of punctual interpretation.
We will now examine whether inceptives can be treated as STATES. When
discussing (256a–b) which are capable of either stative or punctually inceptive
interpretation, we allowed that inceptivity may not be a lexical feature in Arabic,
so one way to avoid the paradox discussed above is to regard ‫( حبل‬Habila) as a
STATE. However, this diminishes the focus which such verbs place on punctual
entry into the stative situation. In the following Example (260a), it is stretching a
point to suggest that the reader should understand that the mother was pregnant
in a stative, hence durative sense, while under the tree, thus I have translated the
verb as ‘conceived’.

(260) a. …‫ وحيث تمخضت بك وأنجبتك‬،‫تحت شجرة التفاح حيث حبلت بك أمك‬

taHta shajarat-i t-tuffaH Haythu
under tree-gen def-apple;coll where
Habilat ’bi-ka
[become]_pregnant;pst-3fsg with-obj.2msg
umm-u-ka, wa-Haythu tamakhkhaD-at
mother-nom-poss.2msg and-where labour;pst-3fsg
bi-ka wa-’anjab-at-ka
with-obj.2msg and-give_birth_to;pst-3fsg-obj.2msg
‘Under the apple tree where your mother conceived [became pregnant
with] you and where she was in labour with you and gave birth to you …’
 (ALAB: Song of Solomon 8: 5)
In constructing an event diagram there is no alternative but to have RT intersect
with ET at a point within the [+durative] nucleus, since we are treating the verb
as a STATE, which is homogeneous or non-dynamic:
(260) b. RT ST
↓ ↓
ET: 0----------------------------------------->
nucleus

This is unsatisfactory in that RT should point to the start or inception of the du-
rative situation, but in this analysis there is no way to ensure that it does, thus
the event diagram must be interpreted as ‘your mother was pregnant with you’
in contravention of pragmatic sense. The problem is even more clearly seen in
(261a), where [+perfective] marking is employed:
 Chapter 10.  Inceptive aspect 217

(261) a. [+perfective] [+past]

‫ماذا لو كانت قد حبلت أثناء شهور حياتها مع زوجها؟‬
mādhā law kān-at qad
what if be;pst-3fsg pfv
Habil-at ’athnā’a shuhūr
[become]_pregnant;pst-3fsg during month;pl
Hayāt-i-hā ma‫ع‬a zawj-i-hā?
life-gen-poss.3fsg with husband-gen-poss.3fsg
‘What if she had become pregnant during the months of living with her
husband?’ (http://alketaba.jeeran.com/story/babel.html, 23 May 2008)

The event diagram in (261b) is consistent with my translation of (261a), based

on the context of the story, which makes it clear that the matter of concern is the
woman’s hypothetical entry into the state of ‘being pregnant’ during ‘the months
of living with her husband’ (a period which has recently ended) and allows that
had she done so, she might still be pregnant at ST:
(261) b. RT ST
↓ ↓
ET: |---------------------------------->
nucleus?
X: |------------------------------------------------------|
‘the months of living with her husband’

However, as for ACTIVITIES where the verb is also [+durative] and [Øtelic], the
[+perfective] verb in (261a) finds no coda to intersect, so it would be expected to
reference the nucleus. Although it is true to say ‘she has become pregnant’ at any
time where the nucleus overlaps with the durative adverbial in X, there is no way
to anchor RT to the beginning of the nucleus, where the emphasis belongs. There-
fore, as in (260a), the nucleus has only an implicit start and this event diagram is
not achievable.
In (261c) and the corresponding event diagram (261d), the nucleus inter-
sected by RT most naturally refers to a period wholly within X:
(261) c. What if she had been pregnant during the months of living with her
husband?
d. RT ST
↓ ↓
ET: <-------------------------->
nucleus
X: |--------------------------------------------------------|
‘the months of living with her husband’
218 The Arabic Verb

I do not offer this as a valid interpretation, however, given the choice of the [+per-
fective] verb which suggests completion: there are more natural ways to express
the situation represented by (261c–d) in Arabic, for example:

(261) e. ‫ماذا لو كانت حبلى أثناء شهور حياتها مع زوجها؟‬

mādhā law kān-at Hublā ’athnā’a
what if be;pst-3fsg pregnant;f during
shuhūr Hayāt-i-hā ma‫ع‬a zawj-i-hā?
month;pl life-gen-poss.3fsg with husband-gen-poss.3fsg
‘What if she had been pregnant during the months of living with her
husband?’

Thus, only the original interpretation of (261a) is viable, which demonstrates that
to characterise this verb as a STATE is unsatisfactory.
However, the event diagram (261b) is almost identical to that presented for an
ACHIEVEMENT in the previous chapter. Is it possible, therefore, to present ‫حبل‬
(Habila) as an ACHIEVEMENT? Let us adjust the event diagram (261f) accordingly
for a verb marked [+dynamic][+telic]:

(261) f. RT ST
↓ ↓
ET: ||------------------------------------->
NUCLEUS CODA
X: |--------------------------------------------------------|
‘the months of living with her husband’

There is now a coda for the [+perfective] verb to intersect and it is still consistent
for RT to be within X, yet anterior with respect to ST. However in this analysis,
it is the nucleus which represents ‘becoming pregnant’ as a punctual event, while
the coda represents the enduring result of that event and thus we need the perfec-
tive verb to focus attention on the nucleus, which is not possible under Olsen’s
analysis. Consequently RT points to the coda, which represents the state of ‘being
pregnant’.
It is even more problematic when the verb is marked [+imperfective] and
a temporally extensive adverbial is employed. The following example, although
somewhat unusual, leaves no doubt as to the meaning of the verb we have been
examining:
 Chapter 10.  Inceptive aspect 219

(262) a. ‫ سنين‬9 ‫ بمعنى ان المرأة كانت تحبل بمولودها لفترة‬.‫ سنين‬9 ‫ تمثل‬،‫ اشهر‬9 ،‫تلك الفترة‬
tilka l-fatra, tis‫ع‬-a ’ashhur,
that;fsg def-period nine-f month;pl
tamthul-u tis‫ع‬-a sin-īn. bi-ma‫ع‬nā
signify;npst.3fsg-ind nine-f year;pl-gen with-meaning
’anna l-mar’a kān-at
that def-woman be;pst-3fsg
taHbal-u bi-mawlūd-i-hā
[be]_pregnant;npst.3fsg-ind with-baby-gen-poss.3fsg
li-fatrat tis‫ع‬-at sin-īn
for-period nine-f year;pl-gen
‘That period, nine months, signifies nine years. Meaning that the woman
was pregnant with her baby for a period of nine years.’
 (http://www.ladeenyon.net/forum/viewtopic.php?f=10&t=4844
 &p=52878, 27 June 2009)

Lest the figurative usage obscure the applicability of the example, I have rewrit-
ten the relevant portion of (262a) as (262b), adjusting the adverbial phrase while
retaining the relevant grammatical constructions:
(262) b. [+imperfective] [+past]
‫ اشهر‬9 ‫كانت المرأة تحبل بمولودها لفترة‬
kān-at l-mar’a taHbal-u
be;pst-3fsg def-woman [be]_pregnant;npst.3fsg-ind
bi-mawlūd-i-hā li-fatrat tis‫ع‬-at ’ashhur
with-baby-gen-poss.3fsg for-period nine-f month;pl
‘The woman was pregnant with her baby for a period of nine months’

We might construct the following event diagram, based on the verb being marked
[+dynamic][+telic]:
(262) c. RT ST
↓ ↓
ET: |-------------------------------------------|----------->
NUCLEUS CODA
X: |-------------------------------------------|
‘nine months’

The [+imperfective] requires that RT intersects with the nucleus, and the dura-
tive adverbial ‘for a period of nine months’ must therefore refer to the nucleus, cf.
(263), for which diagram (262c) is valid:
220 The Arabic Verb

(263) ‫ اشهر‬9 ‫كان يبني بيته لفترة‬

kān-a yabnī bayt-a-hu
be,pst-3msg build;npst.3msg.ind house-acc-poss.3msg
li-fatrat tis‫ع‬-at ’ashhur
for-period nine-f month;pl
‘He was building his house for a period of nine months’

There is no inherent problem with the durativity of the nucleus for an ACHIEVE-
MENT verb, since this is allowable for verbs marked [Ødurative]. What is prob-
lematic is that we have already noted that if ‫( حبل‬Habila) is an ACHIEVEMENT
the nucleus must represent the earlier phase of ‘becoming pregnant’ and the coda
represents the subsequent resulting phase of ‘being pregnant’. This is incompat-
ible with (262b), in which the verb cannot be translated as ‘becoming pregnant’:
it must represent the STATE ‘be pregnant’.
Thus for verbs of this type, neither the category of STATE nor that of
ACHIEVEMENT adequately and consistently describes their behaviour in rela-
tion to grammatical aspect, and I therefore propose to introduce a further category
of INCEPTIVE.

10.2 Extension of Olsen’s scheme for INCEPTIVES

The existence in Arabic of a distinct category of INCEPTIVE which cannot be

explained according to Vendler’s (1967) classification, even as extended by Olsen
(1997), requires that we examine how this category may be consistently incor-
porated into the overall scheme in order to achieve a unified description of lexi-
cal aspect categories in MSA. If a consistent solution is found, it may reasonably
be supposed that it will also be valid cross-linguistically. To recap, an INCEP-
TIVE verb describes both a stative situation and the (necessarily dynamic) entry
into that stative situation. Since discovering and characterising the category of
­INCEPTIVE in Arabic, and inspired by Pinker (2008: 199), I have also identified a
small and restricted group of INCEPTIVES in English, which likewise fall outside
conventional Vendlerian categories. These ‘verbs of posture’ are treated more fully
in Danks (2008), though I will briefly exemplify their usage in Section 10.2.2.
Their existence, and the fact that they may be incorporated consistently into the
scheme I develop here for Arabic, provides evidence for its applicability cross-
linguistically.
Using the example of ‫( حبل‬Habila), ‘be pregnant’ or ‘become pregnant’, it was
noted in the previous section that INCEPTIVES must have both [+dynamic] and
[+durative] features, but that these are not co-existent in the nucleus phase, since the
 Chapter 10.  Inceptive aspect 221

dynamic phase necessarily precedes the non-dynamic durative phase. I will demon-
strate how the [+dynamic] and [+durative] features of INCEPTIVE verbs are applied
to distinct phases.
Olsen’s terms ‘nucleus’ and ‘coda’ are inspired by the field of phonology: spe-
cifically syllable structure. The English syllable ‘cat’ in (264) has onset, nucleus
and coda:
(264) ONSET NUCLEUS CODA
/k/ /æ/ /t/
Cross-linguistically it is a universal that all syllables have a nucleus, but onset
is optional in some languages and coda is also optional where it is allowed at
all (Cairns & Feinstein 1982: 196–197). Olsen’s event time scheme has obligatory
nucleus and optional coda, but I will demonstrate that there is also an optional
onset in event time structure. In fact, Freed (1979: 30) whose work is referred to
by Olsen, talks about a three-phase event time in these terms. Thus event time
structure for a verb might look like this:
(265) |--------------------------------|--------------------------------|------------->
ONSET NUCLEUS CODA

I therefore introduce here a three-phase model of lexical aspect features, in which:

1. an onset phase precedes the nucleus;
2. INCEPTIVES are marked [+dynamic] in onset and [+durative] in nucleus.

However, we have already seen that [+imperfective] marking on a verb causes ref-
erence time to access the nucleus while a [+perfective] marked verb accesses the
coda. So how can the onset be accessed specifically? I will demonstrate that where
onset is present it is also referenced by the [+perfective] marked verb. In fact, I am
not proposing that event time structure for a given verb can have both onset and
coda but rather that they are mutually exclusive and thus there is no conflict as
to which phase is referenced by the [+perfective] marked verb.89 Let us therefore
re-examine ‫( حبل‬Habila) on this basis:

Onset: [+dynamic] Nucleus: [+durative] Coda: Ø

Under this analysis, the [+dynamic] onset represents ‘become pregnant’ and the
[+durative] nucleus represents ‘be pregnant’. Thus the event diagrams for situa-
tions described by the verb resemble a SEMELFACTIVE followed by a STATE.
Let us return to Example (262b), which I have reproduced as (266a):

89. See also Section 10.2.2.

222 The Arabic Verb

(266) a. [+imperfective] [+past]

‫ اشهر‬9 ‫كانت المرأة تحبل بمولودها لفترة‬
kān-at l-mar’a taHbal-u
be;pst-3fsg def-woman [be]_pregnant;npst.3fsg-ind
bi-mawlūd-i-hā li-fatrat tis‫ع‬-at ’ashhur
with-baby-gen-poss.3fsg for-period nine-f month;pl
‘The woman was pregnant with her baby for a period of nine months’
b. RT ST
↓ ↓
ET: ||---------------------------------------------->
ONSET NUCLEUS
X: |-----------------------------------------------|
‘nine months’

The nucleus of the situation, with which RT intersects for a verb marked [+imperfec-
tive], is marked identically to a verb of STATE, i.e. [+durative] [Ødynamic], and there-
fore has stative interpretation. The durative adverbial ‘for a period of nine months’
defines the bounds of the nucleus in this example, though note that without the ad-
verbial phrase the nucleus is unbounded at the end: it is only a matter of pragmatics
that a state of ‘being pregnant’ cannot be indefinite. This pragmatically implicit finish
is analagous to the pragmatically implicit start for ACTIVITIES such as ‘to fight’ and
ACCOMPLISHMENTS such as ‘to build’. In contrast, ‫( حبل‬Habila) has a bounded
start, defined by its onset phase, though since the onset of the situation concludes with
entry into the nucleus phase, the onset is not referenced by the [+imperfective] verb.
Let us re-examine (261a), reproduced as (267a):

(267) a. [+perfective] [+past]

‫ماذا لو كانت قد حبلت أثناء شهور حياتها مع زوجها؟‬
mādhā law kān-at qad
what if be;pst-3fsg pfv
Habil-at ’athnā’a shuhūr
[become]_pregnant;pst-3fsg during month;pl
Hayāt-i-hā ma‫ع‬a zawj-i-hā?
life-gen-poss.3fsg with husband-gen-poss.3fsg
‘What if she had become pregnant during the months of living with her
husband?’

It is now possible to focus on a specific point in time, namely the punctual onset
‘become pregnant’, with the nucleus extending indefinitely and representing the
state ‘be pregnant’, which may or may not still hold at ST. Thus, we can now gener-
ate the event diagram which best represents the situation described:
 Chapter 10.  Inceptive aspect 223

(267) b. RT ST
↓ ↓
ET: ||----------------------------------->
ONSET NUCLEUS
X: |-----------------------------------------------------|
‘the months of living with her husband’

Note that the [+perfective] is drawn to the onset in the absence of a coda, with
which we would otherwise expect it to intersect. We shall return to this observa-
tion later. It should be emphasised that reference to onset only anchors the start
of the period during which it is possible to say ‘she has become pregnant’. Much
in the same way that for a coda to exist there must have been a preceding nucleus,
the completion of which is signalled when the coda is referenced, existence of an
onset requires a following nucleus, the inception of which is signalled when the
onset is referenced.
It remains to be demonstrated that situations where the verb is not marked
for grammatical aspect can also be explained. Here is (256a), repeated as (268a):

(268) a. ً ‫وفي غضون سنة حبلت حنة وأنجبت ابنا‬

wa-fī_ ghuDūn sana Habil-at Hanna
and-within year [become]_pregnant;pst-3fsg Hannah
wa-’anjab-at ’ibn-a-n
and-give_birth_to;pst-3fsg son-acc-indf
‘And within a year Hannah became pregnant and gave birth to a son’
 (ALAB: 1 Samuel 1: 6)

As stated when this example was originally introduced, the adverbial phrase fa-
vours a perfective interpretation. Thus, RT points to the onset, as in (267a). The
nucleus, which represents ‘be pregnant’, again has no intrinsic endpoint, but is
pragmatically bound at its end by the phrase ‘and gave birth’.
(268) b. RT ST
↓ ↓
ET: ||---------------------------------->
ONSET NUCLEUS ↑
X: ‘gave birth’

Whether the whole process or merely the onset took place ‘within a year’ is am-
biguous, but resolution of the ambiguity is not necessary to validate the inter-
pretation of ET. If the adverbial phrase is removed, imperfective interpretation
is also possible:
224 The Arabic Verb

(268) c. ً ‫حبلت حنة وأنجبت ابنا‬

Habil-at Hanna
[become]_pregnant;pst-3fsg Hannah
wa-’anjab-at ’ibn-an
and-give_birth_to;pst-3fsg son-acc.indf
‘Hannah was pregnant and gave birth to a son’

In this case, RT optionally intersects with the nucleus, placing the focus on the
durative state, rather than the dynamic entry into that state:
(268) d. RT ST
↓ ↓
ET: ||---------------------------------->
ONSET NUCLEUS ↑
X: ‘gave birth’

Thus the above analysis of onset and nucleus phases is entirely consistent with
flexible aspectual interpretation where the verb is unmarked for grammatical as-
pect.

