(Phonetics and Phonology) Stephen Anderson, Patricia Keating, Sharon Hargus, Ellen M. Kaisse, Stephen R. Anderson, Patricia A. Keating-Studies in Lexical Phonology. Lexical Phonology-Elsevier Inc, Aca
(Phonetics and Phonology) Stephen Anderson, Patricia Keating, Sharon Hargus, Ellen M. Kaisse, Stephen R. Anderson, Patricia A. Keating-Studies in Lexical Phonology. Lexical Phonology-Elsevier Inc, Aca
Editors
Edited by
Sharon Hargus
Ellen M. Kaisse
Department of Linguistics
University of Washington
Seattle, Washington
Copyright © 1 9 9 3 by A C A D E M I C P R E S S , I N C .
All Rights Reserved.
No part of this publication may be reproduced or transmitted in any form or by any
means, electronic or mechanical, including photocopy, recording, or any information
storage and retrieval system, without permission in writing from the publisher.
Numbers in parentheses indicate the pages on which the authors' contributions begin.
xi
xii Contributors
Sharon Hargus
Ellen M . K a i s s e
REFERENCES
ELLEN M. KAISSE
SHARON HARGUS
Department of Linguistics
University of Washington
Seattle, Washington 98195
1. T H E O R E T I C A L C O M M O N D E N O M I N A T O R 1
1
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
2 Ellen M. Kaisse and Sharon Hargus
2. M O R P H O L O G I C A L Q U E S T I O N S
2.1 B r a c k e t E r a s u r e
2.2 I n t e r a c t i o n i s m
(1)
la -> //[ N [a
N
[If formalized as in (1),] then this phenomenon involves no interaction between morpho-
logical rules and those of the phonology. . . . On the other hand, if the phenomenon is
treated by multiple listings in the lexicon, then the separateness of morphology from pho-
nology is compromised (insofar as this type of lexical choice is a matter of morphology).
3. W O R D - I N T E R N A L P H O N O L O G I C A L D O M A I N S
( m e m b e r s of t h e c a t e g o r y mot), a n d that t h e m o r p h o l o g i c a l w o r d c o r r e s p o n d s to
a h i g h e r - o r d e r p h o n o l o g i c a l constituent, t h e p h o n o l o g i c a l c o m p o u n d , or mot':
(3) mot'
mot mot
S u c h data, o n c e c o n s i d e r e d u n u s u a l a n d p u z z l i n g , h a v e t u r n e d u p in a variety of
l a n g u a g e s , a n d a n a l y s e s similar t o that p r o p o s e d b y Booij a n d R u b a c h for Polish
can b e found in N e s p o r a n d Vogel ( 1 9 8 6 ) , C o h n ( 1 9 8 9 ) , K a n g ( 1 9 9 2 ) , a n d Inkelas
( 1 9 9 0 , this v o l u m e ) . F o r e x a m p l e , K a n g ( 1 9 9 2 ) s h o w s that A n n ' s ( 1 9 8 5 ) four-level
analysis of the lexical p h o n o l o g y of K o r e a n , w h i c h required the l o o p , yields to a
p r o s o d i c analysis with n o e x t r a o r d i n a r y theoretical devices a n d only o n e w o r d -
internal p h o n o l o g i c a l d o m a i n o n c e t h e i n d e p e n d e n c e of m o r p h o l o g i c a l r e p r e s e n -
tations a n d p-structure is r e c o g n i z e d . F o l l o w i n g Inkelas ( 1 9 9 0 ) , the version of
lexical p h o n o l o g y w h i c h r e c o g n i z e s this i n d e p e n d e n c e m a y b e called p r o s o d i c
lexical p h o n o l o g y ( P L P ) .
T h e level o r d e r i n g h y p o t h e s i s ( L O H ) , o n c e c o n s i d e r e d a tenet of classical lexi-
cal p h o n o l o g y , a t t e m p t e d to explain an o b s e r v e d correlation b e t w e e n affix o r d e r
a n d p h o n o l o g i c a l rule d o m a i n s b y p o s i t i n g a strict o r d e r i n g b e t w e e n affixes w h i c h
b e l o n g to different p h o n o l o g i c a l rule d o m a i n s ( t e r m e d LEVELS or STRATA in the
theory of level o r d e r i n g ) . Yet in recent years a n u m b e r of authors h a v e c h a l l e n g e d
the descriptive a d e q u a c y of the correlation in a n u m b e r of l a n g u a g e s (see, e.g.,
Aronoff a n d Sridhar, 1 9 8 7 ; H a r g u s , 1 9 8 8 ; M o h a n a n , 1986; Strauss, 1982). F o r
e x a m p l e , Aronoff a n d S r i d h a r n o t e for E n g l i s h that t h e (stress-affecting) level 1
affix -ity attaches q u i t e p r o d u c t i v e l y to w o r d s c o n t a i n i n g t h e (stress-neutral)
level 2 affix -able, c o u n t e r e x e m p l i f y i n g the L O H . M o r e o v e r , S p e n c e r ( 1 9 8 9 ) h a s
d e m o n s t r a t e d that m u c h of t h e w o r k attributed t o t h e L O H in t h e r e g u l a t i o n of
affix o r d e r in E n g l i s h m u s t b e a c c o m p l i s h e d b y m o r e restrictive m o r p h o l o g i c a l
s u b c a t e g o r i z a t i o n frames. In PLP, with its l o o s e r c o n n e c t i o n s b e t w e e n m o r p h o -
logical a n d p h o n o l o g i c a l r e p r e s e n t a t i o n s , there is n o r o l e for t h e L O H . Instead,
like word- external p h o n o l o g i c a l d o m a i n s , w o r d - i n t e r n a l p-structure in P L P is a s -
s u m e d to b e nested, r e g u l a t e d b y t h e strict layer h y p o t h e s i s ( S L H ) (Selkirk, 1984;
N e s p o r a n d Vogel, 1986), in w h i c h s m a l l e r p h o n o l o g i c a l d o m a i n s are strictly c o n -
tained within larger o n e s .
It is legitimate to a s k w h e t h e r or n o t the S L H is simply a reincarnation of the
L O H . W h i l e t h e o r d e r i n g restrictions i m p o s e d b y t h e L O H a n d t h e S L H are b o t h
hierarchical, P L P a l l o w s that affixes n o t b e u n d e r l y i n g l y specified for d o m a i n b u t
m a y instead CLITICIZE to the p h o n o l o g i c a l d o m a i n of an adjacent m o r p h e m e ,
t h e r e b y a c c o m m o d a t i n g w h a t w e r e p r e v i o u s l y c o n s i d e r e d to b e level o r d e r i n g v i o -
lations. A n Aronoff a n d S r i d h a r - s t y l e analysis of E n g l i s h w o r d s like ungrammat-
8 Ellen M. Kaisse and Sharon Hargus
j | stem
W W w
N [AdjM t u r n
]v a b l e
l Adj itylN
| j stem
w w w
[[[turn] a b i l ] i t y ] !2
4. T H E S T R I C T C Y C L E C O N D I T I O N : I S I T D E R I V A B L E
FROM UNDERSPECIFICATION?
T h e strict c y c l e c o n d i t i o n ( S C C ) , a d e v e l o p m e n t of p r e v i o u s c o n c e p t s intro-
d u c e d b y K i p a r s k y over t w o d e c a d e s , such as the (revised) alternation c o n d i t i o n
( R A C ) , is a c o n s t r a i n t o n the application of certain rules w h i c h p r e v e n t s t h e m from
a p p l y i n g w i t h i n a m o r p h e m e or from a p p l y i n g in e n v i r o n m e n t s w h i c h w e r e al-
r e a d y available o n a p r e v i o u s cycle. L e x i c a l p h o n o l o g i s t s generally a g r e e that
there is a r o b u s t set of p h e n o m e n a w h i c h m o t i v a t e s o m e t h i n g r o u g h l y like t h e
S C C or t h e R A C . In his c o n t r i b u t i o n to this v o l u m e , K i p a r s k y a r g u e s that the S C C
suffers from e m p i r i c a l difficulties w h i l e the R A C suffers from theoretical w e a k -
n e s s e s a n d a r g u e s instead that the p h e n o m e n a they c o v e r c a n b e d e r i v e d from
s e g m e n t a l underspecification. T h e applicability of the S C C to a rule w a s a clear
indication of its lexical status in classical lexical p h o n o l o g y , b u t this h y p o t h e s i s is
u n d e r attack from all directions, i n c l u d i n g that of its originator, for K i p a r s k y n o w
a r g u e s that true lexical rules, word-level rules, a n d postlexical rules m a y all e x -
hibit or fail to exhibit such effects. Iverson, in his c o n t r i b u t i o n , a r g u e s for a return
to the position that strict c y c l e effects are a characteristic s i m p l y of neutralization
rules, w h i l e n o n n e u t r a l i z i n g rules m a y fail to exhibit t h e m even if lexical. B o t h
K i p a r s k y ( 1 9 8 5 ) a n d B o r o w s k y ( 1 9 8 6 , this v o l u m e ) h a v e a r g u e d that w h a t e v e r t h e
precise f o r m of the c o n d i t i o n b l o c k i n g rule application in n o n d e r i v e d e n v i r o n -
m e n t s , it s h o u l d turn off at the last level of the lexicon, the w o r d level, rather than
postlexically. Finally, P o s e r s u g g e s t s in this v o l u m e that the d e r i v e d e n v i r o n m e n t
condition s h o u l d b e d i v i d e d into constraints o n a r e q u i r e m e n t for c o n c a t e n a t i o n
versus a r e q u i r e m e n t of p r e v i o u s application of p h o n o l o g i c a l rule.
K i p a r s k y ( 1 9 8 5 , this v o l u m e ) h a s i n t r o d u c e d a sort of c o n t e x t u a l underspecifi-
cation w h i c h h e uses in his a t t e m p t to a c c o u n t for n o n d e r i v e d e n v i r o n m e n t effects.
10 Ellen M. Kaisse and Sharon Hargus
5. S T R U C T U R E P R E S E R V A T I O N
(8) [+son]
I
[avoice]
[—voice]
T h e r e l e v a n c e of M c F a r l a n d a n d P i e r r e h u m b e r t ' s m e t i c u l o u s interpretation of
structure p r e s e r v a t i o n to H y m a n ' s D a g b a n i c a s e m a y n o w b e a p p a r e n t to the
reader. T h e H i g h T o n e S p r e a d rules, as their n a m e s imply, of c o u r s e c r e a t e linked
structures. If the lexical constraint in D a g b a n i rules out c o n t o u r tones as in (10);
tbu = ' t o n e - b e a r i n g u n i t ' ,
tbu
(11) H L
M
tbu tbu
(12) * T T
AA
(tbu) tbu (tbu)
6. A P P L Y I N G L E X I C A L P H O N O L O G Y T O D I A C H R O N I C C H A N G E
7. R U L E T Y P O L O G Y
8. A N I N V I T A T I O N
NOTES
The order of names is arbitrary; authorship was shared equally. Our remarks owe much
1
REFERENCES
Ahn, S.-C. (1985). The Interplay of Phonology and Morphology in Korean. Doctoral dis-
sertation, University of Illinois, Urbana-Champaign.
Allen, M. (1978). Morphological Investigations. Doctoral dissertation, University of Con-
necticut, Storrs.
18 Ellen M. Kaisse and Sharon Hargus
Connection (S. Inkelas and D. Zee, eds.), pp. 1 2 7 - 1 4 3 . CSLI Publications and Uni-
versity of Chicago Press, Chicago.
Kang, O. (1992). Word-internal prosodic words in Korean. Proceedings of the North East-
ern Linguistics Society 22, 2 4 3 - 2 5 7 .
Kiparsky, P. (1971). Historical linguistics. In A Survey of Linguistic Science (W. O. Ding-
wall, ed.), pp. 5 7 6 - 6 4 9 . University of Maryland, College Park.
Kiparsky, P. (1982). Lexical morphology and phonology. In Linguistics in the Morning
Calm (I.-S. Yang, ed.), pp. 3 - 9 1 . Hanshin Publishing Co., Seoul.
Kiparsky, P. (1983). Word formation and the lexicon. Proceedings of the 1982 Mid-
America Linguistics Conference, pp. 3 - 2 9 .
Kiparsky, P. (1984). On the lexical phonology of Icelandic. In Nordic Prosody III: Papers
from a Symposium (C. C. Elert et al., eds.), pp. 135-164. University of Umea.
Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook 2,
85-138.
Kiparsky, P. (1988). Phonological change. In Linguistics: The Cambridge Survey (F. J.
Newmeyer, ed.), 1, 3 6 3 - 4 1 5 .
Lantolf, J. (1977). Stress subordination: Evidence from Spanish. General Linguistics 17,
8-19.
McFarland, T., and Pierrehumbert, J. (1991). On ich-laut, ach-laut and structure preserva-
tion. Phonology 8, 171 - 1 8 0 .
Mascaro, J. (1976). Catalan Phonology and the Phonological Cycle. Doctoral dissertation,
Massachusetts Institute of Technology, Cambridge.
Mohanan, K. P. (1986). The Theory of Lexical Phonology. Reidel, Dordrecht.
Nespor, M., and Vogel, I. (1986). Prosodic Phonology. Foris, Dordrecht.
Pesetsky, D. (1979). Russian Morphology and Lexical Theory. Unpublished manuscript,
Massachusetts Institute of Technology, Cambridge.
Rubach, J. (1984a). Cyclic and Lexical Phonology: The Structure of Polish. Foris,
Dordrecht.
Rubach, J. (1984b). Rule typology and phonological interference. In Theoretical Issues in
Contrastive Phonology (S. Eliason, ed.), pp. 3 7 - 5 0 . Julius Groos, Stuttgart.
Selkirk, E. O. (1984). Phonology and Syntax: The Relation Between Sound and Structure.
MIT Press, Cambridge, Mass.
Selkirk, E. O. (1986). On derived domains in sentence phonology. Phonology Yearbook 3,
371-405.
Siegel, D. (1979). Topics in English Morphology. Garland, New York.
Spencer, A. (1989). Morphological Theory: An Introduction to Word Structure in Genera-
tive Grammar. Blackwell, Oxford.
Speas, P. (1990). Functional Heads and the Mirror Principle. Unpublished manuscript,
University of Massachusetts, Amherst.
Sproat, R. (1985). On Deriving the Lexicon. Doctoral dissertation, Massachusetts Institute
of Technology, Cambridge.
Strauss, S. (1982). Lexicalist Phonology of English and German. Foris, Dordrecht.
Suner, M. (1975). Spanish adverbs: Support for the phonological cycle? Linguistic Inquiry
6, 6 0 2 - 6 0 5 .
Zee, D., and Inkelas, S. (1991). Clitic groups and the prosodic hierarchy. Proceedings of
the West Coast Conference on Formal Linguistics 5, 5 0 5 - 5 1 9 .
ON THE SIMULTANEITY OF MORPHOLOGICAL
AND PROSODIC STRUCTURE
GEERT BOOIJ*
ROCHELLE LIEBER
* Vakgroep Taalkunde
Vrije Universiteit
1007 MC Amsterdam
The Netherlands
^Department of English
University of New Hampshire
Durham, New Hampshire 03824
1. I N T R O D U C T I O N
23
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
24 Geert Booij and Rochelle Lieber
tures are built from the b o t t o m u p , as follows. A c c o r d i n g to this theory, all mor-
p h e m e s have lexical entries w h i c h indicate their c a t e g o r y and subcategorization
( w h a t category, if any, they attach to, and in w h a t direction), as well as their p h o -
nological representations, lexical c o n c e p t u a l structures ( L C S s ) , and predicate ar-
g u m e n t structures ( P A S s ) . M o r p h e m e s are put t o g e t h e r a c c o r d i n g to their m o r -
2
and Wd.
A n o t h e r p o i n t in p r o s o d i c theory that w e take to b e uncontroversial is that m o r -
p h o l o g i c a l structure a n d p r o s o d i c structure n e e d not a l w a y s b e i s o m o r p h i c . Syl-
lable a n d foot b o u n d a r i e s d o n o t a l w a y s c o i n c i d e with m o r p h e m e b o u n d a r i e s , n o r
d o e s the p h o n o l o g i c a l w o r d a l w a y s m a t c h exactly w i t h the m o r p h o l o g i c a l w o r d
(see B o o i j , 1 9 8 5 ; Booij and R u b a c h , 1984).
It is at this point, however, that w e part c o m p a n y with the a b o v e m e n t i o n e d
theories of p r o s o d i c p h o n o l o g y . B o t h Selkirk ( 1 9 8 4 ) a n d N e s p o r a n d Vogel ( 1 9 8 6 )
a s s u m e that p r o s o d i c structure is built only after construction of w o r d s and sen-
tences has b e e n c o m p l e t e d . Selkirk ( 1 9 8 4 : 8 2 ) d u b s this a SYNTAX-FIRST a p p r o a c h .
Prosodic structure is created in t w o stages. B e l o w the w o r d level, prosodic struc-
ture is built after all m o r p h o l o g i c a l o p e r a t i o n s h a v e b e e n c o m p l e t e d . A b o v e the
w o r d level, p r o s o d i c structure is built as part of the postlexical p h o n o l o g y . N e s p o r
and Vogel ( 1 9 8 6 ) are s o m e w h a t less explicit than Selkirk about the overall orga-
nization of the g r a m m a r , b u t the picture that e m e r g e s from their w o r k is o n e in
w h i c h all p r o s o d i c structure is created as part of the postlexical phonology.
W e argue in w h a t follows that neither of t h e s e m o d e l s is correct. Rather, there
is g o o d reason to believe that m o r p h o l o g i c a l a n d p r o s o d i c structure are built at the
s a m e t i m e , from the b o t t o m u p , so that representations of w o r d s consist of t w o
s i m u l t a n e o u s structurings coexisting on distinct p l a n e s . T h i s a s s u m p t i o n has al-
w a y s b e e n m a d e in the standard version of the theory of lexical p h o n o l o g y , as
Morphological and Prosodic Structure 25
2. E V I D E N C E F O R S I M U L T A N E I T Y
(5) ejsz C * ] A ] A
In s u m , for t h e selection of t h e p r o p e r a l l o m o r p h of b o t h t h e c o m p a r a t i v e a n d t h e
i m p e r a t i v e m o r p h e m e it is crucial that b o t h t h e m o r p h o l o g i c a l a n d t h e p r o s o d i c
structuring of t h e stem b e available. T h e s e facts thus s u p p o r t both t h e theory of
lexical p h o n o l o g y that c l a i m s that p h o n o l o g y a n d m o r p h o l o g y are interspersed,
and t h e c l a i m that is t h e subject of this article, t h e simultaneity thesis.
T h e r e q u i r e m e n t of simultaneity n o t only manifests itself in t h e subcategoriza-
tions of b o u n d m o r p h e m e s in t h e lexicon, b u t also in t h e fact that there are p h o -
nological rules that refer s i m u l t a n e o u s l y t o b o t h types of structuring. L e t us call
s u c h rules B I P L A N A R R U L E S . 6
(8) D U T C H C O M P O U N D STRESS R U L E :
In a configuration [ A B ] , A is strong.N
the constituents that receive the labels " s t r o n g " and " w e a k " are p r o s o d i c cate-
gories (usually called PROSODIC WORDS) w h i c h d o m i n a t e p r o s o d i c categories like
syllable and foot.
T h e stress rule for n o m i n a l c o m p o u n d s of D u t c h is a typical lexical rule, b e -
c a u s e it can also h a v e e x c e p t i o n s (cf. B o o i j , 1977). T h a t is, it c a n n o t simply
b e p a r t of the m a p p i n g p r o c e d u r e that m a p s m o r p h o s y n t a c t i c structure into p r o -
sodic structure. It is, therefore, an instance of a lexical p h o n o l o g i c a l rule that re-
fers s i m u l t a n e o u s l y to the t w o kinds of hierarchical structuring of w o r d s dis-
cussed here.
A final e x a m p l e of a b i p l a n a r rule is the G e r m a n rule of S c h w a Insertion in
n o u n s (Hall, 1989; W i e s e , 1988). T h i s rule inserts the G e r m a n default vowel
s c h w a before an extrasyllabic c o n s o n a n t . F o r instance, the u n d e r l y i n g form of
Uebel ' e v i l ' is / y b l / . W h e n w e syllabify this u n d e r l y i n g form, the III r e m a i n s ex-
trasyllabic, b e c a u s e a c o d a cluster lb\l w o u l d violate the sonority s e q u e n c i n g g e n -
eralization. A s c h w a is then inserted to " s a v e " the HI. A s Hall ( 1 9 8 9 : 8 3 5 ) points
out, this rule only applies to n o u n s : S c h w a Insertion before c o n s o n a n t s also o c c u r s
in adjectives, b u t at a later level, a n d not only before extrasyllabic c o n s o n a n t s .
Therefore, the structural description of this rule h a s to refer s i m u l t a n e o u s l y to
the m o r p h o s y n t a c t i c c a t e g o r y " n o u n " and the p r o s o d i c notion " e x t r a s y l l a b i c
consonant."
T h e c o n c l u s i o n of this section is that b o t h s u b c a t e g o r i z a t i o n s of m o r p h e m e s
and p h o n o l o g i c a l rules s o m e t i m e s h a v e to refer s i m u l t a n e o u s l y to m o r p h o l o g i c a l
a n d p r o s o d i c information, and b o t h thus h a v e a b i p l a n a r character. In the next
section w e s h o w h o w the c o n c e p t of biplanarity can b e u s e d to solve a n u m b e r of
theoretical p r o b l e m s with respect to the interaction of p h o n o l o g y and m o r p h o l o g y .
3. T H E O R E T I C A L C O N S E Q U E N C E S
n o t i o n H E A D O P E R A T I O N as in (9).
the e x a m p l e s in (11).
the W d M I N
prefix w e r e to attach to the m o r p h o l o g i c a l constituent N ° , with con-
c o m i t a n t c o p y i n g of the p h o n e m i c m e l o d y of this constituent, w e w o u l d expect
the derivations illustrated in (12). N o t e that in (12) p r o s o d i c structure is illustrated
a b o v e the p h o n e m i c melody, m o r p h o l o g i c a l structure below.
(12) a. Wd M I N
Wd M I N
b. Wd M I N
Wd M I N
A A A A
o" (j a a a era a a a
AAA AAA
kintalpa m ul a r i
N N N ^ ^ ^ ^
Wd M I N
Wd M I N
Wd M I N
Wd M I N
A A A A
era CTO-CF a a a a a
AA AAA AA AAA
kin tal kintalpa = kintalkintalpa mular mulari = *mularmulan
(13) a. W d M I N
Wd M I N
b. W d M I N
Wd w
A A
AAA
kin tal pa
N N
Wd Wd
W d M I N W d M I N w d M I N w d M I N
A A A A
O" O" CTCTCT O" O" (J (J (J
AA AAA AA AAA
kin tal pa
this prefix is attached. W e a s s u m e , first of all, that the prefix pang- is attached to
a n o u n or verb stem, triggering the p h o n o l o g i c a l rule of sandhi a n d giving rise to
the s i m u l t a n e o u s m o r p h o l o g i c a l a n d p r o s o d i c structure illustrated in (16).
(16) a. F Wd M I N
I A
o" a a
A V
MIN
F W d
I Ao~ (J
A
pa
AA
muu tul
A V
(18) a. b.
r-
A A A A
pa muutul muu tul
- pa-mu-mu:tul
34 Geert Booij and Rochelle Lieber
(20) er ]A ] A
A
[ a (a )]
c w d
(22) -ity h
N o t e that it is not n e c e s s a r y to s u b c a t e g o r i z e -ity for a p r e c e d i n g p r o s o d i c word,
b e c a u s e n o r m a l l y suffixes b e c o m e p a r t of the p r e c e d i n g p r o s o d i c w o r d .
T h e representation of ungrammaticality will thus b e as follows (the asterisk
indicates the d e s i g n a t e d t e r m i n a l e l e m e n t of the p r o s o d i c w o r d grammaticality).
(23)
3 . 3 . Clitics
proclitic = - [ ]
x x
N o t e , h o w e v e r , that s u c h a s u b c a t e g o r i z a t i o n frame is i m p o s s i b l e in t h o s e c a s e s
w h e r e an enclitic is s u b c a t e g o r i z e d for a syntactic host on its right side, unless w e
also a l l o w for s u b c a t e g o r i z a t i o n frames of the following t y p e , with the b o u n d a r y
s y m b o l " = " n o n a d j a c e n t to the c a t e g o r y for w h i c h the clitic is s u b c a t e g o r i z e d .
(29) = enclitic , [ x ] .
x
4. CONCLUSION
ACKNOWLEDGMENT
The second author would like to acknowledge the generous support of the NWO, the
Netherlands Organization for Scientific Research, during the time that this article was
written.
Morphological and Prosodic Structure 41
NOTES
1
Cf. Anderson (1975), Booij (1988), Booij and Rubach (1984), and Cohn (1989). Simi-
lar ideas have been developed in an unpublished dissertation by Inkelas (1989).
2
See Jackendoff (1987), Levin and Rappaport (1986), and Rappaport and Levin (1988)
for discussion of LCS, PAS, and the relationship between them.
3
Note that even the grid-only theory of word stress requires that information about the
syllabification of words be available.
4
It is probably useful at this point to discuss some conceivable alternatives to the analy-
sis proposed here. First, note that the difference in stress pattern between, e.g., sociologie
and preventie cannot be predicted on the basis of the segmental composition of these
words. All present analyses of Dutch stress (e.g., Van der Hulst, 1984; Kager, 1989) assume
that in the normal case main stress falls on the penultimate syllable of words ending in -ie,
and therefore words in -ie with final stress have to be marked diacritically with a feature,
say [ + F ] , that takes care of this. Note, however, that we cannot make use of this feature
[ + F ] instead of stress to select the proper suffix, since it is the distributionally more re-
stricted suffix -ief thai requires that its class of base words be characterized, whereas the
words in -ie that are marked by the feature [ + F ] are those that cooccur with the more
general suffix -isch (note that there is no evidence in Dutch that the distribution of -ief is
determined by a diacritic feature [latinate]).
Observe, furthermore, that we cannot derive the adjectives from nominal stems without
-ie such as sociolog- and prevent-, because in that case the property that distinguishes the
bases of -ief and -isch would not be available, since it is located on the final syllable with
[i]. That is, this is a typical case of word-based morphology.
Another conceivable analysis is based on the idea expressed in Chomsky and Halle
(1968) that morphology precedes phonology, as suggested more recently by Halle and
Vergnaud (1987). The facts discussed here might be analyzed within such as theory as
follows. The morphology attaches both -ief and -isch to nouns in -ie. Prosodic structure is
created cyclically on the basis of the morphological structure of the complex words, and
there is a filter that states that words in which the suffix -ief is preceded by a syllable with
main stress are ill-formed. Note, however, that the final [i] of the base noun that bears main
stress before the suffix is added is deleted by rule before suffixes beginning with [i]. There-
fore the filter could only do its work if it applied before the application of the [i]-deletion
rule. Similarly, the filter would also have to apply before the application of the stress rules
that derive the stress pattern of the adjectives, because otherwise the crucial information
would get lost. That is, the filter cannot function as a prosodic well-formedness condition
on the surface form of these adjectives, as one would expect from filters. One could of
course envision a theory in which filters could be cyclic checking mechanisms, but such a
theory would be far less restrictive than the theory of lexical phonology we assume here; it
would, for example, leave the way open for the ordering of filters after particular rules in a
cycle. Thus, the filter approach that one is forced to accept here, if one rejects the basic
tenet of lexical phonology, seems to be completely ad hoc.
A final alternative analysis of the -iefl-isch facts might seem to be the following. We
might assume a surface filter at the end of the lexicon for checking the stress patterns of
42 Geert Booij and Rochelle Lieber
words with these suffixes, with a postlexical rule deleting [i] before [i]. This alternative is
not correct either, however; such a postlexical filter would incorrectly apply to words end-
ing in [i] followed by the clitic ie [i], for example, wie-ie is 'who he is' (note that -ie forms
one prosodic word with the preceding word).
Alternatively, one might assume a phonological rule that inserts ej in this context.
5
Note, however, that this cannot be a general rule of epenthesis, because ej is only inserted
in comparatives. It is therefore more natural to analyze these facts in terms of two compet-
ing suffixes. If one preferred to assume a phonological insertion rule here, this would still
make the point that lexical phonological rules have to refer to both morphological and
prosodic structure.
Parallel to the discussion above with respect to Dutch, one might consider an alternative
analysis in which a filter forbids the long allomorph to occur after a syllabified consonant.
Again, such a filter could not be a condition on the surface form of these words, because at
the surface all consonants will be syllabified due to the recursive application of syllabifi-
cation procedures.
Note that the examples which we discuss below provide direct evidence against the
6
claim in Cohn (1989:197) that, in languages which have prosodic structure not isomorphic
with morphological structure, the phonology will not refer to morphological structure.
This section is adapted from Lieber, Deconstructing Morphology. Word Formation in
7
Syntactic Theory, with permission from the publisher, the University of Chicago Press.
Copyright © 1992 by the University of Chicago.
Thanks to Patrocinio Schweikart for the Tagalog data. Further evidence that pang-
8
forms are adjectives is that they can occur in the position of modifiers of nouns, as in papel
pang-sulat 'paper for writing'.
McCarthy and Prince (1986) do not state the facts below in terms of morphological
9
subcategorization, so here we are taking the liberty of translating their basic idea into the
morphological framework we have adopted.
M c C a r t h y and Prince (1986) label this reduplication "Verbal Reduplication," but in
10
See also Booij and Rubach (1987) and the references cited there for data concerning
1 2
Dutch clitics.
The general distribution of ie, as with other pronouns, is accounted for by syntactic
1 3
REFERENCES
SHARON HARGUS
Department of Linguistics
University of Washington
Seattle, Washington 98195
1. I N T R O D U C T I O N
2. R E C E N T T H E O R I E S O F P H O N O L O G Y - M O R P H O L O G Y
INTERACTION
1. L a n g u a g e s m a y contain m a x i m a l l y t w o w o r d - i n t e r n a l p h o n o l o g i c a l d o -
m a i n s , to w h i c h m o r p h o l o g i c a l p r o c e s s e s are assigned.
2. Stem-level (cyclic, lexical) p h o n o l o g i c a l rules m a y p r e c e d e word-level, but
not stem-level, m o r p h o l o g y .
2.2. P h o n o l o g y M a y N o t P r e c e d e A n y M o r p h o l o g y
A n u m b e r of p h o n o l o g i s t s (Halle, H a r r i s , a n d V e r g n a u d , 1 9 9 1 ; H a l l e a n d Verg-
n a u d , 1987a, 1987b; O d d e n , this v o l u m e ; S z p y r a , 1987) h a v e recently e s p o u s e d a
resurrection of t h e p r e v i o u s " s t a n d a r d " t h e o r y of p h o n o l o g y - m o r p h o l o g y inter-
action, that p r o p o s e d in SPE a n d further e l a b o r a t e d b y H a l l e ( 1 9 7 3 ) . T h o u g h dif-
fering in details, all t h e s e " n o n i n t e r a c t i o n i s t " m o d e l s posit a m o r p h o l o g i c a l c o m -
p o n e n t w h i c h p r e c e d e s the (possibly level-ordered) p h o n o l o g i c a l c o m p o n e n t . T h e
noninteractionist position is well s u m m a r i z e d b y H a l l e a n d V e r g n a u d ( 1 9 8 7 a : 7 8 ) .
We make the traditional assumption that [morphological] rules are the province of a spe-
cial module, the morphology. In our theory, then, as in SPE, morphology is distinct and
separate from the phonology. Morphology interacts with phonology in that it creates the
objects on which the rules of phonology operate.
O d d e n (this v o l u m e ) p r o v i d e s t h e b e s t d e f e n s e of n o n i n t e r a c t i o n i s m , e x a m i n i n g
m u c h of the often-cited e v i d e n c e for lexical p h o n o l o g y and c o n c l u d i n g that this
e v i d e n c e can b e r e a n a l y z e d in a w a y w h i c h d o e s n o t support it. O d d e n also offers
an e m p i r i c a l a r g u m e n t against interactionism from M a l t e s e .
However, w h i l e s o m e of the e v i d e n c e usually cited for interactionism can a n d
s h o u l d b e r e a n a l y z e d , I s u g g e s t in the following section that not all the e v i d e n c e
can easily b e d i s m i s s e d .
3. P H O N O L O G Y P R E C E D I N G M O R P H O L O G Y
M c C a r t h y and P r i n c e ( 1 9 9 0 ) d i s c u s s a n u m b e r of a n a l y s e s of m o r p h o l o g i c a l
rules w h i c h apply to a " p r o s o d i c a l l y c i r c u m s c r i b e d " portion of a m o r p h o l o g i c a l
48 Sharon Hargus
3.1.1. GERMAN
TABLE 1
MORPHOLOGY REFERRING TO A DERIVED PHONOLOGICAL PROPERTY
English
Case 1 Stress Expletive infix McCarthy (1982)
Case 2 Stress -ize Marchand (1969), Hayes
(personal communication)
Finnish Stress Illative allomorphy Kiparsky (this volume),
Kanerva (1987), Keyser and
Kiparsky (1984)
Lappish Stress Various Bergsland(1976)
allomorphy
Koasati Syllabification Plural formation Kimball (1982), Martin (1988)
Hebrew Metathesis Extraction Bat-El (1986)
German
Case 1 Stress, -erZ-erei Hall (1990), Giegerich (1987)
epenthesis allomorphy
Dutch
Case 1 Stress -ie/isch Booij and Lieber (this volume)
allomorphy
3.1.2. L A P P
3.1.3. HEBREW
3.2. O v e r a p p l i c a t i o n of P h o n o l o g i c a l R u l e s
TABLE 2
3 . 3 . T h e D o m a i n of a P h o n o l o g i c a l R u l e E x c l u d e s a M o r p h o l o g i c a l P r o c e s s
TABLE 3
3.3.1. SUNDANESE
(21) /jiaian/
Nasal Harmony jiaian
Plural infixation ji-ar-alan
Nasal Harmony ji-ar-aian
3.3.2. GERMAN
(25) //pre N g/
Nasal Assimilation
g-Deletion 0
word-level m o r p h o l o g y -ung
3.3.3. LUISENO
(26) p t c k kw q qw ?
s S (s) X x w
v a'
mn rj
1 ry w h
O n e i m p o r t a n t p o i n t on w h i c h analysts d i s a g r e e is the p h o n e m i c status of [s], a
m a r g i n a l p h o n e m e at best. M u n r o a n d B e n s o n ( 1 9 7 3 ) and D a v i s (1976) a n a l y z e
[c] and [s] as a l l o p h o n e s w h i c h are in c o m p l e m e n t a r y distribution. To a c c o u n t for
the distribution of [ c ] / [ s ] , M u n r o a n d B e n s o n f o r m u l a t e the rule of Spirantization
given in (27).
(27)
(28) [s]/ #
/qe:nic7 [qe:nis] ''sqi
squirrel'
/qe:nic-um/ [qe:nicum] ''sqi
squirrels'
/ki:-ca] [kf:ca] ''hhoou s e ( n o m . ) '
/ki:-c/ [kis] ''hoi
house (acc.)'
(29) [§]/ [-cont]
/capomkat/ [capomkat] iiar'
/cacapomkat-um/ [caspumkatum] iiars'
/coka: yla-c/ [ c o k a : ylas] ' w a l k i n g stick (abs.)'
/no-coka: yla/ [ n o s k a : yla] (construct f o r m )
/pu:ci-l/ [pu:cil] 'eye ( a c c . ) '
/ p u : ci-la/ [ p u : sla] 'eye (nom.)'
/yo: vi-c-um-i/ [yo:vismi] 'meadow mice (acc.)'
/po-curo?a/ [pusro?ax] 'his leveling'
/curo?a/ [curo?a]- ' t o level'
58 Sharon Hargus
The surprising thing about the Munro/Benson article is that the authors approach the
subject as if the failure of the c to s rule is found only in connection with one particular
form of reduplication and furthermore only when this produces adjectives.
Reduplicative Plurals
(41) / m i c i - k u : t/ [ m i c k u : t] 'strangled
/mici/- [mici]- 'to strangle'
A n Analysis
Protracted R e d u p l i c a t i o n
a g e n t i v e a n d adjectival -kawut/-ku:t
c a u s a t i v e suffixes -ki, -kixa, -xami, -kixani
t e n s e / a s p e c t suffixes -q(a) -qat, -qus
y
3.3.4. JAVANESE
L i k e L u i s e n o , J a v a n e s e a p p e a r s to distinguish t w o word-internal p h o n o l o g i c a l
d o m a i n s , referred to here as stem- a n d word-level. I p r o p o s e an analysis in w h i c h
Phonology-Morphology Interface 61
slightly from t h o s e of b o t h D u d a s a n d S u h a r n o .
S i n c e m o s t of the rules d i s c u s s e d here involve the vocalic p h o n o l o g y of Java-
n e s e , the vowel i n v e n t o r y is given in (46).
(46) i u
e d o
a
[+high]
(60) a-^ol #
(63) IMPERATIVE
[turn] [turuo] 'sleep' (S22)
[ginan^ar] [ginanjaro] 'rewarded' (S22)
D u d a s d o e s n o t p r o v i d e e x a m p l e s of the c a u s a t i v e i m p e r a t i v e a n d locative i m p e r a -
tive, a n d these suffixes are not in S u h a r n o .
D u d a s notes that " t h e I m p e r a t i v e is the only f o r m a t i o n in the l a n g u a g e w h e r e
a suffixed stem-final / a / a p p e a r s on the surface with [o] as its final s e g m e n t "
(p. 110).
(64) IMPERATIVE
[lurjo] [rjlurjoo] 'go away'
[tako] [nakoo] 'come'
S u h a r n o and D u d a s n o t e various e x c e p t i o n s to ^ - R a i s i n g , m a i n l y in l o a n w o r d s
and place names.
A s n o t e d by S c h l i n d w e i n ( 1 9 8 9 ) , D o u b l i n g m u s t distinguish b e t w e e n stem a n d
suffix, since only the stem r e d u p l i c a t e s in ^ - R a i s i n g . T h i s is easily a c c o m p l i s h e d
via p r o s o d i c c i r c u m s c r i p t i o n of the b a s e ( M c C a r t h y a n d Prince, 1990). However,
the fact that the final vowel of the D o u b l e d form contains [a], rather than [o],
indicates that a - R a i s i n g m u s t counterfeed D o u b l i n g .
/dorja/
M O R P H O L O G Y : suffixation dorjane
Doubling dorja-dorja dorja-dorj ane
PHONOLOGY: ^-Raising o o o
dorjo-dorjo * dorjo-dorjane
3.4. Summary
4. T H E O R E T I C A L I M P L I C A T I O N S
(78) 9-^0 / L
[IMPER]
ACKNOWLEDGMENTS
I thank the following for comments on earlier versions of this paper: Ewa Czaykowska-
Higgins, Tracy Hall, Ellen Kaisse, Paul Kiparsky, Ove Lorentz, David Odden, Bill Poser,
Keren Rice, Yasushi Sato, Pat Shaw, Donca Steriade, Margaret Strong-Jensen, and an
anonymous reviewer.
NOTES
Aronoff notes that his prediction is empirically supported, at least by the reduplicative
1
cases he discusses in the 1988 paper: "To my knowledge, . . . there are no cases in the
literature where a cyclic phonological rule can be shown to misapply in a demonstrably
stem-level reduplication" (p. 6).
The following forms are cited in German orthography, with stress marked with an acute
2
accent.
1 thank Paul Kiparsky and Ove Lorentz for drawing Lapp to my attention.
3
The following words are cited in Bergsland's transcription system, which uses, among
4
other diacritics, acute accents to mark vowel quality distinctions. Stress is therefore indi-
cated by underlining the stressed nuclei. I have also added [.] to mark syllable division (not
provided in Bergsland's transcriptions).
The /h/ fails to be extracted in the above example because it is a tense affix, one of a
5
Additional primary sources consulted on Luiseno are Bright (1965), Kroeber and Grace
6
posed by Marantz.
These are marked S with a page number.
8
Other word-internal instances of the output of a-Raising are due to sharing of vowel
9
of -al nominals could be correctly assigned even if stress followed -al suffixation.
There is a complication to template c which will not be discussed here. See Dudas for
1 4
details.
REFERENCES
Booij, G. (1981). Rule ordering, rule application, and the organization of grammars. In
Phonologica 1980 (W. U. Dressier et al., eds.), pp. 4 5 - 5 6 . Innsbrucker Beitrage zur
Sprachwissenschaft.
Bright, W. (1965). Luiseno phonemics. International Journal of American Linguistics 31,
342-345.
Carrier, J. (1979). The Interaction of Morphological and Phonological Rules in Tagalog:
A Study in the Relationship between Rule Components in Grammar. Doctoral disser-
tation, Massachusetts Institute of Technology, Cambridge.
Carrier-Duncan, J. (1984). Some problems with prosodic accounts of reduplication. In Lan-
guage Sound Structure (M. Aronoff and R. Oehrle, eds.), pp. 2 6 0 - 2 8 6 . MIT Press,
Cambridge, Mass.
Chomsky, N., and Halle, M. (1968). The Sound Pattern of English. Harper and Row, New
York.
Clements, G. N. (1985). The problem of transfer in nonlinear morphology. In Cornell
Working Papers in Linguistics 5, 3 8 - 7 3 .
Cohn, A. (1989). Phonetic evidence for configuration constraints. In Proceedings of the
North Eastern Linguistics Society 1 9 , 6 3 - 7 7 .
Davis, J. F. (1976). Some notes on Luiseno phonology. International Journal of American
Linguistics 42, 192-216.
Dudas, K. (1974). A case of functional opacity: Javanese elative formation. Studies in the
Linguistic Sciences 4, 91 - 1 1 1 .
Dudas, K. (1975). The Phonology and Morphology of Modern Javanese. Doctoral disser-
tation, University of Illinois, Champaign-Urbana.
French, K. M. (1988). Insights into Tagalog Reduplication, Infixation and Stress from Non-
linear Phonology. Summer Institute of Linguistics, Arlington, Tex.
Giegerich, H. (1987). Zur Schwaepenthese im Standarddeutschen. Linguistische Berichte
112,449-469.
Hall, T. (1987). Schwa-Zero Alternations in German. Master's thesis, University of Wash-
ington, Seattle.
Hall, T. (1990). Syllable Structure and Syllable-related Processes in German. Doctoral
dissertation, University of Washington, Seattle.
Halle, M. (1973). Prolegomena to a theory of word-formation. Linguistic Inquiry 4, 3 - 1 6 .
Halle, M., Harris, J., and Vergnaud, J.-R. (1991). A reexamination of the stress erasure
convention and Spanish stress. Linguistic Inquiry 22, 141 - 1 5 9 .
Halle, M., and Vergnaud, J.-R. (1987a). An Essay on Stress. MIT Press, Cambridge, Mass.
Halle, M., and Vergnaud, J.-R. (1987b). Stress and the cycle. Linguistic Inquiry 1 8 , 4 5 - 8 4 .
Harris, J. (1989). The stress erasure convention and cliticization in Spanish. Linguistic
Inquiry 20, 3 3 9 - 3 6 4 .
Hayes, B. (1982a). Extrametricality and English stress. Linguistic Inquiry 13, 2 2 7 - 2 7 6 .
Hayes, B. (1982b). Metrical structure as the organizing principle of Yidiny phonology. In
The Structure of Phonological Representations, part 1 (H. v.d. Hulst and N. Smith,
eds.), pp. 97 - 1 1 0 . Dordrecht, Foris.
Hayes, B. (1990). Precompiled phrasal phonology. In The Phonology-Syntax Connection
(S. Inkelas and D. Zee, eds.), pp. 8 5 - 1 0 8 . CSLI Publications and University of Chi-
cago Press, Chicago.
Hayes, B., and Abad, M. (1989). Reduplication and syllabification in Ilokano. Lingua 77,
331-374.
Phonology-Morphology Interface 73
Innes, G. (1971). A Practical Introduction to Mende. School of Oriental and African Stud-
ies, London.
Kanerva, J. (1987). Morphological integrity and syntax: The evidence from Finnish pos-
sessive suffixes. Language 6 3 , 4 9 8 - 5 2 1 .
Kenstowicz, M. (1986). Multiple linking in Javanese. In Proceedings of the North Eastern
Linguistics Society 16, 2 3 0 - 2 4 8 .
Keyser, S. J., and Kiparsky, P. (1984). Syllable structure in Finnish phonology. In Language
Sound Structure (M. Aronoff et al., eds.), pp. 7 - 3 1 . MIT Press, Cambridge, Mass.
Kimball, G. (1982). Verb pluralization in Koasati. In 1982 Mid-America Linguistics Con-
ference Papers (F. Ingemann, ed.), pp. 4 0 1 - 4 1 1 . Department of Linguistics, Univer-
sity of Kansas, Lawrence.
Kiparsky, P. (1982). Lexical morphology and phonology. In Linguistics in the Morning
Calm (I.-S. Yange, ed.), pp. 3 - 9 1 . Hanshin Publishing Co., Seoul.
Kiparsky, P. (1984). On the lexical phonology of Icelandic. In Nordic Prosody III: Papers
from a Symposium (C. C. Elert et al., eds.), pp. 1 3 5 - 1 6 4 . University of Umea, Umea,
Sweden.
Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook 2,
85-138.
Kroeber, A. L., and Grace, G. (1960). The Sparkman Grammar of Luiseno (University of
California Publications in Linguistics, 16). University of California Press, Berkeley
and Los Angeles.
Marantz, A. (1982). Re reduplication. Linguistic Inquiry 1 3 , 4 3 5 - 4 8 2 .
Marchand, H. (1969). The Categories and Types of Present-Day English Word-Formation,
2nd ed. Beck, Munich.
Martin, J. (1988). Subtractive morphology as dissociation. Proceedings of West Coast Con-
ference on Formal Linguistics 8, 2 2 9 - 2 4 0 .
McCarthy, J. (1982). Prosodic structure and expletive infixation. Language 58, 5 7 4 - 5 9 0 .
McCarthy, J., and Prince, A. (1990). Foot and word in prosodic morphology: The Arabic
broken plural. Natural Language and Linguistic Theory 8, 2 0 9 - 2 8 4 .
Mester, R.-A. (1988). Studies in Tier Structure. Garland, New York.
Miller, W. (1961). Review of Kroeber and Grace, The Sparkman Grammar of Luiseno.
Language 37, 1 8 6 - 1 8 9 .
Mohanan, K. P. (1986). The Theory of Lexical Phonology. Reidel, Dordrecht.
Munro, P., and Benson, J. (1973). Reduplication and rule ordering in Luiseno. International
Journal of America! Linguistics 39, 1 5 - 2 1 .
Murti, M. S. (1984). An Introduction to Sanskrit Linguistics. D. K. Publications, Delhi.
Nespor, M., and Vogel, N. (1986). Prosodic Phonology. Foris, Dordrecht.
Odden, D., and Odden, M. (1986). Ordered reduplication in Kihehe. Linguistic Inquiry 16,
497-503.
Pesetsky, D. (1979). Russian Morphology and Lexical Theory. Unpublished manuscript,
Massachusetts Institute of Technology, Cambridge.
Pesetsky, D. (1985). Morphology and logical form. Linguistic Inquiry 16, 193-246.
Robins, R. H. (1957). Vowel nasality in Sundanese: A phonological and grammatical study.
In Studies in Linguistic Analysis, pp. 8 7 - 1 0 3 . Blackwell, Oxford.
Schlindwein, D. (1989). Reduplication in lexical phonology: Javanese plural reduplication.
In Arizona Phonology Conference, vol. 2 (S. L. Fulmer et al., eds.) (Coyote Papers,
9), pp. 1 1 6 - 1 2 4 . Department of Linguistics, University of Arizona, Tucson.
74 Sharon Hargus
SHARON INKELAS
Department of Linguistics
University of California
Berkeley, California 94720
1. I N T R O D U C T I O N
2. TERMINOLOGY
A n o t o r i o u s l y a m b i g u o u s e l e m e n t in the literature o n p h o n o l o g y a n d m o r -
p h o l o g y is the t e r m PROSODIC, w h i c h c r o p s u p also in the n a m e of the theory
discussed here. O n e application of this t e r m has b e e n to metrical constituents,
such as the m o r a , the syllable, the foot, a n d the w o r d tree. N o t a b l e e x a m p l e s of
this u s a g e are f o u n d in Ito's ( 1 9 8 6 ) p r i n c i p l e of p r o s o d i c licensing, w h i c h requires
every s e g m e n t a l string to be incorporated into units of h i g h e r metrical structure;
in w o r k on p r o s o d i c m o r p h o l o g y by M c C a r t h y and Prince ( 1 9 8 6 , 1990); and in
75
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
76 Sharon Inkelas
3. P R O S O D I C L E X I C A L P H O N O L O G Y
(1) P O S T L E X I C A L PROSODIC H I E R A R C H Y :
Utterance
i
Intonational p h r a s e
Phonological phrase
Phonological word
T h e appeal of this theory c o m e s from its ability to constrain in a natural m a n n e r
the a m o u n t of information m a d e available to e a c h c o m p o n e n t a b o u t the other. A n
i m p o r t a n t step in this direction is the indirect reference h y p o t h e s i s (2), w h i c h p r e -
vents p h o n o l o g i c a l rules from seeing any structure other than the string delimited
by a particular p-constituent.
(2) INDIRECT REFERENCE HYPOTHESIS: p h o n o l o g i c a l rules h a v e access only to
p-structure (i.e., not to m - s t r u c t u r e or c-structure).
Utterance
I
Intonational p h r a s e Post-lexical
Phonological phrase
I
— Phonological word
I
P Lexical
a
T h e parallel b e t w e e n these sublexical constituent types and traditional levels is
close. Just as affixes select for a t t a c h m e n t at a given level in standard lexical p h o -
nology, affixes s u b c a t e g o r i z e for a t t a c h m e n t to a given constituent type in the
m o d e l p r o p o s e d h e r e . (See Section 3.2.3 and Inkelas, 1989, for m o r e details.)
T o s u m u p , the p r o p o s a l is that lexical p h o n o l o g i c a l rule d o m a i n s are to b e
represented neither as metrical constituents n o r as m o r p h o l o g i c a l o n e s , b u t rather
as e l e m e n t s of p-structure, a u n i q u e h i e r a r c h y w h o s e c o v e r a g e generally c o r r e -
s p o n d s to that of m o r p h o l o g i c a l constituents. M e t r i c a l structure exists in a hier-
archy distinct from p-structure. It is a different level of representation, and its units
obey different constraints from those that g o v e r n p-structure. Just as phrasal stress
is assigned b y rules a p p l y i n g within the p h o n o l o g i c a l p h r a s e , so metrical constitu-
ents are built b y p h o n o l o g i c a l rules applying within p-constituents. M e t r i c a l c o n -
stituents are built b y p h o n o l o g i c a l rules, w h e r e a s p-constituents are built b y m o r -
phologically sensitive a l g o r i t h m s — a p r o c e s s to w h i c h I n o w turn.
3.2. L e x i c a l P - C o n s t i t u e n t s
(5) P - C O N S T I T U E N T F O R M A T I O N ALGORITHM ( P C F ) :
Wm -> <-*>m Wp
T h e application of P C F , parallel to an u n m a r k e d p h o n o l o g i c a l p h r a s i n g a l g o -
r i t h m o p e r a t i n g o n syntactic constituents, p r e s u p p o s e s the g e n e r a t i o n of the input
m - s t r u c t u r e . T h o u g h I c a n n o t p r e s e n t t h e full details of a m o r p h o l o g i c a l theory
h e r e , I a s s u m e the e x i s t e n c e of a parallel m - c o n s t i t u e n t f o r m a t i o n w h i c h serves a
function c o m p a r a b l e to Selkirk's ( 1 9 8 2 ) rewrite rules for m o r p h o l o g y .
(11) U n d e r l y i n g representation: s t e m
PCF t s t e m
]p«
p h o n o l o g i c a l rules [stem] p a
PCF [stem] p p
p h o n o l o g i c a l rules [stem] p p
PCF [stem] p a )
p h o n o l o g i c a l rules
3.3.1. COMPOUNDS
S o m e c o m p o u n d s d o b e h a v e as a unit for p h o n o l o g i c a l p u r p o s e s . F o r e x a m p l e ,
c o m p o u n d s in G r e e k ( N e s p o r and Vogel, 1986) form a unitary d o m a i n for stress
rules such that the surface stress of a c o m p o u n d n e e d not reflect the surface stress
of either c o m p o n e n t w o r d . T h e internal m o r p h o l o g i c a l b o u n d a r y of G r e e k c o m -
p o u n d s d o e s not interfere with the d o m a i n of the stress rule.
T h i s split in the b e h a v i o r of c o m p o u n d s for p h o n o l o g i c a l p u r p o s e s has led to
the h y p o t h e s i s that w h i l e c o m p o u n d s of the type found in G r e e k c o r r e s p o n d to a
single p-constituent (12a), c o m p o u n d s like those in Italian c o r r e s p o n d to t w o
constituents (12b).
3 . 3 . 2 . INVISIBILITY
3.3.3. P-SUBCATEGORIZATION
(25) a. U N D E R L Y I N G REPRESENTATION
OF A VISIBLE SUFFIX.' = [[ ] p Suffix] p
b. U N D E R L Y I N G REPRESENTATION
O F A N I N V I S I B L E SUFFIX: =[[ ] ] p Suffix
A F F I X A T I O N O F [[ ] ] al:
p p [ [ h i s t o r i c ] ] al
p p
PCF: —
PHONOLOGICAL RULES APPLY TO: [historic] p al
4. A F F I X A T I O N A N D C Y C L I C I T Y I N P R O S O D I C
LEXICAL PHONOLOGY
4.2. Preaffixal S t e m C y c l e i n S t r a t u m 1
(29) P C F FEEDS A F F I X A T I O N :
INPUT: stem
PCF: [stem]p
F I R S T CYCLE O F P H O N O L O G I C A L R U L E S : [stem] p
A F F I X A T I O N O F [[ ] s u f f i x ] :
p p [stem-suffix] p
S E C O N D CYCLE O F P H O N O L O G I C A L R U L E S : [stem-suffix] p
PCF: —
4.2.1. CARIB
(32) O D D - N U M B E R E D STRESS:
N o . of syllables Stem Gloss
1 wo 'to b e a t '
2 e:ro 'this'
ku:pi 'bathe'
3 eimaka 'to c o m b a p a r t i n g '
tdikuwa 'polishing-stone'
4 d:rawd:ta 'howling monkey'
pd: yawd: ru 'cassava beer'
5 auwanoipono ' c a u s i n g laughter'
ko: kapo: take ' y o u will have m e bitten'
(33) E V E N - N U M B E R E D STRESS:
N o . of syllables Stem Gloss
(34) aki: ma 'tease' ki'n-d: ki: ma: -no ' h e teases h e r '
awo: mi 'to get u p ' ay-a: wo: mi'-i ' y o u m u s t not get
up'
epd: nopi' 'help' ay-e:pd:nopi 'your being helped'
erne: pa 'to teach' ki'n-e:me:pa: -no ' h e teaches t h e m '
etdmboka 'untie' kl'n-e: tdmboka: -no ' s h e unties it'
eyd: to 'call' ki'n-e: yd: to-ya: -tor) 'they call h i m '
kurd: ma 'cure' si-ku: rd: ma-e T cure h i m '
kurd: ma ' l o o k after' ki'-ku: rd: ma-ko ' y o u m u s t look af-
ter m e '
ki'-ku: rd:ma-i 'you m u s t not look
after m e '
i-ku: rd: ma-ko ' y o u m u s t look af-
ter h i m '
Deriving Cyclicity 93
REGULAR IRREGULAR
UNDERLYING
REPRESENTA- karai kdmi
TION:
PCF: [karai] [kami]
INVISIBILITY: [blocked b y m i n i m a l i t y ]
STRESS: [karai]
AFFIXATION: [karaima] [kamiro]
INVISIBILITY: ka [raima] ka [miro]
STRESS:
(late rules) *[ka:raima] [ka: miro]
'blacken' ' c a u s e to b e c o m e
pale r e d '
96 Sharon Inkelas
I N I T I A L V INVISIBILITY: ku [ r i y a r a ] p ka [rawasi] p
'canoe' 'rattle'
[kuriyarari'lp [karawasirilp
I N I T I A L V INVISIBILITY: — —
STRESS ASSIGNMENT: [kuriyarari] p [karawasirilp
PCF: — —
(late rules) [kuriyarari'lp [karawasirilp
'canoe' 'rattle'
98 Sharon Inkelas
[okomori] p [yamaturi]p
V O W E L INVISIBILITY: o [komori] p —
STRESS: o [k6mori] p [yamaturijp
PCF: — —
(late rules) [oko:mori] p [ya:matu:ri]p
'wasp' 'basket'
AFFIXATION: (okomori) m
[okomori]p
V O W E L INVISIBILITY: o [komori'lp
STRESS RULE: o [komorilp
PCF: [okomo] p
V O W E L INVISIBILITY: o [komo] p
'wasp(poss.)' 'wasp'
W e noted earlier that the source from w h i c h P L P derives cyclic rule application
is the p r e s e n c e of p - s u b c a t e g o r i z a t i o n frames in t h e lexical entries of affixes. A n y
stratum associated with affixation will necessarily b e associated with cyclic p h o -
nological rule application.
H o w e v e r , this leaves o p e n the possibility that a particular stratum m i g h t h a v e
n o affixes associated with it. In fact, this is the c a s e with the (postcyclic) lexical
w o r d stratum, as d e s c r i b e d b y K i p a r s k y ( 1 9 8 5 ) a n d Booij and R u b a c h ( 1 9 8 7 ) . T h e
d o m a i n of word-level lexical rules, this s t r a t u m d o e s n o t allow any m o r p h o l o g y
to take p l a c e .
In t e r m s of PLP, w e m a y c a p t u r e the nonaffixational nature of a stratum with a
constraint in u n d e r l y i n g representation against affixes s u b c a t e g o r i z i n g for c o n -
stituents of the a p p r o p r i a t e category. In other w o r d s , Affixation is t u r n e d off at a
particular level of the hierarchy. In t h e c o r r e s p o n d i n g constraint in (54), only the
m - f r a m e is m e n t i o n e d ; the a s s u m p t i o n is that since affixes will a l w a y s h a v e
m - f r a m e s , r e p e a t i n g the constraint o n p-frames w o u l d at best b e r e d u n d a n t . 21
(58) ...
affixational s t r a t u m k (cyclic)
non-affixational s t r a t u m k+1 ( n o n c y c l i c )
non-affixational stratum (noncyclic)
4.3.1. MANAM
CLITICIZATION [?udo?i?i] . p
[?ud6?i?i] . p
Postlexical
phonology
STRESS RULE: [?udo?i?i] . p
4.4. S u m m a r y of Predictions
5. CONCLUSION
ACKNOWLEDGMENTS
During the preparation of this paper I have benefitted from discussions with Sharon
Hargus, Larry Hyman, Richard Sproat, Draga Zee, an anonymous reviewer, and partici-
pants in the Lexical Phonology Workshop at the University of Washington.
NOTES
For arguments against treating the clitic group as a distinct p-constituent, see Buckley
1
(1991), Inkelas (1989), Kanerva (1990), Zee (1988), and Zee and Inkelas (1991). Additions
to this hierarchy have been proposed as well: Condoravdi (1990) has demonstrated that
Modern Greek requires a constituent (the "minimal phrase") between the phonological
phrase and the phonological word; Kanerva (1990) argues from Chichewa for a focal
phrase between the intonational phrase and the phonological phrase.
This move may also be seen as a meeting of prosodic hierarchy theory with Sproat's
2
(1985) proposal that phonological structure and morphological structure are distinct, dif-
fering in the geometry of their bracketing.
In (5) and subsequent examples, angle brackets indicate m-constituency, and square
3
accommodate the apparent need for juncture rules. Another use to which access to internal
structure has been put involves "strict cycle" rules, which putatively apply only across
(internal) morpheme boundaries (Kiparsky, 1982; Mascaro, 1976); but see Kiparsky (this
volume) for a reanalysis not requiring morphological boundary information.
5
Notably excluded from this set are the bracketing paradoxes discussed by, among oth-
ers, Sproat (1985). Certain apparent paradoxes can be handled elegantly in terms of p-
structure; for example, Nespor and Vogel (1986) and Booij and Rubach (1987) have ana-
lyzed certain affixes that do not behave as though they are phonologically part of the word
as exceptionally constituting phonological words on their own. However, I do not at present
see a way to extend this approach to all of the cases discussed by Sproat (see also Harris,
1989), in particular to cases in which a level 2 affix appears to be inside a level 1 affix.
6
It also fails to apply to prefix-final consonants, which Nespor and Vogel account for by
assigning phonological word status to the prefixes in question (see previous footnote).
7
Of course, not all invisibility effects result from the application of a general phono-
logical rule such as that in (17). English, among other languages (e.g., Japanese), possesses
suffixes which are inherently invisible. Especially in a theory which restricts access by
phonological rules to morphological structure, these phenomena pose a challenge; I return
to it in the next section.
8
A problem for postulating a bimoraic minimum in Amele is the large number of mono-
moraic words [e.g., be 'neck (front)', co 'lips', du 'neck (back)', su 'breast (woman's)',
ca 'add', jo 'house']. However, Ito and Hankamer (1989) have argued for Turkish that all
the forms which appear to violate minimal size conditions in that language are underived;
Ito (1990) has made a similar observation for Japanese. If minimality conditions are en-
forced only on (morphologically) complex words, then the Amele examples just cited
would not constitute counterexamples to a bimoraic minimal word condition. Unfortu-
nately, the derived environment hypothesis is difficult to falsify. For example, Amele has a
process of possessive formation which has a zero allomorph in the third person. Whether
we consider the monomoraic possessive co-0 'his/her lips' to be derived or underived
determines the success or failure of a derived environment minimal size condition in the
language. I leave the issue open here, since there is another option: if necessary, we could
always claim that minimal size constraints block phonological rules (such as invisibility
assignment), but not morphological rules. Similar claims about other phonological well-
formedness constraints are found elsewhere in constraint-based phonological theory (Hy-
man, 1991;Paradis, 1988).
9
In this the lexical PCF algorithm conforms to a precedent already set by a number of
postlexical phrasing algorithms in the literature. For example, phrasing algorithms devel-
oped for Greek (Condoravdi, 1990), Hausa (Inkelas, 1988), Korean (Cho, 1990), and
Shanghai (Selkirk and Shen, 1990) possess explicit elsewhere clauses invoked whenever
no more specific source is available to phrase the relevant material.
1 am assuming a general morpheme-based theory of morphology a la Lieber (1980)
1 0
or Kiparsky (1983), in which all affixes, as well as stems and roots, are lexically listed
items as opposed to rules (Anderson, 1986).
Proposals for a notion of phonological subcategorization exist in the literature, but
1 1
none involves the particular kind of phonological constituency at issue here. For example,
Klavans's (1985) phonological subcategorization frames refer to syntactic constituents on
which clitics phonologically lean; Booij and Lieber's prosodic subcategorization frames
106 Sharon Inkelas
(this volume) express the sensitivity of particular affixes to certain metrical properties of
the base. Sproat's (1985) proposal is the closest relative. Separating out issues of linear
precedence from issues of dominance, Sproat proposes to encode the former in the phono-
logical entry and the latter in the morphological entry of an affix. Although Sproat did not
invoke our notion of p-structure per se, his insight of separate phonological and morpho-
logical entries is also central to the proposal made here.
1 2
If roots are inserted as the first step in a derivation, and if morphological sensitivity is
always inward, then root sensitivity to phonological properties of adjacent morphemes
would be ruled out on independent grounds, except perhaps in root-root compounding, as
occurs, for example, in Japanese. Although the compounding process itself imposes pho-
nological properties in that language, I am not aware of any root-specific allomorphy based
on the phonological characteristics of the sister root there, or elsewhere.
1 3
The morphological process not discussed here is compounding; Inkelas (1989) ac-
counts for the two types of compounds mentioned earlier by positing two compounding
rules. One (Mcompounding) refers only to m-structure, and its output feeds right into PCF.
Such compounds correspond to a single p-constituent. The other compounding rule (Pcom-
pounding) refers both to m- and to p-structure, imposing p-structure on its output such that
the two elements of the compound correspond to distinct p-constituents. A more specific
source of p-structure for the compound, Pcompounding overrides PCF exactly as affix-
ation does. Both kinds of compounds, however, are associated with a cycle of phonological
rules, either by virtue of triggering PCF by default, or by virtue of construction-specific
p-structure assignment.
For arguments in favor of such a view, see Inkelas (1989). Evidence that it is necessary
1 4
comes from the ability of bound roots, which lack p-structure, to combine with affixes,
which possess p-frames. The p-frame of the affix interprets the root as its p-sister.
Note that the suffixes -no, -tor}, -se, and -ma always impose length on the preceding
1 5
vowel; the penultimate long vowels of words ending in these suffixes is thus a local effect
and can be overlooked for present purposes.
In possessing a cyclic, structure-preserving stress rule, Carib thus patterns with Diyari
1 6
and Warlpiri, as analyzed in Poser (1989), and with Greek, as analyzed in Steriade (1988).