10.2.1  Feature marking of the onset phase

It was noted earlier that as well as punctually inceptive verbs, Arabic also pos-
sesses verbs which are duratively inceptive. The relevant part of Example (257) is
reproduced as (269a):

(269) a. ‫قد اسو ّدت وجوههم‬

qad iswadd-at wujūh-u-hum
pfv [become]_black;pst-3fsg face;pl-nom-poss.3mpl
‘Their faces have/had blackened’

Just as the event diagram for a punctual inceptive verb resembles a SEMEL-
FACTIVE followed by a state, the diagram for (269) resembles an ACTIVITY
followed by a STATE:
(269) b. RT ST
↓ ↓
ET: |---------------|-------------------------->
ONSET NUCLEUS

The onset is marked [+dynamic] as before, since any entry into a state neces-
sarily involves change of state, according to Passonneau’s (1988: 47) requirement
for ‘kinesis’, but is now also [+durative], while the nucleus is [+durative] but
 Chapter 10.  Inceptive aspect 225

[Ødynamic]­. Note that we must be careful not to impose any aspectual properties
of the alternative English translations of the verb (i)swadda (‫)اسو ّد‬. Whereas ‘be
pregnant’ is absolute, ‘be black’ may be viewed as relative: thus ‘become black’ is
perceived as a durative and continuous process, since, for example, the colour of
toast under a hot grill progresses smoothly through increasing degrees of black-
ness. In my analysis of the event diagram (269b), it is only possible to state (269c)
after completion of the onset, i.e. in nucleus, where the situation is potentially no
longer dynamic:

(269) c. ‫وجوههم سوداء‬

wujūh-u-hum sawdā’
face;pl-nom-poss.3mpl black;f
‘Their faces are black’

It is possible to conceive of situations in which (i)swadda (‫ )اسو ّد‬is consistent with

either a durative or a punctual onset, so it is properly designated [Ødurative] for on-
set. Note that I offer no argument that verbs with [+durative][+dynamic] marking
for onset cannot exist cross-linguistically, only that I have found no proof for their
existence in MSA at this time. Thus, I have chosen to categorise all Arabic verbs which
denote inception of state as INCEPTIVE, with the following combination of features:

Onset: [+dynamic] Nucleus: [+durative] Coda: Ø

This new category has been incorporated into a modified table of verbal catego-
ries, based on a triphasic system consisting of privative features (Table 71):

Table 71.  Olsen’s characterisation of Vendler’s categories extended for INCEPTIVES

CATEGORY ONSET NUCLEUS CODA
STATE Ø [+durative] Ø
ACTIVITY Ø [+durative] [+dynamic] Ø
ACHIEVEMENT Ø [+dynamic] [+telic]
ACCOMPLISHMENT Ø [+durative] [+dynamic] [+telic]
SEMELFACTIVE Ø [+dynamic] Ø
INCEPTIVE [+dynamic] [+durative] Ø

10.2.2  Other feature marking combinations

In this system, onset and coda phases both involve the presence or absence of
marking for single privative features, [+dynamic] and [+telic] respectively, while
nucleus must be marked with either or both of the privative features [+dynamic]
and [+durative]. Consequently, all verbs have a nucleus, but onset and coda ­phases
226 The Arabic Verb

are optional. However, one may postulate that other verb categories consistent
with the constraints thus far established are possible within the system (Table 72),
though we have already noted in the previous section that existence of onset and
coda for the same verb would be problematic.

Table 72.  Other possible feature marking combinations involving a marked onset

CATEGORY ONSET NUCLEUS CODA
[A]? [+dynamic] [+durative] [+dynamic] Ø
[B]? [+dynamic] [+dynamic] Ø
[C]? [+dynamic] [+durative] [+dynamic] [+telic]
[D]? [+dynamic] [+dynamic] [+telic]

We may speculate as to whether these as yet unattested categories of verb ex-

ist and, if so, what they might look like. If INCEPTIVES represent entry into a
STATE, then posited type [B] would represent entry into a SEMELFACTIVE. It
is, however, impossible to conceive of a punctual onset followed by a punctual
nucleus, resembling back-to-back SEMELFACTIVES, as effectively these punc-
tual events must merge into a single SEMELFACTIVE as in the event structure
diagram (270):
(270) ET: ||||
↑↑
ONSET NUCLEUS

The following example in English demonstrates that where a verb is restricted to

having only a true SEMELFACTIVE interpretation, inceptivity is not possible:
(271) *The runaway train began to hit the buffer.

Posited type [D] presents the same problem. Thus only a situation which includes
a durative nucleus can have a punctual onset in the system proposed. If [+dura-
tive] marking is possible for onset, it is conceivable cross-linguistically that verbs
could have durative onset and punctual nucleus, though I consider this unlikely
and moreover, as already stated, I have found no evidence for [+durative] onset
marking in Arabic.
This leaves posited types [A] and [C] in which the nucleus is doubly marked.
We have already noted that for INCEPTIVES the [+perfective] is drawn to the
onset, which is marked [+dynamic], focusing attention on the beginning of the
situation. However, as demonstrated by previous examples of verbs marked
[+telic] in coda, the [+perfective] is drawn to intersect with that coda, focus-
ing on completion. It is my contention, therefore, that a marked onset and a
 Chapter 10.  Inceptive aspect 227

marked coda are most likely mutually incompatible, at least in any language
which exhibits a marked perfective, including English and MSA. However, there
remains no a priori reason why type [A] cannot exist, representing entry into an
ACTIVITY.
At this point, it will be helpful to briefly examine the class of English verbs
which I have identified and categorised as INCEPTIVES, which are all ‘verbs of
posture’. The following examples and event time diagrams demonstrate how I ex-
plain their event time structure in terms of onset and nucleus phases. In (272a–b),
the [+perfective] verb references the onset and thus places the focus on the be-
ginning of the situation, i.e. the act of transition to a sitting position, while the
nucleus which follows represents the ensuing state which still pertains at ST.
(272) a. [+perfective]
Josh has sat down
b. RT ST
↓ ↓
ET: ||------------------------------------------------------------->
ONSET NUCLEUS

In contrast, the [+imperfective] verb in (272c–d) references the nucleus phase,

describing the durative state which necessarily ensues from the act of transition
which must have preceded it:
(272) c. [+imperfective] [+past]
Josh was sitting down
d. RT ST
↓ ↓
ET: ||---------------------------------------->
ONSET NUCLEUS

Thus, crucially, for this class of verb alone it is the nucleus and not a coda phase
which represents the resulting state and therefore it is possible to make a state-
ment such as (272e):
(272) e. Josh has sat down and he is now sitting down.

Where the verb is unmarked for grammatical aspect (272f), either event time
diagram is viable:
(272) f. [+past]
Josh sat down [when he entered the room] / [for the whole time he was
in the room]
228 The Arabic Verb

I provide a fully argued case for this class of verbs as INCEPTIVES in Danks
(2008). However, what is pertinent here is that the English progressive aspect con-
struction be + V-ing, which is marked [+imperfective] and thus references the
nucleus phase, is compatible with these verbs, as in (272c). Since the progressive
aspect is normally only compatible with verbs marked [+dynamic] in nucleus,
it suggests that English, somewhat counter-intuitively, treats durative situations
such as ‘sitting down’ and ‘standing up’ as dynamic rather than static, i.e. as AC-
TIVITIES not STATES, thus these ‘verbs of posture’ are properly INCEPTIVES
OF ACTIVITY.

10.2.3  Extended scheme for lexical aspect categories

Having established that INCEPTIVE OF ACTIVITY is a required lexical aspect

category for English, we must add it to our table of cross-linguistic categories.
Consequently we must strictly designate the class of Arabic inceptives we en-
countered earlier, where the nucleus is [Ødynamic], as INCEPTIVES OF STATE.
Table 73 details the privative feature combinations for all the lexical categories
which have been attested, according to my triphasic extension of Olsen’s (1997)
scheme.

Table 73.  Lexical aspect categories and features: extended triphasic system

CATEGORY ONSET NUCLEUS CODA
STATE Ø [+durative] Ø
ACTIVITY Ø [+durative] [+dynamic] Ø
ACHIEVEMENT Ø [+dynamic] [+telic]
ACCOMPLISHMENT Ø [+durative] [+dynamic] [+telic]
SEMELFACTIVE Ø [+dynamic] Ø
INCEPTIVE OF STATE [+dynamic] [+durative] Ø
INCEPTIVE OF ACTIVITY [+dynamic] [+durative] [+dynamic] Ø

10.3 Inceptive verbs in patterns III and VI

It will now be shown, using several specific examples from patterns III and VI,
that the category of INCEPTIVE OF STATE is applicable to the group of verbs
within these patterns which were identified as involving inception in the previ-
ous chapter.
Recall that many of the verbs assigned to this group involve entry by two or
more parties into a relationship which then persists, for example:
 Chapter 10.  Inceptive aspect 229

(273) ‫ عاهد على‬ [III]

‫ع‬āhada (+ d.o.) ‫ع‬alā (+ i.o.)
‘to promise (s.o.) to do (s.th.)’
The following examples illustrate that this verb is understood both in the sense
of entry into a promise or covenant and that of being in a state of covenant with
another party. Example (274a) is consistent with the event time diagram (274b) in
which the [+perfective] verb references the onset of the event, indicating entry by
the players into a promise or covenant relationship with respect to the president.
It is clear that a durative state of covenant subsequently exists between the parties,
represented by the nucleus phase, which is only bounded by the external event of
the conclusion of the championship (or at least of the players’ participation in it).
(274) a. [+perfective]
‫قد عاهد الالعبون رئيس االتحاد … على احراز البطولة العربية‬
qad ‫ع‬āhad-a l-lā‫ع‬ib-ūn ra’īs-a
pfv promise;pst-3msg def-player-mpl.nom president-acc
l-ittiHād … ‫ع‬alā ’iHrāz-i l-baTūla
def-union on winning[vn:iv]-gen def-championship
l-‫ع‬arabīya
def-Arabic
‘The players have promised the president of the union … that they will
win the Arab championship’ (arabiCorpus: 080999SPOR09)
b. RT ST
↓ ↓
ET: ||---------------------------------------------------->
ONSET NUCLEUS ↑
X: ‘conclusion of the championship’

In (274c) an iterative interpretation is unlikely and it is reasonable to suppose that

what is being described is a state of covenant between Britain and Sharif Hussein
which had been entered into at an unspecified time prior to RT, consistent with
the event time diagram (274d). Indeed it is the stative nature of the relationship
between the parties which renders my translation with ‘was promising’, which is
progressive and thus necessarily dynamic, somewhat clumsy. In the absence of a
verb in English which fulfils the dual function of entry into state and state itself, a
paraphrase might be preferable to maintain good English style.90

90. It was noted in Section 8.4.2.1 that there is no incompatibility between the Arabic [+imper-
fective] and verbs of STATE.
230 The Arabic Verb

(274) c. [+imperfective] [+past]

‫ففي الوقت الذي كانت فيه بريطانيا تعاهد الشريف حسين على تحقيق اإلمبرطورية‬
…‫العربية‬
fa-fī l-waqt alladhi kān- at
so-in def-time rel;msg be;pst-3fsg
fī-hā barīTānīyā tu‫ع‬āhid-u
in-obj.3fsg Britain promise;npst.3fsg-ind
sh-sharīf Hussayn ‫ع‬alā taHqīq-i
def-Sharif Hussein on realisation-gen
l-’imbaraTūrīya il-‫ع‬arabīya …
def-empire def-Arabic
‘So at the time when Britain was promising Sharif Hussein that the Arab
Empire would be established …’
 (http://www.ahewar.org/debat/show.art.asp?aid=25136,
 15 July2009)
d. RT ST
↓ ↓
ET: ||--------------------------------->
ONSET NUCLEUS

The following Example (275), although somewhat archaic, includes another pat-

tern III verb and conveniently contrasts state and entry into state in the same dia-
logue, with the same proviso concerning the use of the progressive in the English
translation:

(275) ‫ قد بايعت يا رسول هللا‬:‫ يا ابن األكوع أال تبايع قلت‬: ‫قال‬
qāl-a: yā ibn-a_l-akwa‫’ ع‬a-lā
say;pst-3msg O Ibn Al-Aqwac q-neg
tubāyi‫ع‬-u qul-tu: qad
pledge_allegiance;npst.2msg-ind say;pst-1sg pfv
bāya‫ع‬-tu yā rasūl-a llāh
pledge_allegiance;pst-1sg O messenger-acc God
‘He said, “O Ibn Al-Aqwac, are you not pledging allegiance?”
I said, “I have pledged allegiance, O messenger of God”’
 (http://www.rasoulallah.net/subject2.asp?hit=1&parent_id=11
 &sub_id=1381, 15 July 2009)
 Chapter 10.  Inceptive aspect 231

Examples (276a–d) include the pattern VI verb ‫( تعارف‬ta‫ع‬ārafa – ‘to get to know

each other/become mutually acquainted’):
(276) a. [+imperfective] [+past]
‫يجمعه مع ذريته وإخوانه… وال شك أنهم يتعارفون في الجنة كما كانوا يتعارفون في الدنيا‬
yajma‫ع‬-u-hu ma‫ع‬a
bring_together;npst.3msg-ind-obj.3msg with
dhurrīyat-i-hi wa-’ikhwān-i-hi …
descendants-gen-poss.3msg and-brother;pl-gen-poss.3msg
wa-lā shakka ’anna-hum
and-neg doubt that-obj.3mpl
yata‫ع‬āraf-ū-n fī l- janna
know_one_another;npst.3m-pl-ind in def-paradise
ka-mā kān-ū yata‫ع‬āraf-ū-n
like-what be;pst-3mpl know_one_another;npst.3m-pl-ind
fī d-dunyā
in def-world
‘He will bring him together with his descendants and his brothers …
and there is no doubt that they will know one another in paradise just
as they used to know one another in this world’
 (http://www.islamweb.net.qa/ver2/Fatwa/ShowFatwa.php?
 Option=FatwaId&lang=A&Id=29906, 15 July 2009)

b. RT ST
↓ ↓
ET: ||--------------------------------->
ONSET NUCLEUS

The context clearly indicates a stative sense of mutual acquaintance in (276a)

where the [+imperfective] verb references the nucleus phase. Note that the event
time diagram (276b) represents the [+past] second instance of the verb in the
sentence. Conversely, in the context of the following Example (276c), the [+per-
fective] verb unambiguously represents dynamic entry into a state of mutual ac-
quaintance which takes place within the closely defined time period specified by
‘the journey to Egypt’:
232 The Arabic Verb

(276) c. [+perfective] [+past]

‫قد تز ّوج االنكليزيان… بعد يوم واحد من وصولهما… وكانا قد تعارفا اثناء االنتقال الى مصر‬
qad tazawwaj-a l-’inglīzīy-ān …
pfv get_married;pst-3msg def-English-du.nom
ba‫ع‬da yawm wāhid min wuSūl-i-himā
after day one from arrival-gen-poss.3du
wa-kān-ā qad ta‫ع‬āraf-ā
and-be;pst-3mdu pfv get_to_know_each_other;pst-3mdu
’athnā’a l-intiqāl ’ilā maSr
during def-journey to Egypt
‘The English couple got married … one day after their arrival … having
got to know one another on the journey to Egypt’
(arabiCorpus: GEN1996: 12959)

d. RT ST
↓ ↓
ET: ||---------------------------------------------------->
ONSET NUCLEUS
X: |----------------------|
‘journey to Egypt’

Finally, I will present evidence in support of inceptivity in a verb which less obvi-
ously involves entry into a new stative relationship:

(277) ‫ حـاسـب‬ [III]