A reviewer aptly points out that correlation between vowel quantity and stress is a
1 7
property of iambic, rather than trochaic, systems (Hayes, 1987, 1991). To the extent that
Carib violates this generalization, the analysis is suspect. However, in defense of the tro-
chaic account proposed here, I note that vowel length is noncontrastive in Carib, assigned
by a late rule that applies after stress foot construction. Moreover, note that vowel length-
ening affects only the first two stress feet. On the assumption that words are exhaustively
parsed by the footing algorithm, then vowel lengthening does not target all stressed syl-
lables. Finally, although vowel lengthening occurs only in open syllables, suggesting that
coda consonants contribute to weight, closed syllables do not interfere with the assignment
of binary feet (moxkd:ro 'they' vs. tuxkusi 'arrow'; ma?md: takd.ra 'species offish' vs.
oxkoto:poti'the cutting of you into pieces'). The conclusion is that we are not dealing with
prototypical quantity-sensitive feet.
In Inkelas (1989) I instead assumed that odd-numbered stems possess an underlyingly
1 8
long vowel and that the extrametricality rule only targeted an initial mora—thus failing to
render the entire first syllable of such forms invisible. But this analysis is problematic in
that vowel length is not otherwise phonemic in the language, and admitting it lexically
Deriving Cyclicity 107
would violate structure preservation (Kiparsky, 1982). By contrast, stress feet are clearly
needed in the lexical phonology. Admitting them in underlying representation is only a
small extension of existing apparatus. For the use of underlying feet in the literature to
capture exceptional stress patterns, see, e.g., Hayes (1981), Hammond (1989), and many
others.
1 9
The syllable canon in Carib is CV(C), such that removing the first CV from a word
will eliminate the stress-bearing capacity of the initial syllable.
W h a t still remains to be explained is why "weak" suffixes, such as those illustrated
2 0
in the examples involving disyllabic stems, do not display the sensitivity of "strong" suf-
fixes to the initial consonant-vowel status of the stem. One possibility, explored in Inkelas
(1989), is to assign to all "weak" suffixes the property of imposing invisibility on initial
consonants. Though it may seem complex at first glance, this account has the advantage of
making the more common, unmarked suffixes duplicate the effects of PCF, which is the
case in, for example, English.
2 1
At worst, it would incorrectly prohibit the representation of clitics, which subcatego-
rize for word-level or higher constituents in the p-hierarchy (Inkelas, 1989; Zee and Inkelas,
1991).
REFERENCES
(S. Inkelas and D. Zee, eds.), pp. 4 7 - 6 1 . CSLI Publications and University of Chicago
Press, Chicago.
Cohn, A. (1989). Stress in Indonesian and bracketing paradoxes. Natural Language and
Linguistic Theory 7, 167 - 216.
Condoravdi, C. (1990). Sandhi rules of Greek and prosodic theory. In The Phonology-
Syntax Connection (S. Inkelas and D. Zee, eds.), pp. 6 3 - 8 3 . CSLI Publications and
University of Chicago Press, Chicago.
Halle, M., and Kenstowicz, M. (1991). The free element condition and cyclic versus non-
cyclic stress. Linguistic Inquiry 2 2 , 4 5 7 - 5 0 1 .
Halle, M., and Mohanan, K. P. (1985). Segmental phonology of Modern English. Linguis-
tic Inquiry 16, 57 - 1 1 6 .
Hammond, M. (1989). Lexical stresses in Macedonian and Polish. Phonology 6, 1 9 - 3 8 .
Hargus, S. (1988). The Lexical Phonology ofSekani. Garland, New York.
Harris, J. (1983). Syllable Structure and Stress in Spanish. MIT Press, Cambridge, Mass.
Harris, J. (1989). The stress erasure convention and cliticization in Spanish. Linguistic
Inquiry 20, 3 3 9 - 3 6 3 .
Hayes, B. (1981). A Metrical Theory of Stress Rules. Doctoral dissertation, Massachusetts
Institute of Technology, Cambridge.
Hayes, B. (1987). A revised parametric metrical theory. Proceedings of the North Eastern
Linguistics Society 17, 2 7 4 - 2 8 9 .
Hayes, B. (1989). The prosodic hierarchy and meter. In Rhythm and Meter (P. Kiparsky
and G. Youmans, eds.), pp. 2 0 1 - 2 6 0 . Academic Press, Orlando.
Hayes, B. (1991). Metrical Stress Theory: Principles and Case Studies. Unpublished
manuscript, University of California, Los Angeles.
Hoff, B. (1968). The Carib Language. Nijhoff, The Hague.
Hyman, L. (1991). Imbrication in Cibemba. Unpublished manuscript, University of Cali-
fornia, Berkeley.
Inkelas, S. (1988). Prosodic effects on syntax: Hausa 'fa'. Proceedings of the West Coast
Conference on Formal Linguistics 7, 3 7 5 - 3 8 9 .
Inkelas, S. (1989). Prosodic Constituency in the Lexicon. Doctoral dissertation, Stanford
University, Stanford, Calif.
Inkelas, S., and Zee, D. (1988). Serbo-Croatian pitch accent: The interaction of tone, stress
and intonation. Language 64, 2 2 7 - 2 4 8 .
Ito, J. (1986). Syllable Theory in Prosodic Phonology. Doctoral dissertation, University of
Massachusetts, Amherst.
Ito, J. (1990). Prosodic minimality in Japanese. In CLS 26-11: Papers from the parasession
on the syllable in phonetics and phonology (K. Deaton, M. Noske, and M. Ziolkowski,
eds.), pp. 213-239.
Ito, J., and Hankamer, J. (1989). Notes on monosyllabism in Turkish. In Phonology at
Santa Cruz 1 (J. Ito and J. Runner, eds.), pp. 6 1 - 7 0 . Syntax Research Center, Univer-
sity of California, Santa Cruz.
Kaisse, E. (1985). Connected Speech. Academic Press, New York.
Kanerva, J. (1990). Focusing on phonological phrases in Chichewa. In The Syntax-Pho-
nology Connection (S. Inkelas and D. Zee, eds.), pp. 1 4 5 - 1 6 1 . CSLI Publications and
University of Chicago Press, Chicago.
Kiparsky, P. (1982). Lexical morphology and phonology. In Linguistics in the Morning
Calm (I.-S. Yang, ed.), pp. 3 - 9 1 . Hanshin Publishing Co., Seoul.
Deriving Cyclicity 109
Kiparsky, P. (1983). Word formation in the lexicon. In Proceedings of the 1982 Mid-
America Linguistics Conference (F. Ingemann, ed.), pp. 3 - 2 9 . University of Kansas,
Lawrence.
Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook 2,
85-138.
Klavans, J. (1985). The independence of syntax and phonology in cliticization. Language
61, 8 5 - 1 2 0 .
Lichtenberk, F. (1983). A Grammar of Manam. University of Hawaii Press, Honolulu.
Lieber, R. (1980). On the Organization of the Lexicon. Doctoral dissertation, Massachu-
setts Institute of Technology, Cambridge.
Mascaro, J. (1976). Catalan Phonology and the Phonological Cycle. Doctoral dissertation,
Massachusetts Institute of Technology. Distributed by the Indiana University Linguis-
tics Club.
McCarthy, J., and Prince, A. (1986). Prosodic Morphology. Unpublished manuscript, Uni-
versity of Massachusetts, Amherst, and Brandeis University, Waltham, Mass.
McCarthy, J., and Prince, A. (1990). Foot and word in prosodic morphology: The Arabic
broken plural. Natural Language and Linguistic Theory 8, 2 0 9 - 2 8 4 .
Mohanan, K. P. (1982). Lexical Phonology. Doctoral dissertation, Massachusetts Institute
of Technology, Cambridge.
Mohanan, K. P. (1986). The Theory of Lexical Phonology. Reidel, Dordrecht.
Nespor, M. (1984). ' T h e phonological word in Italian." In Advances in Non-linear Pho-
nology (H. van der Hulst and N. Smith, eds.), pp. 193-204. Foris, Dordrecht.
Nespor, M., and Vogel, I. (1982). Prosodic domains of external sandhi rules. In The Struc-
ture of Phonological Representations, part 1 (H. van der Hulst and N. Smith, eds.),
pp. 2 2 5 - 2 5 5 . Foris, Dordrecht.
Nespor, M., and Vogel, I. (1986). Prosodic Phonology. Foris, Dordrecht.
Paradis, C. (1988). On constraints and repair strategies. The Linguistic Review 6, 7 1 - 9 7 .
Pesetsky, D. (1979). Russian Morphology and Lexical Theory. Unpublished manuscript,
Massachusetts Institute of Technology, Cambridge.
Poser, W. (1984). The Phonetics and Phonology of Tone and Intonation in Japanese. Doc-
toral dissertation, Massachusetts Institute of Technology, Cambridge.
Poser, W. (1989). The metrical foot in Diyari. Phonology 6, 117 - 1 4 8 .
Pulleyblank, D. (1986). Tone in Lexical Phonology. Reidel, Dordrecht.
Roberts, J. (1987). Amele. Croom Helm, London.
Selkirk, E. (1978). On prosodic structure and its relation to syntactic structure. In Nordic
Prosody II (T. Fretheim, ed.), pp. 11 - 4 0 . TAPIR, Trondheim.
Selkirk, E. (1980). Prosodic domains in phonology: Sanskrit revisited. In Juncture (M.
Aronoff, ed.), pp. 1 0 7 - 1 2 9 . Anma Libri, Saratoga, Calif.
Selkirk, E. (1982). The Syntax of Words. MIT Press, Cambridge, Mass.
Selkirk, E. (1986). On derived domains in sentence phonology. Phonology Yearbook 3,
371-405.
Selkirk, E., and Shen, T. (1990). Prosodic Domains in Shanghai Chinese. In The Pho-
nology-Syntax Connection (S. Inkelas and D. Zee, eds.), pp. 3 1 3 - 3 3 7 . CSLI Publi-
cations and University of Chicago Press, Chicago.
Sproat, R. (1985). On Deriving The Lexicon. Doctoral dissertation, MIT, Cambridge, Mass.
Sproat, R. (1986). Malayalam Compounding: a Non-Stratum Ordered Account. Proceed-
ings of the West Coast Conference on Formal Linguistics 5, 2 6 8 - 2 8 8 .
110 Sharon Inkelas
Steriade, D. (1988). Greek accent: A case for preserving structure. Linguistic Inquiry 19,
271-314.
Vogel, I. (1989). The clitic group in prosodic phonology. To appear in Grammar in Prog-
ress: GLOW Studies for Henk van Riemsdijk. (J. Mascaro and M. Nespor, eds.). Foris,
Dordrecht.
Zee, D. (1988). Sonority Constraints on Prosodic Structure. Doctoral dissertation, Stan-
ford University, Stanford, Calif.
Zee, D., and Inkelas, S. (1990). Prosodically constrained syntax. In The Phonology-Syntax
Connection (S. Inkelas and D. Zee, eds.), CSLI Publications and University of Chi-
cago Press, Chicago.
Zee, D., and Inkelas, S. (1991). The place of Clitics in the Prosodic Hierarchy. Proceedings
of the West Coast Conference on Formal Linguistics 10, 5 0 5 - 5 2 0 .
INTERACTION BETWEEN MODULES IN LEXICAL
PHONOLOGY
DAVID ODDEN
Department of Linguistics
Ohio State University
Columbus, Ohio 43210
1. I N T R O D U C T I O N
(1) ,
LEXICON
u n d e r i v e d lexical
items
level 1 m o r p h o l o g y level 1 p h o n o l o g y
level 2 m o r p h o l o g y level 2 p h o n o l o g y
level n m o r p h o l o g y level n p h o n o l o g y
syntax postlexical p h o n o l o g y
ill
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
112 David Odden
T h i s display e m b o d i e s a n u m b e r of h y p o t h e s e s , the m o s t i m p o r t a n t b e i n g
spelled out in (2).
C l a i m s (2a) and (2b) are the level-ordering c l a i m s , and are simply a s s u m e d here
to b e true. T h e related claim (2c) that p h o n o l o g i c a l levels are the s a m e as m o r p h o -
logical levels has less support, in light of certain b r a c k e t i n g p a r a d o x e s . I also
maintain w i t h o u t c o m m e n t (or c o m m i t m e n t ) the standard a s s u m p t i o n of lexical
p h o n o l o g y that p h o n o l o g i c a l levels are i s o m o r p h i c with m o r p h o l o g i c a l levels,
since the identity of m o r p h o l o g i c a l a n d p h o n o l o g i c a l levels is entirely o r t h o g o n a l
to the question of interaction b e t w e e n p h o n o l o g y and m o r p h o l o g y . C l a i m (2d),
that cyclic b e h a v i o r C A N b e derived from a p h o n o l o g y - m o r p h o l o g y interaction,
is a purely theory-internal c l a i m . A s w e will see, other m o d e l s can derive cyclic
b e h a v i o r w i t h o u t this interaction.
1.1. N o n i n t e r a c t i v e L e x i c a l P h o n o l o g y
(3) SYNTAX
MORPHOLOGY
L e x i c o n of roots
level 1
level 2
level n
PHONOLOGY
Lexical Phonology
level 1
level 2
level n
i
Postlexical P h o n o l o g y
hataf
hataf it
hataf it hum
Interaction in Lexical Phonology 115
SUFFIXATION OF NEGATIVE
(6) L 2 STRESS
L, phonology
hdtaf it hum s
116 David Odden
APOCOPE
STRESS
1.2. T h e I n t e r p r e t a t i o n of B o x e s
2. S Y N T A X A N D P H O N O L O G Y
2 . 1 . O r d e r i n g S a n d h i R u l e s in t h e L e x i c o n
T h e p r o b l e m I c o n s i d e r h e r e is that s o m e rules of K i m a t u u m b i p h o n o l o g y m u s t
b e lexical, b u t they also h a v e access to syntactic structure and p h o n o l o g i c a l p r o p -
erties of s u r r o u n d i n g w o r d s . T h e first rule is the phrase-level rule S h o r t e n i n g ,
w h i c h shortens a long vowel if it is the h e a d of a p h r a s e and is followed b y m a t e -
rial within its p h r a s e . This vowel length alternation is seen in (7).
T h r e e levels can b e m o t i v a t e d in K i m a t u u m b i . L e v e l 1 m o r p h o l o g y c o n s t r u c t s
the stem from the root, derivational affixes s u c h as the causative a n d benefactive,
a n d the stem-final tense inflection. Level 2 verbal m o r p h o l o g y includes addition
of object prefixes, tense-aspect prefixes, a n d subject prefixes. Level 2 n o m i n a l
m o r p h o l o g y is addition of the n o u n class prefixes. At L e v e l 3 , the locative n o m i n a l
prefixes ku-, pa-, a n d mu- are added, a n d in verbs the relative c l a u s e h e a d a g r e e -
m e n t prefixes such as pa- and cha- are a d d e d .
T h e G l i d e F o r m a t i o n rule applies cyclically, in particular interstratal-cyclically
b u t not s t r a t u m - i n t e r n a l - c y c l i c a l l y . T h i s is seen in the contrasting derivations of
mwjjute in (11a), w h i c h h a s the Level 2 prefixes mu a n d /, versus muyuyld in
( l i b ) , w h i c h h a s the level 3 prefix mu a n d the level 2 prefix /. T h e p r o b l e m is that
w e h a v e the s a m e b a s i c configuration of p h o n e m e s in b o t h c a s e s , b u t d e t e r m i n i n g
w h i c h of the v o w e l s b e c o m e s a glide requires k n o w i n g the level at w h i c h e a c h
m o r p h e m e is m a d e available to the p h o n o l o g y .
(11) a. [my-j-ute]
mu-j-ute Input to L 2
NA Glide Formation
in the f r o g s '
(17) a- d e m o n s t r a t i v e prefix
na 'with, and'
n(cheche) initial syllabic nasal of 'four'
n(tupu) initial syllabic nasal of 'is n o t '
ka- subord. verb prefix
maladu 'tomorrow'
kjtiwj 'how'
namanj 'what'
rial 'who'
gani 'which'
bulj 'how'
kill 'what'
gaaku 'what kind'
gu- et al. d e m o n s t r a t i v e prefix
wa- et al. associative prefix
gu- et al. n u m e r a l prefix
mu-, pa-, ku- locative prefixes
(20) utijlf kuaanjy —> ut\l\ kwaanjy ' y o u s h o u l d run to the firewood'
util\kyaan]ii —> iit\l\ kyaanjy ' y o u r a n to the firewood'
2.2. Precompilation
N L P has n o p r o b l e m s with this state of affairs, since in that theory the output
of the syntax is fully available to lexical p h o n o l o g y , and therefore these sandhi
rules can b e lexical. T h e q u e s t i o n r e m a i n s w h e t h e r o n e can h a n d l e these data,
retaining the s u p p o s e d o r d e r i n g b e t w e e n syntax and lexical phonology, by m o d i -
fying L P in s o m e way. H a y e s ( 1 9 9 0 ) , discussing similar p r o b l e m s , including the
S h o r t e n i n g - G l i d e F o r m a t i o n p a r a d o x of K i m a t u u m b i , p r o p o s e s a different m o d i -
fication of p h o n o l o g y , n a m e l y p r e c o m p i l a t i o n theory. In p r e c o m p i l a t i o n theory a
w o r d m a y have m u l t i p l e lexical derivations; h e n c e multiple outputs e m e r g e from
the lexicon for e a c h w o r d . E a c h of t h e s e derivations is t a g g e d for a property re-
ferred to as a lexical instantiation frame. L a n g u a g e s m a y define sets of instantia-
tion frames w h i c h serve as c o n t e x t for lexical p h o n o l o g i c a l rules. U p o n entering
the postlexical p h o n o l o g y , the frame definitions of the l a n g u a g e are consulted, and
the syntactic, m o r p h o l o g i c a l , and p h o n o l o g i c a l properties of the w o r d in its sen-
tence are c h e c k e d ; out of the various derivations g e n e r a t e d in the lexicon, the
correct form is then inserted into the sentence, and the string is submitted to the
Interaction in Lexical Phonology 123
T h e definition of F r a m e 1 is given as ( 2 2 ) .
(22) F R A M E 1: [ , .
X .. [ X ] Y] Y * 0
andangyo NA Shortening
k\-andangyo k\andaangyo Level 2 prefixing
kyaandangyo kyaandaangyo Output of lexical phonology
V V/ [ [ F r a m e 2]
[+D]
F r a m e 2 is defined as in ( 2 5 ) . 1
tax, m o r p h o l o g y , a n d p h o n o l o g y of s u r r o u n d i n g w o r d s , then E L P m a k e s n o p r e -
diction a b o u t information relations b e t w e e n syntax and p h o n o l o g y ; rather, it d e -
scribes these relations b y different m e a n s from NLP, but the forms g e n e r a b l e in
both theories are the s a m e .
If the theories are w e a k l y equivalent, then w e m u s t turn to s e c o n d a r y consid-
erations such as e l e g a n c e a n d c o m p u t a t i o n a l c o m p l e x i t y to evaluate the theories.
T h e r e is a c o m p u t a t i o n a l a d v a n t a g e to a l l o w i n g lexical rules to directly see the
o u t p u t of syntax. In p r e c o m p i l a t i o n theory, there can b e m u l t i p l e frames, and
frames can overlap, so e a c h frame definition d o u b l e s the n u m b e r of derivations
n e c e s s a r y for a form. F o r instance, if there are t w o frames defined in a l a n g u a g e ,
then four derivations are required for a w o r d , o n e for F r a m e 1, o n e for F r a m e 2,
o n e for F r a m e s 1 a n d 2, a n d o n e for the e l s e w h e r e form. In general, w h e n there
are n frames, w e n e e d 2 derivations. In the c a s e of K i m a t u u m b i (as discussed in
n
O d d e n , 1992), there are eight external sandhi rules w h i c h are p a r t of the lexical
p h o n o l o g y , so 2 5 6 parallel derivations are n e e d e d for e a c h w o r d . O n the other
h a n d , in the theory o r d e r i n g lexical p h o n o l o g y after syntax, only o n e derivation is
required, since the rules simply inspect the w o r d - e x t e r n a l c o n t e x t to d e t e r m i n e
w h e t h e r their c o n d i t i o n s for application are satisfied. P r e c o m p i l a t i o n theory thus
entails m o r e c o m p l e x c o m p u t a t i o n s .
3. M O R P H O L O G Y A N D P H O N O L O G Y
3.1. 1-SinguIar A l l o m o r p h y in A r a b i c
(27) G L I D E ELISION: c
—con • 0/V- —V
+ son
V O W E L FUSION: O O
C L O S E D SYLLABLE SHORTENING:
qaday-tu T settled'
/qadiy-un/ qadin 'a judge (nom.)'
/?al-qadiy-un/ ?al-qddi 'the judge (nom.)'
qadiy-u-1 s g . / qddi-ya 'my judge'
T h i s is w h a t L P predicts c o u l d h a p p e n , b u t it w o u l d s e e m to b e a p r o b l e m for N L P ,
w h i c h requires all m o r p h e m e s to b e c o n c a t e n a t e d before any p h o n o l o g i c a l rules
apply. H o w e v e r , it is s h o w n b e l o w that this c a s e C A N b e h a n d l e d in the noninter-
active theory.
(36) D A N I S H IMPERATIVE T R U N C A T I O N
9-^0/ J
[Imper]
(37) [Infin]-*0/ J
[Imper]
O f c o u r s e , in t h i n k i n g a b o u t this restriction t o c o n c a t e n a t i o n , o n e s h o u l d k e e p in
m i n d that n o n l i n e a r r e p r e s e n t a t i o n s m a y m a k e it LOOK like a m o r p h o l o g i c a l p r o -
c e s s is d o i n g m e t a t h e s i s o r infixing, o r g e m i n a t i n g c o n s o n a n t s . A s M c C a r t h y
( 1 9 7 9 ) s h o w s , this is j u s t a n illusion.
T h e c o m p l e t e s e g r e g a t i o n of m o r p h o l o g i c a l a n d p h o n o l o g i c a l o p e r a t i o n s m a k e s
strong c l a i m s a b o u t m o r p h o l o g y . A c o n s e q u e n c e of this p r i n c i p l e is that w e rule
o u t o n theoretical g r o u n d s s u p p o s e d c a s e s of " P r o c e s s M o r p h o l o g y " ( M a t t h e w s ,
1974, inter alia), w h e r e m o r p h o l o g i c a l rules p e r f o r m p h o n o l o g i c a l c h a n g e s . E x -
a m p l e s of p r o c e s s m o r p h o l o g y m u s t b e r e a n a l y z e d as t w o p r o c e s s e s , n a m e l y
p u r e l y m o r p h o l o g i c a l c o n c a t e n a t i o n , p l u s a p h o n o l o g i c a l o p e r a t i o n . F o r instance,
the m o r p h o l o g i c a l p a r t of G e r m a n U m l a u t is s i m p l y stating t h e c o n d i t i o n s for
a d d i n g a n affix o r set of affixes. It h a p p e n s that t h e p h o n o l o g i c a l c o n t e n t of this
suffix is o r c o n t a i n s a floating v o w e l feature [ — b a c k ] . T h e p h o n o l o g y is then
r e s p o n s i b l e for l i n k i n g that feature t o t h e a p p r o p r i a t e v o w e l .
3.2.1. REDUPLICATION
S u c h a division of l a b o r r e m o v e s t h e K i h e h e k i n d of r e d u p l i c a t i o n from t h e
p o o l of s u p p o r t for the interactive m o d e l . T h e p r o b l e m of K i h e h e is that t h e entity
w h i c h r e d u p l i c a t e s is t h e stem, w h i c h is t h e o u t p u t of L e v e l 1 m o r p h o l o g y a n d
thus g e n e r a l l y e x c l u d e s t h e object prefix o r t h e infinitive prefix, w h i c h a r e at
L e v e l 2 . H o w e v e r , in c a s e s o m e prefixal e l e m e n t h a s fused syllabically w i t h t h e
initial s t e m syllable, t h e prefix s e g m e n t s g e t c o p i e d as well.
(40) L 2 to underlying
telek a
ku 'a bit' cook tense
infinitive (redup.) (stem)
L 9 co L, to Lj phonology
A
a a a
ku
UA
telek a
mapping, copy
Li telek a
to
0)
A
a a a
ku
UA
telek a
L2
L ? to Lo to sequencing
A A
a a a a a a
li k A H A A
UA UA
ku telek a telek a
mapping, copy
L, co sequencing
^9 <JU
A A
a a
A\t\
• a a
M
kwit a
A d d i n g the p r o s o d i c affix co constitutes the entire m o r p h o l o g y of reduplication;
the ability to a d d this affix or the selection of a specific reduplication m o r p h e m e
as t h e realization of a m o r p h o l o g i c a l c a t e g o r y is not in any w a y affected b y p r i o r
application of p h o n o l o g i c a l rules in K i h e h e , n o r is it in any other l a n g u a g e . T h e
interesting a n d characteristic w o r k involved in reduplication is largely d o n e b y the
p h o n o l o g y , w h i c h r e c e i v e s this d e g e n e r a t e representation, n a m e l y a s e q u e n c e of
real s e g m e n t s p l u s s o m e s e g m e n t a l l y e m p t y p r o s o d i c t e m p l a t e , a n d the p h o n o l o g y
h a s the responsibility for filling in that t e m p l a t e . In K i h e h e a n d cases like it, this
takes p l a c e after certain p h o n o l o g i c a l r u l e s . 2
3.2.2. TRUNCATION
(50) p a r a m - k a underlying
C V C V C C V
a a a
nominative destructuring
C V C V C
a a a
defaults, resyllabification
C V C V C V
\ \ If
a a a
a a a
V V C V / V
V V V V
maqha i diphthongization
V V C V
maqha i a defaults
V V C V
T h e r e a r e still u n t o u c h e d c a s e s w h e r e p h o n o l o g i c a l r e a n a l y s i s is n o t p o s s i b l e .
T h e best k n o w n c a s e is the v e r b - t o - n o u n derivational suffix -al in E n g l i s h , w h i c h ,
it is often said, c a n attach o n l y to final-stressed stems.
(60) 9->0f V
3.4. Precyclicity o r P r e c o m p i l a t i o n ?
(63) STRESS:
V -> [+stress] / C ((VC) VC )]
0
1
(64) APOCOPE:
V->0/ CV
hdtaf-na input to L 2
hatdf-na Stress
NA Apocope
(69) htaftuuna ' y o u (pi.) snatched u s ' (htdftu ' y o u (pi.) s n a t c h e d ' )
htafniekom ' w e s n a t c h e d y o u (pi.)' (htdfna ' w e s n a t c h e d ' )
ma hatfuus 'they didn't s n a t c h ' (hdtfu 'they s n a t c h e d ' )
N o t e that this derived length attracts stress, so w e get hatfuus, not * hatfuus. There-
fore, B o u n d a r y L e n g t h e n i n g p r e c e d e s Stress o n Level 2.
N o w w e arrive at the p a r a d o x in ( 7 1 ) . T h e p r o b l e m is that verb stems w h i c h e n d
with a vowel, such as ?dra, m u s t l e n g t h e n their final vowel before a Level 2 suffix
as predicted, a n d this l e n g t h e n i n g m u s t take p l a c e o n Level 1 before L stress x
?arda+na Stress
NA Apocope
*[?araana]
T h e s e facts c a n b e a c c o m m o d a t e d in a n y t h e o r y w h e r e all m o r p h e m e s a r e c o n -
c a t e n a t e d before a n y p h o n o l o g i c a l rules apply, p r o v i d i n g that w e allow s o m e
rules, a n d in particular, B o u n d a r y L e n g t h e n i n g , to a p p l y precyclically, so that
Level 2 suffixes c a n b e seen a n d t h u s trigger l e n g t h e n i n g before Level 1 p h o n o -
logical rules apply. In fact, w e c a n also a c c o u n t for this p r o b l e m if w e treat
L e n g t h e n i n g as a p r e c o m p i l e d rule, w h i c h w o u l d b e written as (74).
(75) ?ara
F r a m e 1 is defined a s in ( 7 6 ) .
(76) {Object 1
F r a m e 1: [VERB J [Negative J
4. SUMMARY
NOTES
Sharon Hargus has suggested the possibility of eliminating the negative condition "not
1
preceded by H " in this rule by giving the morphemes in question an underlying initial H,
which deletes after a stem H. This possibility can be ruled out on two grounds. First, the
citation form of the relevant morphemes lacks the H, cf. mu-kjkdlaango 'in the frying pan':
in the citation form, there is no preceding H, so no reason to delete the putative underlying
H of /my-kjkalaango/. Second, ITI is subject to a syntactic condition that the toneless stem
which conditions ITI and the morpheme which undergoes the rule cannot be separated by
a righthand S-bracket. So, despite the fact that muundu 'person' has no H tones and the
following morpheme mu is one of the morphemes undergoing ITI, ITI does not apply in
the sentence naabik\tee nama \yaydapim\lwe na muyndu] my-kjkdlaango 'I put the meat
which was bought by the person in the frying pan' because the determinant and focus are
in different clauses. If we construe ITI as deletion of an underlying H, we must further
expand the environment for that rule, so that if the H is separated from the preceding word
by ] , then the initial H tone must also be deleted, even if there is no preceding H.
s
most likely have the same form as the NLP analysis. Since reduplication may (under resyl-
labification, as in reduplication of kwiita or kulwiitd) copy an object prefix or the infinitive
prefix ku, reduplication follows prefixation of these morphemes. The problem is identifying
the substring subject to reduplication. Given inputs such as REDUP•+ku[teleka] and RE-
DUP+ku[lu[teleka]] which reduplicate as kutelekateleka and kulutelekateleka, the morpho-
logical constituent which copies is the stem. But by the assumptions of LP, the internal
morphological structure of the verb is not recoverable. Morphological structure simply
does not suffice to identify the correct substring which reduplicates in the case of kulwii-
talwiita, where the structure which is copied includes nonstem material (Iw, the object
prefix). Therefore, in the LP account, some prosodic structure must form the basis for
Interaction in Lexical Phonology 143
identifying the structures subject to reduplication, and this structure must include the stem
syllables but may not include prefix syllables except when prefix syllables fuse with stem
syllables by phonological rules.
3
The reason that postlexical deletion, especially any deletion applying after phonologi-
cal rules, should be phonological is that the morphological component is, in the theory of
LP, part of the lexicon. Of course, one could expand LP in such a way that "postlexical
morphology" is not a theoretical anomaly, but such a move would seriously undermine the
motivation for distinguishing between the lexical and postlexical components.
4
It is beyond the scope of this article to present a complete theory of rules of this type,
but it is important to know something about how such rules are constrained. It is assumed
here that rules may either delete the segmental and prosodic material under a specified
prosodic constituent, or may delete the segmental material under the constituent leaving
prosodic structure intact. Rules of the former type are written as simply deleting the rele-
vant prosodic constituent. Rules of the latter type are written as deletion of the prosodic
element, with the additional annotation that "prosodic structure is preserved."
5
It is often assumed that i, not /, is the vowel which arises from default rules in Korean.
However, i has a restricted distribution in Korean: no morphemes ends in i except for the
demonstratives ki and ni. The relevant generalization is that i cannot be prepausal—the
demonstratives can never be prepausal. A similar constraint on i appears in Tigrinya; epen-
thesis inserts i in word-final position, but the vowel is realized phonetically as i in that
position. Chung (1991) provides other arguments that word-finally, / and not / arises by
default.
6
The forms denial and trial do not conform to this generalization; note, however, that
the stress-final generalization is falsified by burial.
REFERENCES
Chung, Y. H. (1991). The Lexical Tone System of North Kyungsang Korean. Doctoral dis-
sertation, Ohio State University, Columbus.
Clements, G. N. (1986). The problem of transfer in non-linear phonology. Cornell Working
Papers in Linguistics 7, 1 - 3 6 .
Halle, M., and Vergnaud, J.-R. (1987). An Essay on Stress. MIT Press, Cambridge, Mass.
Hayes, B. (1990). Precompiled phrasal phonology. In The Phonology-Syntax Connection
(S. Inkelas and D. Zee, eds.), pp. 8 5 - 1 0 8 . CSLI Publications and University of Chi-
cago Press, Chicago.
Hulst, H. van der (1984). Syllable Structure and Stress in Dutch. Foris, Dordrecht.
Hyman, L. (1990). Conceptual Issues in the Comparative Study of the Bantu Verb Stem.
Paper presented at the 21st Conference on African Linguistics, University of Georgia,
Athens.
Inkelas, S. (1989). Prosodic Constituency in the Lexicon. Doctoral dissertation, Stanford
University, Stanford, Calif.
Kaisse, E., and Shaw, P. (1985). On the theory of lexical phonology. Phonology Yearbook
2, 1-30.
Kiparsky, P. (1982). Lexical phonology and morphology. In Linguistics in the Morning
Calm (I. S. Yang, ed.), pp. 3 - 9 1 . Hanshin, Seoul.
Kisseberth, C , and Abasheikh, M. (1976). On the interaction of phonology and mor-
phology: a Chi-mwi: ni example. Studies in African Linguistics 7 , 3 1 — 110.
Marantz, A. (1982). Re reduplication. Linguistic Inquiry 1 3 , 4 3 5 - 4 8 2 .
Martin, J. (1988). Subtractive morphology as dissociation. Proceedings of the West Coast
Conference on Formal Linguistics 8, 2 2 9 - 2 4 0 .