Hāsaba (+ d.o.)
‘to call (s.o.) to account / hold (s.o.) responsible’

There is a suggestion in Wehr’s (1994) entry for this verb that it may be used both
in the sense of a single instance of ‘calling to account’ and in that of a durative pe-
riod of ‘holding responsible’. However, we must ensure that this is not an artefact
of translation and that both senses are indeed possible in Arabic usage. Consider
the following sentences which were submitted to native speaker informants:

(278) a. ‫ حاسب أحمد المدير مرات كثيرة‬ {YYY}

Hāsab-a ’aHmad-u l-mudīr
call_to_account;pst-3msg Ahmad-nom def-director
marr-āt kathīr-a
time-pl many-f
‘Ahmad called the director to account many times’
(Questionnaire: 24)
 Chapter 10.  Inceptive aspect 233

b. ‫ حاسب أحمد المدير طوال سنين كثيرة‬ {Y?Y}

Hāsab-a ’aHmad-u l-mudīr
hold_responsible;pst-3msg Ahmad-nom def-director
Tiwāla sin-īn kathīr-a
during year-pl.gen many-f
‘Ahmad held the director responsible for many years’
(Questionnaire: 10)
c. ‫ كان أحمد يحاسب المدير طوال سنين كثيرة‬ {YYY}
kān-a ’aHmad-u yuHāsib-u
be;pst-3msg Ahmad-nom hold_responsible;npst.3msg-ind
l-mudīr Tiwāla sin-īn kathīr-a
def-director during year-gen.pl many-f
‘Ahmad held [was holding] the director responsible for many years’
 (Questionnaire: 43)
d. ‫ ما زال أحمد يحاسب المدير‬ {YYY}
mā zāl-a yuHāsib-u
neg cease;pst-3msg hold_responsible;npst.3msg-ind
’aHmad-u l-mudīr
Ahmad-nom def-director
‘Ahmad still holds the director responsible’  (Questionnaire: 30)
e. ‫ يحاسب أحمد المدير منذ الحادث‬ {Y??}
yuHāsib-u ’aHmad-u
hold_responsible;npst.3msg-ind Ahmad-nom
l-mudīr mundhu l-Hādith
def-director since def-incident
‘Ahmad is holding the director responsible since the incident’
(Questionnaire: 13)
f. ‫ حاسب أحمد المدير منذ الحادث‬ {NYY}
Hāsab-a ’aHmad-u l-mudīr
call_to_account;pst-3msg Ahmad-nom def-director
mundhu l-Hādith
since def-incident
‘Ahmad called the director to account [held the director responsible]
since the incident’ (Questionnaire: 16)
g. *‫ يحاسب أحمد المدير حتى نهاية األزمة‬ {N?N}
yuHāsib-u ’aHmad-u
hold_responsible;npst.3msg-ind Ahmad-nom
l-mudīr Hattā nihāyat-i l-’azma
def-director until end-gen def-crisis
‘Ahmad is holding the director responsible until the end of the crisis’
 (Questionnaire: 18)
234 The Arabic Verb

Example (278a), which was acceptable to all three informants, is unambiguously

iterative due to the adverbial phrase, hence my translation with ‘called to account’.
However, despite the temporally extensive adverbial in (278b), in which the verb
is unmarked for grammatical aspect, and in (278c), in which it is marked [+im-
perfective], these sentences are both capable of interpretation either as a continu-
ous state of ‘holding responsible’ or iterative events of ‘calling to account’. Thus in
the remaining examples I have attempted to minimise ambiguity by combining
the verb with different adverbial phrases, though with limited success, in part due
to the lack of agreement between informants except for Example (278d) which
remains somewhat ambiguous. Example (278e) shows some evidence for a stative
interpretation, though it was not universally acceptable and somewhat surpris-
ingly the informant who rated it as fully grammatical disagreed with the other
two informants on the equivalent sentence containing the s-stem verb (278f).
Moreover the responses for (278g) appear incompatible with those for (278e). I
am therefore forced to conclude that the evidence from native speakers is unreli-
able and inconclusive for this verb. However, although it is difficult to find unam-
biguous usages in the corpus, in the following Example (279) the context gives a
strong indication that the [+imperfective] verb is to be understood in a durative
rather than an iterative sense:
(279)
[+imperfective] [+past]
‫قال صدام خالل الحديث انه كان يحاسب أسرته أكثر من اآلخرين‬
qāl-a Saddām khilāla l-Hadīth

say;pst-3msg Saddam during def-interview
’anna-hu yuHāsib-u
that-obj.3msg hold_responsible;npst.3msg-ind
’usrat-a-hu ’akthar min al-’ākhar-īn
family-acc-poss.3msg much;cmpr from def-other;mpl-gen

‘Saddam said during the interview that he held [was holding] his family
responsible more than the others’
 (http://www.paleyad.com/vb/showthread.php?p=45299, 16 July 2009)

Unlike the previous examples, where the nucleus phase clearly represents a state,
it is possible that the situation described by the nucleus phase of this and other
similar verbs may be dynamic, in which case they are INCEPTIVES OF AC-
TIVITY. As previously discussed, in the absence of an equivalent of the pro-
gressive aspect which is such a useful tool for ascertaining dynamicity in Eng-
lish, it would be necessary to test demonstrably durative examples of each verb
with a range of adverbials such as ‘deliberately’, ‘carefully’ or ‘gradually’, none
of which are compatible with static situations. This might be a fruitful area for
further research, though at present it will suffice to conclude that these verbs are
 Chapter 10.  Inceptive aspect 235

I­ NCEPTIVES, whether of ACTIVITY or of STATE. As such, we note that they

are [+dynamic] in onset, [+durative] and possibly [+dynamic] in nucleus, but
unmarked for telicity in coda.

10.4 Summary

We have demonstrated that it is necessary to establish a category of INCEPTIVE

verbs cross-linguistically and have further subdivided this category into INCEP-
TIVES OF STATE and INCEPTIVES OF ACTIVITY. Each of these lexical as-
pect categories has been described in terms of the privative features elaborated
by Olsen (1997), but within a triphasic scheme which includes an optional onset,
obligatory nucleus and optional coda, the presence of an onset only being permit-
ted when the coda is absent and vice-versa.
The pattern III and pattern VI verbs which were identified in the previous
chapter as having meanings involving dynamic entry into durative situations, and
were thus considered anomalous within the existing scheme, have been explained
satisfactorily within the newly defined INCEPTIVE categories. As such, they have
been found to be unmarked for telicity in coda, i.e. [Øtelic].
This completes the work of categorising C1āC2 sequence verbs for lexical as-
pect: conclusions will be drawn in Chapter 12.
 chapter 11

The passive in patterns III and VI

Several times in earlier chapters the question of passivisability, especially for

verbs in patterns III and VI, has been raised. In Section 7.1, it was noted that the
only pattern III word form without long ā is the s-stem passive verb (C1ūC2iC3)
and the same is true of pattern VI (tuC1ūC2iC3). We therefore concluded in Sec-
tion 7.2.2.1 that “it is specifically the long ā which is basic to the pattern”. Here
I will seek to question this conclusion, since, if it is valid, we might expect a
degree of incompatibility of these patterns with passivisation by vowel melody
change. Furthermore, it was suggested in Section 7.3.2.1 that, due to the mutual-
reciprocal­ semantics common to these patterns, there is often no true patient,
even when the verb is formally transitive. This precludes both actional passive
formation by vowel melody change and statal passives expressed with the pas-
sive participle. However, does the reluctance to passivise extend to other verbs
in these patterns also?
These are the questions which I propose to investigate briefly in this chapter,
indicating possible avenues for further research where appropriate. In addition to
issues concerning passivisation, I will also comment on the verbal noun forms for
patterns III and VI.

11.1 Passive formation by vowel melody change

Recall that in Section 3.2.1 we introduced the concept of three morphemic tiers

in Arabic word formation. The templatic tier, whether treated as a CV skeleton
or as a prosodic template, provides the base structure for a given verbal pattern,
into which root consonants are inserted together with a vowel melody, i.e. the se-
quence of vowels which produces a fully vocalised stem. Recall also that the ‫ضـ ّمـة‬
(Damma) or u vowel is characteristic of the passive, occurring in the sequence u‑i
for the s-stem and u‑ a for the p-stem verb. However, in the derived patterns there
may be three or even four syllables for which vowels must be supplied and, since
the vowel melodies for s- and p-stems propagate differently across the template,
the resulting forms for patterns III and VI are as shown in Table 74.
238 The Arabic Verb

Table 74.  Pattern III and pattern VI passive verb forms

s-stem passive p-stem passive
pattern III C1ūC2iC3 yuC1āC2aC3
pattern VI tuC1ūC2iC3 yuta C1āC2aC3

Thus it is in the s-stem verb alone in both patterns III and VI that the long vowel
between C1 and C2 is ū. Although I have already presented evidence from nomi-
nal forms in Chapter 7 which suggests that it is specifically the ā vowel, resulting
from the combination of the active vowel melody with the template, which is
characteristic of verbs in patterns III and VI, by examining the compatibility of
verbs in these patterns with the actional passive, we may be able to investigate
further whether it is the specific vowel or vowel lengthening in general which is
important.

11.1.1 Corpus data evidence

Unlike the other verbal patterns, in which the passive vowel melody is indistin-
guishable from its active counterpart in unvowelled text, the vowel lengthen-
ing in patterns III and VI conveniently results in an orthographic distinction.
Thus, for example, we may use the corpus search string ‫( كـوتـب‬kwtb) to return
all instances of the passive s-stem for the verb ‫( كـاتـب‬kātaba) and similarly for
the pattern VI equivalent. Thus the corpus token counts in this section were all
obtained by searching the arabiCorpus newspaper corpus91 for the appropriate
s-stem passive forms.
The verbs in Examples (280)–(284) all returned zero token counts:

(280) ‫ كـاتب‬ [III]

kātaba (+ d.o.)
‘to correspond with (s.o.)’
(281) ‫ قـاتل‬ [III]
qātala (+ d.o.)
‘to fight [with] (s.o.)’
(282) ‫ سـافر‬ [III]
sāfara
‘to travel, set out on a journey’

91. Accessed 23–24 July 2009.

 Chapter 11.  The passive in patterns III and VI 239

(283) ‫ تـكـاتب‬ [VI]

takātaba
‘to correspond together’
(284) ‫ تـقـاتل‬ [VI]
taqātala
‘to fight together’
Clearly the last three examples are intransitive, thus passivisation is impossible,
but the zero token counts for Examples (280)–(281) provide evidence that transi-
tive verbs of this type, where the direct object is a co-participant in the action and
hence not patient, also do not form an actional passive. In Section 5.2.2.1, recog-
nising that the entities represented by grammatical subject and direct object may
participate in the action described by a verb asymmetrically whilst still involving
a degree of mutuality, I classified the verb in (285) as mutual:
(285) ‫ عـاون في‬ [III]
‫ع‬āwana (+ d.o.) fī (+ i.o.)
‘to help, assist (s.o.) in/with (s.th.)’

On the basis of the English translation, we might expect this verb to passivise
readily, but the corpus count again returned zero tokens, suggesting that it also
resists formation of the actional passive.
In Section 7.3.1.1 it was noted that, recent developments in the language ex-
cepted, the actional passive construction in MSA does not permit the type of
agentive by-phrase which is optionally present in the English passive construc-
tion. Thus the Arabic actional passive construction does not merely bring the
patient into the foreground relative to the agent by making the patient subject, it
removes reference to the agent altogether. In terms of valency, one of the verbal
arguments is deleted, such that the numeric valency of the verb in the actional
passive is reduced by one.
It appears, therefore, that it is this deletion of the agent which is resisted by
verbs such as Example (285), consistent with the emphasis on agency and process,
as opposed to patiency and result, noted in Section 7.3.2 both for pattern III and
pattern VI verbs and for noun templates containing long ā. Resistance to forming
an actional passive also extends to many transitive verbs in these patterns with no
obvious implication of participation in the action on the part of the direct object.
Thus (286), previously encountered in Section 9.1.2.4.2, also returned a corpus
token count of zero.

(286) ‫ نـاهـز‬ [III]

nāhaza (+ d.o.)
‘to attain, reach, seize (s.th.)’
240 The Arabic Verb

However, it is not true to say that actional passives are completely excluded for
transitive pattern III verbs. The verb in Example (287a) was used to illustrate verbs
of surprise in Section 9.1.2.1 and is very commonly encountered in the actional
passive form, returning 2,652 corpus tokens. It is perhaps noteworthy, however,
that the equivalent of the English paraphrase using the active in (287b), in which
the subordinate clause is subject, is not viable in Arabic, rendering use of the pas-
sive unavoidable.
(287) a. ‫فوجئت بأن عددا كبيرا من المصريات لم يذهبن الى الهرم‬
fūji’-tu bi-’anna ‫ع‬adad-an
surprise;pst.pass-1sg with-that number-acc.indf
kabīr-an min al-maSrī-āt lam
large-acc.indf from def-Egyptian-fpl neg
yadhhab-na ’ilā l-haram
go;npst;3-fpl to def-pyramid
‘I was surprised that a large number of Egyptian women did not go to the
pyramid’ (arabiCorpus: 021799AMOD02)
b. ‘That a large number of Egyptian women did not go to the pyramid
surprised me’

11.1.2 Native speaker evidence

Although investigation of passivisability was not the primary purpose when ob-
taining questionnaire responses from native speaker informants, some of the data
collected will be of interest here. The responses of native speakers for six transitive
verbs which we have previously encountered in other examples are summarised
in Table 75. The sentences used to elicit the responses may be consulted in Appen-
dix III. Note that (Questionnaire: 21) largely confirms the findings for the same
verb in (287a) above. However, it is clear that acceptability of the actional passive
is highly variable, both between different verbs and for different speakers assess-
ing the same verb. If passivisability by this mechanism were a formal property of
the patterns, we might expect a greater degree of unanimity between informants.

Table 75.  Actional passive: Native speaker informant responses

verb transliteration translation responses questionnaire reference
‫بارك‬ bāraka ‘to bless’ {YNY} 44
‫جاور‬ jāwara ‘to adjoin’ {NYN}  3
‫جاوز‬ jāwaza ‘to surpass’ {NNN} 14
‫حاسب‬ Hāsaba ‘to hold responsible’ {YYY} 52
‫فاجأ‬ fāja’a ‘to surprise’ {Y?Y} 21
‫قاتـل‬ qātala ‘to fight’ {?N?}  5
 Chapter 11.  The passive in patterns III and VI 241

11.1.3  Conclusion

Although we may conclude from the corpus data that there is evidence that most
pattern III and pattern VI verbs do not readily form an actional passive, this con-
struction is not only allowable but common for a minority of verbs. On the basis
of this data and incorporating the limited native speaker responses I have gath-
ered, I tentatively conclude that actional passive formation is most likely a mat-
ter of the semantics of the individual verb, rather than a formal property of the
patterns themselves. However, this does not preclude the possibility that further
research into passivisability by the mechanism of vowel melody might uncover
other properties shared by verbs which either do or do not enter into this con-
struction. That the ū vowel itself is not entirely incompatible with the patterns
suggests that it is vowel lengthening rather than vowel identity in the pattern tem-
plates which is indicative of atelicity, a matter which will be discussed further in
the concluding chapter.

11.2 Passive participle formation

The passive participles for patterns III and VI take the forms muC1āC2aC3 and
mutaC1āC2aC3 respectively. As such, they exhibit the long ā which we have come
to associate with agency and process.92 Since the nature of the passive participle
or ‘done’ form is that it designates patiency and result, and if our analysis thus far
is correct, this raises the possibility of a conflict between form and meaning. We
will therefore examine evidence regarding passive participle formation and usage
for patterns III and VI.

11.2.1  Dictionary evidence

Entries for all pattern III and pattern VI verbs attested in Wehr (1994) were ex-
amined and instances of the corresponding passive participle were recorded. We
must proceed with caution in assessing this data, since Wehr (1994: xiii) states
that “participles … are listed as separate items only when their meaning is not
immediately obvious for the verb, particularly where a substantival or adjectival
translation is possible”. However, the frequency with which the passive participles
of these patterns are listed is remarkably low compared, for example, with pattern
II passive participles or indeed with pattern III and pattern VI active participles.

92. See Section 7.3.2.1.

242 The Arabic Verb

Ryding (2005: 209, 549) confirms their infrequency, all her examples being among
those counted in Table 76 and listed in Appendix IV.

Table 76.  Pattern III and pattern VI passive participles

Pattern III Pattern VI
passive participles   15    6
total verbs 465 389
percentage 3.2% 1.5%

Of the 21 attested word forms matching the templates, many are atypical in some
way of passive participles generally. Example (288) has a specialised meaning
which suggests use of a lexical gap for a recent coinage:

(288) ‫ مـعـاقـَب‬ [ppt:III]

mu‫ع‬āqab
‘suspended [player/club (sport)]’ [verb: ‘to punish (s.o.)’]

The grammatical term in Example (289) is also specialised and, although its prov-
enance from the verbal meaning is clear, it may notably only be used as a noun,
having no adjectival resultative usage.

(289) ‫ مـنـادى‬ [ppt:III]

munādan
‘noun in the vocative’ [verb: ‘to call out (s.th.)’]

Also used exclusively nominally, Examples (290) and (291), whilst undeniably

representing patients of the corresponding verbs, are effectively agentive nouns
of occupation:

(290) ‫ مـشـا َور‬ [ppt:III]

mushāwar
‘advisor/consultant’ [verb: ‘to seek advice from / consult (s.o.)’]
(291) ‫ مـيـا َوم‬ [ppt:III]
muyāwam
‘day labourer’ [verb: ‘to hire by the day’]

However, the following Example (292) not only has an active sense but represents
performance of the corresponding verb itself, which, being intransitive in this
usage, defies definition of any patient or result upon it, suggesting that the active
participle would be a more appropriate form:
 Chapter 11.  The passive in patterns III and VI 243

(292) ‫ مـتـدا َول‬ [ppt:VI]

mutadāwal
‘current/circulating’ [verb: ‘to be current, in circulation’]

It is unclear for Examples (293) and (294) exactly how they represent the result of
performing the corresponding verb, while clearly sharing a root meaning with it:

(293) ‫ مـتـقاضـى‬ [ppt:VI]

mutaqāDan
‘subject to prosecution’ [verb: ‘to carry on a lawsuit (together)’]
(294) ‫ مـتـنـا َول‬ [ppt:VI]
mutanāwal
‘attainable / within reach’ [verb: ‘to reach for (s.th.)’]

However, Example (295a) is probably unrelated to the corresponding pattern

III verb, instead sharing a basic meaning with several nominal/adjectival forms
(295b) which appear to be derivatives of an absent pattern I verb:

(295) a. ‫ مـوا َرب‬ [ppt:III]?

muwārab
‘ajar’ [verb: ‘to double-cross, outsmart (s.o.)’]
b. ‫ وراب‬ ‫ ورب‬ ‫ موروب‬
wirāb warb mawrūb
‘obliqueness’ ‘obliqueness/slant’ ‘slanting/ajar’

11.2.2  Native speaker evidence

With the aforementioned proviso that an investigation into passivisability was

not the primary aim of conducting native speaker questionnaires, the data which
were obtained are presented in Table 77:

Table 77.  Passive participle: Native speaker informant responses

Verb Transliteration Translation Responses Questionnaire reference
‫بارك‬ bāraka ‘to bless’ {?Y?} 36
‫جاور‬ jāwara ‘to adjoin’ {Y?Y}  7
‫جاوز‬ jāwaza ‘to surpass’ {YNY} 11
‫حاسب‬ Hāsaba ‘to hold responsible’ {?Y?} 38
‫فاجأ‬ fāja’a ‘to surprise’ {NNN} 46
‫قاتـل‬ qātala ‘to fight’ {YYN} 20
244 The Arabic Verb

Once again, the response from native speaker informants is confused, with high
variability between informants for the same test sentences. The only passive par-
ticiple which elicited consistent results was that of the verb ‫( فاجأ‬fāja’a), used as
an example of verbs of surprise in Section 9.1.2.1. Some comments follow on the
variability of responses:
1. Because of the highly systematic nature of Arabic morphology it is possible
for a native speaker (or indeed any learner with knowledge of the templates)
to reproducibly generate words which they have never encountered in real
language.93 Thus since all the passive participles presented in the question-
naire were generated by me from actual verbs according to the appropriate
template, the reaction of native speakers may err towards considering them
valid words, even if their semantics suggest that their use might be problem-
atic. If so, it is possible that native speakers (or at least those surveyed) are
insufficiently critical of well-formed but semantically doubtful words for their
responses to be helpful. Further testing with more contextually embedded
examples might alleviate this problem.
2. Native Arabic speakers are accustomed to reading unvowelled texts. The pas-
sive participles in the questionnaire were specifically marked with critical
vowelling to avoid confusion with the active participles which differ only by
one vowel. However, informal discussion with one of the participants sub-
sequent to processing the questionnaire results revealed that he may have
unconsciously ‘corrected’ my vowelling and mentally substituted the active
participle. Whilst it was too late to correct the methodology in this study, I
would suggest that any future investigation might be less subject to error if the
examples were presented orally, where the vowel alternation would be more
readily apparent.
3. Possibly related to the previous point, when asked what the meaning of the
passive participle of ‫( جاور‬jāwara – ‘to adjoin’) might be, at least one infor-
mant was unable to separate it semantically from the active participle. In any
further survey, it is suggested that participants might be asked to define the
words being tested.
4. Of the passive participles tested here, only that from the verb ‫( بارك‬bāraka),
which was investigated in Section 9.1.2.4.2, is actually listed by Wehr (1994).
The passive participle ‫( مـبـا َرك‬mubārak), being the name of the former Presi-
dent of Egypt, is a word which will sound familiar to Arabic speakers. Thus the
finding that two informants found its use unacceptable may be more a matter
of the context in which it was presented than its acceptability as a valid form.

93. See also Cuvalay-Haak (1997: 89).

 Chapter 11.  The passive in patterns III and VI 245

Thus data from native speaker informants are inconclusive and further research
with a focus on the passive participle and improved methodology is indicated.

11.2.3  Corpus data evidence

It is unfortunately impossible to distinguish in an unvowelled text between the

majority of past participle forms in patterns III and VI and the corresponding
active participles, since they differ only in the vowelling on C2. The exception
to this generalisation is where C3 is weak as in (289) and (293), where the pas-
sive participle is written with final alif maqSūra (‫)ى‬. However, even a search for
this kind of passive participle is problematic, since many Arabic newspapers and
other printed sources fail to distinguish systematically or consistently between
alif maqSūra and final yā (‫)ي‬, which risks confusion with the definite form of the
corresponding active participle, and there is also a broken plural form from a base
pattern noun with weak final consonant which may be similarly confused. Thus it
is not a simple matter to conduct an exhaustive corpus search.
Data obtained from arabiCorpus do reveal that Examples (289) and (293),
while attested in the dictionary, are absent from the corpus.94 However, while I
am convinced that the majority of the matches returned by corpus searches for
passive participle strings are actually returning active participles, the sheer vol-
ume of data makes it impossible to examine each instance individually in context,
which is the only recourse in the absence of formal orthographic distinction. For
example, for the verb ‫( جاور‬jāwara), a search for the string ‫( مجاور‬mjāwr) returned
6,898 tokens: even after discounting over 6,000 instances of the pattern III verbal
noun, which also contains the string, this leaves over 800 instances which could
be of either participle.

11.2.4  Conclusion

Passive participle forms from patterns III and VI are undoubtedly rare. Where
they do occur, many appear to have a meaning which is compatible in some way
with agency. Further research with a specific focus on the passive participle and
a targeted methodology is suggested. However, preliminary study indicates that
instances of conflict between formal realisation with ā, which has otherwise been
shown to have connotations of agency and process, and resultative passive mean-
ing are minimal.

94. Data in this section obtained from arabiCorpus newspaper sub-corpus, searched 27–28
July 2009.
246 The Arabic Verb

11.3 Verbal noun forms

Although not directly related to passivisation, I will briefly discuss the verbal noun
forms here, both because of the formal similarity of two of the three templates to
the passive participles and because this is another area in which further research
is suggested, since there is an interesting form-meaning interaction which bears
investigation.
In Section 7.3, four verbal noun forms are listed for pattern III, only two of
which, muC1āC2aC3a and C1iC2āC3, are in common use, with 263 and 64 diction-
ary entries respectively, including 46 where both exist for the same verb, while
the only pattern VI verbal noun form is mutaC1āC2aC3a. Thus the most common
form for pattern III and the only form for pattern VI closely resemble the cor-
responding passive participles, differing only in that the verbal nouns show femi-
nine inflection. Since the verbal noun denotes performance of the action of the
verb or an instance thereof, the notion of long ā as denoting agency and process
is entirely compatible. The obvious Saussurean research questions are whether
there is any semantic significance in the form which the pattern III noun takes for
a given verb, any correlation with other properties of the verb and, if so, whether
any light may thus be shed on the significance of the position of long ā in the tem-
plate. Put simply, one verbal noun retains the C1āC2 sequence of the correspond-
ing verb and the other does not and Saussurean structuralism dictates that this
alternation be accompanied by an accompanying alternation in meaning, though
Ryding (2005: 506) allows that where they coexist, they may have “either equiva-
lent or slightly different meanings”, while Badawi et al. (2004: 79) state that “the
principle of selection is not clear and is best left as a lexical item”. The question
of the alternate forms is beyond the scope of the present research, thus further
investigation is required.

11.4 Summary

Whilst both actional and statal or resultative passive constructions seem to sit
somewhat uneasily with the pattern III and pattern VI verbs, for reasons which
have been outlined above, further research is needed into passivisability for these
patterns. However, we can proceed to draw final conclusions in the following chap-
ter on the following basis: it is vowel lengthening in the C1vC2 sequence which
is characteristic of patterns III and VI, and hence the formal marker of atelicity;
active vowelling with ā shows a far greater degree of compatibility with these pat-
terns as a whole, thus passive ū is rarely encountered and the association of ā with
agency and process seen in nominal forms is maintained in most verbal usage.
 chapter 12

Conclusions

12.1 Overview

From the outset, it was noted that Arabic possesses a highly systematised mor-
phology, most clearly observed in its verbal patterns, which is described and its
instances enumerated in Chapter 2. From a Saussurean perspective, a language
which specifies morphological form so clearly and rigidly is an ideal candidate for
investigation of form-meaning relationships. Data were presented demonstrating
that the distribution of verb forms by pattern is not random and that patterns
morphologically derived from one another have a tendency to co-occur for the
same root. However, the nature of Arabic morphology is also that it is highly com-
plex and multi-layered, challenging traditional descriptions of the morpheme. We
observed in Chapter 3 that there is a lively debate in progress concerning the ba-
sis of derivation in Arabic, but concluded that there is evidence that morphemic
input into a given word form takes place at three levels: root, prosodic or CV
skeleton, and vowel melody (McCarthy & Prince 1990a). Thus formal morphemic
description is achievable.
However, turning to semantics, it became clear in Chapter 4 that while at-
tempts have frequently been made to characterise the meanings of the verbal pat-
terns, no definite conclusions have been drawn which suggest that the Saussurean
systematic link between signifiant and signifié has been firmly established for
any of the 19 patterns available to triliteral and quadriliteral roots. In Chapter 5,
­examining the morphologically related patterns III and VI, the most consistently
recognised meanings of mutuality and reciprocity were effectively presented as
accepted thesis. However, although this constitutes the most successful expla-
nation of the meaning of these patterns to date, between a quarter and a third
of these verbs were shown to be semantically antithetical: lexical exceptions to
­mutual-reciprocal meaning.
Aware that transitivisation and detransitivisation (or more properly valency
change) have been suggested as derivational properties elsewhere in the verbal
system, a hierarchical valency approach was developed in Chapter 6 to examine
the ta- prefix morpheme which, amongst other pairs, derives pattern VI from pat-
tern III. Despite the success with which this morpheme was shown to reduce tran-
sitivity, a similar approach found no evidence that pattern III derivation from the
base meaning of pattern I shows any consistent component of valency change.
248 The Arabic Verb

Noting that formal parallels have been drawn between pattern III (and VI)
vowel lengthening and that seen in some broken nominal plurals, the notion of
verbal plurality was explored in Chapter 7. More promising, however, was the
formal similarity noted in numerous nominal templates, suggesting that the long
ā characteristic of pattern III and pattern VI verbal forms is associated elsewhere
in the language system with agentivity, process and temporal complexity. This led
us to posit that lexical aspect might be key to understanding and characterising
patterns III and VI. Thus, after a consistent and workable model of aspectuality
was established in Chapter 8, a detailed examination of lexical aspect in patterns
III and VI was undertaken in Chapter 9.
Having established that a new lexical aspect category of INCEPTIVE is re-
quired to explain a significant number of Arabic verbs, including many in pat-
terns III and VI, our model of aspectuality was further developed in Chapter 10,
and it was discovered that English also possesses a restricted set of verbs of pos-
ture which must properly be described as INCEPTIVE. Finally, we have been
treating the long ā vowel as a single morph, whereas templatic/prosodic theory
suggests that it is actually the surface realisation of two morphs on different
tiers, namely the template containing the Cvv first syllable which contributes
vowel lengthening and the active vowel melody which contributes the specific
vowel identity. Thus in Chapter 11 we briefly explored data on passivisability in
order to investigate whether there is any inconsistency in regarding the long ā,
a product of the combination of vowel lengthening with active vowel melody, as
a specific characteristic of patterns III and VI in contrast with vowel lengthen-
ing alone.
It will now be appropriate to draw specific conclusions concerning form and
meaning in verbal patterns III and VI on the basis of the evidence presented in
the preceding chapters and finally to suggest some possible implications for the
language system of MSA as a whole and further avenues of research.

12.2 The characteristics of patterns III and VI

12.2.1  Form

We may now formalise the conclusions drawn from the evidence concerning
form presented in Chapter 11 and place them in the context of prosodic templatic
morphology, isolating the templatic contribution of vowel lengthening from the
identity of the vowel itself.
 Chapter 12.  Conclusions 249

1. The formal characteristic of Arabic verbal pattern III is a triconsonantal

templatic morph with a bimoraic (heavy) first syllable of the type Cvv,
manifested with the specific vocalisation supplied by the appropriate vowel
melody, including lengthening of the vowel following the first consonant
supplied by the root.
2. The formal characteristics of Arabic verbal pattern VI are the triconso-
nantal templatic morph of pattern III and additional prefixed ta- morph,
manifested with the specific vocalisation supplied by the appropriate vowel
melody, including lengthening of the vowel following the first consonant
supplied by the root.

12.2.2  Meaning

An interim summary of telicity in patterns III and VI was presented in Table 70 in

Section 9.2, pending further investigation of verbs identified as potentially incep-
tive. We are now in a position to incorporate the findings of Chapter 10 into that
data summary, noting from Table 78 that, whatever measure is employed, char-
acterisation of patterns III and VI as atelic has a success rate of over 99%. To state
this in terms of lexical exceptions, only five true exceptions are recorded, only one
of which is demonstrably in contemporary usage.

Table 78.  Final categorisation of pattern III and pattern VI verbs by telicity

[Øtelic] [+telic] excluded
dictionary count ,    945 (99.2%)    5 (0.5%)    3 (0.3%)
corpus type count ,    658 (99.5%)    1 (0.2%)    2 (0.3%)
corpus token count 184,739 (99.8%) 108 (0.1%) 193 (0.1%)

Thus, to state our conclusions regarding meaning simply:

1. The form of Arabic pattern III verbs gives rise to atelic meaning.
2. That component of the form of Arabic pattern VI verbs which is common to
pattern III likewise gives rise to atelic meaning.

12.2.3  Relating form and meaning

Combining the preceding statements on form and meaning, isolating the specific
formal characteristic concerned and unifying reference to patterns III and VI:
250 The Arabic Verb

The Arabic triconsonantal verbal templatic morph with a bimoraic (heavy)

first syllable of the type Cvv gives rise to atelic meaning.

This is the irreducible finding of this research: we may wish to phrase it less tech-
nically for the non-Arabist:
The Arabic vowel lengthening verbal patterns give rise to atelic meaning.

12.3 Directions for further research

The following topics are presented as areas for further investigation which re-
late directly to aspect and the vowel-lengthening patterns. Other avenues for
research, for example extension of the hierarchical valency reduction scheme
(Chapter 6) to the other ta-prefixed patterns V and QII, have been suggested and
noted in passing.