Matthews, P. (1974). Morphology: An Introduction to the Theory of Word Formation.
Cambridge University Press, Cambridge.
McCarthy, J. (1979). Formal Problems in Semitic Phonology and Morphology. Doctoral
dissertation, Massachusetts Institute of Technology, Cambridge.
Mohanan, K. P. (1986). The Theory of Lexical Phonology. Reidel, Dordrecht.
Odden, D. (1987). Kimatuumbi phrasal phonology. Phonology Yearbook 4, 1 3 - 3 6 .
Odden, D. (1990a). Syntax, lexical rules and postlexical rules in Kimatuumbi. In The
Phonology-syntax Connection (S. Inkelas and D. Zee, eds.), pp. 2 5 9 - 2 7 7 . CSLI Pub-
lications and University of Chicago Press, Chicago.
Odden, D. (1990b). Phonology and its interaction with syntax and morphology. Studies in
the Linguistic Sciences 20, 6 9 - 1 0 8 .
Odden, D. (1992). Kimatuumbi Phonology and Morphology. Unpublished manuscript,
Ohio State University, Columbus.
Odden, D., and Odden, M. (1986). Ordered reduplication in Kihehe. Linguistic Inquiry 16,
497-503.
THE STRUCTURE OF THE SLAVE
(NORTHERN ATHABASKAN) VERB
KEREN D.RICE
Department of Linguistics
University of Toronto
Toronto, Ontario, Canada M5S 1A1
1. INTRODUCTION
145
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
146 Keren D. Rice
2. T H E S T R U C T U R E O F T H E S L A V E V E R B
2 . 1 . T h e C o n t e n t of t h e T e m p l a t i c P o s i t i o n s
2.2. Inflection o r D e r i v a t i o n ?
2.5. T h e D e r i v a t i o n of t h e V e r b W o r d
3. P R E V I O U S T R E A T M E N T S O F W O R D F O R M A T I O N
4. T H E S Y N T A X O F T H E S L A V E V E R B
4.1. T h e F u n c t i o n a l N a t u r e of t h e C o n j u n c t M o r p h e m e s
T h e first line is not an iterative form. In the s e c o n d line, the first w o r d includes the
iterative m o r p h e m e na, w h i l e this m o r p h e m e is absent in the second word, with
iterative m e a n i n g c o n t r i b u t e d solely by the v e r b stem. In the c a s e of imperfective,
perfective, a n d optative, or p r i m a r y aspect m o r p h e m e s , these m o r p h e m e s are
obligatory a n d aspect is not d e t e r m i n a b l e b y stem form alone. B y the criterion of
obligatoriness, then, the iterative d o e s n o t qualify as functional.
T h e distributive m o r p h e m e is like the iterative m o r p h e m e in that the m e a n i n g
of distributivity can b e given solely b y stem form or b y c h o i c e of conjugation.
that the voice e l e m e n t c o m b i n e s with the verb stem lexically, with voice and tran-
sitivity alternations d e t e r m i n e d in the lexicon. S e e also M c D o n o u g h ( 1 9 9 0 ) ,
Speas (1991).
The Structure of the Slave Verb 157
4.2. T h e O r d e r i n g of t h e F u n c t i o n a l I t e m s
(8) subject a g r e e m e n t - t e n s e - a s p e c t - o b j e c t a g r e e m e n t - v o i c e - v e r b
158 Keren D. Rice
4 . 3 . T h e S t r u c t u r e of t h e F u n c t i o n a l C o m p l e x
(9)
(10) CP
SubAgr V
PrimAsp SubAgr
SecAsp PrimAsp
Gen SecAsp
/X
DOAgr Gen
/ \
VP DOAgr
/\
VP NP
A s
V Adv
Prev Root,
4.4. T h e S t r u c t u r e of t h e VP
4.5. S u m m a r y
5. T H E U N D E R L Y I N G R E P R E S E N T A T I O N O F T H E V E R B
(11) voice-root] R o o t
(12) [h]
voice M R o o t [jtfJpreverb 'preach, bark'
E a c h m o r p h e m e is labeled for category. W h e n this phrasal unit is inserted into t h e
larger syntactic structure, the m o r p h e m e s are correctly placed. N o further stipu-
lation of position is required, a s it is a direct c o n s e q u e n c e of the phrasal structure,
w h i c h in turn is p r e d i c t e d from scopal p r o p e r t i e s .
Lexical entries c a n also include g e n d e r material a n d direct objects. F o r instance,
the verb ' s c a r e ' h a s a g e n d e r m o r p h e m e a n d 'tell a l i e ' a direct object that m u s t
o c c u r with t h e verb s t e m . 1 2
162 Keren D. Rice
6. T H E P H O N O L O G I C A L S T R U C T U R E O F T H E S L A V E V E R B
6 . 1 . E v i d e n c e for D o m a i n s o f R u l e A p p l i c a t i o n
6.2. D e r i v i n g t h e D o m a i n s
(16) a. right e d g e of X m a x
, X is f u n c t i o n a l — p h o n o l o g i c a l p h r a s e
b. right e d g e of X ° , X is lexical (noun, verb, adverb, p o s t p o s i t i o n ) — w o r d
c. right e d g e of a g r e e m e n t — s m a l l w o r d
( ) phrase, X P
( )( )() ( ) w o r d , X ° , lex
[ ][ ][] ( )( )[ ] small word,
X § a r
7. SUMMARY
position of the verb root. In o r d e r to derive the surface position of the root, I
s u g g e s t e d that raising m o v e s it from its position within the verb p h r a s e in order to
assign it c a t e g o r y status.
T h e traditional single ' w o r d ' status of the verb in Slave is in this a c c o u n t a
derived property. T h e e n d - b a s e d a l g o r i t h m defines the verb as a single unit by
seeking out the right e d g e of phrasal projections of functional categories. T h e
p h o n o l o g i c a l d o m a i n s within the w o r d are d e t e r m i n e d b y first m a r k i n g the right
e d g e of major c a t e g o r y lexical items a n d s e c o n d m a r k i n g the right e d g e of a g r e e -
m e n t m o r p h e m e s . T h e verb " w o r d " is thus not a lexical c o n s t r u c t but a p h o n o -
logical o n e , and verb affixes are syntactically w o r d s .
8. C O M P A R I S O N W I T H L E X I C A L P H O N O L O G Y
different in the different d o m a i n s . T h u s the fact that the stem and the disjunct
prefixes share similar rules is not surprising [as it is in lexical p h o n o l o g y , as this
a p p e a r s to b e a violation of either the strong d o m a i n h y p o t h e s i s (Kiparsky, 1984)
or the c o n t i n u o u s stratum h y p o t h e s i s ( M o h a n a n , 1 9 8 6 ) ] — t h e y share p r o p e r t i e s
b e c a u s e their s e g m e n t a l a n d metrical p r o p e r t i e s are similar, a n d t h e s e p r o p e r t i e s
are similar b e c a u s e they are m a j o r - c a t e g o r y lexical i t e m s as o p p o s e d to functional
i t e m s . In a lexical p h o n o l o g y m o d e l , s o m e of the basic tenets of lexical p h o n o l o g y
m u s t b e rejected since rules apply to d i s c o n t i n u o u s d o m a i n s .
W h a t general c o n c l u s i o n s c a n b e d r a w n with respect to lexical p h o n o l o g y ? T h i s
is a difficult q u e s t i o n to answer, as I have s u g g e s t e d that a l a n g u a g e that a p p e a r s
to h a v e c o m p l e x m o r p h o l o g y , a n d t h u s a p p e a r s to b e a g o o d test c a s e for lexical
p h o n o l o g y , d o e s not in fact have c o m p l e x m o r p h o l o g y . T h e surface c o m p l e x i t i e s
in the m o r p h o l o g y result from derived rather than u n d e r l y i n g p r o p e r t i e s . T h u s ,
any c o n c l u s i o n s r e g a r d i n g the interaction of p h o n o l o g y a n d m o r p h o l o g y in w o r d
formation say n o t h i n g at all about lexical p h o n o l o g y . H o w e v e r , the Slave findings
p e r h a p s force a reevaluation of other l a n g u a g e s that l o o k particularly t r o u b l e s o m e
for lexical p h o n o l o g y . If they receive similar r e a n a l y s e s , it m a y b e that l a n g u a g e s
in w h i c h w o r d formation is truly m o r p h o l o g i c a l rather than resulting from p h o -
nological d o m a i n a s s i g n m e n t are in fact well a c c o u n t e d for b y the m o d e l .
T h e p h o n o l o g i c a l m o d e l that is briefly d e s c r i b e d in this p a p e r also c a n b e
v i e w e d as p r o v i d i n g s u p p o r t for s o m e of the tenets of lexical p h o n o l o g y . K i p a r s k y
( 1 9 8 4 ) p r o p o s e s the strong d o m a i n h y p o t h e s i s , w h i c h allows the g r a m m a r of a
l a n g u a g e to stipulate w h e r e a rule c e a s e s to apply, b u t n o t to turn a rule on. All
rules are thus potentially a p p l i c a b l e at the first level and apply there if p e r m i t t e d
by other p r i n c i p l e s such as the strict cycle condition a n d structure preservation. In
the m o d e l that I h a v e p r o p o s e d , this is precisely the c a s e : all rules are a p p l i c a b l e
at the small w o r d and fail to apply there if their structural description is not m e t
(i.e., the foot is not p r e s e n t yet) or if they violate structure preservation (e.g.,
insertion of glottal stop at the e d g e of a p h o n o l o g i c a l p h r a s e ) . T h u s , in a sense this
m o d e l s u p p o r t s m a n y of the principles of lexical p h o n o l o g y .
9. CONCLUSION
ACKNOWLEDGMENTS
I have several people to thank for their helpful discussion of this article: Leslie Saxon,
Peggy Speas, Eloise Jelinek, Sharon Hargus, Elizabeth Cowper, Diane Massam, Aryeh
Faltz, the participants in the lexical phonology conference, and the reviewer of an earlier
version of the article.
NOTES
Inkelas (1989) proposes a version of lexical phonology in which cases of lack of iso-
1
3
1 use orthography in most cases. The following symbols should be noted, gh is voiced
velar fricative, dh a voiced dental fricative, th a voiceless dental fricative, an acute accent
represents a high tone, a hook under a vowel represents nasalization. The symbol n, a
palatal nasal, is an abstract representation for a morpheme which may surface as a high
front vowel, as nasalization on a vowel, or as voicing on a continuant, depending on pho-
nological and morphological context.
4
w- is the reflex in Hare, a Slave dialect, of *s; and y- is the reflex of *gh.
5
It is interesting that inflectional and derivational affixes also appear in a marked order
in the Slave noun. In particular, the inflectional morpheme indicating possessive agreement
is phonologically closer to the stem than the derivational augmentative and diminutive
morphemes. See Rice (1991a) for some discussion.
6
T h e inflection/derivation question disappears in some ways since I treat the verb as
phrasal rather than lexical in nature. The so-called derivational items are treated as lexical
categories and the inflectional items as functional categories.
7
This position thus represents an abandonment of the strong version of the lexicalist
hypothesis, which requires that all word formation take place in the lexicon. Instead, in-
flectional morphology is part of syntax proper and lexical operations are restricted to deri-
vational morphology. This assumption alone rules out the type of lexical phonology model
proposed by Hargus (1988), as in that model the verb word is formed in its entirety in the
lexicon.
8
1 use the term F U N C T I O N A L where Anderson (1982, 1988) uses the term I N F L E C T I O N A L .
9
In addition, the classifiers have unique phonological properties.
10
T h i s ordering is similar to that found by Bybee (1985) in her survey of morpheme
ordering in fifty languages; however, Bybee's survey is based on surface morpheme order
and Speas's on a more abstract underlying order, so they are not directly comparable.
1 1
The rudiments of this analysis come from work by Palma dos Santos (1991).
1 2
It is useful to summarize the types of word formation found in the Slave verb. First,
voice elements can be added to the root in the lexicon. Otherwise no lexical word formation
is found. Syntactic "word formation" arises from the operation of Raising. Finally, as dis-
cussed in Section 6, phonological "word formation" results from the imposition of pro-
sodic structure on the syntactic structure.
13
T h i s process is also often thought of as deletion on the functional domain (e.g., Rice,
1989). Whether epenthesis or deletion is the preferred analysis does not affect the claim
that the morpheme patterns differently depending on its position in the word.
1 4
Similar restrictions are found within nouns and postpositions, which also constitute
small words by the definition given.
REFERENCES
RICHARD SPROAT
Linguistics Research Department
AT&T Bell Laboratories
Murray Hill, New Jersey 07974
1. INTRODUCTION
Another consequence of BE, as pointed out in Pesetsky (1979), is that it derives the prin-
ciple of Lexical Integrity, first proposed in Chomsky (1970). The Lexical Integrity Hy-
pothesis says that syntactic rules cannot have access to the internal structure of words. It
173
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
174 Richard Sproat
prevents, for example, a pronoun taking father in fatherless as its antecedent, exempts the
self'm self-destruction from the syntactic conditions governing anaphor binding, and pre-
vents -al and -ion from being attached to refuse and destroy in the syntax. Given that
morphological operations take place in the lexicon, and that internal brackets are erased
by BE in the output of the lexicon, it follows that syntactic operations will be blind to the
internal structure created by morphological operations.
2. S E N S I T I V I T Y O F P H O N E T I C I M P L E M E N T A T I O N R U L E S
TO LEXICAL BOUNDARIES
0 1 2 3 4 5 6
Boundary strength
Figure 1. Sensitivity of phonetic implementation rules to boundary strength, (a), speaker CS;
(b), speaker CC.
Lighter/I/
#
Retraction of tongue dorsum for l\l (mm)
C
\ C
-69
I
-70
V
I
-71
>
Darker W
i r
3 4
Boundary strength
Figure 1. Continued.
(3) Speaker t P
CS t -8.33
6 = <0.0005
cc t = -6.19
6 <0.001
AD t = -2.87
6 <0.05
DB f = -5.27
6
<0.005
RS t = -1.25
6
0.26
(4) Speaker t P
CS h = 6.67 <0.001
cc - 4.00 <0.01
AD *6 = 2.52 <0.05
DB *6 = 1.74 0.13
RS *6 - 2.14 0.076
(5) a. (o b. y
03 Q)
beel equator
3. P R A G M A T I C S A N D W O R D - I N T E R N A L S T R U C T U R E
(6) a. Patty is a definite [Kal Kan cat]. Every day she waits for it.
Television a d v e r t i s e m e n t for Kal K a n , January 2 8 , 1987)
b. There's a [Thurber story] about his maid.. .
H e a r d in c o n v e r s a t i o n , S e p t e m b e r 7, 1988
c. We went up to [Constable country]', we stayed in the village he was
born in.
H e a r d in c o n v e r s a t i o n , O c t o b e r 1 1 , 1988
d. / refer you to the [Schachter paper]', he's very proud of it. . .
S p e a k e r in r e s p o n s e to a question at N o r t h E a s t e r n Linguistics Society,
N o v e m b e r 12, 1988
e. Well, action is still needed. If we 're to finish the job, Reagan's Regiments
will have to become the [Bush Brigades]. Soon he'll be the chief, and
he '11 need you every bit as much as I did.
R. R e a g a n , farewell speech, J a n u a r y 1 1 , 1989, reported in A s s o c i a t e d
Press N e w s w i r e
f. Millions of [Oprah Winfrey fans] were thoroughly confused last week
when, during her show, she emotionally denied and denounced a vile
rumor about herself.
M i k e R o y k o , M a y 2 2 , 1989, cited b y M c C a w l e y , 1989, as an e x a m p l e of
reflexive u s a g e — i . e . , not as an e x a m p l e of an a n a p h o r i c island
g. / had a [paperroute] once but my boss said I took too long deliv-
erin' 'em.
" L . A . L a w , " 1987
h. I'm a [mystery-story buff] and read (and watch on PBS) a lot of them.
N o r t h w e s t e r n University electronic bulletin board, January, 1989
i. We asked [Saab 9000-CD owners] about its road-handling. ..
Television ad for S a a b , M a r c h 12, 1989
j. For a [SYNTAX slot], I'd rather see someone with more extensive
coursework in it.
H e a r d d u r i n g a discussion of various subdisciplines of linguistics, w h e r e
the s p e a k e r w a s c o n t r a s t i n g syntax with other subdisciplines, J a n u a r y
18, 1987
k. At the same time as coffee beans were introduced, the Arabs made
changes in [coffee preparation] that greatly improved its flavor.
Schapira, J., Schapira, D . , a n d Schapira, K., The book of coffee and tea,
1982, p . 7
1. A : Are we ciderless? B : Yes we're [ciderless]. You should have told
me—I would have brought some.
In c o n v e r s a t i o n , J a n u a r y 2 3 , 1987
m. Do [parental] reactions affect their children?
H e a r d in c o n v e r s a t i o n by Jill Burstein, M a r c h 15, 1990
Looking into Words 183
n. "I heard someone say," he began, "that you are a [New Zealander]. I
was out there as a small boy."
M a r s h , N., Night at the Vulcan, 1 9 5 1 , p . 2 0 7
o. Our neighbors, who are sort of [New York City-ites], they have jobs
there . . .
H e a r d in conversation, D e c e m b e r 30, 1990
3 . 1 . W h y R e f e r e n c e i n t o W o r d s Is O f t e n Infelicitous
(8) a. Fred is a cowboy. #He says they can be difficult to look after.
b. #John wants to be a fireman because he likes putting them out.
In an e x p e r i m e n t r e p o r t e d in M c K o o n et al. ( 1 9 9 0 ) a n d W a r d et al. ( 1 9 9 1 ) , s u b -
j e c t s read b l o c k s of e x a m p l e d i s c o u r s e s such as the t w o r e p r e s e n t e d in (9) a n d
w e r e s u b s e q u e n t l y p r e s e n t e d w i t h lists of test w o r d s w h e r e they w e r e required to
indicate as rapidly a n d accurately as possible w h e t h e r or not the test w o r d w a s in
o n e of t h e p r e c e d i n g d i s c o u r s e s . In cases w h e r e a w o r d w a s a modifier, such as
hostile in (9b), subjects r e s p o n d e d significantly m o r e slowly to the test item hos-
tile than w h e n that w o r d h a d a p p e a r e d in a predicate as in (9a); in the t w o dis-
c o u r s e s in (9), the c o m p l e m e n t a r y results w e r e found for intolerable. S o there is
g o o d psycholinguistic e v i d e n c e that modifier position is i n d e e d less accessible
than at least o n e other syntactic position, a n d w e suggest in W a r d et al. (1991) that
this fact is a n o t h e r factor in d e t e r m i n i n g the relative infelicity of reference into
words.
N o t e that the a b o v e f a c t o r s — s e m a n t i c opacity and the effect of the t y p e of
syntactic function (or position) of s o m e word-internal e l e m e n t s — d o h i n g e on
properties of the m o r p h o l o g i c a l c o n s t r u c t i o n s involved. B u t there is n o m o r p h o -
syntactic PROHIBITION against reference into w o r d s . Rather, various properties of
Looking into Words 185
3.2. I n c r e a s i n g t h e Felicity of R e f e r e n c e i n t o W o r d s
(10) a. TOPICAL: Sam likes the outdoor life. Having grown up in rural Ken-
tucky, he knows a lot about nature and is an expert at fishing and shoot-
ing. He goes on hunting trips as often as he can. He used to hunt just
small game, like rabbit and quail. However, lately he's taken up deer
hunting. He thinks that they are really exciting to track.
b. N O N T O P I C A L : Sam has many interests in the outdoors. He's an avid
skier, and each winter he takes about a month off from work to ski in
Colorado. In the summertime, he visits his parents in Montana where
he has a chance to do some mountain climbing. Lately, he's taken up
deer hunting. He thinks that they are really exciting to track.
3.3. Summary
(11) Mary has been fatherless for years; he died when she was five.
(12) John's gradual self-destruction by drug use was heartrending to his family.
(13) A : You're not English, are you? B : No, I was brought up there . . .
(Gaslight)
4. S O M E C O N C L U S I O N S
ACKNOWLEDGMENTS
I thank Sharon Hargus and an anonymous reviewer for helpful comments on this article.
Looking into Words 191
NOTES
1
Not a low-level phonological rule, contra the analysis of l\l variation in Halle and
Mohanan (1985) and Mohanan (1986). We make the common assumption (cf. Liberman
and Pierrehumbert, 1984) that phonetic implementation rules introduce continuous varia-
tion whereas phonological rules of any kind, including low-level ones, introduce categori-
cal variation.
2
N o t e that all speakers pronounced beel equator as a compound—i.e., deaccenting
equator—except speaker AD, who did place some prominence on equator in some of the
examples; this accenting difference did not appear to show up in the form of a different
trend for speaker AD.
3
If it is not visible, then the whole compound beel equator should count as a tetrasylla-
ble word, and the rime of beel should be shorter in this case than in the disyllabic beeling,
by Lehiste's argument.
4
On the interpretation of Booij and Lieber (this volume) and apparently also Cohn
(1989), metrical structure—prosodic structure, in their terms (see Inkelas, 1989, for a dis-
cussion of the terms prosodic and metrical)—is the only structure to which lexical pho-
nological rules can refer. On that view, then, the only representation of morphological
structure that the phonology sees is the metrical structure, so if BE were to apply so as to
render morphological information invisible to phonological structure, it would have to
erase metrical structure. Of course, metrically weak morphemes such as affixes might be
incorporated into whatever metrical structure they attach to, so that the resulting structure
is indistinguishable from a metrical structure derived from a monomorphemic input. So
BE might in such cases follow from metrical considerations alone. On the other hand, there
is no reason to assume on purely metrical grounds that the phonological wordhood of the
components of a compound should be erased.
5
1 note in passing that Booij (1983) uses metrical structure to account for some Dutch
data which appear to compromise lexical integrity. Also Mohanan (1986:60 n. 10) inti-
mates that such a compromise might be possible since "postlexical phonological opera-
tions have access to the phonological structures exiting from the lexicon (e.g. stress trees,
syllable trees)."
6
1 continue to use Postal's original terminology and talk of reference into words. How-
ever, in doing so I am being rather terminologically loose: the correct way to describe the
situation is to say that an anaphor is used to refer to a discourse entity evoked by a mor-
pheme which is morphologically contained within another word.
7
There are a number of possible reasons for this: as Paul Kiparsky has noted (personal
communication) pronouns are closed-class items and as such are generally barred from
undergoing morphological derivation. Of course, as an anonymous reviewer points out,
there are derived forms of prepositions such as thereto, thereabout. However, it is fair to
say that such formations are not productive in present-day English.
A somewhat related possibility is the more general statement that functional categories
and their projections cannot serve as the basis of morphological derivation in English. This
would link the unavailability of *r/z<?ra-hater with the unacceptability of *the The Bronx
hater (Fabb, 1984); note that on recent views of phrase structure, in particular Abney
(1987), a fully specified NP is considered to be a projection of a functional category (in
this case the determiner the), and therefore the full DP The Bronx could not be morpho-
192 Richard Sproat
logically contained within the compound by the proposed prohibition. See Sproat (1985,
1988) for a third possible account.
Finally, note that if a language lacks whatever grammatical constraint of English pre-
vents pronouns from occurring within words, then we would expect pronouns which do
occur, say as the lefthand member of a compound, to be able to corefer with antecedents
outside the word. This is exactly what we find in Sanskrit (Gillon, 1990).
Of course, it is necessary to check that the subject is resolving the reference correctly;
8
see McKoon et al. (1993) and Ward et al. (1991) for details of the methods used for veri-
fying that reference was resolved. Note also that topicality was shown in the cited work to
have a significant (though reduced) effect on reference to full noun phrases in discourses
similar to (10) (e.g., lately he's taken up hunting deer).
T h e fact that BE prevents -al and -ion from attaching to refuse and destroy in the
9
view intimated in Booij and Rubach (1987) that bracketings constructed at stratum II are
not erased by BE, since BE only applies at cyclic strata and stratum II is a noncyclic stra-
tum. A further compatible view is that of Hammond (1984). Hammond argues that BE
does not apply to compounds, although it applies to other morphological constructions. In
general, any version of lexical phonology and morphology which assumes a sufficiently
weak interpretation of BE will be compatible with more or fewer of the data presented here.
For a particularly lucid discussion of the various arguments for and against lexical
1 4
Again, Mohanan's proposal that BE apply in the syntax was intended to rule out pho-
1 7
REFERENCES
Abney, S. (1987). The English Noun Phrase in its Sentential Aspect. Doctoral dissertation,
Massachusetts Institute of Technology, Cambridge.
Allen, M. (1978). Morphological Investigations. Doctoral dissertation, University of Con-
necticut, Storrs.
Baker, M. (1988). Incorporation: A Theory of Grammatical Function Changing. Univer-
sity of Chicago Press, Chicago.
Booij, G. (1983). Coordination and reduction in complex words: A case for prosodic
phonology. Vrije Universiteit Working Papers in Linguistics 3, Vrije Universiteit,
Amsterdam.
Booij, G. and Rubach, J. (1987). Postcyclic versus postlexical rules in lexical phonology.
Linguistic Inquiry 18, 1 - 4 4 .
Chomsky, N. (1970). Remarks on nominalization. In Readings in English Transforma-
tional Grammar (R. Jacobs and P. Rosenbaum, eds.), pp. 1 8 4 - 2 2 1 . Ginn, Waltham,
Mass.
Chomsky, N., and Halle, M. (1968). The Sound Pattern of English. Harper and Row, New
York.
Cohn, A. (1989). Stress in Indonesian and bracketing paradoxes. Natural Language and
Linguistic Theory 7, 167-216.
Corum, C. (1973). Anaphoric peninsulas. In Papers from the Regional Meeting of the Chi-
cago Linguistic Society 9, 8 9 - 9 7 .
Di Sciullo, A. M., and Williams, E. (1987). On the Definition of Word. MIT Press, Cam-
bridge, Mass.
Downing, P. (1977). On the creation and use of English compound nouns. Language 53,
810-842.
Fabb, N. (1984). Syntactic Affixation. Doctoral dissertation, Massachusetts Institute of
Technology, Cambridge.
Gillon, B. (1990). Sanskrit word formation and context free rules. Toronto Working Papers
in Linguistics 11(2).
Halle, M., and Mohanan, K. P. (1985). Segmental phonology of Modern English. Linguis-
tic Inquiry 16, 57 - 1 1 6 .
Hammond, M. (1984). Level ordering, inflection, and the righthand head rule. MIT Work-
ing Papers in Linguistics 7, 3 3 - 5 2 .
Hargus, S. (1985). The Lexical Phonology of Sekani. Doctoral dissertation, University of
California, Berkeley.
Inkelas, S. (1989). Prosodic Constituency in the Lexicon. Doctoral dissertation, Stanford
University, Stanford, Calif.
Kiparsky, P. (1982a). Lexical morphology and phonology. In Linguistics in the Morning
Calm, (I.-S. Yang, ed.), pp. 3 - 9 1 . Hanshin, Seoul.
Kiparsky, P. (1982b). Word formation and the lexicon. Mid-America Linguistics Confer-
ence 3 - 2 9 .
Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook 2,
85-138.
194 Richard Sproat
Lakoff, G., and Ross, J. (1972). A note on anaphoric islands and causatives. Linguistic
Inquiry 3, 1 2 1 - 1 2 5 .
Le Roux, C. (1988). On the Interface of Morphology and Syntax: Evidence from Verb-
Particle Combinations in Afrikaans. Master's thesis, University of Stellenbosch, Stel-
lenbosch, South Africa.
Lehiste, I. (1972). The timing of utterances and linguistic boundaries. Journal of the Acous-
tical Society of America 51, 2 0 1 8 - 2 0 2 4 .
Lehiste, I. (1980). Phonetic manifestation of syntactic structure in English. Annual Bulletin
of the Research Institute of Logopaedics and Phoniatrics 14, 1 - 2 7 .
Levi, J. (1978). The Syntax and Semantics of Complex Nominals. Academic Press, New
York.
Liberman, M., and Pierrehumbert, J. (1984). Intonational invariants under changes in pitch
range and length. In Language Sound Structure. (M. Aronoff and R. Oehrle, eds.),
pp. 1 5 7 - 2 3 3 . MIT Press, Cambridge, Mass.
Lieber, R. (1984). Grammatical rules and sublexical elements. In Papers from the Para-
session on Lexical Semantics, pp. 187-199. Chicago Linguistic Society, Chicago.
Lieber, R. (1991). Deconstructing Morphology: Word Formation in a Government-Binding
Syntax. The University of Chicago Press, Chicago.
McCawley, J. (1989). 1989 Linguistic Flea Circus. Unpublished manuscript, University of
Chicago.
McKoon, G , Ratcliff, R., Ward, G , and Sproat, R. (1990). Structural and Discourse Ma-
nipulations of Salience in the Interpretations on Anaphora. Paper presented at the
Third Annual CUNY Conference on Human Sentence Processing, New York, N.Y.
McKoon, G , Ward, G , Sproat, R., and Ratcliff, R. (1993). Morphosyntactic and pragmatic
factors affecting the accessibility of discourse entities. Journal of Memory and Lan-
guage 32, 1 - 2 0 .
Mohanan, K. P. (1982). Lexical Phonology. Doctoral dissertation, Massachusetts Institute
of Technology, Cambridge.
Mohanan, K. P. (1986). The Theory of Lexical Phonology. Reidel, Dordrecht.
Nespor, M., and Vogel, I. (1986). Prosodic Phonology. Foris, Dordrecht.
Pesetsky, D. (1979). Russian Morphology and Lexical Theory. Unpublished manuscript,
Massachusetts Institute of Technology, Cambridge.
Pierrehumbert, J., and Beckman, M. (1988). Japanese Tone Structure. MIT Press, Cam-
bridge, Mass.
Postal, P. (1969). Anaphoric islands. Papers from the Regional Meeting of the Chicago
Linguistic Society 5, 2 0 5 - 2 3 9 .
Reinhart, T. (1983). Anaphora and Semantic Interpretation. University of Chicago Press,
Chicago.
Selkirk, E. (1980). The role of prosodic categories in English word stress. Linguistic In-
quiry 11, 5 6 3 - 6 0 6 .
Shibatani, M., and Kageyama, T. (1988). Word formation in a modular theory of grammar.
Language 64, 451 - 4 8 4 .
Siegel, D. (1974). Topics in English Morphology. Doctoral dissertation, Massachusetts In-
stitute of Technology, Cambridge.
Silverman, K. (1988). Utterance-internal prosodic boundaries. Proceedings of the Austra-
lian International Conference on Speech Science and Technology 2, 8 6 - 9 1 .
Looking into Words 195
TONI BOROWSKY
Department of Linguistics
University of Sydney
Sydney, New South Wales 2006, Australia
1. INTRODUCTION
199
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
200 Toni Borowsky
(1) - LEXICON -
- PHONOLOGY-
Morphological Lexical
operations, phonology,
Stem/Level 1 Level 1
Morphological
Lexical
operations,
phonology
Word/Level 2
Postlexical
phonology
(2) - LEXICON
PHONOLOGY-
Morphological
Lexical
operations,
phonology
Level /
Morphological
Lexical
operations,
phonology
Level n
Postlexical
phonology
2. T H E W O R D C Y C L E
(4)
centr + al
AA
sen tr dl
(5) o-
A
center sen (tr) Level 1
a o~
Level 2: word cycle
sen tr Phonology: Sonorant Syllabification
(7 <J
A it
[sen t r] [sen
An
t r + irj]
CT O"
2.1. G e r m a n Morphophonemics
patterns in (6).
Similarly, in word-final position, a n d before Level 2 affixes, [g] deletes after [rj],
as well as w h e n it c a n n o t b e syllabified (Hall, 1989b).
(7) L E V E L 1:
diphthong + ier + en 'to d i p h t h o n g i z e '
tang + ier + en 'to affect'
laryng + al 'laryngeal'
versus
L E V E L 2:
Spreng + ung 'explosion'
sing 4- e '(I) s i n g '
Jung + en 'boys (nom.)'
Jung + ling '(a) y o u t h '
(9) f-»jc/[+back]
(10) a.
Kuhchen [ku:90n] 'little c o w '
Pfauchen [ p ao£on]
f
'little p e a c o c k '
Tauchen [taocon] 'little r o p e '
b. [x]
Kuchen [ku:xsn] 'cake'
pfauchen [ p aoxon]
f
'to hiss'
tauchen [taoxan] 'to dive'
(11) -son
+cont
-voice
V ,C
u- 1
[back] [+high]
(12)
S t e m / L e v e l 1: hoch kuh kuchen
hoch hoch Umlaut
(Lieber,
1989)
W o r d / L e v e l 2: hoch hoch kuh kuchen
[x] [x] Fricative
Assimilation
Morphology: hoch+lich kuh+chen
[hox] [hoclic] [ku:can] [ku: xan]
2.2. S u m m a r y
3. E N G L I S H A L L O P H O N I C R U L E S
3.1. Belfast D e n t a l i z a t i o n
O n the basis of these properties, as well as the fact that its pattern is exactly that
of the clearcut Word-level rules, I assign it to the W o r d level instead, p a c e Hall.