12.3.1  Nominal aspect

Aspect is a cross-linguistic grammatical category applicable to verbs and as such

has been the topic of several of the preceding chapters. However, we noted in
Chapter 7 that the striking formal similarity between the Arabic atelic verbal pat-
terns III and VI and a number of nominal patterns within the language system
is accompanied by commonalities of meaning. Whilst it is not unexpected that
nominal forms which are actually verbal participles carry an aspectual burden,
and indeed in English grammatical aspect is expressed in the form auxiliary +
participle, there appears to be an extension of the concept of aspect in Arabic into
other nominal forms, such that those which are formally similar to verbal pattern
III evoke a sense of atelicity, imperfectivity (or temporal complexity) and dura-
tivity, for example the nouns of instrument and occupation encountered in Sec-
tion 7.3.1. However, there are also cross-linguistic implications: if lexical aspect
is a valid nominal category in Arabic, might it be applicable to other languages,
even to those such as English where lexical aspect rarely has morphological reali-
sation? For example, taking the near-synonymous terms in (296a−c), might their
meanings be nuanced by the understanding that the first displays agency and the
second patiency whilst the third is neutral?
(296) a. student
b. trainee
c. pupil
 Chapter 12.  Conclusions 251

Furthermore, there is an open-endedness about (296a) which suggests atelicity,

whereas the clear difference between Examples (297a–b) demonstrates that the
situation represented by (296b) is telic, i.e. it is temporary, pending the endpoint
of completion of the training.
(297) a. The patient was attended by a trainee nurse.
b. The patient was attended by a trained nurse.

I therefore envisage research which explores the validity of morphologically ex-

pressed lexical aspect in the Arabic noun and, if fruitful, extends the concept of
aspect in the noun cross-linguistically.

12.3.2  Aspect and passivisability

Although the need for further research into compatibility of patterns III and VI
with both actional and resultative passives was noted in Chapter 11, there are
wider implications both for the language system of MSA and cross-linguistically.
Concerning the passive in English, German and Russian, Beedham (2005: 57) con-
cludes “that the passive is an aspect”, by which he means a grammatical aspect.
Whilst we have observed an interaction in MSA between verbal lexical aspect
and passivisability, are there grounds for considering either or both of the vowel
melody actional passive and the participial resultative passive to be manifestations
of grammatical aspect? In examining the resultative passive, we may wish to re-
member that the Arabic passive participle is essentially the ‘noun of done’, with the
implication of telicity which that carries. Having already established the usefulness
of Olsen’s (1997) scheme for explaining the interactions between lexical and gram-
matical aspect, I would wish to conduct any examination of the Arabic passive
as aspect in the same theoretical context, conscious, moreover, that if Beedham’s
claims are valid then any application of Olsen’s scheme may reasonably be expect-
ed to be capable of explaining the passive in the languages he has studied also.

12.3.3  Inceptivity of state and of activity

It was demonstrated in Chapter 10 that both INCEPTIVES OF STATE and IN-

CEPTIVES OF ACTIVITY are predicted by my extension to Olsen’s (1997) lexi-
cal aspect categories. I have shown that the former are present in Arabic and the
latter in English, and, whilst allowing that Arabic may exhibit both categories,
further research which establishes a conclusive test to distinguish ACTIVITY
from STATE is needed to confirm or deny this. Furthermore, those with expertise
in other languages may wish to investigate which of these categories, if either, is
252 The Arabic Verb

attested elsewhere and indeed whether there is any evidence that there is a cross-
linguistic need to introduce DURATIVE INCEPTIVE categories, i.e. for verbs
marked [+durative] in onset.

12.3.4  Alternative verbal noun forms

In that form gives rise to meaning, the formation of verbal nouns from pat-
terns I – III according to alternative templates95 is a clear invitation to research the
meanings of these templates. Since pattern I verbs give rise to a plethora of differ-
ent and diverse verbal noun forms, and since the semantics of pattern III have al-
ready been thoroughly investigated, I would propose starting with pattern III and
then applying any insights gained to the other patterns in turn. Since the only two
common templates for pattern III verbal nouns differ according to whether the
long ā vowel is in C1āC2 or C2āC3 position, a possible avenue to explore is whether
there are parallels in other nominal forms which show this alternation.96

12.3.5  Defining atelicity

Whilst the evidence leading to the conclusion that vowel lengthening in the ver-
bal patterns corresponds with atelic meaning is overwhelming, there remains an
inconsistency in our analysis: specifically, what we have discovered is not that
verbs in these patterns are atelic but that they show absence of [+telic] privative
feature marking. Recall that in Olsen’s (1997) scheme the opposition is between
[+telic] privative feature marking and [Øtelic] zero marking and not between the
equipollent features [+telic] and [−telic] (or [+atelic]). Olsen simply does not al-
low in her analysis that atelicity may be a marked feature. How, then, do we re-
solve this inconsistency?
The obvious solution is to assert that vowel lengthening in patterns III and VI
corresponds with [Øtelic] meaning. However, this is problematic for two related
reasons: the first stems from markedness theory and the second from Saussurean
structuralism. It is clear that pattern III is formally marked relative to the base
pattern I, i.e. it is more complex morphologically, specifically on account of vowel
lengthening in the first syllable of the s-stem.97 Frequency data for pattern III

95. Wright (1967: I.116) notes rare alternatives for other patterns also.
96. See the tables in Section 7.3.
97. Greenberg’s (1966: 26) “zero expression of the unmarked category”. Greenberg (1966: 29)
also notes “the lesser degree of morphological irregularity in marked forms”, referring to the
lack of alternative vowelling in the Arabic derived verbal patterns.
 Chapter 12.  Conclusions 253

also support the notion that it is a marked form. As such, we would expect it
to correspond with a typologically marked category, whereas in Olsen’s analysis
[Øtelic] designates verbs unmarked for telicity, i.e. in the default condition for
telicity. Thus to propose that pattern III is marked for the condition [Øtelic] is
arguably self-contradictory since we are claiming that it is marked for unmarked-
ness. The related Saussurean objection becomes clear when we note that many
pattern I verbs, including examples used in previous chapters, also belong to the
lexical aspect categories which are unmarked for telicity. If the morphologically
unaugmented pattern I is in widespread use for verbs with [Øtelic] meanings,
then the additional formal complexity of vowel lengthening might be considered
redundant: it ‘adds’ form without necessarily adding meaning.
Can the inconsistency be resolved by allowing [+atelic] privative feature
marking? This proposal counters the objections raised in the previous para-
graph if it can be demonstrated that verbs in patterns III and VI possess uncan-
cellable atelic meaning. There is no a priori reason why Olsen’s scheme cannot
be extended to incorporate lexical aspect categories in which [+atelic] feature
marking contrasts with zero marking in the same way that [+imperfective] and
[+perfective] privative features for grammatical aspect are set in opposition
with zero marking rather than with one another. However, we would need to be
convinced that new lexical aspect categories are required in Arabic which are
distinguished from those already identified on the grounds of inherent atelicity
which may not be pragmatically cancelled. Demonstrating her point with Eng-
lish examples in which an atelic ACTIVITY is rendered telic by context, Olsen
(1997: 19) maintains that “[t]elicity, that is, the [+telic] feature, should … be
part of the semantic representation of lexical aspect; atelicity … should not”. It
is not difficult to find examples in Arabic, such as (298) which contains the fa-
miliar pattern III verb ‫( قاتل‬qātala), in which atelicity is cancelled pragmatically
in compositional context and an endpoint introduced, thus demonstrating that
[Øtelic] is the correct analysis.

(298) ‫المعارضة العراقية لن تقاتل حتى الموت‬

al-mu‫ع‬āraDa al-‫ع‬irāqīya lan
def-opposition def-Iraqi neg.fut
tuqātil-a Hattā l-mawt
fight;npst.3fsg-sbjv until def-death
‘The Iraqi opposition will not fight to the death’
(arabiCorpus: GEN1996: 4022 )

Having ruled out the above solutions, I tentatively propose a third in which the
morphological form of pattern III makes a positive contribution to meaning while
retaining the analysis of these verbs as [Øtelic]. We established in Chapter 6 that
254 The Arabic Verb

the prefixed ta- morph is detransitivising,98 contributing the property of valency

reduction or minimisation to pattern VI with very few exceptions. Thus it has no
meaning in itself but rather modifies existing meaning: we may conceptualise its
derivational effect as cancelling a syntactically required element of the verb or re-
ducing the directness with which the element is addressed by the verb. In a similar
manner, I propose that the vowel lengthening derivation of pattern III cancels or
blocks [+telic] marking. Of course, this only becomes obvious when the pattern
III verb has a pattern I equivalent from the same root which is marked [+telic].
Let us consider again the verbs in (299a−c):

(299) a. ‫ قـتل‬ [I]

qatala (+ d.o.)
‘to kill (s.o.)’
b. ‫ قاتل‬ [III]
qātala (+ d.o.)
‘to fight [with] (s.o.)’
c. ‫ تقاتل‬ [VI]
taqātala
‘to fight one another’

In previous chapters we have noted that (299b), in common with many, though
not all, pattern III verbs, implies mutuality of participation in the action of the
verb by both subject and grammatical object and that (299c) incorporates the
participants together in the performance of the action with explicit reciprocity.
However, were we to encounter (299c) for the first time, unaware of the existence
of (299b), might we not expect the reciprocal meaning to be ‘to kill one another’
rather than ‘to fight one another’? I contend that it is specifically because the man-
ner of derivation is as in (300) that (299c) cannot have such a [+telic] mean-
ing: the vowel lengthening morph which derives pattern III from the root has
‘blocked’ the [+telic] feature of the root meaning and it cannot therefore resurface
in pattern VI which also contains the ‘blocking’ morph.
(300) root meaning -----------------> [III] -----------------------> [VI]
vowel ta- prefix morph
lengthening
templatic morph

I offer one further piece of evidence: although in Section 5.3 we dismissed the

need to establish a separate semantic classification for pattern III of ‘conative’,

98. At least in these patterns, pending investigation of the pattern II–V and QI–QII pairs.
 Chapter 12.  Conclusions 255

the nature of verbs with conative meaning is that they are ACTIVITIES and thus
­necessarily designated [Øtelic]. Thus it is entirely consistent with the examples
given in Chapter 5 that pattern III derivation of an ACTIVITY from a root mean-
ing which is either an ACHIEVEMENT or an ACCOMPLISHMENT will be
accompanied by a sense of conativity when the [+telic] marking of the root is
‘blocked’, whereas derivation from a root which is already [Øtelic] implies no
conativity. Otherwise formulated, if a root meaning designates a [+telic] result
(ACHIEVEMENT or ACCOMPLISHMENT), the pattern III derivative verb des-
ignates the [Øtelic] ACTIVITY of attempting to bring about that result, whereas
if no resulting endpoint is designated by the root meaning, no sense of conativity
is required to interpret the derived verb as an ACTIVITY.
Clearly, further research is required to support this ‘blocking’ hypothesis, in
particular a detailed examination of pattern III meanings in relation to their pat-
tern I counterparts.

12.4 Summary

Beginning with the Saussurean concept of the linguistic sign and the belief
that within it signifiant and signifié are inextricably linked, and also inspired by
­Beedham’s premise that lexical exceptions are a means to gaining insight into
the very grammatical rules they appear to violate, we have investigated two of
the verbal patterns of Modern Standard Arabic, namely those displaying vowel
lengthening, with respect to their form and meaning. The chapters which precede
the discovery that it is atelic aspectual meaning which characterises these patterns
place into the wider context the significance of this finding.
Whether this research will have much didactic application to Arabic or im-
pact upon future reference grammars of the language is uncertain. However, it is
undoubtedly of far greater interest to the linguistic scholar. Firstly, it firmly estab-
lishes the existence of derivationally realised lexical aspect, as distinct from the
inflectional grammatical aspect commonly recognised in Arabic. Secondly, if not
entirely serving to validate Beedham’s method, due to its significant departures
from his original methodology, it upholds the value of searching for systematic-
ity within apparent chaos. Moreover, the discovery of the inceptive lexical aspect
category in Arabic did ultimately stem from identifying a class of exceptions and
seeking to explain them. Thirdly, in establishing sound arguments for both adopt-
ing and extending Olsen’s model of aspect, significant contributions have been
made to advancing a Vendlerian understanding of lexical aspect categories cross-
linguistically.
 Bibliography

Allerton, D. J. 1982. Valency and the English verb. London: Academic Press.
Al-Qahtani, Duleim M. 2003. A dictionary of Arabic verbs. Beirut: Librairie du Liban.
Al-Qahtani, Duleim M. 2005. Semantic valence of Arabic verbs. Beirut: Librairie du Liban.
Al-Saqi, Fadil Mustafa. 1970. Ism al-fācil bayna al-ismiyah wa-al-ficliyah. Cairo: University of
Cairo MA dissertation.
Al-Tarouti, Ahmed F. 1991. Temporality in Arabic grammar and discourse. Los Angeles: UCLA
PhD dissertation.
Al-Tarouti, Ahmed F. 2001. “Dimensions of aspect.” Scientific Journal of King Faisal University
(Humanities & Management Sciences) 2(1): 197–230.
Bache, Carl. 1982. “Aspect and Aktionsart: Towards a semantic distinction.” Journal of Linguis-
tics 18(1): 57–72.
Badawi, Elsaid, Carter, Michael G. and Gully, Adrian. 2004. Modern written Arabic: A compre-
hensive grammar. London: Routledge.
Bagemihl, Bruce. 1989. “The crossing constraint and ‘backwards languages’. ” Natural Language
& Linguistic Theory 7(4): 481–549.
Bahloul, Maher. 2008. Structure and function of the Arabic verb. London: Routledge.
Bakalla, M. H. 1979. The morphological and phonological components of the Arabic verb. Lon-
don: Longman.
Bat-El, Outi. 2003. “Semitic verb structure within a universal perspective.” In Shimron (ed.),
29–59.
Bauer, Laurie. 2003. Introducing linguistic morphology. Edinburgh: Edinburgh University Press.
Beard, Robert. 1995. Lexeme-morpheme base morphology: A general theory of inflection and
word formation. Albany: State University of New York.
Beedham, Christopher. 1982. The passive aspect in English, German and Russian. Tübingen:
Gunter Narr.
Beedham, Christopher. 2005. Language and meaning: The structural creation of reality. Amster-
dam: John Benjamins.
Benmamoun, Elabbas. 1999. “Arabic morphology: The central role of the imperfective.” Lingua
108: 175–201.
Benmamoun, Elabbas. 2003a. “Reciprocals as plurals in Arabic.” In Lecarme (ed.), 53–62.
Benmamoun, Elabbas. 2003b. “The role of the imperfective template in Arabic morpology.”
In Shimron (ed.), 99–114.
Bennett, Patrick R. 1998. Comparative Semitic linguistics: A manual. Winona Lake: Eisenbrauns.
Berg, Thomas and Abd-El-Jawad, Hassan. 1996. “The unfolding of suprasegmental representa-
tions: A cross-linguistic perspective.” Journal of Linguistics 32(2): 291–324.
Bergen, Benjamin K. 2004. “The psychological reality of phonaesthemes.” Language 80(2):
290–311.
258 The Arabic Verb