Let us then c o m p a r e the derivations of the forms in (16). At Level 1, the rule
d o e s not apply at a l l — i t is b l o c k e d by structure p r e s e r v a t i o n as well as the S C C .
At Level 2 it applies on the w o r d cycle to g e n e r a t e the correct f o r m s in the first
and last c a s e . S i n c e m o r p h o l o g y follows p h o n o l o g y , the rule d o e s not get an o p -
portunity to apply on the W o r d cycle in the m i d d l e c a s e . After the affix -er is
attached to wide, the rule d o e s not apply. T h u s wider h a s a n o n d e n t a l d.
3.2. A i t k e n ' s L a w
(18) ]d d]
agreed [9gri:d] greed [grid]
kneed [i:] need [nid]
brewed [u:] brood [brud]
stayed [e:] staid [sted]
toed [o:] toad [tod]
gnawed [o:] node [nod]
3 . 3 . L o n d o n [AU] ~ [DU]
(21) [AU]->[DU]/ 1L 1 3
(24) [io]->[e:]/ #
T h u s w e fix this rule on the W o r d level. In o r d e r to derive the forms in (25), the
rule m u s t apply o n t h e W o r d c y c l e before any Word-level affixation, as follows.
4. T H E W O R D L E V E L
4 . 1 . R u l e s W h i c h A p p l y t o Affixes: / - i n g /
phonology.
T h e usual a s s u m p t i o n a b o u t the p h o n o l o g i c a l b e h a v i o r of b o u n d m o r p h e m e s
like / - i n g / is that p h o n o l o g y should take p l a c e only after they are attached to the
w o r d . G i v e n this, w e h a v e n o r e a s o n to e x p e c t that the g should not b e h a v e like
any final g at Level 1 — t h a t is, it should b e e x t r a m e t r i c a l o n its o w n cycle a n d
r e s c u e d b y syllabification into the onset of the following syllable [[[swrn^] iNg]
est] on the next. O f c o u r s e it is not. T h e form ^[swirjirjgost] is u n g r a m m a t i c a l .
G i v e n this, let us s u p p o s e instead that the g is not e x t r a m e t r i c a l . T h e n w e can
derive the correct result if w e e n s u r e the derivation is cyclic, as s u g g e s t e d in B o -
r o w s k y ( 1 9 8 6 ) — t h a t is, the g will d i s a p p e a r on the Iswiq+ingl cycle. T h i s is
directly c o u n t e r to all o t h e r c l a i m s a b o u t t h e W o r d level in t h e literature (Booij
and R u b a c h , 1987; H a l l e a n d K e n s t o w i c z , 1989; H a l l e and M o h a n a n , 1985; Halle
a n d Vergnaud, 1987), w h e r e it is a r g u e s that the W o r d level is n o n c y c l i c .
H o w e v e r , if the derivation is cyclic, there are still p r o b l e m s . W e h a v e seen that
o n e of the p r o p e r t i e s of t h e s e w o r d - c y c l e rules is t h e fact that if their structural
description is m e t on a c y c l e created after affixation at Level 2, the rule d o e s not
apply. T h a t is, in t h o s e c a s e s , G e r m a n Fricative A s s i m i l a t i o n and Belfast D e n t a l -
ization, in w h i c h the trigger a n d the target of the rule a p p e a r e d in different m o r -
p h e m e s , the rule never applied. T h e rules only applied tautolexically. If the deri-
vation w e r e cyclic, then the rule should get a n o t h e r c h a n c e to apply o n the c y c l e
at w h i c h the trigger and target b o t h occur. B u t this is not the c a s e ; neither of these
t w o rules applies after the w o r d c y c l e . N o t e that the s a m e situation arises if the
1 7
4.2. R u l e s W h i c h S e e m N o t t o A p p l y at L e v e l 2 : N a s a l A s s i m i l a t i o n
and Stress
attached after the stress rules a p p l y . T h a t is, the failure of certain stress rules to
18
the fact that the o n e class of affixes never h a s any effect on previously assigned
stresses w h i l e the other d o e s . It d o e s not follow from cyclicity. It is j u s t as p o s s i b l e
that earlier stresses could b e w i p e d out o n a n o n c y c l i c d o m a i n as on a cyclic
d o m a i n . N o t e too that the fact that it is a l w a y s the second-level affixes w h i c h are
stress neutral and the first-level affixes stress sensitive is u n e x p l a i n e d so long as
s o m e stress rules c a n a p p l y at the s e c o n d level. It c o u l d in p r i n c i p l e b e the other
w a y r o u n d given any other lexical p h o n o l o g y m o d e l — b u t it is not.
If w e a s s u m e that the stress rules apply to e a c h m o r p h e m e individually on the
w o r d c y c l e , it follows that these affixes will a l w a y s b e stress-neutral with respect
2 0
attributing the differences b e t w e e n the t w o levels to the fact that the levels are
c h a r a c t e r i z e d b y distinct sets of rules and, m o r e recently, to the c l a i m that the
W o r d level is n o n c y c l i c . H o w e v e r , neither of these ideas gives a satisfactory e x -
planation of the o b s e r v e d differences in S t e m forms and W o r d forms. T h e hy-
p o t h e s i s that certain rules are restricted to certain levels d o e s n o t e x p l a i n w h y t h e
forms should b e superficially so different. To say that the d e g e m i n a t i o n rule, or
the stress rule, or constraints on the structure of a w e l l - f o r m e d syllable turn off
after the first level is to explain n o n e of t h e s e things. It leaves aside as well the
fact that those rules w h i c h d o apply at the s e c o n d level are still not a c c o m p l i s h i n g
the s a m e sort of e d g e - e r a s i n g function as the rules of the first level. T h e output of
the Word-level p h o n o l o g y a n d m o r p h o l o g y n e v e r l o o k s like an u n d e r i v e d w o r d .
N e i t h e r can the c l a i m that the w o r d level is n o n c y c l i c , as h a s b e e n s u g g e s t e d by
Halle a n d M o h a n a n ( 1 9 8 5 ) , Booij a n d R u b a c h ( 1 9 8 7 ) , or K i p a r s k y (1985) p r o v i d e
an a d e q u a t e e x p l a n a t i o n . In fact, as w e will see below, the c l a i m is descriptively
i n a d e q u a t e . If the W o r d level is n o n c y c l i c , rules should take p l a c e across the b o a r d
within the p r o s c r i b e d d o m a i n , as in the postlexical p h o n o l o g y , and not b e re-
stricted to smaller d o m a i n s internal to the p r o s c r i b e d d o m a i n , as it s e e m s to b e at
the W o r d level.
A l t h o u g h it m a y apply cyclically inside nested p r o s o d i c d o m a i n s , postlexical
p h o n o l o g y is r e p u t e d to h a v e the p r o p e r t y of a p p l y i n g the rules across the b o a r d
within e a c h of these d o m a i n s , i g n o r i n g c o n s t i t u e n c y information internal to that
p r o s o d i c d o m a i n . T h u s , for e x a m p l e , the F l a p p i n g rule c h a n g e s / t , d / to [D] in foot-
m e d i a l position w h e r e v e r its structural description is m e t within s o m e larger p r o -
sodic d o m a i n . T h u s it applies inside w o r d s , as well as inside s o m e phrasal struc-
tures: repe[D]i[D]iveness, mor[D]ality, go [D]omorrow, ge[D] Anne. Word-level
p h o n o l o g y s h o w s n o t h i n g like this taking place. W h a t w e have seen instead is that
m o r p h o l o g i c a l b o u n d a r i e s at the W o r d level create i n d e p e n d e n t p h o n o l o g i c a l d o -
m a i n s internal to the d o m a i n of the w h o l e w o r d and that rules respect t h o s e inter-
nal d o m a i n s — t h e y apply tautolexically.
Let us a s s u m e that the first or stem level of the lexicon contains, inter alia, a list
of stems and a list of affixes, as well as the rules w h i c h relate e a c h stem to the
others in the list (or, alternately v i e w e d , the rules d e r i v e e a c h s t e m from another).
W h a t this m e a n s is that e a c h form derived at Level 1 exists as an i n d e p e n d e n t
form in the l i s t . A s s u m e as well that the W o r d level is distinguished by the fact
23
that every possible derived form is actively derived, and these lexical entries are
220 Toni Borowsky
not i n d e p e n d e n t l y listed. S o , in this lexicon, all the stems, as well as both classes
of affixes, are listed, but n o d e r i v e d Word-level f o r m s o c c u r in the list. If w e as-
s u m e all this to b e the case, the set of affixes w h i c h select for S t e m s (i.e., the
Level 1 affixes) can b e identified w i t h o u t explicit m a r k i n g . T h e s e affixes are the
o n e s w h i c h o c c u r in the listed f o r m s . T h e Word-level affixes o c c u r in the lexical
lists only as separate b o u n d m o r p h e m e s . T h e y d o not also a p p e a r in the listed
w o r d s . T h e t w o classes of affixes c a n b e differentiated by structure preservation,
as it w e r e . T h e set of affixes w h i c h also o c c u r as c o m p o n e n t s of listed w o r d s will
b e the Level 1/Stem-level affixes.
T h e fact that the S t e m level is cyclic is d u e to the fact that the existing lexical
items are derived from o n e another. T h u s presidential is derived from /president
+ ial/, a n d presidentiality is d e r i v e d from /presidential + ity/. Similarly, the fact
that the level is s t r u c t u r e - p r e s e r v i n g follows: relating f o r m s in existing i t e m s al-
lows n o r o o m for innovation in the d e r i v a t i o n s . Clearly, n o n e w s e g m e n t s or struc-
tures c o u l d b e i n t r o d u c e d . (I refer the r e a d e r to Sproat, 1985, w h o m a k e s the s a m e
observations.) N o t e , lastly, that the S C C , b y w h i c h structure-changing rules are
b l o c k e d in n o n d e r i v e d e n v i r o n m e n t s , is also derived: listed entries will b l o c k any
internal c h a n g e s d u e to t h e e l s e w h e r e condition (see Kiparsky, 1982), and thus
the only p e r m i s s i b l e c h a n g e s are o n e s at e d g e s w h e r e n e w material is b e i n g
incorporated.
N o n e of these characteristics carry over to the W o r d level. W e d o not readily
a s s u m e that m o r p h o l o g i c a l l y c o m p l e x w o r d s at this level m u s t b e l i s t e d — a t least
not in the w a y that conjoin, conjunctive, conjunction, for e x a m p l e , are listed. T h e
first level of the lexicon is not c o n s i d e r e d to b e " p r o d u c t i v e " b e c a u s e it is
2 4
5. A N A L T E R N A T I V E P R O P O S A L
(27) Level 1: a
k
lorj
k (g) lor) (g)
Syllabification
Final C extrametricality
kA Morphology—extrametricality lost
lor) g + er
a a
k A
lor) g er
Syllabification
Level 2:
Word cycle
lor) g 8 —• 0
0
a
A
lor] + ir/g Morphology
a a
AA
Iov iv (g) Syllabification, etc.
H o w d o e s this idea explain the facts I h a v e b e e n d i s c u s s i n g ? If syllabification
from the w o r d cycle is m a i n t a i n e d t h r o u g h o u t the W o r d level, then any rule w h i c h
is sensitive to syllabification will apply in syllables w h i c h h a v e not b e e n restruc-
tured. F o r e x a m p l e , a rule w h i c h I h a v e c l a i m e d m u s t apply on the w o r d c y c l e
could in fact apply at any t i m e at the w o r d level b e c a u s e the syllable structure at
this level will not b e c h a n g e d even after suffixation. T h u s , so long as w e allow
syllabification to take p l a c e either before or at the w o r d cycle, the other rules will
apply w i t h o u t p r o b l e m .
(28)
Level 1: po (/)
hA h
po I + ar ro (11) Syllabification
cr
Level 2:
A
pole polar
A
roll Syllabification
AA
roll + er
Syllabification
Rule (21)
Recall that the m o s t serious a r g u m e n t against the idea that the W o r d level w a s
cyclic after the w o r d cycle w e r e t h o s e cases in w h i c h the addition of an affix
On the Word Level 223
A
way (d) Syllabification
Level 2:
Word cycle:
/IN
way d Syllabification
/IN I I
Morphology: way d + dr Syllabification
NA Belfast Dentalization
N/1
V C ] V
w i d e r Belfast Dentalization
r o 11 i ng London [^u] / [vu]
224 Toni Borowsky
(31) * o-
K
v ] c
agree d Aitken's law
paw s London pause/paws (Appendix)
(32) *
A
VC f V
II I
u n e thical
6. A P P E N D I X : F U R T H E R E X A M P L E S O F W O R D - L E V E L R U L E S
6.1. ^ - T e n s i n g
(33) ce T e n s i n g :
nasals
ce - > E / _ fricatives
v o i c e d stops
[E]
graph graphic
psychopath psychopathic
mass, massive massive
class, classy classical, classify
classing
Level 2 :
Word cycle cl[E]ss r^-Tensing
Cycle 2 cl[E]ss + y
6.2. V o w e l R o u n d i n g i n A d e l a i d e E n g l i s h
In the A d e l a i d e dialect of A u s t r a l i a n E n g l i s h as d e s c r i b e d b y S i m p s o n ( 1 9 8 0 ) ,
there is a rule similar to the L o n d o n rule in (28). C e r t a i n v o w e l s are r o u n d e d in
the e n v i r o n m e n t of a tautosyllabic /.
6.3. L o n d o n pause—paws
T h e affixes d o not close the syllables for the rule. Instead the rule applies as if the
final vowel of the stem w a s in an o p e n syllable, itself final. T h e rule thus applies
on the w o r d cycle before any Level 2 m o r p h o l o g y .
T h e r e are a few other rules w h i c h a p p e a r to m e to b e e x a m p l e s of w o r d - c y c l e
rules; however, I d o not h a v e e n o u g h information a b o u t t h e m to be absolutely
sure. T h e s e rules exhibit the s a m e sort of patterning s h o w n by all the previous
examples.
P r e - / B r e a k i n g (Wells, 1982) m a k e s a m o n o s y l l a b i c w o r d disyllabic before /.
A s usual, the effects of this rule carry o v e r to affixed forms in w h i c h the / appears
n o w to b e in the onset of the affix syllable.
ACKNOWLEDGMENTS
I thank Geert Booij, Allan James, John Harris, Harry van der Hulst, John McCarthy,
Norval Smith, and three anonymous reviewers for their comments on earlier versions of
this paper. John Harris, Paul Kiparsky, and John McCarthy also provided me with addi-
tional examples of word-level rules. I am also grateful to students in my seminars at the
University of Delaware and Rijksuniversiteit te Leiden, in particular Barbara Bullock and
On the Word Level 229
Mark Verhijde, as well as to audiences at the Lexical Phonology Workshop, Seattle; the
University of Texas, Austin; and the University of Massachusetts, Amherst. The article was
completed while I had the honor of being a fellow at the Netherlands Institute for Advanced
Study (NIAS), and I am grateful indeed to the Institute for providing me with such a pleas-
ant and wonderful environment.
NOTES
1
Three general types are distinguishable:
deletion is unconvincing because it is so small: that the relevant words should simply be
marked as exceptions to g-deletion. Yet the observation that the facts pattern in one direc-
tion and not in the opposite direction is a valid one. g does not delete before the Level 1
affixes. This is significant. Marking forms as exceptions when their pattern is predictable,
even if there are only a few such forms, misses the point. True exceptions would not fall
into a pattern and we would get a random set of variations.
7
The formulation of the rule is not crucial and does not affect my point, since it is the
pattern of distribution I am concerned with. The correct analysis seems to me to be one
similar to Hall's, in which the fricative is unspecified and gets [back] by rule and is other-
wise nonback.
8
If you think about it, all the rules seen so far—English as well as German—have a
similar property. The crucial factor for all these rules is the fact that they are blind to any
material outside their immediate domain. Most of the rules I have discussed so far have the
property of applying only within the Word built in, by dint of the ] in their conditioning
environment. (Note it may equally well be [.) This seems to be similar to saying these
processes only apply tautomorphemically. The sole difference is that in these rules the word
edge is the conditioning environment.
9
In some dialects of German the rule is postlexical and in fact the forms in (10) are all
pronounced with [x] as expected.
10
N o t e that in these dialects these words are pronounced without a vowel in the suffix:
[elamentriy], [saenitriy].
1 1
Harris formulates the rule as a blank-filling rule, which gets him around the SCC
violation. However, on his analysis it remains an odd rule since it applies O N L Y at Level 1,
which is peculiar by virtue of the fact that it is clearly not structure-preserving, as he him-
self notes. I think this is dubious since it is quite unlike any of the other rules of English
which are restricted to Level 1. These are a small set of highly morphologized rules, e.g.,
Trisyllabic Laxing, Velar Softening, and Spirantization. Belfast Dentalization is quite un-
like any of these rules.
I t is not entirely clear what the effect of stress on Aitken's law is. The stress condition
1 2
is perhaps more complicated than is presented. As John McCarthy has pointed out to me,
the only unstressed vowels in English are schwas, which would not undergo the rule. The
only cases of nonschwa unstressed vowels occur word finally. He notes as well that there is
no interaction of cyclic stress assignment with the rule which supports the claim that it is a
Word-level rule.
1 3
See Harris (1990) for arguments leading to his assigning this rule to Level 2.
1 4
Booij and Rubach (1987) discuss Dutch Syllable Final devoicing ( [ — son] —>
[ — voice] / $) as an example of what they refer to as a word-level rule which applies
non(post)cyclically. That this rule is not cyclic is illustrated by alternations like held [t]
'hero' vs. heldin [d] 'heroine'. They argue further that the rule is lexical on the grounds
that it precedes two other clearly postlexical processes, Voicing Assimilation and Resyllab-
ification. This argument, based on ordering of rules, does not conclusively place Dutch
Final Devoicing in the lexicon, however, and I assume that this rule and others like it (e.g.,
English r-deletion) are very early postlexical rules which apply in the first postlexical do-
main of phonological word.
A study of the few rules associated with the rest of the Level 2 affixes reveals only
1 5
On the Word Level 231
one possible candidate which could show some phonological interaction between affix and
stem or affix and affix. That is the cluster of rules, Voicing Assimilation and epenthesis/
deletion, found with the inflectional affixes -ed and -es. I suggest, however, that these pro-
cesses are postlexical and do not take place at Level 2 at all. We know that they must apply
postlexically in certain syntactic structures: the dog's bone, the cat's milk, the horse's
mouth; as well as after postlexical reduction processes: the cat's gonna bite, the dog's
gonna scratch, the horse's gonna snort, etc. There is no evidence to force us to apply the
rules at Level 2 when the affixes are attached.
1 6
T h e same observation holds for the l-ungl affix in German. Bloomfield (1930) sug-
gested that the German affix /-chen/ should be treated as an independent form for the pur-
poses of the rule which determines the distribution of [x] and [c] exactly as I suggest in
this system. The affix must cycle independently through the phonology in order to derive
the [c] from some underlying underspecified fricative.
1 7
Another possibility which has been suggested to me is the one to be discussed in
Section 5.
1 8
Note that I exclude the compound stress rule or the nuclear stress rule. When I refer
to the regular stress rules I mean the word-stress rules. For an account of the different
compound stress phenomena compatible with my view, see, e.g., Sproat (1985), Sel-
kirk (1982).
1 9
Perhaps also /-ly/, l-yl and /-ish/—all of which retain their full vowels. It could be
that these affixes have tense vowels but are not stressed, -ize is another affix which seems
stressed to me. This affix is generally problematic, and it is sometimes claimed that it has
a dual class membership (Kiparsky, 1982; Selkirk, 1982). I personally believe this affix is
really a Level 1 affix and not stress-neutral at all—as evidenced by words like democratize,
automatize, catholicize, and others. I suggest that the affix is itself stressed and thus shifts
stress off adjacent stressed syllables. It does not have any effect on stress in a word like
standardization because there is no stress adjacent to its own stress. Stress cannot shift to
the last syllable of standard because that would cause a clash.
2 0
Note that it does not follow that all Level 2 affixes should be stressed just because
each of them has been cycled on. For example, suffixes like -ness, -less, and even -dom
with internal [a]s do not get stressed (e.g., stresslessness, mercilessness have only one
stress per word). On the other hand, affixes like un-, re-, and -hood are stressed even though
their stresses do not appear to affect the rest of the stresses of the word—e.g., reroute,
unfortunate, boyhood. Similarly, affixes consisting of one consonant, say, either would be
repaired by syllabification by the insertion of a [a], or they could escape syllabification till
later. Repairing them by means of epenthesis would have the effect of making the English
inflectional affixes l-dl and l-zl into /-ad/ and /-dz/. This could be why we cannot decide
which is the correct underlying representation for these affixes.
2 1
The only potential edge-erasing phonology we find is CV syllabification, which we
assume takes place automatically and continuously throughout the derivation (ltd, 1986).
However, even if syllabification were to be suspended, there is no evidence which would
force the syllabification of stem final Cs to apply at the word level. It could be done on the
first postlexical phonological domain, as the Dutch final devoicing facts mentioned in
note 14 suggest.
2 2
With the exception of Sproat (1985), who makes most of these observations.
232 Toni Borowsky
2 3
Following Kiparsky (1982) we could assume that the output of every cycle is a "lexi-
cal entry." This could be interpreted to mean that each form derived at Level 1 is "listed"
in a strange sense (not exactly that of Jackendoff): the lexical identity rule for each form.
2 4
Of course this is not true. We C A N produce new words by means of the morphological
and phonological rules at the first level. The lexicon is not merely a list of existing words.
But it does seem that by some measure of productivity the Word level is more so than the
Stem level.
2 5
Suppose that to be productive in the sense of Word-level productive means that the
affix is used on line in the actual creation of complex words. That is, Stem-level forms may
be analyzable but when you use them in production you just do a lexical look-up of the
whole word. It is lexically listed. On the other hand, when you use a Word-level affixed
form, you actually put it together during production. It is not available as a prepackaged
form. Psycholinguistic data, as far as I know, have shown that Level 1 phonological and
morphological processes are never affected in speech errors and so on, while Level 2 pro-
cesses are. We could suppose therefore that the base of affixation in these forms is not an
underlying representation but an actually occurring word. This would support my picture
of the lexicon. I am grateful to Donca Steriade for making these points to me.
2 6
A similar suggestion was made by an anonymous reviewer.
2 7
Conservative South African English.
2 8
It may be possible to redefine these rules as foot rules, similar to English Flapping.
Even on this assumption, though, we cannot maintain the idea that lexical syllabification is
limited to structure-building processes only. However, until more is known about the rules
in question, I leave these speculations for future research.
2 9
English and German are not the only languages which have a word cycle. For some
word-level rules in Arabic dialects see Dunlap (1987), who discusses two rules: Palestinian
Backing (from Younes, 1984) and Bedouin Hijazi Liquid Emphaticization (from Al-
Mozaini, 1982). Booij and Rubach (1987) refer to Rochet's (1973) discussion of Loi de
position and other examples from French dialects. Danish Grave Assimilation appears to
me to be a similar type of rule (see Borowsky, Ito, and Mester, 1984, and references cited
therein). Most of the well-known word-final cluster simplification rules, especially in Indo-
European languages (Catalan: see, e.g., Kiparsky, 1985; Danish: Ito, 1984; Icelandic: e.g.,
Kiparsky, 1985; Ito, 1986; etc.), appear to be examples of the same Word-level prosodic
licensing phenomena as the English cases in (3).
30
R e c a l l that London English is a nonrhotic dialect: lore and law are homophonous.
REFERENCES
Rochet, B. (1973). On the Status of the Word in French Phonology. International Review
of Applied Linguistics in Language Teaching 25, 187-196.
Rubach, J. (1990). Final devoicing and cyclic syllabification in German. Linguistic Inquiry
21, 7 9 - 9 4 .
Selkirk, E. O. (1982). The Syntax of Words. MIT Press, Cambridge, Mass.
Simpson, J. (1980 Cyclic Syllabification and a First Cycle Rule of Vowel-rounding in Some
Dialects of Australian English. Unpublished manuscript, Massachusetts Institute of
Technology, Cambridge.
Sledd, J. (1958). Some questions of English phonology. Language 34, 2 5 2 - 2 5 8 .
Sproat, R. (1985). On Deriving the Lexicon. Doctoral dissertation, Massachusetts Institute
of Technology, Cambridge.
Wells, J. (1982). Accents of English, 3 vols. Cambridge University Press, Cambridge.
Younes, M. (1984). Emphasis and the low vowels in Palestinian Arabic. In Working Papers
in Cognitive Science (J. McCarthy and A. Woodbury, eds.).
STRUCTURE PRESERVATION AND
POSTLEXICAL TONOLOGY IN DAGBANI
LARRY M. HYMAN
Department of Linguistics
University of California
Berkeley, California 94720
1. I N T R O D U C T I O N
235
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
236 Larry M. Hyman
2. L E X I C A L T O N O L O G Y
A s a p r e l i m i n a r y to u n d e r s t a n d i n g the n o m i n a l t o n o l o g y of D a g b a n i , the e x -
a m p l e s in (1) are d e s i g n e d to illustrate that the vast majority of n o u n s e n d with an
overt ( n o u n class) suffix. 5
(7) a. a a a b. a a a
L H H H © H //
H H H L
i
L
as f o r m a l i z e d in (10).
(10) (<x)
H L
I I I
H H H
I I
H L
[pag-a] H-tone Anticipation (9a)
N
H
H-tone Spreading (10)
V
H L
[pag-a] [san-a( )]
!
Surface representation
(6b) w o u l d b e r e p r e s e n t e d as in (12b).
(12) a CT cr t>. !
yi-li 'stranger's house'
I I v v
H L H H L H
T h e floating L following such n o u n s as sdn-d ' s t r a n g e r ' h a s n o effect before
p a u s e . However, as p o i n t e d out b y W i l s o n ( 1 9 7 0 : 4 0 9 ) , n o u n s such as pdg-d and
sdn-d h a v e different surface realizations w h e n a c c o m p a n i e d b y the vocative into-
nation w h i c h involves a l e n g t h e n e d final v o w e l .
3. P O S T L E X I C A L T O N O L O G Y
V II
H L H
b. san-a kodu —• sdn(-d) kodu 'stranger's banana'
I I II
H L LH
C. wa-hu kodu —• wd-h(u) kodu 'horse's banana'
I I II
L H L H
d. kparj-a kodu • kpdrj-'(d) kodu 'guineafowl's banana'
HL H LH
T h e input tones indicated on pdg-d and sdn-d in (14a,b) are t h o s e lexically derived
in (11). In the outputs in (14), the final v o w e l of the p o s s e s s o r n o u n is s h o w n in
p a r e n t h e s e s since, as w a s already said, it is subject to reduction, even deletion.
Structure Preservation in Dagbani 241
'du b e c o m e s L. Finally, the H on the -hit suffix in (14c) is lost, either as a result
of the vowel r e d u c t i o n p r o c e s s or by t h e rule of L - t o n e S p r e a d i n g in (7a).
T h e intrigue of the a b o v e c o m e s from the c o m p a r i s o n of (14a) and (14b).
W h e r e a s H T S j will derive the correct o u t p u t in (15a), it will not p r o d u c e the
correct output in ( 1 5 b ) .
H LH
b. san(-a) kodu *[san kodu] (cf. [san kodu])
V\ I I
H L L H
H T S j applies b e t w e e n the t w o n o u n s in (15a) to derive the i n t e r m e d i a t e form
s h o w n in the right c o l u m n , w h i c h t h e n u n d e r g o e s final l o w e r i n g of H t o L. In
!
L L H
b. sa: rj kar-li [sa:T| kar-li] 'big stranger'
H L H
C. war kar-li [war kar-li] 'big horse'
L L H
d. kpa:rj kar-li [kpa.T| kar-li] 'big guineafowl'
HL L H
H L H L
b. a a gives rise to (d) d
H L L H L
A
L H L H L H
b . bd —• bd 'father' bd kodu
i i n I
H H H L H L L H
W h i l e these t w o n o u n s s h o w a single t o n e in isolation, L versus H respectively, it
is clear from the genitive f o r m s to the right that they actually h a v e a H - t o n e suffix.
In (18a) this H is a n c h o r e d o n t o the L of kodu ' b a n a n a ' , w h i c h is then delinked,
creating i n t e r m e d i a t e ko'du, w h o s e H then is lowered to kodu before p a u s e . In
!
© — - 0 / L
, A
H L
(20) a. b.
H © L H (©)
1 L
v v I I i v I
H L H H L L H
HTS, san-a akarma
pag-a akarma
M / II I i v I
H L L H
H L H
HTS. san-a akarma
pag-a akarma
H L H H © L H
Contour Simplification san-a akarma
H © L H
phonetic representation [pa^ akarma] [san akarma]
syllable. T h u s , if there is n o © - d e l e t i o n , H T S w o u l d h a v e t o b e f o r m u l a t e d as in
2
A L, //
244 Larry M. Hyman
T h e need for this rule is seen in the last stage of the derivation in (21b), since the
output is not *[san a k a r m a ] .
T h i s , then, c o m p l e t e s the presentation of the t w o postlexical p h e n o m e n a
d u b b e d H T S and H T S . In the n e x t section I c o n s i d e r different interpretations of
] 2
these facts.
4. D I S C U S S I O N
any case be c o l l a p s e d . 16
A s a s e c o n d difference, H T S t c a u s e s the L to delink,
w h i l e H T S d o e s not, thereby f o r m i n g a H L c o n t o u r o t h e r w i s e unattested in the
2
language.
T o slightly restate these findings, there are three logical analyses, as s u m m a -
rized b e l o w in (23).
the fact that the properties of these t w o rules differ slightly m u s t simply
b e stipulated.
b. H T S j a n d H T S constitute a single rule w h i c h h a s t w o separate ap-
2
Unfortunately for this analysis, HTS and H T S apply in exactly the SAME place
{ 2
in the p h o n o l o g y .
A s a first d e m o n s t r a t i o n of this fact, c o n s i d e r the derivations in (24).
v
(24) a. kddu 'banana' versus kddu titd-li 'big banana'
II v
L H
LH
b. pdg(-d) kodu titd-li 'woman's big banana'
H L H
C. sdn(-d) kddu titd-li 11
'stranger's big banana'
H L L H
kddu, first yielding kddu, then kddu. In (24c), the lexical H of the n o u n stem san
' s t r a n g e r ' first s p r e a d s o n t o the suffixal -a, delinking its L, a n d then spreads
again o n t o the first syllable of kddu ( p r o d u c i n g kddu, w h i c h then simplifies b y
L - a b s o r p t i o n to kddu, as seen).
T h e derivation in (24b) s h o w s that the s a m e w o r d m a y b e affected b y b o t h
H T S j a n d H T S . N o w c o m p a r e ( 2 5 a ) , w h e r e H T S and H T S b o t h c r o s s a w o r d
2 x 2
V ^ - ? " 1 I
H L L H
b. sdn(-d) war kar-li 'stranger's big horse'
W 1 II
H L L L H
T h e derivation in (25b) s h o w s the lexical H of the n o u n stem san s p r e a d i n g o n t o
the suffix -a, followed b y a s e c o n d s p r e a d i n g of this H o n t o the following n o u n
stem war ' h o r s e ' . C o m b i n i n g these o b s e r v a t i o n s , w e arrive at the c o n c l u s i o n that
both H T S j and H T S are free to apply either within or across w o r d s .
2
i
L L
v I
H
b. pdg(-d) zag-si bd 'a woman refused them'
H L H
C. sdn(-d) zdg-si bd 'a stranger refused them'
V W I
H L L H
(27) ( B r a n c h i n g ) subject + v e r b :
a. doo rjwun jie pdg(-d) ] Su zdg-si bd 'a man who saw a woman refused them'
H L H
b. doo rjwun jie sdn(-d) ]Su zdg-si bd 'a man who saw a stranger refused them'
H L L H
a. d ti pdg(-d) kodu 19
'he gave a woman a banana'
I I \L^I
L L H LH
b. d ti sdn(-d) kodu 'he gave a stranger a banana'
I I VV1I
L L H L L H
(29) Object + a d v e r b (sdhld 'yesterday'):
a. d jie pdg(-d) sohld 'he saw a woman yesterday'
L L H L H
b. d jie sdn(-d) sohld 'he saw a stranger yesterday'
I I VV1 I
L L H L L H
Structure Preservation in Dagbani 247
I I
L L H L LH
W 1 II
b. d ti sdn(-d) d kddu 'he gave a woman your banana'
I I
L L H L L LH
tive H - L o w e r i n g , is s u m m a r i z e d in (31a).
(33) a. d jie wdb t-ttd-li —• o jie wdb ttid-li: 'did he see a big elephant?'
| | | \j/ I I I \|/ & -8* 'elephant')
db
L L H H L L H L
b. d jie zdb !
t4td-li —• d jie zdb !
tkd-li: 'did he see a big hair?'