Berjaoui, Nasser. 2007. A Moroccan Arabic secret language: The x...xinCa family. Munich:
LINCOM Europa.
Biberstein Kazimirski, Albert de. 1960. Dictionnaire arabe-française: Contenant toutes les raci-
nes de la langue arabe. Paris: Maisonneuve.
Bickel, Balthasar, Comrie, Bernard and Haspelmath, Martin. 2008. “Leipzig glossing rules.”
http://www.eva.mpg.de/lingua/pdf/ LGR09_02_23.pdf (3 March 2009).
Binnick, Robert I. 1991. Time and the verb: A guide to tense and aspect. Oxford: Oxford Uni-
versity Press.
Bohas, Georges. 2006. “The organisation of the lexicon in Arabic and other Semitic languages.”
In Perspectives on Arabic linguistics XVI: Papers from the sixteenth annual symposium on
Arabic linguistics, Cambridge, March 2002, Sami Boudelaa (ed.), 1–37. Amsterdam: John
Benjamins.
Bolinger, Dwight. 1968. Aspects of Language. New York: Harcourt Brace Jovanovich.
Boudelaa, Sami and Marslen-Wilson, William D. 2001. “Morphological units in the Arabic
mental lexicon.” Cognition 81(1): 65–92.
Brockelmann, Karl. 1904. [Socin’s] Arabische Grammatik: Paradigmen, Literatur, Übungsstücke
und Glossar. Berlin: Reuther & Reichard.
Buckley, Ronald. 2004. Modern literary Arabic. Beirut: Librairie du Liban.
Butler, Christopher. 1985. Statistics in linguistics. Oxford: Basil Blackwell.
Cachia, Pierre. 1973. The monitor: A dictionary of Arabic grammatical terms: Arabic–English,
English–Arabic. Beirut: Librairie du Liban.
Cairns, Charles E. and Feinstein, Mark H. 1982. “Markedness and the theory of syllable struc-
ture.” Linguistic Inquiry 13(2): 193–225.
Cantineau, Jean. 1934. Le dialecte arabe de Palmyre. Beirut: Institut français de Damas.
Chouémi, Moustapha. 1966. Le verbe dans le Coran: Racines et formes. Paris: C. Klincksieck.
Cohen, Marcel. 1924. Le système verbal sémitique et l’expression du temps. Paris: Leroux.
Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related prob-
lems. Cambridge: Cambridge University Press.
Comrie, Bernard. 1989. Language universals and linguistic typology. Chicago: University of
Chicago Press.
Connor-Linton, Jeff. 2003. “Chi square tutorial.” http://www.georgetown.edu/faculty/ballc/
webtools/web_chi_tut.html (26 February 2007).
Cowan, David. 1958. An introduction to modern literary Arabic. Cambridge: Cambridge Uni-
versity Press.
Cusic, David Dowell. 1981. Verbal plurality and aspect. Palo Alto, CA: Stanford University PhD
dissertation.
Cuvalay-Haak, Martine. 1997. The verb in literary and colloquial Arabic. Berlin: Mouton de
Gruyter.
Dahl, Östen and Talmoudi, Fathi. 1979. “Qad and laqad: tense/aspect and pragmatics in Ara-
bic.” In Aspectology: Papers from the 5th Scandinavian conference of linguistics, Frostavallen,
Apr. 27–29, 1979, Thore Pettersson (ed.), 51–68. Stockholm: Almqvist & Wiksell.
Danks, Warwick. 2007. “Verbal derivation in Modern Standard Arabic: Towards a unified
description of form X.” Paper presented at the Second Postgraduate Conference in Theo-
retical and Applied Linguistics, June 25, 2007, Newcastle University.
Danks, Warwick. 2008. “Standing up to scrutiny: Inceptive verbs in Arabic and English.” Paper
presented to the Linguistics Institute of St Andrews, November 25, 2008, University of
St Andrews.
 Bibliography 259

Davis, Stuart and Zawaydeh, Bushra Adnan. 2001. “Arabic hypocoristics and the status of the
consonantal root.” Linguistic Inquiry 32(3): 512–520.
Dickins, James and Watson, Janet C. E. 2009. “Transitivity.” In Versteegh (gen. ed.), 528–535.
Dixon, Robert M. W. 2000. “A typology of causatives.” In Changing valency, Robert M. W. Dixon
and Alexandra Y. Aikhenvald, 30–83. Cambridge: Cambridge University Press.
Dixon, Robert M. W. 2005. A semantic approach to English grammar. Oxford: Oxford University
Press.
Doron, Edit. 2003. “Agency and voice: The semantics of the Semitic templates.” Natural Lan-
guage Semantics 11(1): 1–67.
Dowty, David R. 1979. Word meaning and Montague grammar. Dordrecht: D. Reidel.
Drozdík, Ladislav. 1998. “Modern written Arabic in Hans Wehr’s Arabisches Wörterbuch für
die Schriftsprache der Gegenwart (1st–5th editions).” Asian and African Studies 7(2):
211–218.
Eisele, John C. 1999. Arabic verbs in time: Tense and aspect in Cairene Arabic. Wiesbaden:
Harrassaowitz.
El-Kassas, Dina. 2007. “Surface-syntactic constructions controlled by Arabic ‘psychic’ verbs.”
http://www.dinaelkassas.com/fichiers/ELKASSAS_SSyntC_final%20 version.pdf (20 March
2009).
El-Tikaina, Ibrahim S. H. 1983. A lexical approach to the Arabic verb conjugations. Seattle:
University of Washington PhD dissertation.
Fassi Fehri, Abdelkader. 1987. Anti-causatives in Arabic, causativity and affectedness. Rabat:
Mohammed V University.
Fassi Fehri, Abdelkader. 2003. “Verbal plurality, transitivity, and causativity.” In Lecarme (ed.),
151–185.
Fassi Fehri, Abdelkader. 2004. “Temporal/aspectual interaction and variation across Arabic
heights.” In The syntax of time, Jacqueline Guéron and Jacqueline Lecarme (eds), 235–257.
Cambridge, MA: MIT Press.
Fischer, Wolfdietrich. 2002. Grammatik des klassischen Arabisch. Wiesbaden Harrassaowitz.
Fleisch, Henri. 1944. Les verbes à allongement vocalique interne en Sémitique. Paris: Institut
d’ Ethnologie.
Fleisch, Henri. 1974. “Sur l’aspect dans le verbe en arabe classique.” Arabica 21(1): 1–19.
Fleisch, Henri. 1979. Traité de philologie arabe 2: Pronoms, morphologie verbale, particules.
Beirut: Dar al-Machreq.
Freed, Alice F. 1979. The semantics of English aspectual complementation. Dordrecht: Reidel.
Gafos, Adamantios I. 2002. “An argument for a stem-based view of Arabic morphology: Double
verbs revisited.” In Perspectives on Arabic linguistics XIII–XIV: Papers from the thirteenth
and fourteenth annual symposia on Arabic linguistics, Dilworth B. Parkinson and Elabbas
Benmamoun (eds), 59–86. Amsterdam: John Benjamins.
Gaudefroy-Demombynes, Maurice and Régis Blachere. 1952. Grammaire de l’Arabe classique.
Paris: Maisonneuve.
Goetze, Albrecht. 1942. “The so-called intensive of the Semitic languages.” Journal of the Ameri-
can Oriental Society 62(1): 1–8.
Gray, Louis H. 1934. Introduction to Semitic comparative linguistics. New York: Columbia Uni-
versity Press.
Greenberg, Joseph H. 1950. “The patterning of root morphemes in Semitic.” Word 6(2):
162–181.
260 The Arabic Verb

Greenberg, Joseph H. 1966. Language universals with special reference to feature hierarchies. The
Hague: Mouton.
Greenberg, Joseph H. 1991. “The Semitic “intensive” as verbal plurality.” In Kaye (ed.), 577–587.
Hallman, Peter. 2005. “Interfaces in Arabic.” http://individual.utoronto.ca/hallman/Interfaces.
pdf (29 January 2007).
Hammond, Michael. 1988. “Templatic transfer in Arabic broken plurals.” Natural Language &
Linguistic Theory 6(2): 247–270.
Harter, Leon H. 1964. “A new table of percentage points of the chi-square distribution.”
Biometrika 51(1–2): 231–239.
Heard, James. 1827. A practical grammar of the Russian language. St Petersburg: Printed for the
author and sold by Sleunine, London.
Heath, Jeffrey. 1987. Ablaut and ambiguity: Phonology of a Moroccan Arabic dialect. Albany:
State University of New York Press.
Heath, Jeffrey. 2003. “Arabic derivational ablaut, processing strategies, and consonantal “roots”. ”
In Shimron (ed.), 115–129.
Herbst, Thomas. 1999. “English valency structures – A first sketch.” http://webdoc.gwdg. de/
edoc/ia/eese/artic99/herbst/main1.html (27 August 2009).
Hoeksema, J. 1985. [1984] Categorial morphology. New York: Garland.
Holes, Clive. 1990. Gulf Arabic. London: Routledge.
Holes, Clive. 2004. Modern Arabic: Structures, functions and varieties (Revised Edn.). Washing-
ton, DC: Georgetown University Press.
Hopper, Paul J. and Thompson, Sandra A. 1980. “Transitivity in grammar and discourse.”
Language 56(2): 251–299.
Huehnergard, John. 2000. A grammar of Akkadian. Winona Lake: Eisenbrauns.
Jakobson, Roman. 1971. “Shifters, verbal categories and the Russian verb.” In Selected Writings
II: Word and Language, Roman Jakobson, 130–147. The Hague: Mouton.
Jespersen, O. 1924. The philosophy of grammar. London: George Allen & Unwin.
Kaye, Alan S. (ed.). 1991. Semitic studies: In honor of Wolf Leslau on the occasion of his eighty-
fifth birthday, November 14th, 1991. Wiesbaden: Harrassowitz.
Kenny, Anthony. 1963. Action, emotion and will. London: Routledge.
Koerner, E. F. Konrad. 1999. “Three Saussures – One ‘structuralist’ avant la lettre.” In Langue
and parole in synchronic and diachronic perspective: Selected proceedings of the XXXIst an-
nual meeting of the Societas Linguistica Europaea, St Andrews, 1998, Christopher Beedham
(ed.), 19–34. Pergamon: Oxford.
Larcher, Pierre. 1999. “Vues “nouvelles” sur la dérivation lexicale en arabe classique”. In Tradi-
tion and innovation: Norm and deviation in Arabic and Semitic linguistics, Lutz Edzard and
Mohammed Nekroumi (eds), 103–123. Wiesbaden: Harrassowitz.
Lecarme, Jacqueline (ed.). 2003. Research in Afroasiatic grammar II: Selected papers from the
fifth conference on Afroasiatic languages, Paris, 2000. Amsterdam: John Benjamins.
Leemhuis, Frederik. 1977. The D and H stems in Koranic Arabic: A comparative study of the
function and meaning of the faccala and ‘afcala forms in Koranic usage. Leiden: Brill.
Levin, Aryeh. 1991. “What is meant by al-mafcūl al-muţlaq?” In Kaye (ed.), 917–926.
Lyons, John. 1977. Semantics, Vol. 2. Cambridge: Cambridge University Press.
McCarthy, John J. 1981. “A prosodic theory of nonconcatenative morphology.” Linguistic
Inquiry 12: 373–418.
McCarthy, John J. 1985. [1979]. Formal problems in Semitic phonology and morphology. New
York: Garland.
 Bibliography 261

McCarthy, John J. and Prince, Alan. 1990a. “Prosodic morphology and templatic morphol-
ogy.” In Perspectives on Arabic linguistics II: Papers from the second annual symposium
on Arabic linguistics, Mushira Eid and John McCarthy (eds), 1–54. Amsterdam: John
Benjamins.
McCarthy, John J. and Prince, Alan. 1990b. “Foot and word in prosodic morphology: The Ara-
bic broken plural.” Natural Language & Linguistic Theory 8(2): 209–283.
MacDonald, John. 1963. “The Arabic derived verb themes: A study in form and meaning.”
Islamic Quarterly 7: 96–116.
McEnery, Tony and Wilson, Andrew. 2001. Corpus linguistics: An introduction. Edinburgh:
Edinburgh University Press.
Mahfoudi, Abdessatar. 2007. “The place of the etymon and the phonetic matrix in the Arabic
mental lexicon.” Journal of Arabic and Islamic Studies 7(4): 74–102.
Meillet, Antoine. 1893. “Les lois du langage I: Les lois phonétiques.” Revue Internationale de
Sociologie 1: 311–312.
Mohammad, Mahmoud D. 1983. The semantics of tense and aspect in English and modern stan-
dard Arabic. Georgetown: Georgetown University PhD dissertation.
Moscati, Sabatino. 1964. An introduction to the comparative grammar of the Semitic languages:
Phonology and morphology. Wiesbaden: Harrassowitz.
Mourelatos, Alexander P. D. 1978. “Events, processes and states.” Linguistics and Philosophy 2:
415–434.
Mughazy, Mustafa. 2005. “Rethinking lexical aspect in Egyptian Arabic.” In Perspectives on
Arabic linguistics XVII–XVIII: Papers from the seventeenth and eighteenth annual symposia
on Arabic linguistics, Mohammad T. Alhawary and Elabbas Benmamoun (eds), 133–172.
Amsterdam: John Benjamins.
Mullins, Christopher. 2005. “On the origin and inherent meaning of the L-stem.” https://www.
indiana.edu/~iulcwp/pdfs/05-mullins.pdf (22 October 2008).
Newman, Francis W. 1895. Handbook of modern Arabic. Nottingham: Stevenson, Bailey &
Smith.
O’Leary, De Lacy. 1969. Comparative grammar of the Semitic languages. Amsterdam: Philo
Press.
Olsen, Mari Broman. 1997. A semantic and pragmatic model of lexical and grammatical aspect.
New York: Garland.
Passonneau, Rebecca J. 1988. A computational model of the semantics of tense and aspect.
Computational Linguistics 14(2). 44–60.
Pinker, Steven. 2008. The stuff of thought: Language as a window into human nature. London:
Penguin.
Poutsma, H. 1926. A grammar of late modern English. Groningen: Noordhof.
Prunet, Jean-François. 2006. “External evidence and the Semitic root.” Morphology 16(1):
41–67.
Prunet, Jean-François, Béland, Renée and Idrissi, Ali. 2000. “The mental representation of
Semitic words.” Linguistic Inquiry 31(4): 609–648.
Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey and Svartvik, Jan.1985. A grammar of
contemporary English. London: Longman.
Ratcliffe, Robert R. 1997. “Prosodic templates in a word-based morphological analysis of Ara-
bic.” In Perspectives on Arabic linguistics X: Papers from the tenth annual symposium on
Arabic linguistics, Mushira Eid and Robert R. Ratcliffe (eds.), 147–171. Amsterdam: John
­Benjamins.
262 The Arabic Verb

Reig, Daniel. 1983. La conjugaison arabe. Paris: Maisonneuve et Larose.

Reinhardt, Carl. 1894. Ein arabischer dialekt gesprochen in cOmān und Zanzibar: nach prak-
tischen gesichtspunkten für das Seminar für orientalische sprachen in Berlin. Stuttgart:
W. Spemann.
Richardson, John. 1969. [1776]. A grammar of the Arabick language. Menston: Scolar Press.
Ryder, Stuart A., II. 1974. The D-stem in western Semitic. The Hague: Mouton.
Ryding, Karin C. 2005. A reference grammar of modern standard Arabic. Cambridge: Cam-
bridge University Press.
Ryle, Gilbert. 1949. The concept of mind. Chicago: University of Chicago Press.
Saad, George N. 1982. Transitivity, causation and passivization: A semantic-syntactic study of the
verb in classical Arabic. London: Kegan Paul.
Saussure, Ferdinand de. 1966. [1916, in French]. Course in general linguistics. Eds. Charles Bally
and Albert Sechehaye. Tr. Wade Baskin. New York: McGraw-Hill.
Schulz, Eckehard. 2004. A student grammar of modern standard Arabic. Cambridge: Cambridge
University Press.
Selkirk, Elisabeth O. 1980. “On prosodic structure and its relation to syntactic structure.” In
Nordic Prosody II, T. Fretheim (ed.), 111–140. Trondheim: Tapir.
Shimron, Joseph (ed.). 2003a. Language processing and acquisition in languages of Semitic, root-
based, morphology. Amsterdam: John Benjamins.
Shimron, Joseph. 2003b. “Semitic languages: Are they really root-based?” In Shimron (ed.),
1–28.
Smith, Carlota S. 1997 [1991]. The parameter of aspect. Dordrecht: Kluwer Academic.
Soltan, Usama. 2009. “Transitivity: Object.” In Versteegh (gen. ed.), 535–542.
Sproat, Richard W. 1992. Morphology and computation. Cambridge, MA: MIT.
Swadesh, Morris. 1946. “Chitimacha.” In Linguistic structures of native America, Cornelius
Osgood and Harry Hoijer (eds), 312–336. New York: Viking Fund.
Tesnière, Lucien. 1959. Élements de syntaxe structurale. Paris: C. Klincksieck.
Thatcher, G. W. 1922. Arabic grammar of the written language. London: Nutt.
Thomson, Godfrey H. and Thompson, J. Ridley. 1915. “Outlines of a method for the quantita-
tive analysis of writing vocabularies.” British Journal of Psychology 8: 52–69.
Ussishkin, Adam. 1999. “The inadequacy of the consonantal root: Modern Hebrew denominal
verbs and output-output correspondence.” Phonology 16(3): 401–442.
Ussishkin, Adam. 2000. “Root-and-pattern morphology without roots or patterns.” http://dingo.
sbs.arizona.edu/~ussishkin/papers/PDF/UssishkinNELS30Paper.pdf (10 December 2008).
Ussishkin, Adam. 2005. “A fixed prosodic theory of nonconcatenative templatic morphology.”
Natural Language and Linguistic Theory 23(1): 169–218.
Ussishkin, Adam. 2006. “Semitic morphology: Root-based or word-based?” Morphology 16(1):
37–40.
Vendler, Zeno. 1957. “Verbs and times.” Philosophical Review 66(2): 143–160.
Vendler, Zeno. 1967. Linguistics in philosophy. Ithaca, NY: Cornell University Press.
Versteegh, Kees, (gen. ed.). 2009. Encyclopedia of Arabic language and linguistics, Vol. 4. Leiden:
Brill.
Verkuyl, Henk J. 1993. A theory of aspectuality: The interaction between temporal and atemporal
structures. Cambridge: Cambridge University Press.
Watson, Janet C. E. 2002. The phonology and morphology of Arabic. Oxford: Oxford University
Press.
 Bibliography 263

Watson, Janet C. E. 2006. “Arabic morphology: Diminutive verbs and diminutive nouns in
San’ani Arabic.” Morphology 16(2): 189–204.
Wehr, Hans, tr. J. Milton Cowan. 1994. A dictionary of modern written Arabic: Arabic–English.
Ithaca, NY: Spoken Language Services.
Wickens, G. M. 1980. Arabic grammar: A first workbook. Cambridge: Cambridge University
Press.
Wright, William. 1967. [1859–1862]. A grammar of the Arabic language. Cambridge: Cam-
bridge University Press.
Zaborski, Andrzej. 1997. “Qātala and qattala in Semitic and Hamitosemitic.” Rocznik Orienta-
listyczny 50: 257–262.
Zandvoort, R. W. 1962. “Is ‘aspect’ an English verbal category?” In Contributions to English
syntax and philology: Gothenburg Studies in English XIV, F. Behre (ed.), 1–20. Göteborg:
Almqvist and Wiksell.

Additional online resources

Arabic life application Bible [ALAB]. 1988. International Bible Society. http://www.biblegateway.
com/versions/?action=getVersionInfo&vid=28 (May 2008–Sept. 2009).
arabiCorpus. Dilworth B. Parkinson (creator). Brigham Young University. http:// arabicorpus.
byu.edu (June 2007–Sept. 2009).
Google. http://www.google.co.uk (May 2008–Sept. 2009).
Holy Bible: New international version – UK [NIV]. 1984. International Bible Society. http://
www.biblegateway.com/versions/?action=getVersionInfo&vid=64 (May 2008–Sept. 2009).
Oxford English dictionary [OED]. Oxford University Press. http://dictionary.oed.com
(Oct. 2006–Sept. 2009).
 appendix i

Data tables

The following data files are included as digital appendices, available via the following URL:
http://hdl.handle.net/10023/961

VERBAL OCCURRENCES LISTED BY ROOT AND PATTERN

Triliteral: Roots_and_patterns_tri.pdf
(198 pages)
Quadriliteral: Roots_and_patterns_quad.pdf
(14 pages)

CLASSIFICATION OF VERBS BY SEMANTIC CATEGORIES

(see Appendix II for category labels)
Pattern III (with pattern VI absent): Pattern_III_not_VI_sem.pdf
(9 pages)
Pattern VI (with pattern III absent): Pattern_VI_not_III_sem.pdf
(6 pages)
Patterns III and VI co-occurrent: Patterns_III_and_VI_sem.pdf
(10 pages)

CLASSIFICATION OF VERBS BY VALENCY STRUCTURE CODES

(see Table 41, Section 6.3 for description of codes)
Pattern III (with pattern VI absent): Pattern_III_not_VI_val.pdf
(14 pages)
Pattern VI (with pattern III absent): Pattern_VI_not_III_val.pdf
(9 pages)
Patterns III and VI co-occurrent: Patterns_III_and_VI_val.pdf
(15 pages)

CLASSIFICATION OF VERBS BY ASPECTUAL CATEGORY

(including arabiCorpus token counts according to the methodology
described in Section 9.1.1.1)
Pattern III (with pattern VI absent): Pattern_III_not_VI_asp.pdf
(15 pages)
Pattern VI (with pattern III absent): Pattern_VI_not_III_asp.pdf
(10 pages)
Patterns III and VI co-occurrent: Patterns_III_and_VI_asp.pdf
(19 pages)

VERBAL NOUNS
Pattern III muC1āC2aC3a: Pattern_III_muCaaCaCa.pdf
(11 pages)
Pattern III C1iC2āC3: Pattern_III_CiCaaC.pdf
(5 pages)
 appendix ii

Semantic category labels

Label Transitive? Valency Gloss

structure code
CAU Y 2B Causative upon d.o.
CAU- 2A Causative upon i.o.
CAU± Y 3B Causative upon d.o. with respect to i.o.
CON+ Y 2B Conative with d.o.
CRP± 2A Causative reflexive passive with i.o. as agent
ESR 1B Estimative, reflexive
ESR± 2A Estimative, reflexive with respect to i.o.
EST Y 2B Estimative of d.o.
EST- 2A Estimative of i.o.
IMP± 1A Impersonal, monovalent with i.o.
INT 1B Intransitive
INT± 2A Intransitive with i.o.
INT±± 3A Intransitive with two i.os
MUT Y 2B Implied mutuality with d.o.
MUT- 2A Implied mutuality with i.o.
MUT+ Y 3C Implied mutuality with d.o. and additional d.o.
MUT± Y 3B Implied mutuality with d.o. and additional i.o.
MUT-± 3A Implied mutuality with i.o. and additional i.o.
PAS 1B Semantically passive
PAS+ Y 2B Semantically passive, with d.o.
PAS± 2A Semantically passive, with i.o.
SIM 1B Simulative
SIM+ Y 2B Simulative towards d.o.
SIM± 2A Simulative towards i.o.
REC 1B Explicit reciprocity, usually non-singular subject
REC+ Y 2B Explicit reciprocity with d.o.
REC± 2A Explicit reciprocity with i.o.
REF 1B Reflexive
REF+ Y 2B Reflexive effect with respect to d.o.
REF± 2A Reflexive effect with respect to i.o.
REF++ Y 3C Reflexive effect with respect to two d.os
REF+± Y 3B Reflexive effect with respect to d.o. and i.o.
268 The Arabic Verb

Label Transitive? Valency Gloss

structure code
STA 1B Stative
STA+ Y 2B Stative with respect to d.o.
STA± 2A Stative with respect to i.o.
STC 1B Conative towards state
TRA Y 2B Transitive with d.o.
TRA- 2A Semantically transitive, with i.o.
TRA+ Y 3C Ditransitive with two d.os
TRA± Y 3B Transitive with d.o. and additional i.o.
TRA-± 3A Semantically transitive, with two i.os
TRA+± Y 4C Transitive with two d.os and i.o.

Notes:
i. Only the categories actually attested in the attached data tables are listed here;
ii. Valency structure codes refer to Table 41 in Section 6.3;
iii. d.o. = direct object; i.o. = indirect object (with preposition).
 appendix iii

Native speaker questionnaire

iii.1 Background

Three native speaker informants were approached to complete the questionnaire reproduced
here in Section III.4. At the time, each informant was a postgraduate student and/or an Ara-
bic language tutor at the University of St Andrews. All informants were male native Arabic
speakers under 35 with significant educational background in Arabic as well as in English. In
accordance with University policy, permission for this research was obtained from the Ethics
Committee of the School of Modern Languages (Application Ref. ML4979). To comply with
the permission granted, all original completed questionnaires and other documents identifying
the original participants have been destroyed and only the summarised data are recorded here.

iii.2 Procedure

The informants were presented with the information and consent form reproduced in Sec-
tion III.3 and were talked through the exercise before being given the questionnaire to com-
plete in their own time. Following completion, the informants were visited again and their
responses were discussed to resolve any difficulties and to ensure as far as possible that their
assessments were based solely on judgements of grammaticality and not on any extralinguistic
considerations or merely on preferences of word order. Some of the sentences presented for
assessment contained verbs and constructions other than those of specific interest in order
both to assess the level of agreement between informants and to minimise any conscious or
unconscious anticipation by informants of the responses expected.

iii.3 Information and consent form

The Arabic Verb

Information for Questionnaire Participants
 
Dear
Thank you for volunteering to help me with my PhD research. My thesis involves investigation
of the verb forms of Modern Standard Arabic, and I am concentrating particularly on Forms III
and VI (faa’ala and tafaa’ala). I have made some interesting observations and formulated ex-
planations, but I am now at the point where I need to obtain some input from native Arabic
speakers.
270 The Arabic Verb

I have compiled a questionnaire consisting of around 60 short sentences to be assessed for

naturalness and grammaticality. There are no right or wrong answers and I am interested in
your ‘gut reaction’ as a native speaker, rather than any deep analysis. However, I would welcome
your comments on why particular sentences seem wrong. Please mark one of the following for
each example.
 Good, natural Arabic
? Doubtful or not natural, but possible
* Not grammatical or possible
I will arrange for you to return the questionnaire to me electronically or on paper and we can
discuss any points you wish to raise, or suggestions you may have. At that time, I may also ask
you if you are willing to take part in a similar, follow-up questionnaire.
The responses you give will only be seen by myself and my research supervisors. No data, in-
cluding any background information you may give me, will be identifiable by name in any
published material and original questionnaires will be destroyed once the research project is
completed. You are free to withdraw from participation at any time without giving any explana-
tion and if you do not feel able to give an opinion on a particular example, you may omit it.
Please complete the declaration below, confirming your willingness to participate in the ques-
tionnaire.
Many thanks for your participation!

Warwick Danks
School of Modern Languages, University of St Andrews

I have read the ‘Information for Questionnaire Participants’, and give my consent to taking part
in this research study. I am over 18 years of age.

Signed: Name: Date:

‫ ‬ ‫‪Appendix III.  Native speaker questionnaire 271‬‬

‫‪iii.4 Questionnaire‬‬

‫‪Comment‬‬ ‫‪*?‬‬
‫كان بيتنا يجاور المكتبة‬ ‫‪1‬‬
‫ما زال هللا يبارك شعبه‬ ‫‪2‬‬
‫جوو َرت المكتبة‬ ‫‪3‬‬
‫حاسب أحمد المدير‬ ‫‪4‬‬
‫قوتل صديقه‬ ‫‪5‬‬
‫قاتل أحمد صديقه طوال عشرين دقيقة‬ ‫‪6‬‬
‫المكتبة مجا َورة‬ ‫‪7‬‬
‫فاجأ أحمد صديقه مرات كثيرة‬ ‫‪8‬‬
‫جاور بيتنا المكتبة مرات كثيرة‬ ‫‪9‬‬
‫حاسب أحمد المدير طوال سنين كثيرة‬ ‫‪10‬‬
‫كان عدد سكان المغرب مجا َوزاً‬ ‫‪11‬‬
‫فاجأ أحمد صديقه طوال عشرين دقيقة‬ ‫‪12‬‬
‫يحاسب أحمد المدير منذ الحادث‬ ‫‪13‬‬
‫جووز عدد سكان المغرب‬ ‫‪14‬‬
‫ما زال أحمد يفاجئ صديقه‬ ‫‪15‬‬
‫حاسب أحمد المدير منذ الحادث‬ ‫‪16‬‬
‫سافر أحمد من الجزائر‬ ‫‪17‬‬
‫يحاسب أحمد المدير حتى نهاية األزمة‬ ‫‪18‬‬
‫فُتح الباب‬ ‫‪19‬‬
‫كان صديقه مقاتَال‬ ‫‪20‬‬
‫فوجئ صديقه‬ ‫‪21‬‬
‫فتح أحمد الباب طوال عشرين دقيقة‬ ‫‪22‬‬
‫كان أحمد يسافر الى القاهرة طوال يومين‬ ‫‪23‬‬
‫حاسب أحمد المدير مرات كثيرة‬ ‫‪24‬‬
‫سافر أحمد الى القاهرة حتى مغرب الشمس‬ ‫‪25‬‬
‫يجاوز عدد سكان القاهرة عدد سكان المغرب منذ سنين كثيرة‬ ‫‪26‬‬
‫قاتل أحمد صديقه مرات كثيرة‬ ‫‪27‬‬
‫يفاجئ أحمد صديقه منذ كان طفالً‬ ‫‪28‬‬
‫جاوز عدد سكان القاهرة عدد سكان المغرب‬ ‫‪29‬‬
‫ما زال أحمد يحاسب المدير‬ ‫‪30‬‬
‫يفتح أحمد الباب منذ الساعة السابعة‬ ‫‪31‬‬
‫كان أحمد يفاجئ صديقه طوال عشرين دقيقة‬ ‫‪32‬‬
‫‪272 The Arabic Verb‬‬

‫يجاوز عدد سكان القاهرة عدد سكان المغرب اآلن‬ ‫‪33‬‬

‫سافر أحمد الى القاهرة‬ ‫‪34‬‬
‫كان أحمد يفتح الباب طوال عشرين دقيقة‬ ‫‪35‬‬
‫كان ابراهيم مبا َركا ً‬ ‫‪36‬‬
‫قاتل أحمد صديقه‬ ‫‪37‬‬
‫كان المدير محا َسبا ً‬ ‫‪38‬‬
‫كان عدد سكان القاهرة يجاوز عدد سكان المغرب حتى أواخر‬
‫‪39‬‬
‫القرن‬
‫فاجأ أحمد صديقه‬ ‫‪40‬‬
‫كان هللا يبارك ابراهيم طوال عشرين سنة‬ ‫‪41‬‬
‫فتح أحمد الباب‬ ‫‪42‬‬
‫كان أحمد يحاسب المدير طوال سنين كثيرة‬ ‫‪43‬‬
‫بورك ابراهيم‬ ‫‪44‬‬
‫بارك هللا ابراهيم طوال عشرين سنة‬ ‫‪45‬‬
‫كان صديقه مفا َجأ ً‬ ‫‪46‬‬
‫جاوز عدد سكان القاهرة عدد سكان المغرب حتى أواخر القرن‬ ‫‪47‬‬
‫يفتح أحمد الباب حتى الساعة السابعة‬ ‫‪48‬‬
‫جاور بيتنا المكتبة‬ ‫‪49‬‬
‫كان هللا يبارك ابراهيم مرات كثيرة‬ ‫‪50‬‬
‫كان الباب مفتوحا ً‬ ‫‪51‬‬
‫حوسب المدير‬ ‫‪52‬‬
‫بارك هللا ابراهيم مرات كثيرة‬ ‫‪53‬‬
‫كان أحمد يسافر طوال يومين‬ ‫‪54‬‬
‫فتح أحمد الباب مرات كثيرة‬ ‫‪55‬‬
‫جاوز عدد سكان القاهرة العدد في المغرب مرات كثيرة‬ ‫‪56‬‬
‫يحاسب أحمد المدير بينما يبحث في المشكلة‬ ‫‪57‬‬
‫سافر أحمد الى القاهرة مرات كثيرة‬ ‫‪58‬‬
‫جاوز عدد سكان القاهرة العدد في المغرب مرات كثيرة‬ ‫‪56‬‬
‫يحاسب أحمد المدير بينما يبحث في المشكلة‬ ‫‪57‬‬
‫سافر أحمد الى القاهرة مرات كثيرة‬ ‫‪58‬‬
‫كان هللا يبارك ابراهيم حتى أواخر أيّامه‬ ‫‪59‬‬
‫كان أحمد يقاتل صديقه طوال عشرين دقيقة‬ ‫‪60‬‬
 Appendix III.  Native speaker questionnaire 273

iii.5 Data summary

The responses of informants were transferred to the table below using the following notation:
 Good, natural Arabic {Y}
? Doubtful or not natural, but possible {?}
* Not grammatical or possible {N}*