L L H L H L L H L L
(34) a. d jie sdn(-d) kodu —• d jie sdn(-d) kodu: 'did he see the
| | \ | | | | | \ ' \ " ' ] | stranger's banana?'
L L H L L H L L H L L L
that m u s t follow H T S . 2
I I
HL H
b. /zab-gu/ 'hair' —• zdb-gu —• *zdb-gu
I I
HL H
In ( 3 5 b ) , if H l o w e r i n g applies first to this form, then H T S will incorrectly derive
!
2
(36) a. a (a) a b. O O O
U4"1 U4"l
H L L H L L
A s seen, the m i d d l e T B U m u s t b e optional, b e c a u s e H T S will also apply across
2
5. CONCLUSION
I V I V
H H H H H
After this u n l i n k e d H associates to the leftmost T B U , H T S applies lexically to
derive the H - H s e q u e n c e . W i t h a lexical application w e see that a H m a y m i g r a t e
three syllables to the right of its initial link, s p r e a d i n g o n c e lexically, a s e c o n d
t i m e by H T S , and a third time by H T S [recall ( 2 4 b ) ] . T h e lexical application is
x 2
II II v
H H H L H L
W h a t is clear is that if it is from lexical H T S , then this form c a n n o t then trigger
H T S ! postlexically, b e c a u s e of the floating L t o n e that b l o c k s H T S . Instead, it x
ACKNOWLEDGMENTS
Materials for this article were gathered in the field methods course given in the Spring
of 1988 at the University of Southern California. I am indebted to Mr. Abdul Saedu, who
served as informant, and to the members of that course for their participation in the discov-
ery of the tone system of Dagbani. I would like to thank Drs. W. A. A. Wilson and A.
Naden for sending me their work on Dagbani and related Gur languages and for their com-
ments and general help during our initial analysis of these materials. For previous work on
Dagbani tonology, see especially Wilson (1970). Finally, thanks to Ellen Kaisse and an
anonymous Academic Press referee for comments.
NOTES
JFor example, Harris (1987) argues that structure preservation is not respected in the
lexical phonology in Southeastern Bantu, while Kaisse (1990) and Rice (1990) give ex-
amples where structure preservation persists into the postlexical phonology. Given the con-
clusion I reach in this study, it may be that some of the claimed lexical/postlexical distinc-
tions are simply "tendencies" that are reflected in many, but not all, languages.
2
The verbal tonology has also been studied and analyzed and supports the conclusions
reached on the basis of the nominal tonology.
3
The following tone marks are written over the first (or only) V of each syllable: ' =
H(igh) tone,' = L(ow) tone, = HL (falling) tone; marks downstep tone.
A !
4
1 might note that the drop from H to H is considerably greater in interval than in other
!
languages I have studied. As we shall see, if only one TBU occurs with the H tone before !
is present but is subject to final vowel reduction. To underscore this point, consider the
corresponding genitive and pre-adjective realizations of jieb-gd 'crocodile' and wdb-gu
'elephant', whose isolation forms were originally cited in (3a,b):
W jieb-g1
yi-li 'crocodile's house'
wdb-gi •'yi-li 'elephant's house'
In (i) the vowel of the suffixes -gd and -gu is replaced by a very short, perhaps epenthetic
centralized vowel following the Igl of the suffix. In (ii), on the other hand, both this vowel
and the Igl are missing, again illustrating that a noun appears bare (without suffix) when
modified by an adjective.
9
This contrasts with the situation in other Gur languages such as Moore and Lama,
where Kenstowicz et al. (1988) have shown the need for a L tone genitive particle.
10
T h o u g h irrelevant to the analysis, it is possible that at this postlexical level the two
H's would have already conflated as a single H linked to the two syllables.
1 1
In Section 5 I shall consider an alternative whereby the H of the suffix comes in
unlinked, associates first to the noun stem, and then spreads rightward.
1 2
According to the common practice of filling a gap in the pattern, one might propose
that nouns such as /pag/ are actually underlyingly LH. Though not an insurmountable prob-
lem, there are at least two reasons for not doing this. First, there is the question of why a
H-tone suffix should cause a LH contour to become H, rather than simplifying to L [as in
fact happens elsewhere in the language by the rule of contour simplification in (22)]. Sec-
ond, there is the problem that this H does not surface when followed directly by a L
tone—cf. (16a).
13
I t cannot literally be from H T S ] , as conceived up to this point, since the L of kddu
would then be expected to delink, ultimately yielding *sdn kodu.
( 19b) could also apply to delete the free L tone before pause, since in that position it
1 4
also has no surface effect. On the other hand, free L's are not deleted before H tone, since
they are needed to condition downstep.
Cf. Wilson (1970:414): "A sequence HL other than V(C)CV becomes HH before L."
15
The effect of this rule is to guarantee that a HL falling tone will surface only if it is followed
by a H tone, a constraint that is found in a number of other tone languages, e.g., Kinande,
Luganda, etc. There is an (intonational) exception to this, however; see (34).
I f HTS] did apply across the free L tone in (15b) and (16d), for instance, we would
1 6
obtain the incorrect outputs *[san kodu] (from intermediate *sdn ko'du) and * [kparj kar-
li] (from intermediate *kpdjj kdr-'li), respectively.
1 7
After H T S applies to this form and others cited below, a H - H L sequence is obtained,
2
e.g., intermediate kddu in (24c). As transcribed, this fall is simplified by contour simplifi-
cation (22).
18
T h i s rule of L-tone Spreading (LTS) spreads rightwards onto a H TBU if the latter is
in turn also followed by a H TBU, i.e., L - H - H becomes L - L - H . The H - H sequence either
may be a single H that has spread by HTS j , or it may be a separate H feature from the next
morpheme. In either case the rule applies. (For discussion of the relevance of such ex-
Structure Preservation in Dagbani 253
amples to the feature geometry of tone within a parametric framework, see Hyman and
Pulleyblank, 1988.)
1 9
I n this example [and also in (29a)], only HTS j can apply, since after the initial spread-
ing of H, the intermediate output of 'banana' (ko'du) does not have a following L TBU For
H T S to apply. (Recall that this intermediate representation becomes kodu before pause.)
2
2 0
T h i s differs slightly from Wilson's (1970:414) statement of the rule ["Before the ?
[interrogative] marker a final H tone or H tone sequence is lowered to L back as far as the
last Downstep of the utterance (if any)"] because of the data in (32d).
2 1
(32b) also shows that when a form would normally take a glottal stop prepausally in
a declarative utterance (cf. Hyman, 1989), the glottal stop is not observed within the cor-
responding interrogative.
22
T h e r e is an alternative to H-Lowering whereby a boundary interrogative L delinks the
H('s) on the appropriate TBU(s). Assuming that the affected sequences of H's have all been
fused into a single H-tone feature at the CG level, the rule of H-Lowering simply targets
the last H of the domain. In this alternative, this H would have to link to single tonal node,
or else it is not clear how a boundary L can dislodge it from a sequence of TBUs. For this
reason, and also for the reason that we know we need intonational rules that modify lexical
tones (see, e.g., Hyman, 1990), I stick with the analysis given above.
2 3
This seems to be what Wilson's (1970:413) rule is intended to do, which changes a
[ + h ] [ - h] [ - h] sequence into [ + h ] [ + h ] [ + h ] , independent of whether the second [ - h]
is linked or not.
2 4
Since the discussion of SP centers around a single tonal constraint, what is needed is
a demonstration that more than one such constraint turns off at the same arbitrary point in
the postlexical phonology. Unfortunately, this is hard to demonstrate. It can be pointed out
that the constraint against LH contours is never violated in Dagbani. However, none of the
three rules that were said to follow H T S (L-absorption, interrogative H-Lowering, H -
2
!
REFERENCES
Goad, H. (1988). Tone in the Dagbani Noun Phrase. Unpublished manuscript, University
of Southern California, Los Angeles.
Harris, J. (1987). Non-structure-preserving rules in lexical phonology: Southeastern Bantu
harmony. Lingua 73, 2 5 5 - 2 9 2 .
Hyman, L. M. (1989). The phonology of final glottal stops. In Proceedings of the West
Coast Conference on Linguistics 18, 113 - 1 3 0 .
Hyman, L. M. (1990). Boundary tonology and the prosodic hierarchy. In The Phonology-
Syntax Connection (S. Inkelas and D. Zee, eds.), pp. 1 0 9 - 1 2 5 . CSLI Publications and
University of Chicago Press, Chicago.
Hyman, L. M., and Pulleyblank, D. (1988). On feature copying: parameters of tone rules.
In Language, Speech and Mind (L. M. Hyman and C. N. Li, eds.), pp. 3 8 - 4 8 . Rout-
ledge, London.
254 Larry M. Hyman
Kaisse, E. M. (1985). Connected Speech: The Interaction of Syntax and Phonology. Aca-
demic Press, Orlando.
Kaisse, E. M. (1990). Toward a typology of postlexical rules. In The Phono logy-Syntax
Connection (S. Inkelas and D. Zee, eds.), pp. 1 2 7 - 1 4 3 . CSLI Publications and Uni-
versity of Chicago Press, Chicago.
Kenstowicz, M., Nikiema, E., and Ourso, M. (1988). Tonal polarity in two Gur languages.
Studies in the Linguistic Sciences 18, 7 7 - 1 0 3 .
Kiparsky, P. (1984). On the lexical phonology of Icelandic. In Nordic Prosody III (C.-C.
Elert et al., eds.), pp. 135-162. University of Umea, Umea, Sweden.
Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook 2,
85-138.
Lieberman, M., and Pierrehumbert, J. (1984). Intonational invariance under changes in
pitch range and length. In Language Sound Structure (M. Aronoff and R. T. Oehrle,
eds.), pp. 1 5 7 - 2 3 3 . MIT Press, Cambridge, Mass.
Rialland, A. (1980). Marques de ponctuation et d'integration dans l'enonce en gurma. Bul-
letin de la Societe de Linguistique de Paris 7 5 , 4 1 5 - 4 3 2 .
Rice, K. D. (1990). Predicting rule domains in the phrasal phonology. In The Phonology-
Syntax Connection (S. Inkelas and D. Zee, eds.), pp. 2 8 9 - 3 1 2 .
Wilson, W. A. A. 1970. External tonal sandhi in Dagbani. African Language Studies 11,
405-416.
(POST) LEXICAL RULE APPLICATION
GREGORY K. IVERSON
Department of Linguistics
University of Wisconsin—Milwaukee
Milwaukee, Wisconsin 53201
1. INTRODUCTION
T h e t h e o r y of lexical p h o n o l o g y d e v e l o p e d b y K i p a r s k y ( 1 9 8 2 ) p r o p o s e s a fun-
d a m e n t a l distinction b e t w e e n LEXICAL and POSTLEXICAL r u l e s — t h e latter b u t n o t
the f o r m e r apply across the b o a r d , that is, w i t h o u t r e g a r d for derivational h i s t o r y
or m o r p h o l o g i c a l c o m p o s i t i o n ; are typically e x c e p t i o n l e s s ; a n d c o m m o n l y p r e s e n t
only n o n n e u t r a l i z i n g or " a l l o p h o n i c " effects. A major descriptive feature of the
theory as originally c o n c e i v e d is that a given p h o n o l o g i c a l rule h a s m e m b e r s h i p
in o n e or the other of t h e s e t w o c a t e g o r i e s , b u t n o t both. K i p a r s k y ( 1 9 8 5 ) a r g u e s
that in certain c a s e s , however, specifically nasal c o n s o n a n t assimilation in C a t a l a n
a n d o b s t r u e n t voice assimilation in R u s s i a n , a single rule m u s t b e a c c o r d e d post-
lexical as well as lexical status. T h i s relaxation of the t h e o r y is n e c e s s i t a t e d for
Catalan, h e m a i n t a i n s , by the a p p a r e n t o r d e r i n g p a r a d o x that in s o m e d e r i v a t i o n s
nasal assimilation m u s t b o t h p r e c e d e and follow another, p r e s u m a b l y lexical, rule.
In the p r e s e n t article it is s h o w n that the C a t a l a n o r d e r i n g p a r a d o x falls a s i d e u n d e r
m o r e g e n e r a l c o n c e p t i o n s of rule interaction a n d feature representation, a n d that
the t w o rules actually b o t h apply only within the postlexical d o m a i n . T h i s result
in turn h a s i m p o r t a n t i m p l i c a t i o n s in lexical p h o n o l o g y for i m p l e m e n t a t i o n of the
derived e n v i r o n m e n t constraint on p h o n o l o g i c a l p r o c e s s e s s u c h as palatalization
in K o r e a n , w h o s e effects are s o m e t i m e s derivationally restricted, s o m e t i m e s not.
D e s p i t e a p p e a r a n c e s to the contrary, rules like t h e s e m a y also apply within a single
c o m p o n e n t of the p h o n o l o g y if it is only structure-preserving applications of rules
255
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
256 Gregory K. Iverson
2. CATALAN
pan+s Default P l a c e F e a t u r e s
[b£rj bim pans]
(4) c c 1
I , Jo-
Root Root
Place
— sonorant
(6) a. [ + nasal] +continuant b . [+nasal] [ — sonorant]
Place Place
Labial Labial
[—distributed] [-(-distributed]
([mf]) ([mp])
(Post) Lexical Rule Application 261
Place
[+anterior]
3. KOREAN
(9) p t c k s (Lax)
ph t h gh k h (Aspirated)
p' t' c' k' s' (Tense)
Of these, the plosives /t t / c o n v e r t to t h e affricates / c c / w h e n in position before
h h
4. C O N T E X T - S E N S I T I V I T Y I N U N D E R S P E C I F I C A T I O N
j j
Coronal Coronal
[+high] [-high]
p h e m i c structures c o n t a i n i n g any of [t t t ' ] before [i], e.g., [madi] < / m a t i / ' k n o t ' ,
h
(16) C O N T E X T - S E N S I T I V E UNDERSPECIFICATION :
It t t 7
h
[ - h i g h ] before / i / , o t h e r w i s e unspecified (IT T T 7 )
n
Ic c c 7 :
h
unspecified (IT T T 7 ) before / i / , o t h e r w i s e [ + h i g h ]
h
/ss7: unspecified ( / S S 7 )
268 Gregory K. Iverson
be followed by [i].
It is clear, however, that the only reason [t t t ' ] in these e n v i r o n m e n t s are
h
plicity relative to the lexical representation of [c c c ' ] before [i] are only illusory,
h
as these are offset by the c o m p l e m e n t a r y listing of [ — high] for the plain stops in
the palatalizing e n v i r o n m e n t . If [t t t ' ] w e r e unspecified irrespective of context,
h
(17) C O N T R A S T I V E SPECIFICATION:
Ii t t 7
h
[-high]
/c c c ' /
h
[+high]
/ss7 unspecified
(18) C O N T E X T - F R E E UNDERSPECIFICATION :
/tt t7
h
unspecified
/cc c7 h
[+high]
Is s 7 unspecified
5. CONCLUSION
NOTES
That the nasal assimilates only as far as laminopalatal place when before (palatal) [A]
1
is puzzling and presumably requires a special restriction. Catalan does have phonemic
(Post) Lexical Rule Application 271
/ n / , but simple avoidance of merger with this segment cannot be the whole story since the
same nasal does merge with phonemic Iml in so[m] pocs 'they are few'.
2
Except that labial nasals do assimilate before labiodentals; see below.
3
With no Root node, the melodically empty C will be stray-erased at the end of the
derivation.
4
The syntactic conditions on syllabification in Catalan would seem to be fully express-
able in terms of parametric variation per the theory of syntax-phonology interaction de-
veloped in Kaisse (1985:186ff.).
5
The postlexical derivation of velar nasals makes unnecessary the special lexical treat-
ment of them Kiparsky ( 1 9 8 5 : 1 0 1 - 1 0 3 ) proposes, because the velar nasal under present
assumptions does not exist except at the postlexical level. Variation in the retention of
postnasal [k] (as with [p] and [t]) is then due generally to the syntactic conditions on syl-
labification, though under the same stylistic conditions the velar stop is somewhat more
likely to be retained than the labial or dental (Gonzales, 1989).
6
The blocking of rule (5b) in this case will follow from the elsewhere condition. This
predictable precedence of the more specific rule (5a) over (5b) assures that the labiodental
articulation specified for the fricative IfI percolates onto the nasal Iml in an Imfl cluster,
rather than the (default) bilabial articulation of Iml onto IfI.
7
Though not with a palatal (cf. note 1).
8
Korean palatalization has been described from various perspectives by Kim-Renaud
(1974), S.-G. Kim (1976), C.-W. Kim and Ahn (1983), Ahn (1985, 1988), Sohn (1987),
Iverson (1987), Iverson and Wheeler (1988), and Cho and Sells (1991).
9
Presumably It' I would undergo this process too, except that the only member of the
tense stop series which occurs finally in stems is the velar IV!I (Chung, 1980).
10
Palatalization occasionally results in homophony, e.g., [maji] derives from either
/ m a t + i / 'the eldest' or /maci/ 'hempen paper', [kac i] from either / k a t + i / 'together' or
h h
/ k a c i / 'value'.
h
1 1
Besides Is s 7 , the alveolar sonorants also evince allophonic palatalization between
([mujii] < /mun-f-i/ 'door (subj.)') as well as within morphemes ([huAAun] < /hullyun/
'magnificent'). Cho and Sells (1991) suggest that morpheme-internal instances of It t t 7 h
"palatalize," too—but without affricating—so that the alveolar stops in (1 la) would also
reflect a degree of superficial coarticulation with HI. But the effect here seems to be no
more than is universally instantiated in such contexts, and in any case is not nearly as
categorically salient as among the coronal sonorants and fricatives.
1 2
As pointed out to me by Sang-Cheol Ahn, there is some variation among speakers as
to the extent of palatalization with respect to tense / s 7 , i.e., /s'i/ seems to vary individually
between essentially unpalatalized ([s'i]) and sharply palatalized ([s'i]) articulations. Since
the other potentially palatalizable tense obstruent, / t 7 , happens not to occur in relevant
derived environments (cf. note 9), palatalization for some speakers appears to exclude the
tense obstruents in general.
1 3
A variant of the Cho and Sells (1991) two-rule approach is outlined by Kiparsky (this
volume) in which palatalization per se would always be allophonic, applying first in the
derived environments of the lexical phonology, then without morphological restriction in
the postlexical phonology. But before exiting the lexicon, the intermediate postalveolar
stops (It l \'I) produced by the rule in derived environments would merge with the pho-
h
nemic affricates (Ic c c 7 ) by a separate rule applying at the word level. The violations of
h
272 Gregory K. Iverson
rules, which also predict features based on properties of segments found elsewhere in the
morpheme (e.g., all the features of I si are redundant in initial triconsonantal clusters of
English even though Is/ would otherwise be specified [+continuant]). Here, however, there
is no interaction with derivation, since classical sequence redundancy rules apply strictly
within morphemes; and neither does the representational savings they provide have to be
undone by marking of the feature for blocking effect purposes in complementary environ-
ments (Itl need not be specified [-continuant] in any environment even if Isl in clusters is
also unspecified for that feature because there are no lexical exceptions to the cluster gen-
eralization). Cf. Ao (1991) for a possible case in which a sequence redundancy rule none-
theless does appear to apply after a rule of phonological derivation, though the effect there
is still inert, i.e., neither feeding nor bleeding.
2 1
In terms of the definition in note 15, supplying the feature [+high] to underspecified
ITI is neutralizing because [+high] coronals (Ic c c 7 , also underspecified for [continuant])
h
exist in the input to the rule. Supplying [+high] to Is s'i (specified as [+continuant]),
however, is not neutralizing because no [+high, +continuant] consonants exist in the input
to the rule. A similar characterization of feature redundancy is outlined in Steriade (1987).
T h i s account is based on Iverson (1989), in which laryngeal segments are configured
2 2
with specifications for laryngeal features and for [continuant], but without distinctive place
of articulation features.
2 3
The representation of Itl as unspecified in all environments has the consequence that
Korean morphemes with /. . . t i . . . / are no more complex (in fact, one feature less so) than
those with /. . . c i . . ./. Similarly, invariant radical underspecification of vowel quantity in
English does not distinguish the long vowels in paint and pain (both with /VV/), whereas
the context-sensitive approach outlined in (14) would specify the vowel as long in paint
but not in pain. In the absence of evidence showing Korean /. . . t i . . ./ to be more marked
than /. . . ci . . ./, or of any suggesting the vowel in paint is in some way more remarkable
than that in pain, this consequence seems quite correct.
2 4
This property also holds of underspecified segments for which rule application
is blocked by virtue of exceptional marking elsewhere in the word. For example, the
loanword in Hungarian biiro [byro:] 'bureau' contravenes the regular front/back vowel
harmony pattern of the language (Ringen, 1988); its exceptionality can be encoded by
assigning the specification [ — back] just to the word's leftmost vowel in underlying repre-
sentation. In regular cases, [ — back] is a property of either all the vowels in the morpheme
(/OrOm, [ — back]/ - » [orom] orom 'joy') or none of them (/vArOs/ —> [varos] varos 'city',
with [+back] throughout by default). In irregular biiro, however, the lexical attachment of
[ — back] just to the first vowel (or to the second in similarly disharmonic sofdr [sof0: r]
'chauffeur') causes the other vowel's backness value to be supplied by default rather than
harmony since, in consequence of the morpheme's exceptional lexical marking, the spread-
ing within it of [ — back] would be neutralizing and so inapplicable in this nonderived
context.
REFERENCES
Ahn, S.-C. (1985). The Interplay of Phonology and Morphology in Korean. Doctoral dis-
sertation, University of Illinois, Urbana.
Ahn, S.-C. (1988). Lexicality vs. postlexicality in Korean palatalization. In Linguistics in
274 Gregory K. Iverson
the Morning Calm 2 (The Linguistic Society of Korea, ed.), pp. 2 4 9 - 2 6 3 . Hanshin,
Seoul.
Anderson, S. R. (1974). The Organization of Phonology. Academic Press, New York.
Ao, B. (1991). Kikongo nasal harmony and context-sensitive underspecification. Linguistic
Inquiry 22, 193-196.
Booij, G., and Rubach, J. (1987). Postcyclic versus postlexical rules in lexical phonology.
Linguistic Inquiry 18, 1 - 4 4 .
Cho, Y.-M., and Sells, P. (1991). A Lexical Account of Phrasal Suffixes in Korean. Unpub-
lished manuscript, Stanford University, Stanford, Calif.
Chung, K. (1980). Neutralization in Korean: A Functional View. Doctoral dissertation,
University of Texas, Austin.
Clements, G. N., and Keyser, S. J. (1983). CV Phonology: A Generative Theory of the
Syllable. MIT Press, Cambridge, Mass.
de Chene, B. (1988). Japanese x-epenthesis and Inflectional Boundary Buffer-Segments.
Paper presented at the winter meeting of the Linguistic Society of America, New
Orleans.
Goldsmith, J. (1990). Autosegmental and Metrical Phonology. Blackwell, Oxford.
Gonzalez, M. (1989). Catalan Nasal Assimilation: Lexical or Postlexical? Unpublished
manuscript, University of Massachusetts, Amherst.
Gorecka, A. (1989). Are Front Vowels Coronal? Paper presented at the winter meeting of
the Linguistic Society of America, Washington, D.C.
Hayes, B. (1986). Inalterability in CV phonology. Language 62, 3 2 1 - 3 5 1 .
Hayes, B. (1989). Compensatory lengthening in moraic phonology. Linguistic Inquiry 20,
253-306.
Hualde, J. (1989). The strict cycle condition and noncyclic rules. Linguistic Inquiry 20,
675-680.
Hume, E. (1989). Blocking Effects in Korean Umlaut: Evidence for the Coronality of Front
Vowels. Paper presented at the winter meeting of the Linguistic Society of America,
Washington, D.C.
Ito, J. (1986). Syllable Theory in Prosodic Phonology. Doctoral dissertation, University of
Massachusetts, Amherst,
ltd, J. (1989). A prosodic theory of epenthesis. Natural Language & Linguistic Theory 7,
217-259.
Iverson, G. (1987). The Revised Alternation Condition in lexical phonology. Nordic Jour-
nal of Linguistics 10, 151 - 1 6 4 .
Iverson, G. (1989). On the category Supralaryngeal. Phonology 6, 2 8 5 - 3 0 3 .
Iverson, G , and Kim, K.-H. (1987). Underspecification and hierarchical feature represen-
tation in Korean consonantal phonology. In Papers From the Parasession on Autoseg-
mental and Metrical Phonology, pp. 1 8 2 - 1 9 8 . Chicago Linguistic Society, Chicago.
Iverson, G , and Wheeler, D. (1988). Blocking and the elsewhere condition. In Theoretical
Morphology: Approaches in Modern Linguistics (M. Hammond and M. Noonan,
eds.), pp. 3 2 5 - 3 3 8 . Academic Press, San Diego.
Kaisse, E. (1985). Connected Speech: The Interaction of Syntax and Phonology. Academic
Press, Orlando.
Kaisse, E. (1986). Locating Turkish devoicing. Proceedings of the West Coast Conference
on Formal Linguistics 5, 119-128.
(Post) Lexical Rule Application 275
Kim, C.-W, and S.-C. Ahn. (1983). Palatalization in Korean Revisited. Paper presented at
the winter meeting of the Linguistic Society of America, Minneapolis.
Kim, S.-G. (1976). Palatalization in Korean. Doctoral dissertation, University of Texas,
Austin.
Kim-Renaud, Y.-K. (1974). Korean Consonantal Phonology. Doctoral dissertation, Uni-
versity of Hawaii, Manoa.
Kiparsky, P. (1973). Abstractness, opacity, and global rules. In Three Dimensions of Lin-
guistic Theory (O. Fujimura, ed.), pp. 5 7 - 8 6 . TEC, Tokyo.
Kiparsky, P. (1982). Lexical morphology and phonology. In Linguistics in the Morning
Calm (I.-S. Yang, ed.), pp. 3 - 9 1 . Hanshin, Seoul.
Kiparsky, P. (1984). On the lexical phonology of Icelandic. Nordic Prosody III (C.-C. Elert,
ed.), pp. 1 3 5 - 1 6 4 . Almqvist and Wiksell, Stockholm.
Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook 2,
85-138.
Koutsoudas, A., Sanders, G , and Noll, C. (1974). The application of phonological rules.
Language 50, 1-28.
Macfarland, T., and Pierrehumbert, J. (1991). On ich-Laut, ach-Laut and structure preser-
vation. Phonology 8, 171-180.
Mascaro, J. (1976). Catalan Phonology and the Phonological Cycle. Doctoral dissertation,
Massachusetts Institute of Technology, Cambridge.
McCarthy, J. (1986). OCP effects: gemination and antigemination. Linguistic Inquiry 17,
207-264.
Myers, S. (1987). Vowel shortening in English. Natural Language & Linguistic Theory 5,
485-518.
Paradis, C , and Prunet, J.-F. (1989). On coronal transparency. Phonology 6, 3 1 7 - 3 4 8 .
Plapp, R. (1990). The Geometry of Russian Voice Assimilation. Paper presented at the Min-
nesota Conference on Language and Linguistics, Minneapolis.
Ringen, C. (1988). Transparency in Hungarian vowel harmony. Phonology 5, 3 2 7 - 3 4 2 .
Sagey, E. (1986). The Representation of Features and Relations in Non-linear Phonology.
Doctoral dissertation, Massachusetts Institute of Technology, Cambridge.
Sanders, G. (1990). On the analysis and implications of Maori verb alternations. Lingua
80, 149-196.
Sapir, J. D. (1965). A Grammar of Diola-Fogny (West African Language Monographs 3).
Cambridge University Press, London.
Sohn, H.-S. (1987). Underspecification in Korean Phonology. Doctoral dissertation, Uni-
versity of Illinois, Urbana.
Steriade, D. (1987). Redundant values. In Papers from the Parasession on Autosegmental
and Metrical Phonology, pp. 3 3 9 - 3 6 2 . Chicago Linguistics Society, Chicago.
Yip, M. (1987). English vowel epenthesis. Natural Language & Linguistic Theory 5,
463-484.
Yip, M. (1991). Coronals, consonant clusters and the coda condition. In The Special Status
of Coronals: Internal and External Evidence (C. Paradis and J.-F. Prunet, eds.),
pp. 6 1 - 7 8 . Academic Press, San Diego.
BLOCKING IN NONDERIVED ENVIRONMENTS
PAUL KIPARSKY
Department of Linguistics
Stanford University
Stanford, California 94305
1. T H E P R O B L E M
277
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
278 Paul Kiparsky
2. N D E B IS N O T S P E C I F I C T O C Y C L I C O R L E X I C A L R U L E S
(1) COALESCENCE:
ea, ed —> ee
da —> 66
oa —> oo
(3) ^-DELETION: a a a
t—• 0
2.2. N D E B in W o r d - L e v e l a n d Postlexical R u l e s
(11) t-^sl—i
deletion p r o c e s s triggers d e g e m i n a t i o n . 7
3. N D E B E F F E C T S E X P L A I N E D A W A Y
I a d o p t the following a s s u m p t i o n s .
(20) a. Underspecification (Kiparsky, 1982; A r c h a n g e l i and Pulleyblank,
1989).
b. Default rules m a y be o r d e r e d to apply either cyclically, at the w o r d
level, or postlexically (Pulleyblank, 1986; R i c e , 1988).
c. T h e optimal g r a m m a r is the simplest ( C h o m s k y and Halle, 1968).
d. S t r u c t u r e - c h a n g i n g rules are to b e d e c o m p o s e d into deletion (delink-
ing) plus structure-building (Poser, 1982; M a s c a r o 1987; C h o , 1990).
For p u r p o s e s of the a r g u m e n t , underspecification is u n d e r s t o o d in the strictest
sense, that is, w h a t Steriade ( 1 9 8 7 a ) calls RADICAL UNDERSPECIFICATION, includ-
ing the a s s u m p t i o n of strict binarity of feature specifications in u n d e r l y i n g lexical
representations. In e a c h e n v i r o n m e n t , w e can h a v e at m o s t [OFJ and [ a F ] , w h e r e
[ — a F ] is the value assigned by the m o s t specific rule ( l a n g u a g e - p a r t i c u l a r or uni-
versal) w h i c h is a p p l i c a b l e in that e n v i r o n m e n t . (That is, the relevant " e n v i r o n -
m e n t s " are defined by the rule system, including the m a r k e d n e s s rules of universal
g r a m m a r ) . T h e e s s e n c e of m y p r o p o s a l could be m a i n t a i n e d even u n d e r w e a k e r
versions of underspecification, however, as is s h o w n below.
A corollary of (20c) and (20d) is that rules will b e structure-building if possible
(i.e., unless positive e v i d e n c e requires positing a deletion rule). A corollary of
(20c) and (20a) is that u n d e r l y i n g representations are m i n i m a l l y specified.
M y principal thesis is:
(21) N D E B is the result of structure-building rules applying to underspecified
representations.
286 Paul Kiparsky
(22) a.
I +coronal
—> [ + c o n t i n u a n t ] / i
+obstruent_
b. [ + o b s t r u e n t ] —> [ - c o n t i n u a n t ]
Binarity can b e m a i n t a i n e d if e a c h e n v i r o n m e n t s h o w s at m o s t a t w o - w a y o p -
position. T h i s is clearly the c a s e for continuancy, as s h o w n in (26), w h e r e the
c o l u m n s represent, respectively, s e g m e n t s alternating b e t w e e n t a n d s, fixed t a n d y
fixed s.
(It is these palatalized but unaffricated stops that the analysis p r o p o s e d b y Iverson
and W h e e l e r neglects.)
finally.
(33) a. * [ . . . § a . . . ] , * [ . . . nu . . . ] , * [ . . . fo . . . ] , * [ . . . 9a.. .]
b. * [ . . . s i . . . ] , * [ . . . n i . . . ] , * [ . . . t i . . . ] , * [ . . . c i . . .]
(34) /t, t / h
/c, c /h
/ s , n, 1/
Before i [+ant] [Oant] [Oant]
Elsewhere [Pant] [ - ant] [Oant]
(35) t \ c 5 s §_
High - + - + - +
Anterior + - - - + -
D e l a y e d release — — 4- + — —
(36) [ + c o r o n a l ] —> [ + a n t e r i o r ]
forest-and field-nom. b e g r e e n - d e c l .
4. V A C U O U S L Y D E R I V E D E N V I R O N M E N T S C O U N T
AS UNDERIVED
4.1. Finnish
by the past tense suffix / (38a). Certain final vowels, including are truncated
before it ( 3 8 b ) . 1 4
4.2. I c e l a n d i c
4.3. Catalan
T h a reason Presuffixal Vowel L o w e r i n g can simply apply before any suffix is that,
unless the suffix is prestressed, the stem vowel will s i m p l y get r e d u c e d and the
effects of Vowel L o w e r i n g will b e u n d e t e c t a b l e .
M a s c a r o also a r g u e s for N D E B in a rule h e m o t i v a t e s for C a t a l a n w h i c h takes
o to u a n d e to Q } Let us refer to this rule as M i d Vowel R e d u c t i o n . Certain
1
5. S T R U C T U R E - C H A N G I N G R U L E S A R E N O T S U B J E C T T O N D E B
(47) v v c c v v v c c v
- N N I
a t o a t o
(49) v v c c v +c v v c c v +c
- N I I I
a t o +n a t o +n
- N i l
a t e a t e
- \l N I
a t e +n a t e +n
(52) a. c v c c v + c c v c v + c v
II I IN I I I I I
h a t u + t o m a +n a
b. c v c c v + c c v c
I I I I III
h a t u + t o m a
G e m i n a t e s followed by t a u t o m o r p h e m i c closed syllables [such as the first tt in
hottentotti; see (15)] m u s t then b e lexically p r e a s s o c i a t e d with the coda. C o n s e -
quently they will not u n d e r g o gradation. A s before, the possibility of such lexical
Blocking in Nonderived Environments 297
5.2. C h u m a s h S i b i l a n t H a r m o n y
A n a p p a r e n t d e r i v e d - e n v i r o n m e n t effect in a s t r u c t u r e - c h a n g i n g rule of C h u -
m a s h , a l a n g u a g e formerly s p o k e n a l o n g the central California coast, h a s b e e n
p r e s e n t e d as a p r o b l e m for the present p r o p o s a l by P o s e r ( 1 9 9 0 ) . M y reanalysis
d r a w s on his fuller earlier t r e a t m e n t (Poser, 1982), and on his source, A p p l e -
gate ( 1 9 7 2 ) .
In C h u m a s h , a p r o c e s s of Sibilant H a r m o n y m a k e s sibilants and affricates agree
in laminality ([ ± anterior] in Poser, 1982, [ ± distributed] in Poser, 1990, and here)
with the r i g h t m o s t sibilant or affricate in the word. T h e rule is s t r u c t u r e - c h a n g i n g
and assimilates u n d e r l y i n g apicals to l a m i n a l s (53) as well as u n d e r l y i n g l a m i n a l s
to a p i c a l s .
20
A n o t h e r rule of C h u m a s h , P r e - C o r o n a l L a m i n a l i z a t i o n ( P C L ) (56), m a k e s a
sibilant laminal i m m e d i a t e l y before a n o n s t r i d e n t c o r o n a l (t, /, n):
(62) c c c c
(63) c c
Coronal Coronal
[-Strident] [-Strident]
[ + Strident] [ +Strident]
[ +Distributed]
PCL —
— (blocked) S+is+Lu+SiSiN
Distributed — s+is+Lu + SiSiN
Spreading
Default wasti s+waTti+lok'in+us s+is+lu+sisin
O n this analysis, C h u m a s h is c o m p a t i b l e with m y c l a i m that N D E B occurs only
in n o n - s t r u c t u r e - c h a n g i n g r u l e s .
2 3
5.3. C h a m o r r o V o w e l L o w e r i n g
b. i, u e l s e w h e r e .
(69) a. hungan ' y e s ' , asut ' b l u e ' , listu ' q u i c k '
( [ + h i g h ] w h e r e [ — high] is e x p e c t e d by (66a))
b. neni ' b a b y ' , ispeyus ' m i r r o r ' , sitbesa ' b e e r '
([ — high] w h e r e [ + h i g h ] is e x p e c t e d b y (66b))
W e s e e m to h a v e arrived at a contradiction.
T h e p r o b l e m is only apparent. T h e height alternation is the result of t w o sepa-
rate rules, a structure-building l o w e r i n g rule (with [ + h i g h ] default) conditioned
b y syllable structure, and a s t r u c t u r e - c h a n g i n g raising rule c o n d i t i o n e d by stress.
Blocking in Nonderived Environments 301
(H)) [ [ - h i g h ] in closed s y l l a b l e s ] , . ,
[ - l o w ] -> < \ " \ (non-structure-changing)
[ [ + h i g h ] (elsewhere) J
(73) l\l Id
Stressed c l o s e d syllable [+high] [Ohigh]
Stressed o p e n syllable [Ohigh] [ — high]
U n s t r e s s e d syllable [Ohigh] —
6. N D E B E F F E C T S I N P R O S O D I C R U L E S
(75) * *
(a){(bc)(def))
Blocking in Nonderived Environments 303
(77) a = b
(80) a . * * * * *
(Mo) (non ga) (he) (la) -h (Monon ga) (he) (la)
* * * * *
(Ha) (cken) (sack) —> (Hacken) (sack)
304 Paul Kiparsky
6.2. Q u a n t i t y
A s Ito points out, such stems m u s t b e lexically specified as short. Lexically speci-
fied s u b m i n i m a l feet will not b e l e n g t h e n e d t h r o u g h the structure-building assign-
m e n t of the canonical b i m o r a i c foot. F o r instance, a s s u m i n g a m o r a i c representa-
tion of length, su w o u l d h a v e the u n d e r l y i n g representation (83),
(83) F
su
w h i c h e s c a p e s the m i n i m a l foot r e q u i r e m e n t in virtue of its lexically specified
metrical structure. A s t e m c o n f o r m i n g to t h e canonical pattern, such as suu ' n u m -
b e r ' , has n o metrical structure in the lexicon a n d gets assigned a full t w o - m o r a
foot b y the o r d i n a r y p r o s o d i c p a r s i n g p r o c e s s e s o p e r a t i n g o n / s u / . Finally, if
truncation and reduplication s u p e r i m p o s e their o w n foot patterns on the lexically
specified p r o s o d i c structure of the s t e m (on the principle that affix properties
s u p e r s e d e stem properties), they will effectively neutralize u n d e r l y i n g lexical
quantity.
6.3. D i s h a r m o n y
(84) e k o — + ~ e k o - + e k o
-A -A -A
-A +A-A
(86) te ' lo te ~ lo
— |
-A +A -A
(88) a. [karamelli + a]
b. [kara] [melli + d\
7. CONCLUSIONS
P r e v i o u s t r e a t m e n t s of N D E B h a v e not s u c c e e d e d in r e c o n c i l i n g e m p i r i c a l c o v -
e r a g e with theoretical adequacy. T h e R A C p r o v i d e s a descriptively fairly a c c u r a t e
c i r c u m s c r i p t i o n of t h e p h e n o m e n o n , e x c e p t that formulating t h e constraint as a
categorical prohibition of absolute neutralization is p r o b a b l y too strong. T h e R A C
is, however, clearly unsatisfactory as a p r i n c i p l e of g r a m m a r , b e c a u s e of t h e for-
m a l indefinability of the class of " n e u t r a l i z a t i o n r u l e s , " the undesirability of hav-
ing different p r i n c i p l e s for obligatory a n d optional rules, a n d the d u b i o u s status of
the c o n c e p t of " d e r i v e d e n v i r o n m e n t . " T h e S C C , on the other h a n d , is preferable
o n g e n e r a l theoretical g r o u n d s b u t s i m p l y fails t o m a t c h t h e facts in m a n y specific
instances. In this article I h a v e p r e s e n t e d a n e w interpretation w h i c h resolves this
d i l e m m a . It m a k e s essential u s e of underspecification and of d e c o m p o s i t i o n of
s t r u c t u r e - c h a n g i n g rules. Its m a i n a d v a n t a g e s are that it r e d u c e s the b l o c k i n g ef-
fects to i n d e p e n d e n t p r i n c i p l e s of g r a m m a r , predicts their restriction to structure-
building rules a n d to n o n v a c u o u s rule applications, recaptures the e m p i r i c a l
generalizations b e h i n d the original restriction of the alternation condition to
308 Paul Kiparsky
ACKNOWLEDGMENTS
I would like to thank Young-mee Cho for her helpful advice at several stages in the
preparation of this article. It has also benefited from the pointed and constructive comments
of Ellen Kaisse, John McCarthy, and an anonymous reviewer. I take responsibility for any
remaining errors.
NOTES
1
See also Giegerich (1988) for a different proposal to this effect.
2
Recently, Hammond (1991) has also argued for an acquisition-based account of SCC
effects, based on a "morphological WYSIWYG" principle. I hope to return to Hammond's
proposal on another occasion.
3
Consonant Gradation, as well as other cases which establish the same general point, is
discussed at greater length below.
4
Unless otherwise stated, I assume the analysis of Finnish word phonology presented in
Keyser and Kiparsky (1984), complemented by the morphological analysis of Kanerva
(1987). The reader is referred to those sources for the evidence supporting the rules and
underlying forms which figure in the discussion below. I should add that I know of no
reason to believe that these rules might belong to different levels of the lexical phonology,
or indeed, that Finnish even has more than one level.
5
The -h- is obligatory after a primary stress and after a long vowel, e.g., maa+han 'into
the ground', mi+hin 'into what', veesee+hen 'into the toilet' (words of the latter type,
with noncontracted unstressed long vowels, are loans, and they are often treated as con-
tracted in nonstandard speech, e.g. veesee+seen). Otherwise, - h- is deleted, except in ar-
chaic, poetic, or dialectal usage.
6
T h e deletion cannot be cyclic because it it is blocked by any following suffix, e.g.,
/vaatteCe+n/ —» vaatteen.
7
Detailed motivation for this analysis is presented in Keyser and Kiparsky (1984).
8
The reader points out that these data could be reconciled with cyclic application of
Consonant Gradation on the assumption that final consonants are cyclically extrametrical,
and that rule (8) is ordered before it. Even on these assumptions, Consonant Gradation
would have to apply at the word level as well (and extrametricality would have to assumed
be turned off there), because of such cases as (16) and (49). Therefore, this alternative does
not call into question the status of Consonant Gradation as a word-level rule subject
to NDEB.
9
T h e data in (18) represent the actual pronunciation. Postlexical gemination is normally
Blocking in Nonderived Environments 309
not shown in the spelling (menepds, itsekin). For the evidence that clitics like -kin are
syntactic elements and not attached lexically see Kanerva (1987).
1 0
On the Zee/Inkelas assumptions mentioned above, so is application across clitic
boundary. Moreover, on those assumptions both the fact that the ruki rule stopped applying
across clitic boundary and the fact that it stopped applying across word boundary in clas-
sical Sanskrit can be seen as consequences of a single change, namely, that the rule became
confined to the lexical phonology. This development represents a characteristic trajectory
of phonological rules (Zee, this volume).
1 1
As shown by the failure of sandhi gemination, and by forms like gen.sg. / n u k k e + n /
nuke+n (not /nukkeCe+n/ *nukkee+n) and part.sg. /nukke+ta/ nukke+a (not /nukke-
C e + t a / *nuket+ta), there is no protective "ghost consonant" after them, as in (16) and in
(50) below.
1 2
T h e analysis is similar to that of English palatalization, which applies both lexically
and postlexically, but with the lexical palatals undergoing (word-level?) spirantization, e.g.,
/edition/ [s] vs. /hit you/ [c] (Borowsky, 1986).
1 3
Similar examples can be given for other endings beginning with -/.
1 4
It is specifically the first vowel which is truncated, as shown by sequences of dissimi-
lar vowels, e.g., /tunte+i + vat/ tunsivat 'they knew', /vaati+eCe/ vaade 'demand'.
1 5
John McCarthy points out that even simple cases like /koti/ koti should constitute
vacuously derived environments for t —> s, since raising would apply vacuously to the final
vowel.
1 6
It is immaterial for the point whether Stress Assignment is done along the lines pro-
posed by Harris (1983, 1987) for Spanish, or following the somewhat different analysis
developed by Alsina.
1 7
Actually, this rule is formulated more generally as applying to the low vowels D, £ as
well, but in Mascaro's analysis these get raised to mid vowels in unstressed position any-
way by the second Vowel Reduction rule, which brooks no exceptions whatever.
1 8
A similar case of vacuously derived environments apparently counting as derived for
the SCC was noted by Steriade (1987b note 23) for Yakan (data from Behrens, 1973:25).
Here unstressed a is raised to / in derived stems; on a Mascarovian account, the latter
condition could be eliminated.
1 9
Note also that if Consonant Gradation is not deletion but lack of association, it auto-
matically follows that gradated consonants cannot undergo /-Deletion [rule (3)], in such
words as hatu + ton. As John McCarthy has pointed out to me, ordering /-Deletion before
Consonant Gradation would be problematic because the structural description of t-Deletion
crucially refers to a combined prosodic and melodic context, which is not available until
after Consonant Gradation, on the analysis suggested here.
2 0
Geminate sibilants are realized as s , s .
h h
2 1
Some cases which at first blush seem to be exceptions to PCL could, if the proposed
analysis is on the right track, be viewed as exceptionally undergoing the shared feature
convention in the lexical phonology in spite of being heteromorphemic: s+lu+skumu 'it
branches into four', s+netus 'he does it to him' (Applegate 1972:120).
2 2
O u r formulation supposes that [distributed] is dependent on [strident].
2 3
Applegate has many other examples of s ~ s alternations in wasti-type, words (1972:
347, 351, 352, 367, 375), but none that clearly involve vmy/w?-type words. The derivation
/ha p+xoslo? + s/ —> apxoslos 'you blow your nose' is given on p. 522, but no */xoslo/ is
310 Paul Kiparsky
actually cited (is / h a + s + n o x s / asnoxs 'nose' somehow related?). The analysis I have pro-
posed would actually predict that underlying /st/ does not undergo Sibilant Harmony, since
the multiple link established by the shared feature convention would block spreading.
2 4
Chung describes a pattern of optionality in the application of the high-mid alternation
in vowels with secondary stress, which is of great theoretical interest but not directly
relevant here.
25
K a g e r ' s version of the FEC furthermore covers cases like original(-ity), which were
problematic for the SCC account.
2 6
From this perspective, the distinction between strong and weak FEC effects is akin to
that between stress-neutral and stress-dominant affixes. It is no accident, then, that clitic-
triggered stress-assignment always displays strong FEC effects (Arabic, Manam, Greek),
and that weak FEC effects are associated with stress subordination.
2 7
A similar proposal was made by Bruce Hayes in a talk presented at a phonology work-
shop at the University of Massachusetts, Amherst, in 1983, of which I have seen only the
handout.
28
C o n t r a Harris (1989:358).
2 9
Adopting the notation of Kager (1989), I delimit feet by parentheses and extrametrical
elements by angled brackets.
3 0
With the no-crossing constraint blocking any spread across it.
3 1
Finnish linguists sometimes characterize such long words as "quasi-compounds."
There is no morphological support for this, in cases like (88) at least, but the intuition
nevertheless has a real basis, to which my proposal also does justice, in that compounds
similarly have at least two stress feet.
3 2
This would be the unmarked situation, though different harmonic processes could
potentially be restricted to different spans as well, a possibility which is apparently realized
in some Turkic languages.
REFERENCES
Hume, E. (1990). Front vowels, palatal consonants, and the rule of Umlaut in Korean.
Proceedings of the North Eastern Linguistic Society 21.
Inkelas, S. (1989). Prosodic Constituency in the Lexicon. Doctoral dissertation, Stanford
University, Stanford, Calif.
Ito, J. (1986). Syllable Theory in Prosodic Phonology. Doctoral dissertation, University of
Massachusetts, Amherst.
Ito, J. (1990). Prosodic minimality in Japanese. Syntax Research Center, Cowell College,
University of California, Santa Cruz, Technical Report, SRC-90-04.
Iverson, G. K., and Wheeler, D. W. (1988). Blocking and the elsewhere condition. In Theo-
retical Morphology (M. Hammond and M. Noonan, eds.), pp. 3 2 5 - 3 3 8 . Academic
Press, San Diego.
Kager, R. (1989). A Metrical Theory of Stress and Destressing in English and Dutch. Foris,
Dordrecht.
Kaisse, E. (1986). Locating Turkish devoicing. Proceedings of the West Coast Conference
on Formal Linguistics 5, 119 - 1 2 8 .
Kanerva, J. (1987). Morphological integrity and syntax: The evidence from Finnish pos-
sessive suffixes. Language 6 3 , 4 9 8 - 5 2 1 .
Kean, M.-L. (1974). The strict cycle in phonology. Linguistic Inquiry 5, 1 7 9 - 2 0 3 .
Keyser, S. J., and Kiparsky, P. (1984). Syllable structure in Finnish phonology. In Language
Sound Structure (M. Aronoff and R. T. Oehrle, eds.), pp. 7 - 3 1 . MIT Press, Cam-
bridge, Mass.
Kiparsky, P. (1973). Phonological representations. In Three Dimensions of Linguistic
Theory (Osamu Fujimura, ed.). TEC, Tokyo.
Kiparsky, P. (1982). Lexical phonology and morphology. In Linguistics in the Morning
Calm (Linguistic Society of Korea, ed.), pp. 3 - 9 1 . Hanshin, Seoul.
Kiparsky, P. (1984). On the lexical phonology of Icelandic. In Nordic Prosody III(C. Elert,
I. Johansson, and E. Strangert, eds.), pp. 135-164. Almqvist & Wiksell, Stockholm.
Kiparsky, P. (1985). Some consequences of lexical phonology. Phonology Yearbook 2,
82-138.
Kiparsky, P. (1992). Underspecification and Harmony Systems. Unpublished manuscript,
Stanford University, Stanford, Calif.
Lahiri, A., and Evers, V. (1989). Palatalization and coronality. In The special status
of Coronals: Internal and External Evidence (C. Paradis and J.-F. Prunet, eds.),
pp. 7 9 - 1 0 0 . Academic Press, San Diego.
Levergood, B. (1984). Rule governed vowel harmony and the strict cycle. Proceedings of
the North Eastern Linguistic Society 14, 2 7 5 - 2 9 3 .
Lombardi, L., and McCarthy, J. (1991). Prosodic circumscription in Choctaw morphology.
Phonology 8, 3 7 - 7 2 .
Macdonell, A. (1968). Vedic Grammar. Indological Book House, Varanasi.
Macfarland, T , and Pierrehumbert, J. (1991). On ich-Laut, ach-Laut and structure preser-
vation. Phonology 8, 171 - 1 8 0 .
Mascaro, J. (1976). Catalan Phonology. Doctoral dissertation, Massachusetts Institute of
Technology, Cambridge.
Mascaro, J. (1987). A Reduction and Spreading Theory of Voicing and Other Sound Effects.
Unpublished manuscript, Barcelona.
Blocking in Nonderived Environments 313
McCarthy, J. (1986). OCP effects: Gemination and antigemination. Linguistic Inquiry 17,
207-263.
Oresnik, J. (1977). Modern Icelandic u-umlaut from the descriptive point of view. Gripla
2, 151-182. Reprinted in Studies in the Phonology and Morphology of Modem Ice-
landic (J. Oresnik, ed., 1985). Buske, Hamburg.
Ottosson, K. (1988). Fragments of the Lexical Morphology and Phonology of Icelandic.
Unpublished manuscript, University of Maryland, College Park.
Poser, W. (1982). Phonological representations and action-at-a-distance. In The Structure
of Phonological Representations, part 2 (H. van der Hulst and N. Smith, eds.),
pp. 1 2 1 - 1 5 8 . Foris, Dordrecht.
Poser, W. (1986). Diyari stress, metrical structure assignment, and the nature of metrical
representation. Proceedings of the West Coast Conference on Formal Linguistics 5,
178-191.
Poser, W. (1989). The metrical foot in Diyari. Phonology 6, 1 1 7 - 1 4 8 .
Poser, W. (1990). Are Strict Cycle Effects Derivable? Unpublished manuscript, Stanford
University, Stanford, Calif.
Prince, A. (1985). Improving tree theory. Proceedings of the Berkeley Linguistic Society
11,471-490.
Pulleyblank, D. (1986). Tone in Lexical Phonology. Reidel, Dordrecht.
Ramsey, S. R. (1978). Accent and Morphology in Korean Dialects. The Society of Korean
Linguistics, Seoul.
Rice, K. (1988). Continuant voicing in Slave (Northern Athapaskan): The cyclic applica-
tion of default rules. In Theoretical Morphology (M. Hammond and M. Noonan, eds.),
pp. 3 7 1 - 3 8 8 . Academic Press, San Diego.
Rubach, J. (1984). Cyclic and Lexical Phonology. Foris, Dordrecht.
Selkirk, E. (1980). Prosodic domains in phonology: Sanskrit revisited. In Juncture (M.
Aronoff and M.-L. Kean, eds.), pp. 1 0 7 - 1 2 9 . Anma Libri, Saratoga, Calif.
Shaw, P. (1985). Lexical phonology. Phonology Yearbook!, 1 7 1 - 2 0 0 .
Steriade, D. (1982). Greek Prosodies and the Nature of Syllabification. Doctoral disserta-
tion, Massachusetts Institute of Technology, Cambridge.
Steriade, D. (1987a). Redundant values. Papers from the Regional Meeting of the Chicago
Linguistic Society 24, 3 3 9 - 3 6 2 .
Steriade, D. (1987b). Vowel Tiers and Geminate Blockage. Unpublished manuscript, Mas-
sachusetts Institute of Technology, Cambridge.
Steriade, D. (1988). Greek accent: a case for preserving structure. Linguistic Inquiry 19,
271-314.
Wackernagel, J. (1895). Altindische Grammatik, vol. 1. Vandenhoeck & Ruprecht,
Gottingen.
Whitney, W. D. (1887). Sanskrit Grammar. Harvard University Press, Cambridge, Mass.
Zwicky, A. (1986). The general case: Basic form versus default form. Proceedings of the
Berkeley Linguistics Society 12, 305 - 1 4 .
ARE STRICT CYCLE EFFECTS DERIVABLE?
WILLIAM J. POSER
Department of Linguistics
Stanford University
Stanford, California 94305
1. D E R I V E D E N V I R O N M E N T E F F E C T S
(1) underlying
nominative essive representation gloss
a. last lasina laTi 'glass'
b. koti kotina koti 'home'
c. vesi vetend veTE 'water'
d. kuusi kuusena kuusE 'fir'
315
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
316 William J. Poser
(2) F I N A L E-RAISING:
[]->[+high]/ #
[]-»[-high]
2. C H U M A S H S I B I L A N T H A R M O N Y
( 4 ) illustrates the fact that the third p e r s o n subject prefix surfaces as [s] w h e n n o
other sibilant follows. But w h e n the past tense suffix / w a s / is a d d e d as in ( 5 ) , Isl
b e c o m e s Isl.
Are Strict Cycle Effects Derivable? 317
3. C H U M A S H P R E - C O R O N A L L A M I N A L I Z A T I O N
Fusion is an operation which takes identical primary content nodes and fuses them pro-
vided that the nodes are non-distinct; i.e., both nodes do not dominate different secondary
nodes. We assume that fusion is headed in that the secondary features of the triggering
segment are maintained.
m a s h Sibilant H a r m o n y orders P C L b e t w e e n D i s t r i b u t e d D e l i n k i n g , w h i c h d e -
links n o n r i g h t m o s t specifications of the feature [dist], a n d D i s t r i b u t e d S p r e a d i n g ,
w h i c h s p r e a d s t h e r e m a i n i n g r i g h t m o s t specification of [dist] leftward.
(11) O R D E R OF RULES:
Distributed Delinking
Pre-Coronal Laminalization
Distributed Spreading
NOTES
1
Let me take this opportunity to correct an error in Poser (1982), where I glossed has-
xintilawas as 'his former Indian name'. Actually, this form is the adjective meaning 'his
former gentile' and has the given meaning only when combined with the noun masti
"name", xintila is a loan from Spanish gentil 'gentile, heathen'.
2
I n Poser (1982) I treated the distinction between Chumash Isl and III as a distinction
between [4-ant] and [ — ant]. Here I adopt the proposal of Lieber (1987:147) that it should
be characterized as a distinction between [ — dist] and [H-dist], that is, as a distinction be-
tween apical and laminal.
3
Lieber (1987) mistakenly credits herself with this proposal and attributes to Poser
(1982) the metrical view of which it is a critique.
4
A full formalization of this analysis requires a decision as to how to handle the trans-
parency of the nonstrident coronals, which fortunately does not appear to be relevant to the
points at issue here. For discussion of the transparency issue see Shaw (1991).
REFERENCES
Mascaro, J. (1976). Catalan Phonology and the Phonological Cycle. Doctoral dissertation,
Massachusetts Institute of Technology, Cambridge.
Poser, W. J. (1982). Phonological representation and action-at-a-distance. In The Structure
of Phonological Representations, part 2 (H. van der Hulst and N. Smith, eds.),
pp. 1 2 1 - 1 5 8 . Foris, Dordrecht.
Shaw, P. A. (1991). Consonant harmony systems: The special status of coronal harmony.
In Phonetics and Phonology, Vol. 2 (C. Paradis and J.-F. Prunet, eds.), pp. 1 2 5 - 1 5 7 .
Academic Press, San Diego.
THE CHRONOLOGY AND STATUS
OF ANGLIAN SMOOTHING
B. ELAN DRESHER
Department of Linguistics
University of Toronto
Toronto, Ontario, Canada M5S 1A1
1. I N T R O D U C T I O N
T h e O l d E n g l i s h s o u n d c h a n g e k n o w n as S m o o t h i n g h a s b e e n p r o b l e m a t i c in
t e r m s of b o t h its c h r o n o l o g y and status. Descriptively, S m o o t h i n g is a n a m e given
to the p r o c e s s w h e r e b y , in certain dialects, d i p h t h o n g s w e r e m o n o p h t h o n g i z e d
before the velar c o n s o n a n t s k, g, and x, w h e n t h e s e followed either directly or with
an i n t e r v e n i n g r or /. T h e o b s e r v e d c h a n g e s are listed in (1).
(1) cea ce
e *" ' "r\
1
+cons
=/ +son 4-obst
[ 4 - stress] _ - nasal_ +back
I b e g i n b y r e v i e w i n g s o m e p r o b l e m s in d e t e r m i n i n g the c h r o n o l o g y of S m o o t h -
ing. T h e s e p r o b l e m s are w o r t h c o n s i d e r i n g , even t h o u g h the c h r o n o l o g y of
S m o o t h i n g relative to o t h e r rules is in fact easily d e t e r m i n e d b y l o o k i n g at the
d o c u m e n t s . N e v e r t h e l e s s , m a n y traditional g r a m m a r i a n s did n o t believe w h a t the
texts a p p e a r to b e s h o w i n g , b e c a u s e the interaction of S m o o t h i n g with s o m e other
rules w a s not w h a t it should h a v e b e e n in t e r m s of their theory of l a n g u a g e c h a n g e ,
w h i c h itself w a s c o n n e c t e d to their t h e o r y of s y n c h r o n i c g r a m m a r . S o a study of
this c a s e b r i n g s o u t clearly the i n t e r d e p e n d e n c e b e t w e e n s y n c h r o n i c t h e o r y a n d
the interpretation of d i a c h r o n i c d e v e l o p m e n t s , as well as turning u p an interesting
p r o b l e m to w h i c h w e c a n apply c u r r e n t theories. L e t us turn, then, to the facts of
this case.
2. T H E C H R O N O L O G Y O F S M O O T H I N G
(2) ce cea
e eo * I C V
i io [+stress] [+back]
(3) a. b.
weorc dcegas Earlier forms
were — Smoothing
— dceagas Back Mutation
(4) a. b.
dcegas sprecan Earlier forms
dceagas spreocan Back Mutation
dcegas sprecan Smoothing
spreocan Analogy
Monophthongs Diphthongs
Nouns 67 5
W e a k verbs 15 0
S t r o n g verbs and derivatives 3 69
Anglian Smoothing 329
a. S t r o n g verbs
sprecan Original Form
— Smoothing
spreocan Back Mutation
b. W e a k verbs and n o u n s
cwcecade dcegas Original F o r m
cwceacade dceagas Back Mutation
cwcecade dcegas Smoothing
3. M O R P H O L O G I C A L C O N D I T I O N S
(8) a. BREAKING:
->
ce cea
e eo y
i Jo [+stress]
=/
[+stress]
Anglian Smoothing 331
b. RETRACTION:
+son
ce —» a I
+back
[+stress] L
(9) 3->0/V C £ 0
Morphology Phonology
Roots, Stems B r e a k , Retrct, B c k M u t E-stem
N o u n affixes / - M o n o , /-Mut, S m o o t h Level
Word
Verb A G R B r e a k , Retrct, B c k M u t
Level
4. R U L E S I N N O N L I N E A R P H O N O L O G Y
(20) a. RETRACTION:
I I I I
A I d • A I d
I \ I
[BACK] [BACK]
Anglian Smoothing 335
b. BREAKING:
I I /\ I
A I d A V / d
I \ i
[BACK] [BACK]
Similarly, ^ - R e s t o r a t i o n and B a c k M u t a t i o n exemplify the t w o styles of p r o p a g a -
tion of [ B A C K ] from a following vowel.
(21) a. #-RESTORATION:
f A t u
I I
f A t u
[BACK] \y
b. B A C K MUTATION:
[BACK]
[V] [V] [V]
I I / A
A V t u
f A t u - [V]
[BACK]
L e t us turn n o w [BACK]to the rules w h i c h apply only at the E - s t e m level of the lexical
p h o n o l o g y . W e o b s e r v e i m m e d i a t e l y that / - M o n o p h t h o n g i z a t i o n a n d /-Mutation
can n o w b e s u b s u m e d u n d e r a single generalization, namely, d e l i n k a specification
[ B A C K ] to the left of lil. T h i s will h a v e the effect of fronting a b a c k v o w e l a n d
voiding the s e c o n d part of a d i p h t h o n g . (Recall that the s e c o n d e l e m e n t of a diph-
t h o n g o c c u p i e s n o slot of its o w n [i.e., h a s n o m o r a i c v a l u e ] , so the loss of [ B A C K ]
c a u s e s it to m e l d into the front vowel.) W e will call this unified rule / - U m l a u t ; its
effects are illustrated in (22).
(22) /-UMLAUT:
a. /-Mutation: b. /-Monophthongization:
A
h A l d i sc A V r p i
N
[BACK] 0
N
[BACK] 0
336 B. Elan Dresher
N u m b e r of t o k e n s with
Form ce e a
mceht 18 3
dceg 41 6 1
asagas — 1
mcegan 5 1
*plcegian 2 1
ah — 44
/\ I s A x
s A V x
N [BACK]
[BACK]
T h i s d e l i n k i n g of a multiply-linked feature is a k i n d of O C P effect, in that it sim-
plifies the transition from a front to a b a c k e l e m e n t . T h i s is n o t an u n n a t u r a l
6
p r o c e s s , a n d so it is n o t n e c e s s a r y to s u p p o s e that S m o o t h i n g before a b a c k c o n -
sonant is p a r a d o x i c a l or p r o b l e m a t i c .
7
5. R U L E S O R C O N S T R A I N T S
[BACK]
6. L E X I C A L P H O N O L O G Y A N D D I A C H R O N Y
7. CONCLUSION
ACKNOWLEDGMENTS
I would like to thank Peter Avery, Alana Johns, Keren Rice, Tom Wilson, and an anony-
mous referee for helpful comments.
NOTES
For more on the language and texts, see Brunner (1965), Campbell (1959), Dresher
l
(1978), Hogg (1992), Kuhn (1965), Luick (1964), and Toon (1983).
340 B. Elan Dresher
2
N o t e that Old English texts typically spell the low diphthong [aea] as ea. There is little
doubt that this was purely an orthographic convention, and for ease of exposition, I indicate
it as cea.
3
See Kiparsky (1988) for a review of Neogrammarian and contemporary issues in pho-
nological change.
4
T h e diagram allows for cyclic levels, but I have not shown this; the Word Level espe-
cially may not be cyclic.
5
The representations in (20) and subsequently are schematic, and much structure is
omitted. Gussenhoven and van de Weijer (1990) consider the spreading feature to be
[+dorsal].
6
Strictly speaking, this is what McCarthy (1989) calls anti-spreading, which often works
together with the OCP to enforce cooccurrence restrictions.
7
An anonymous referee suggests that Smoothing may have begun in fast speech, where
the listener might be likely to interpret the second part of the diphthong as a mere formant
transition to the consonant. Translating this perception backward to phonemic representa-
tions gives the effect of monophthongization. It is noteworthy in this connection that Brun-
ner (1965) proposed that Smoothing is orthographic only, because the backness of the sec-
ond part of the diphthong was already conveyed by the following consonant. I see no
reason, however, to limit this effect to the orthography.
8
We thus return in a way to Luick's suggestion that Back Mutation was prevented before
velar consonants.
9
See Singh (1987), Paradis (1988), Goldsmith (1990: Ch. 6), and Myers (1991) for
recent proposals.
REFERENCES
ELLEN M. KAISSE
Department of Linguistics
University of Washington
Seattle, Washington 98195
1. INTRODUCTION
343
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
344 Ellen M. Kaisse
a. In the absence of counterevidence, assign the smallest number of strata as the do-
main of a rule.
b. In the absence of counterevidence, assign the highest possible stratum as the do-
main of a rule (where 'lowest' = stratum 1).