1 YYY 16 NYY 31 Y?Y 46 NNN

2 YYY 17 YYY 32 ??N 47 YYN
3 YNN 18 N?N 33 NNN 48 YYN
4 YYY 19 YYY 34 YYY 49 N?Y
5 N?? 20 YYN 35 N?N 50 ?YY
6 YYY 21 ?YY 36 Y?? 51 YYY
7 Y?Y 22 ?YY 37 YYY 52 YYY
8 YYY 23 NYN 38 ?Y? 53 YY?
9 NNN 24 YYY 39 YYY 54 NNN
10 Y?Y 25 N?N 40 YYY 55 YYY
11 YN? 26 ?NN 41 NYY 56 ?YY
12 Y?N 27 YYY 42 YYY 57 N?N*
13 Y?? 28 YNN 43 YYY 58 YYY
14 NNN 29 ?YY 44 NYY 59 ?YY
15 YYY 30 YYY 45 YYY 60 N?Y

* On further investigation, this sentence was found to have been rejected by the informants
on grounds other than the use of the verb of interest.
 appendix iv

Passive participles in patterns III and VI

The following is an exhaustive list of pattern III and pattern VI verbs giving rise to passive par-
ticiple forms according to Wehr (1994):

Pattern III Pattern VI

‫آخذ‬ ākhadha’ ‫تبادل‬ tabādala
‫بارك‬ bāraka ‫تداول‬ tadāwala
‫خالط‬ khālaTa ‫تقادى‬ taqādā
‫شاهد‬ shāhada ‫تنازع‬ tanāza ‫ع‬a
‫شاور‬ shāwara ‫تناول‬ tanāwala
‫صادر‬ Sādara ‫توارث‬ tawāratha
‫ضاعف‬ Dā‫ع‬afa
‫طالب‬ Tālaba
‫عافى‬ ‫ع‬āfā
‫عاقب‬ ‫ع‬āqaba
‫الحظ‬ lāHaDHa
‫نادى‬ nādā
‫نازع‬ nāza ‫ع‬a
‫وارب‬ wāraba
‫ياوم‬ yāwama
 Name index

A Comrie, B.  103, 157, 160–162, Hammond, M.  43

Abd-El-Jawad, H.  47 164, 167–168 Harter, L. H.  33
Allerton, D. J.  108–110 Connor-Linton, J.  32 Heard, J.  168
Al-Qahtani, D. M.  17, 26–29, 78 Cowan, D.  64, 66, 68, 71, 161 Heath, J.  40–41, 45, 48
Al-Saqi, F. M.  146 Cusic, D. D.  136 Herbst, T.  108–109
Al-Tarouti, A. F.  146, 160 Cuvalay-Haak, M.  79–80, 244 Hoeksema, J.  163
Aristotle  163 Holes, C.  16, 45, 63–65, 67,
D 69–70, 72, 95, 101, 124, 136,
B Dahl, Ö.  161 145, 148, 150, 159–160
Bache, C.  157, 164 Danks, W.  23, 28, 36, 75 Hopper, P. J.  129–130
Badawi, E.  18, 64–65, 67, 69–70, Davis, S.  46, 48–49 Huehnergard, J.  75
103–106, 116, 161, 246 Dickins, J.  106–107
Bagemihl, B.  48 Dixon, R. M. W.  104, 136 J
Bahloul, M.  78, 160 Doron, E.  74–76 Jakobson, R.  160
Bakalla, M. H.  78 Dowty, D. R.  163 Jespersen, O.  5
Bat-El, O.  40–41 Drozdík, L.  24, 26, 29
Bauer, L.  16, 39, 44, 53 K
Beard, R.  43 E Kazimirski, B.  51
Beedham, C.  2–13, 15, 76, 83, Eisele, J. C.  161 Kenny, A.  163
132, 147, 149, 155, 158, 251, 255 El-Kassas, D.  106 Koerner, E. F. K.  3, 81
Benmamoun, E.  40–41, 134, El-Tikaina, I. S. H.  79
137–140, 142–143, 161 L
Bennett, P. R.  74 F Larcher, P.  40
Berg, T.  47 Fassi Fehri, A.  79–80, 85, 123, Leemhuis, F.  22, 77, 80
Bergen, B. K.  53 134–137, 139, 141, 161 Levin, A.  104
Binnick, R. I.  215 Feinstein, M. H.  221 Lyons, J.  157, 160
Blachère, R.  101 Fischer, W.  101
Bohas, G.  40, 51–55 Fleisch, H.  95, 159 M
Bolinger, D.  53 McCarthy, J. J.  17, 19, 22, 26–29,
Boudelaa, S.  55 G 36, 43, 47, 55–62, 112, 123, 131,
Brockelmann, K.  40, 101 Gafos, A.  41, 45 134, 138, 140, 147, 247
Buckley, R.  65, 91, 101, 159 Gaudefroy-Demombynes, M. MacDonald, J.  22–23
Butler, C.  32, 35 101 McEnery, T.  191
de Goeje, M. J.  71 Mahfoudi, A.  47, 55
C Goetze, A.  77 Marslen-Wilson, W. D.  55
Cachia, P.  104–105, 146 Gray, L. H.  75 Meillet, A.  3, 81
Cairns, C. E.  221 Greenberg, J. H.  40, 80, 134–136, Mohammad, M. D.  78, 160
Cantineau, J.  101 252 Moscati, S.  74
Caspari, C. P.  71 Mourelatos, A. P. D.  163
Chouémi, M.  80 H Mughazy, M.  183, 187
Cohen, M.  74 Hallman, P.  123 Mullins, C.  123
278 The Arabic Verb

N Ryding, K. C.  64–65, 67, 69, 72, U

Newman, F. W.  66, 68 103–104, 149, 161–162, 242, Ussishkin, A.  39, 41–42, 46
246
O Ryle, G.  163 V
O’Leary, D.  74–75 Vendler, Z.  157, 162–165, 168–169,
Olsen, M. B.  12, 157, 163, S 182, 187, 202, 255
165–187, 189, 202, 211, 213, Saad, G. N.  103–104 Verkuyl, H. J.  158, 163
215–216, 220–221, 225, 235, Saussure, F. de  2–6, 23, 53, 78,
251–253, 255 81, 83, 120, 129, 246–247, W
253, 255 Watson, J. C. E.  46, 49–51, 55,
P Schulz, E.  65, 67, 69, 146, 149, 78, 106–107, 112, 152
Parkinson, D. B.  13 151–152 Wehr, H.  15, 24–27, 29, 51, 84,
Passonneau, R. J.  186–187, 224 Shimron, J.  39–40, 51 86, 102, 107, 117, 124–125,
Poutsma, H.  170, 208 Soltan, U.  107 135–136, 190, 200, 205, 208,
Prince, A.  17, 19, 22, 26–29, 36, Sproat, R. W.  39 232, 241, 244
43, 47, 55–62, 112, 123, 131, Swadesh, M.  134 Wickens, G. M.  64, 67, 69, 78
134, 140, 147, 247 Wilson, A.  191
Prunet, J-F.  46, 49–50, 53, 55 T Wright, W.  17–18, 21, 64, 66,
Talmoudi, F.  161 68, 71, 74–75, 104, 107, 123,
R Tesnière, L.  108, 110 132, 135–136, 144–145, 149,
Ratcliffe, R. R.  41–45 Thatcher, G. W.  66, 68, 71 159–161, 252
Reig, D.  16 Thompson, J. R.  191
Reinhardt, C.  85 Thompson, S. A.  129–130 Z
Richardson, J.  64, 66, 68, 71 Thomson, G. H.  191 Zaborski, A.  123
Robertson Smith, W.  71 Zandvoort, R. W.  157
Ryder, S. A.  76–78 Zawaydeh, B. A.  46, 48–49
 Subject index

A C deverbals  79
absolute  66, 68 C1āC2 sequence  132–133, 143, diachrony  5–7, 27, 53, 116
accomplishments  163–164, 151, 246 dialects of Arabic
168–169 C2āC3 sequence  143–144, 151 Ammani  48
achievements  163, 169 case grammar  78 Bedouin  48
activities  163–164, 169, 200, 214 causatives  66–67, 69, 71, 75–76, Egyptian  183, 187
adjectives  145, 152 78–79, 118–119, 136–137 Hijazi  48
adjuncts  109 causativisation  123 Jordanian  47–48
adverbs  176–177, 186–187, 195, chess analogy  5–6 Khartoum  107
234 chi-square test  32–34 Moroccan  48
affixes  16, 43 Chitimacha  134 Palmyre (Syria)  101
agency  153–155, 239, 246, 250 citation form  17, 19 San’ani  49–50
agent moyen  95 Classical Arabic  72, 103 dictionary  24–25
Akkadian  75 Classicisms  27 didactic grammars  71, 73
Aktionsart  157–158 Coda  166, 168, 211, 218 diminutives  49–50
ambitransitivity  105, 109 cognate noun phrase  104 distributive  135
anti-causatives  79 colloquialisms  27 durative see durativity
antithesis  8 complements  108–109 continuatively  170, 208
antonyms  115 conativity  66–67, 101–102, durativity  153, 156, 168–169, 172,
aphasia see metathesis 254–255 175, 180, 185, 201, 214
Arab grammarians  77, 104 contextual deletion  109 dynamicity  45, 95, 165, 168–169,
Arabic, varieties of continuous  69, 162 172, 175, 179, 183, 190, 215, 234
see also dialects co-operation  68, 89
Classical see Classical Arabic CV skeleton  56, 60–61 E
Modern Standard see eductive  69
Modern Standard Arabic D effective  67, 69
arabiCorpus  13, 116, 118 data empty morphs  44–45
ascriptive  67 collection  24–27, 84–87, epenthetic vowel  19–20, 57
aspect  74 100, 110–112 equilibrium  7
compositional  158, 164, 166 corpus  12–13, 116, 191–195, ergative  69, 136
grammatical  158, 166 198, 201, 203, 211, 249 estimative  66–69, 71
lexical  12, 158, 162–166, 248 dictionary  13, 24–37, 87, etymons  51–52, 54–55
nominal  250–251 90–100, 107,110–114, 190, exceptions, lexical
atelicity  155, 197, 241, 246, 195, 198, 201, 203, 211, 249 see lexical exceptions
249–251, 255 declarative  66, 71 explicit specification approach
delocutives  67, 79, 202–203 71–73
B denominatives  26, 66–68, extensive  66–67, 70, 72
back-formation  90 70–72, 79, 92, 202–203 extrametricality  58
bayna  118–119 derivation, basis of  39
benefactive  69 desire  68 F
broken plurals  11, 59, 133–134, detransitivisation  10, 62, 112– factitive  66–67, 71, 75
137–140, 143–144 113, 116, 122, 141–142, 247, 254 feature marking  166, 224–226
280 The Arabic Verb

foot L nomen patientis  145, 154–155

prosodic see prosodic foot language nominal forms  12
trochaic  58 as a system  3 noun
form and meaning  2–5, 9, 62, change  7 of fragmentation  144
120, 132, 142, 247 lexical exceptions  78, 83, 90, of instrument  143
frequentative  66, 72 114–121, 142, 210, 249, 255 of intensity  144
method of  132 of occupation  144
G lexical gaps  30 trimoraic stems  59
gaps lexical issues  24–27 verbal see verbal nouns
lexical see lexical gaps lexicon, distribution in  23, 37, 51 nucleus  166
synchronic  27 ludlings  48
gemination  24, 72, 75, 77, O
79–80, 135 M onset  221, 224, 226
see also reduplication Mandarin  187
Google  13 markedness  135, 138, 252–253 P
grammarians  63–64 matrices  51, 53–54 pan-Semitic comparison  9
Arab  77, 104 maximal stem constraint  59 paradigms, verbal  16–17, 19–21,
grammatical rules  4 meaning 52
see form and meaning participles  5, 133, 143–144, 155,
H medial vowelling variants  241–246, 250–251
habitual  152, 162 18–19, 45, 93–100, 102 passive  5, 12, 56, 66, 68–69, 92,
Hebrew  74, 76 mediopassive  67, 69 147–149, 154–155, 197–198
Modern  39, 41, 47 metathesis  46–48 passivisation (passivisability)
scriptures  26 methodology  4–13, 24, 28, 12, 237, 251
Hegelian triad  8, 132 32–33, 132, 191–193 patiency  67, 119, 241, 250
homomorphs  26–27, 100, 125 middle voice  68–69, 148 pattern I variants
homonyms  26–27, 193 mimophones  53–54 see medial vowelling variants
hypocoristics  48–49 Modern Standard Arabic  24 patterns
moras  56 co-occurrence of  31–37, 83,
I morphemes  23, 40–44, 61–63, 92–100, 190–191
imperfect  16–17, 66 122, 131 derived  18–19
inceptives discontinuous  40 quadriliteral  21, 28–30, 37
durative  252 tiers  56 triliteral  28–36
of activity  228, 234–235, 251 morphology perfect  16–17
of state  228–235, 251 derivational  55, 84, 134 English  166–167
inchoative  148 nominal  133, 143–153 petition  68
indirect object  10, 86, 88, 110, templatic  43, 74 phi coefficient  33–35
114, 117 verbal  249 phonaesthemes  53
inflections, verbal  16–17 morphosemantics  64, 75 phonological word  56
ingressives  187, 213 mutuality  10, 66, 68–69, 83, plurals, broken
intensive (intensitive)  66–67, 115–118, 121, 154–155, 201, 254 see broken plurals
70, 72, 75–76, 78–80, 135–137 asymmetric  88–89 plurality  11, 80
intransitivity  67, 69, 85, 87, 92, potentiality  67
104, 115–116, 127–128, 140 N prefixes  16, 61–62
irregularity  10 n- prefix  27 pretence  68–69
iterative  66, 69, 72, 152, 156, 162, native speakers  13, 102, 169, 207, privative features  166, 224–225
165, 234 234, 240, 243–245 process  153, 239, 246
neuter  68–69 productivity  29–30
nomen agentis  145 progressive  162, 167, 170, 228
 Subject index 281

prosodic (templatic) analysis sign  2, 53, 255 transitivity  45, 67, 69, 75, 86,
56–62, 83, 123, 131 significance, statistical 91, 95
prosodic foot  56 see statistical significance through a preposition  105
psycholinguistic evidence  simulative  69, 91, 93, 98 type counts  191–193
46–47 slips of the tongue  47–48 typology  15, 80
statistical significance  32–33
Q states  45, 66, 69, 86–87, 95, 163, U
qad and laqad  161, 178 215–217 unaccusative  69, 76
quantitative analysis  29 stage-level  168–169
questionnaires  13, 169, 207, stativity see states V
232–234, 243–244 stems valency  10, 247
Qur’an  24, 26–27, 80 p-stem  16–22, 41, 45, 133, 146, hierarchical  129
159–162 increase  75–76, 123–124
R s-stem  16–21, 45, 133, 159–162 reduction or minimisation
radicals  51 minimal  58–59 112–117, 120–121, 254
reciprocity  10, 66, 68–69, 83, trimoraic see noun reversed hierarchical  115
112, 115–116, 118, 121, 154–156, verbal  60–62 structure codes  111
201, 254 stem-final syllables  58 tetravalency  110
reductionism  70–71 structuralism  2–3, 63, 83 zero-valency  110
reduplication  80, 136, 152 see also Saussure, F. de verbal nouns  104, 133, 143–144,
see also gemination suffixes  16 149–151, 252
reflexive  23, 66–69 syllable verbs
regularity  10 stem-final see stem-final biliteral verbs  17–18, 28, 39
repetitive  67, 80, 165 syllables irregular  15
requestative  69 structure theory  166 middle see middle voice
resultative  67, 69 types  56–58 of communication and speech
root, consonantal  16–17, 29, synchrony  5–7, 78 106
39–40, 42–44, 46–52, 55–56, synonyms  120 of giving  198–201
63, 79 synthesis  12 of posture in English  227–228
root-and-template model  43 system, linguistic  3, 7, 81 of surprise  194–198
rules, grammatical vocalised base  40–42, 50
see grammatical rules T vowels
t- affix  21–23 epenthetic see epenthetic
S ta- prefix  10, 21–23, 61, 83, vowel
sawfa  161 112–114, 120–122, 123, 141–142, lengthening  11
scientific method  8 247, 254 melody  22, 56, 60, 237–241,
semantics  11, 23, 64–81, 83–102, telicity  155, 164, 168–169, 175, 249
106, 247 179, 196, 204–210, 235 variants see medial vowelling
change in  27 temporal complexity  153, 156 variants
semelfactives  168–169, 198, tense  74, 159–162, 166
202–203 thesis, antithesis and synthesis  8 W
semi-causative  75 time schemata  162 word games  47–48
Semitic  22, 40–42, 46, 48–49, token counts  191–193
54, 73–81, 134 transitivisation  67, 125–126,
140, 247