Principles la,b are of course not the only ones conceivable. Alternatively, one might think
of the following principles: (a) In the absence of counterevidence, assign the maximum
number of strata as the domain of a rule, (b) In the absence of counterevidence, assign the
lowest stratum as the domain of a rule.
as well, or, instead, forsaking the older, earlier strata. T h e first type of descent
should result in " s a n d w i c h " t y p e rule application, w h e r e a n e w l y e x t e n d e d rule
can apply b o t h before a n d after a n o t h e r rule. T h i s is a characteristic part of the
" s c a t t e r e d " rule p h e n o m e n o n . T h e s e c o n d t y p e of d e s c e n t will result in a rule
w h i c h w e d e s c r i b e as h a v i n g " m o v e d " or " r e o r d e r e d , " for instance the C y p r i o t
c a s e of Section 3.
2. E X T E N S I O N D O W N W A R D O F M O R P H O L O G I C A L D O M A I N :
KASKA ^-DELETION
(3) I. stem
S-Deletion
II. conjunct prefixes
III. conjunct prefixes
IV. conjunct prefixes
V. disjunct prefixes
sition in the utterance is not available within the lexicon. W e see the logical c o n -
nection b e t w e e n generalization a n d d o w n w a r d m o v e m e n t in later sections.
3. R U L E R E O R D E R I N G A N D G E N E R A L I Z A T I O N IN
CYPRIOT GREEK
(5)
FRICATIVE + STOP
lexical postlexical
t e 6 n 1 r t 0 d n 1 r
vc — + - — + + + +
cnt - 4- - + 4- - - +
son - - - 4- + + _ _ - + 4- 4-
lat - 4- - - - - - 4- -
nas + - - - - - + - -
(12) R E D U N D A N C Y RULES:
a. 4-son —> + v c
f.
b
- r -lat -son j
—» — v c
- s o n —> -cont!
J
L~ n a s
i —» 4-cont
c. 4-nas —> —cont -son
4-vc
d. + l a t —> - c o n t
-lat
e. —» 4-cont
4- s o n
Rule Reordering and Generalization 351
3 . 2 . T h e O b s t r u e n t i z a t i o n of lyl in C y p r i o t
Informally,
(13) y^WIC
(15) O B S T R U E N T I Z A T I O N O F y:
~— c o n s
— obst +cons
+cont —> +obst
+hi — cont
_-bk _ vc
(17) /spiti + a/
spitya Glide Formation
spitkya O b s t r u e n t i z a t i o n of y
spinel C o n t i n u a n t Dissimilation
Obstruentization
Postlexicon: Dissimilation
Rule Reordering and Generalization 353
3 . 3 . M o t i v a t i o n for C o n t i n u a n t D i s s i m i l a t i o n in t h e Postlexical C o m p o n e n t
(23) a. U N G E N E R A L I Z E D FORM:
-fobst +obst
- voice — voice —> + c o n t —cont
acont Pcont
b. G E N E R A L I Z E D FORM:
a c o n t p c o n t —» + c o n t - c o n t
Instead, w e say that the rule has always h a d the " g e n e r a l " form of (23b). T h e
class of s e g m e n t s affected b y the rule b e c o m e s larger u n d e r generalization not
b e c a u s e the rule has c h a n g e d , but b e c a u s e of the characteristics of the stratum it
has m o v e d to. O u r a r g u m e n t follows the path laid d o w n in K i p a r s k y (1985). Ki-
p a r s k y a r g u e s that R u s s i a n Voicing A s s i m i l a t i o n applies m o r e b r o a d l y in its post-
lexical applications than its lexical o n e s b e c a u s e values for voice b e c o m e assign-
able to a n d s p r e a d a b l e from nondistinctively voiced sonorants. Similarly, w e say
that C y p r i o t C o n t i n u a n t D i s s i m i l a t i o n affects a n d is triggered b y the n o n d i s t i n c -
tively c o n t i n u a n t s e g m e n t s w h e n it m o v e s to a later stratum. T h e innovation in m y
a p p r o a c h is to r e c o g n i z e this b r o a d e n i n g as a p r i m a r y m e c h a n i s m of rule gener-
alization. It m a y well b e that not all rule generalizations can b e r e d u c e d to this
t y p e — r u l e generalization b y the actual r e m o v a l of features from rules m a y yet b e
Rule Reordering and Generalization 355
4. A S C A T T E R E D R U L E IN S W I S S G E R M A N
(25) a. UNGENERALIZED:
V + son
-high -» [+low] / +cor
+back - nasal
- long — lateral
b. GENERALIZED:
V +cor
-high - » [ + low] / - nasal
+back — lateral
- long_
(26) r 1 n t s
Nasal - - + 0 0
Lateral - + - 0 0
Rule Reordering and Generalization 357
5. L I Q U I D D E L E T I O N I N S A M O T H R A K I G R E E K
(29) 4-son
4-cor - > 0 / V .
-nas
(30) r 1 n
Nasal — — +
Lateral — 4- —
(31) r 1 n
Nasal - +
Lateral — +
(32) 1 r
Lateral 4- —
Rule Reordering and Generalization 359
6. CONCLUSION
NOTES
The other case of rule scattering in the literature known to me is Bley-Vroman's (1975)
1
treatment of Old Norse. In that language, umlaut of vowels in extra-heavy syllables applies
before a vowel syncope rule. The generalized umlaut of vowels in any type of syllable
applies after syncope. The treatment of this phenomenon in terms of rule movement is not
obvious to me but might emerge from a close study of the form syllabification takes at
different strata in Old Norse or of syllabification's ordering with respect to umlaut. The
Quebecois French case discussed at the end of this section also involves loss of a closed
syllable requirement.
2
In Sekani transcriptions, a vowel unmarked for tone is high, HI is an allomorph of the
lsg prefix used in the perfective of verbs having a zero or -h- classifier ('elf'). The grave
accent preceding the conjunct prefix indicates a floating tone, realized as Low in Sekani
and as High in Kaska.
3
1 am grateful to Doug Pulleyblank for directing me to this example. An anonymous
referee suggests another interpretation of Dumas's facts, namely that the vowel lengthening
rule has become optionally cyclic, so that it can optionally lengthen vowels that precede
voiced fricatives that are word final prior to suffixation.
4
The spirant partner of lil is / 8 / , not Isl. And indeed Isl acts asymmetrically with respect
to other continuants. It cannot despirantize itself and causes despirantization of any frica-
tive which precedes it, in violation of the continuancy template. Despirantization before
Isl is a very old rule, one found in Attic Greek. It is therefore unclear whether phenomena
involving Isl should be used as evidence in discussions of Continuant Dissimilation, and I
will avoid such examples.
5
Actually, the phonemic status of voiced stops is a more vexed question than my simple
statement indicates, and Modern Greek scholars have argued over it at length. Voiced stops
do appear as the result of voicing spread from a preceding nasal, and when that nasal is
optionally lost, an apparently unconditioned voiced stop can appear on the surface. Voiced
stops also appear in loanwords, though an etymologically incorrect nasal occasionally ap-
pears along with them. I cannot hope to resolve the issue here, but the reader may be
362 Ellen M. Kaisse
assured that the distribution of voiced stops is defective in Modern Greek and that credible
analyses where these segments are lacking abound. Moreover, the evidence for the derived
character of voiced stops is particularly weighty in Cypriot, where they are invariably pre-
ceded by nasals.
6
An anonymous reviewer reminds me that Idl does not spirantize after /l/, though / b /
and Igl do. The complication may result from the fact that III and Idl are homorganic; the
linking convention may prevent spirantization of stops sharing a place node with an adja-
cent consonant.
7
Data in this and the following sections are taken from the excellent work of Newton
(1972a,b), except where otherwise noted. My analysis of some rules differs. I discuss the
spread of consonantality in Kaisse (1992).
8
Pat Shaw points out to me that the emergence of y from this rule as a voiceless conso-
nant indicates that the rule applies before the redundant values for voicing of glides and
stops have been assigned. Obstruentization of y is therefore likely to be a lexical rule in
Cypriot, and the value of voicing can be omitted from the actual structural change of the
rule. The placement of Obstruentization in the lexicon is consistent with my claim that
Continuant Dissimilation, which follows Obstruentization, is postlexical. It also suggests
weakly that not only did Continuant Dissimilation move down, but that Obstruentization
moved up. However, the movement of Obstruentization into the lexical component could
of course have been independent of its reordering with Continuant Dissimilation. As we
have said, the normal course of development for any new rule is for it to move into the
lexicon.
9
There are still a few controversial cases where rule insertion appears the best analysis,
the most celebrated being Lachmann's Law.
10
N a s a l s never seem to participate in the continuancy template; in fact, fricatives are
disfavored after presumably [ - continuant] nasals, since the normal historical development
of aspirated and voiced stops to fricatives did not occur after nasals. I conclude that the
value for continuancy for nasals is never specified within the phonology of Greek.
The Gulf War prevented my planned visit to Cyprus in January 1991. I apologize to
1 1
REFERENCES
DRAGA ZEC
Department of Modern Languages and Linguistics
Cornell University
Ithaca, New York 14853
1. INTRODUCTION
365
Phonetics and Phonology, Volume 4 Copyright © 1993 by Academic Press, Inc.
Studies in Lexical Phonology All rights of reproduction in any form reserved.
366 Draga Zee
(1) Level 1
Cyclic
Level 2
Level 3 Phonological word
Postcyclic
Level 4 Phonological phrase
con, however, the p h o n o l o g i c a l w o r d can only a p p e a r in its smaller size, that is,
without clitics.
2. A C C E N T R E T R A C T I O N A S A T O N E - S P R E A D I N G R U L E
c. L o n g rising: raazlika
IV I
LH L
d. L o n g falling: raadnica
H L
In (7), the pitch p r o p e r t i e s of e a c h of the four accents are r e p r e s e n t e d in t e r m s
of H i g h a n d L o w t o n e s . A s noted in L e h i s t e a n d Ivic ( 1 9 8 6 ) , pitch p r o m i n e n c e is
found on the only syllable b e a r i n g a falling accent a n d on the a c c e n t e d and the
p o s t a c c e n t u a l syllables of the rising accent. T h i s is c a p t u r e d in (7) in t e r m s of the
distribution of the H i g h tones: a falling a c c e n t is associated with a H i g h tone
linked to a single syllable, and a rising accent with a H i g h linked to t w o c o n s e c u -
tive syllables. T h e leftmost syllable linked to a H i g h tone bears stress, w h i c h is
m a r k e d with an asterisk.
T h e r e p r e s e n t a t i o n in (7) is s u g g e s t i v e in several respects. If w e factor out vowel
length, the four accents can be r e d u c e d to j u s t t w o — t h e falling H ( L ) and the rising
( L ) H H ( L ) t y p e . M o r e o v e r , the t w o a c c e n t t y p e s can b e k e p t distinct even if the
L o w tones are left out of the structure: the falling accents can simply b e c h a r a c -
terized as singly linked H i g h s , and the rising accents as d o u b l y linked H i g h s .
(8) a. Short rising: marama
V
H
b. Short falling: kucica
I
H
c. L o n g rising: raazlika
V
H
d. L o n g falling: raadnica
i
H
370 Draga Zee
(9) SPREADING: u ji
\ l
H
H
c. L o n g rising: raazlika
I
H
c. L o n g falling: raadnica
b. neraadnik 'nonworker'
N
H
2.2. O l d S t o k a v i a n D i a l e c t s
d. L o n g falling: raadnica
i v
H L
372 Draga Zee
marama
(14) a. Short falling
kucica
b. Short falling
raazlika
c. Short falling
H
c. L o n g falling: raadnica
(15) F O R M S W I T H LEXICAL T O N E :
a. tone ' s i n k - 3 S g P r e s ' (imperfective)
I
H
b. potone ' s i n k - 3 S g P r e s ' (perfective)
(16) L E X I C A L L Y T O N E L E S S FORMS:
a. tonu ' s i n k - 3 S g A o r i s t ' (imperfective)
H
b. potonu 4
s i n k - 3 S g A o r i s t ' (perfective)
H
Rule Domains and Phonological Change 373
3. P O S T C Y C L I C A P P L I C A T I O N S O F S P R E A D I N G
apply either within the lexical or within the postlexical version of the p h o n o l o g i c a l
w o r d . T h e p h o n o l o g i c a l w o r d in its postlexical size is referred to here, informally,
as the clitic g r o u p ; this t e r m is taken to refer to a d o m a i n size rather than to an
actual p r o s o d i c constituent.
Before t u r n i n g to the p o s t c y c l i c d o m a i n s of S p r e a d i n g , I will digress briefly to
c o m m e n t on t h e p r o s o d i c p r o p e r t i e s of S e r b o - C r o a t i a n w o r d s . F o l l o w i n g Inkelas
( 1 9 8 9 ) , I classify lexical i t e m s in S e r b o - C r o a t i a n into those that are prosodically
salient and t h o s e that lack p r o s o d i c salience. Prosodically salient forms will m a p
into p h o n o l o g i c a l w o r d s . F o r m s that lack p r o s o d i c salience are all clitics in S e r b o -
Croatian; they include certain p e r s o n a l p r o n o u n s , prepositions, conjunctions, and
particles. S i n c e S e r b o - C r o a t i a n clitics s u b c a t e g o r i z e for the p h o n o l o g i c a l word,
they will b e i n c o r p o r a t e d into this particular p r o s o d i c constituent, f o r m i n g an en-
larged d o m a i n .
I first e x a m i n e the p o s t c y c l i c d o m a i n s of S p r e a d i n g for the three N e o - S t o k a v i a n
dialects to b e d i s c u s s e d h e r e a n d then c o m p a r e t h e m with those of Initial H i g h
Insertion. It will b e s h o w n that the d o m a i n s of the t w o rules c o i n c i d e in e a c h of
the dialects to b e e x a m i n e d below. T h e d o m a i n s of Initial H i g h Insertion will in
fact play an i m p o r t a n t role in d e t e r m i n i n g the d o m a i n s of S p r e a d i n g for e a c h of
the dialects.
3.1. D o m a i n s of Spreading
the p h o n o l o g i c a l p h r a s e .
In N S 1 , a word-initial H i g h s p r e a d s o n t o the p r e c e d i n g clitic, as in (18).
H H
N N
H H
see(lsg) house(Acc)
entered Aux in house
T see a house.
'He got into the house.'
T o n e s p r e a d i n g o n t o a proclitic is a h i g h l y general p h e n o m e n o n in this dialect.
In o r d e r to illustrate this, w e g i v e a few m o r e e x a m p l e s (taken from V u k o v i c ,
1940). In (21a), t o n e s p r e a d s o n t o a p r e p o s i t i o n from a p r e c e d i n g adjective;
and in (21b), a H i g h linked to the initial syllable of a verb spreads to the
complementizer.
N N
H H
b . da vidiim 'that (I) see'
N
H
M o r e o v e r , in N S 1 w e also e n c o u n t e r applications of S p r e a d i n g within the p h o -
nological p h r a s e . E x a m p l e s in (22) are t a k e n from V u k o v i c ( 1 9 4 0 ) , and t h o s e in
(23) from R u z i c i c ( 1 9 2 7 ) .
(22) a. dvabrata 'two brothers'
N
b . tries ijedna kuca 'thirty one houses'
N
C. cetiri litre 'four liters'
N
H
'my brother (Vocative)'
d. moj brate
N
H
e. sve Ijeto 'all summer'
N
H
'all night'
f. svu nooc
N
H
N
g. svu dugu nooc 'all (the) long night'
H H
376 Draga Zee
\J
H
b . dvije kuce 'two houses'
N
H
C. tri knjige 'three books'
N
H
d. pet dinaaraa 'five dinars'
N
H
It should b e p o i n t e d out that S p r e a d i n g will not apply b e t w e e n j u s t any t w o w o r d s
in a p h o n o l o g i c a l p h r a s e . In all e x a m p l e s that h a v e b e e n reported, the target of
S p r e a d i n g is either a n u m e r a l or a nonclitic p r o n o u n . T h e s e forms receive tone
optionally in the dialect d e s c r i b e d b y V u k o v i c and R u z i c i c as well as in other
dialects d i s c u s s e d in this article. E x a m p l e s in (24), taken from V u k o v i c , illustrate
this point. In (24a) the p r o n o u n sav air is toneless, a n d in (24b) it is associated
4
N
H
b . svi su . . . 'all are.
N
H
d. deseet koosaa 'ten scathes'
H H
ispred kuce 4
in front of (the) h o u s e '
H
In this dialect, therefore, w e e n c o u n t e r n o alternation of the sort illustrated in
(20), as s h o w n b y the following e x a m p l e s .
(26) a. victim kucu
I I
H H
see( 1 sg) house(Acc)
T see a house.'
b. usao je u kucu
\l I
H H
entered Aux in house
'He got into the house.'
Finally, N S 3 p o s s e s s e s alternant f o r m s like t h o s e listed in ( 2 7 ) - ( 2 8 ) (taken
from N i k o l i c , 1970, a n d M o s k o v l j e v i c , 1 9 2 7 - 2 8 ) , w h i c h strongly suggests that
S p r e a d i n g applies b o t h within the p h o n o l o g i c a l w o r d a n d within the clitic g r o u p .
1 3
I
H
b . u banju 'to (the) spa'
N
H
(28) a. u Sapcu 'in Sabac'
I
H
b . u Sapcu 'in Sabac'
N
H
Variation that c h a r a c t e r i z e s N S 3 will b e attributed to the optionality of rule appli-
cation: in this dialect S p r e a d i n g applies obligatorily within the p h o n o l o g i c a l w o r d ,
but optionally within the clitic g r o u p . 1 4
378 Draga Zee
(29)
NS1 NS2 NS3
3 . 2 . C o m p a r i s o n w i t h Initial H i g h I n s e r t i o n
H
If the proclitic is disyllabic, t o n e is assigned to its first syllable. This follows
from the w a y the rule is f o r m u l a t e d in (17): t o n e is assigned to the leftmost m o r a
of the relevant d o m a i n .
(31) N S 1: iza graada ' b e h i n d (the) t o w n '
H
In N S 2 , however, the rule ignores the clitic and operates on the word-ini-
tial m o r a .
Rule Domains and Phonological Change 379
I I \ \
H H H H
see(lsg) city (Acc) went Aux to city
T see a city' 'He went to the city.'
In N S 3 , w e e n c o u n t e r alternant f o r m s of the following kind, w h i c h I attrib-
ute to optional rule application in the lexical, a l t h o u g h not in the postlexical,
domain. 1 5
H H
(36)
NS1 NS2 NS3
Lexical No Yes Yes(Opt)
phonological word
(phonological word)
Postlexical Yes No Yes
phonological word
(clitic g r o u p )
3 . 3 . I n t e r a c t i o n of S p r e a d i n g a n d Initial H i g h I n s e r t i o n
3.3.1. INNS3
I
H
b . u pasnjaaku 'in (the) pasture'
N
H
T h e alternation s h o w n in (37) follows from the optionality of rule application:
Initial H i g h Insertion is optional within the lexical p h o n o l o g i c a l word. M o r e o v e r ,
w e also find three-way alternations as in (38) (again from N i k o l i c , 1970); (38c) is
d u e to the optionality of S p r e a d i n g within the postlexical p h o n o l o g i c a l w o r d (i.e.,
the clitic g r o u p ) .
H
b . iz salaasa 'from (the) field'
N
H
C. u salaasu 'in (the) field'
I
H
(39) iz salaasa
H
N H
H
salaasa salaasu
Lexical phonological word: salaasa
Spreading
salaasa salaasu
I I
Initial H i g h I j
H H
H H
Spreading
iz salaasa
H
Initial H i g h iz salaasa
I
I
Rule Domains and Phonological Change 381
3.3.2. INNSI
N
H
This difference is p r e d i c t e d u n d e r the analysis p r o p o s e d h e r e . In N S 1 , Initial H i g h
Insertion applies only within the postlexical p h o n o l o g i c a l w o r d , that is, within the
clitic g r o u p . In (40a), this rule will apply to the entire prepositional p h r a s e , w h i c h
forms a postlexical p h o n o l o g i c a l word, a n d will link the H i g h to the proclitic za;
and in (40b) it will apply only to the form nooc, w h i c h by itself constitutes a
postlexical p h o n o l o g i c a l word. T h e toneless form svu, w h i c h is an a p p r o p r i a t e
target for S p r e a d i n g , u n d e r g o e s this rule within the p h o n o l o g i c a l p h r a s e , that is,
at the point at w h i c h it shares c o n s t i t u e n c y with nooc. T h e s e steps are s h o w n
explicitly in the following derivation.
I N
H H
Spreading
Initial H i g h za nooc nooc
i i
j i
H H
382 Draga Zee
H H
Spreading svu nooc
4. C Y C L I C A P P L I C A T I O N O F S P R E A D I N G
(42)
H
Rule Domains and Phonological Change 383
(43) a. a b. a
I A
I I
H H
4.1.1. T O N E LINKING R U L E
H H H
384 Draga Zee
H H H
H H H
malin-e
PI. Nom. jelen-i kolen-a
I
H
H malin-a H
Gen. jelen-a kolen-a
H H H
Dat. jelen-ima malin-ama kolen-ima
H H H
Acc. kolen-a
jelen-e malin-e
I
H
Instr. H
jelen-ima H
malin-ama kolen-ima
H H H
Loc. jelen-ima malin-ama kolen-ima
H H H
H I H
H
Rule Domains and Phonological Change 385
H H H
Loc. junaak-u gitaar-i naceel-u
I I I
H H H
H H H
Gen. junaak-a gitaar-a naceel-a
H H H
Dat. junaac-ima gitaar-ama naceel-ima
H H H
Acc. junaak-e gitaar-e naceel-a
I I I
H H H
Instr. junaac-ima gitaar-ama naceel-ima
H H H
Loc. junaac-ima gitaar-ama naceel-ima
H H H
T h i s rule will link a H i g h tone to the final syllable of the stem if this syllable is
light, as in (44). If the stem-final syllable is heavy, the tonal rule will fail to apply
on the stem c y c l e ; but it will reapply on the next h i g h e r cycle, assigning t o n e to
the suffix, as in (45). T h e relevant derivations are given in (48).
386 Draga Zee
(48)
C Y C L E 1: [junaak] [jelen]
H H
Tone Linking — [jelen]
H
C Y C L E 2: [[junaak] a] [[jelen] a]
H I
H
Tone Linking [[junaak] a]
T h e tonal rule operates only o n those forms w h i c h are associated with lexical
tone. In contrast, toneless f o r m s receive a H i g h t o n e on their initial syllable, b y
virtue of Initial H i g h Insertion (discussed in Section 2.3), as s h o w n in the deriva-
tions of t w o toneless m a s c u l i n e f o r m s , oblaak ' c l o u d ' and dever ' b r o t h e r - i n - l a w ' .
(49)
CYCLIC:
C Y C L E 1: [oblaak] [dever]
Tone Linking — —
C Y C L E 2: [oblaak] a] [dever] a]
Tone Linking — —
POSTCYCLIC: Initial H i g h [[oblaak] a] [[dever] a]
H H
H H
H H
(52) a. jad-a 4
woe(gen.sg.)'
b. dvoor-a 'court(gen.sg.)'
4.1.2. M E T A T O N Y
(54) junaak t
hero(nom.sg.)'
(55) [junaak] a
C Y C L E 2: [[ junaak]a] [ [ junaak]ask]
H H
i
Yer Delinking: \[junaak]k] [ [ junaak]ask]
I I
H H
Metatony: [\junaak\k] [[ junaak]ask]
\ 4 \ 4
H H
other rules
4.2. T o n e L i n k i n g a n d S y l l a b l e S t r u c t u r e
(59) fx s
I
H
390 Draga Zee
(60) ix]
H
T h i s rule will b e able to o p e r a t e only if the domain-final m o r a is also the h e a d
m o r a , that is, if it p o s s e s s e s an s label. If the r i g h t m o s t m o r a h a p p e n s to b e the
n o n h e a d m o r a , p o s s e s s i n g a w label, the rule will fail to apply, since the m o r a
targeted b y the rule will not b e able to receive tone. T h u s , b y assigning t o n e to
only a subset of m o r a s , w e c a n derive the fact that the H i g h that u n d e r g o e s T o n e
L i n k i n g can b e linked to a light but not to a h e a v y s y l l a b l e . M o r e o v e r , the con-
23
(61) a. a a b. *a a C. *a a
I I A I A I
|x s |JL S H'S M-w m ' M^s
MW
\ l X
J
H H H
(62)
H H H
b . jelen ( n o m . s g . ) , jelen-a (gen.sg.) versus jeleen-a (gen.pl.) ' d e e r '
l I l
H H H
c. bubreg ( n o m . s g . ) , bubreg-a (gen.sg.) versus bubreeg-a (gen.pl.) ' k i d n e y '
H H H
(63)
t
H
T h e delinked H i g h tone will then a s s u m e its shifted position by virtue of M e -
tatony, the relinking rule p r o p o s e d in the previous section. In (65) are given the
derivations of the genitive plural forms jeleena and naarooda.
(65) CYCLE i: lone Linking jelen naarod
H
CYCLE 2.- jelen]a naarod]a \
Gem GcnP
I I
H H
naarod]
OUTPUT:
H
jeleena H
"I naarooda
H I
H
T h i s analysis, however, d o e s not extend to the N e o - S t o k a v i a n dialects, w h i c h
exhibit a different pattern of the genitive shift: t o . . . shifts to a p r e c e d i n g light
syllable, but not to a p r e c e d i n g ! avy syllable. T h u s , -rms in (66) retract their
H i g h in the genitive plural, w h i l e i ' se in (67) fail to d o s o .
; 2 6
H H H
b. malin-a (nom.sg.) versus maliin-a (gen.pl.) ' r a s p b e r r y '
I I
H H
c. kolen-o (nom.sg.) versus koleen-a (gen.pl.) ' k n e e '
H H
d. republi-a (nom.sg.) versus republiik-a (gen.pl.) ' r e p u b l i c '
H H
e. kategorij-a (nom.sg.) versus kategoriij-a (gen.pl.) ' c a t e g o r y '
H H
(67)
a. naarod (nom.sg.), naarod-a (gen.sg.) versus naarood-a (gen.pl.) ' p e o p l e '
I I I
H H H
Rule Domains and Phonological Change 393
H H H
c. ucioonic-a ( n o m . s g . ) versus uciooniic-a (gen.pl.) ' c l a s s r o o m '
H H
d. organizaacij-a ( n o m . s g . ) versus organizaaciij-a (gen.pl.) 'organization'
H H
N
H
CYCLE 2 : jelen } a naarod 1 a
N
H
GenPl Delinking(NS)
jelen ] a
OUTPUT: H naarooda
jeleena
H
394 Draga Zee
5. S P R E A D I N G A N D T H E S T R O N G D O M A I N H Y P O T H E S I S
(70)
NS1 NS2 NS3
(71)
NS1 NS2 NS3
In N S 1 , o n t h e o t h e r h a n d , the p h o n o l o g i c a l w o r d is c r e a t e d postlexically, w h i c h
results in the set of d o m a i n s listed in (73).
6. T H E R O U T E O F C H A N G E
Postlexical p h o n o l o g i c a l word / /
phonciugical phrase /
(77)
I N
H H
well me(dat) come
b . svi da ste mi [ dobro J [ dosli ] 'welcome all of you'
I \ lN
H H H
all Comp Aux me(dat) well come
(80) a. [biiloje]. 'there was .
H
was Aux
b. [pa je biilo ] . . . 'and there was . . . '
N
H
and Aux was
N
H
b . pred rat 'before (the) war'
ACKNOWLEDGMENTS
I am grateful to Sharon Inkelas, Ellen Kaisse, Paul Kiparsky, and an anonymous referee
for most helpful comments on earlier versions of this article.
NOTES
1
Serbo-Croatian is divided into three major dialects, the Stokavian, the Cakavian, and
the Kajkavian. I cover the accentual situation in the Stokavian dialects, with the exception
of those characterized by only one accent. The subdivisions within the Stokavian group of
dialects adopted here are those proposed in Ivic (1958, 1985).
2
1 assume that rule domains are provided by morphological constituents at the cyclic
levels, and by prosodic constituents at the postcyclic levels.
3
Here I follow Inkelas (1989), where clitics are analyzed as prosodically dependent
elements which subcategorize for prosodic domains. Serbo-Croatian clitics, both proclitics
and enclitics, subcategorize for the domain of the phonological word.
4
T h i s proposal is made in Zee (1988), to account for the applications of several post-
cyclic rules in Bulgarian. Kanerva (1989) adopts this perspective in his analysis of clitics
in Chichewa. Section 5 of the present article provides further arguments for this view of
the phonological word. This position is originally due to Selkirk (1986), who assumes that
clitics are included into the phonological word.
5
According to King (1973), a new rule is necessarily added at the end of a grammar,
that is, as the latest in the set of ordered rules (for other views in the debate on rule addition
see Halle, 1962; King, 1969; Kiparsky, 1965; as well as Kaisse, this volume). However,
while King focuses on the ordering of a newly added rule, my focus is on the new rule's
domain.
6
T h e situation in the Old Stokavian is somewhat more complex. While the more con-
servative Old Stokavian dialects are indeed characterized only by the falling accents (as in
the regional variety described in Stevanovic, 1940), the less conservative Old Stokavian
dialects may also possess the rising accents. According to Ivic (1985), the crucial difference
Rule Domains and Phonological Change 401
between the Old Stokavian and the Neo-Stokavian dialects is in the distribution of the pitch
accents, regardless of the number of accents they may possess. The less conservative Old
Stokavian dialects are discussed in Section 6.
7
Vowel length is marked throughout the article as gemination. This is the only departure
from the standard orthography.
8
Syllable weight in Serbo-Croatian is determined by vowel length: a syllable with a
short vowel is light, and a syllable containing a long vowel is heavy. Consonants do not
contribute to syllable weight. Furthermore, Serbo-Croatian does not have diphthongs; two
unlike adjacent vowels belong to different syllables and are separated by a hiatus (Lehiste
and Ivic, 1967).
9
T h a t the prefix ne- is not associated with a High tone if it escapes the effect of Spread-
ing is shown by the pair of forms pusaac 'smoker' and nepusaac 'nonsmoker', both having
lexical tone on their final syllables; the lexical High then spreads to the preceding syllable,
which in pusaac is also the initial syllable. The prefix ne- in nepusaac remains toneless.
I n Inkelas and Zee (1988), lexical tones are prelinked in the underlying form. In Zee
1 0
(1988), lexical tone is assigned by a cyclic tonal rule. This article adopts the latter analysis,
which is presented in Section 4.
1 1
In order to make sure that rule (17) applies only to toneless forms, we may invoke the
OCP in its passive version (McCarthy, 1986), which will prevent the application of (17) to
a form already linked to a High tone. Generally, Serbo-Croatian words may be associated
with at most one High tone.
N S 1 is represented by what is known as the Vukovian norm, described in Danicic
1 2
(1925); this norm is based on the east Herzegovian dialect (see Vukovic, 1940, and Ruzicic,
1927, for descriptions of two of its regional varieties). NS2 corresponds to the eastern
standard, spoken predominantly in Belgrade (Miletic, 1952). NS3 is spoken in the regions
of Srem and Macva (see the descriptions of regional varieties in Moskovljevic 1927-28
and Nikolic 1953-54, 1966, 1970).
1 3
1 have found no convincing evidence that Spreading applies within the phonological
phrase in this dialect. Nikolic (1970:51) lists a single case of this type, sto hektaara '(one)
hundred nectars' in which the High on the first syllable of hektara spreads onto sto.
1 4
I f Spreading applies lexically, it will not reapply in the postlexical component. The
failure of Spreading to apply more than once can be attributed to inalterability (Hayes,
1986): given the statement of the rule in (9), only a singly linked High will undergo
Spreading.
1 5
Note that in NS3 Initial High Insertion is optional lexically, but obligatory postlexi-
cally. In other words, a phonological word has to be assigned tone within the phonological
word domain. It was suggested by an anonymous referee that Initial High Insertion is a
metrical rather than a tonal rule (see Halle, 1971) since the High assigned by this rule
coincides with stress, and that it is metrical structure rather than tone that is obligatorily
assigned within the phonological word. However, the High assigned by Initial High Inser-
tion may undergo Spreading, as shown in Section 3.3, and in this case it will not coincide
with stress. In fact, a metrical analysis would have to posit a stress shift in this case that
would replicate Spreading. This demonstrates that a tonal analysis is to be preferred over a
metrical one.
1 have found no cases of the application of Initial High Insertion within the domain of
1 6
the phonological phrase. This is to be expected, I believe, since Initial High Insertion ap-
402 Draga Zee
plies only within toneless domains (see note 11), and a phonological phrase can never be
toneless by virtue of the fact that it has to contain at least one content word, which will
have received tone by this point.
1 7
Here I assume the moraic theory of syllable structure with a direct representation of
moras, as in Hyman (1984), Hayes (1989b), Ito (1989), and Zee (1988), among others. I
also adopt the labeling convention of marking the leftmost mora as s(trong), and the re-
maining moras as w(eak); the ^-labeled mora is interpreted here as the head of the syllable
it belongs to (for details see Zee, 1988).
1 8
The forms in (44) and (45), as well as all other forms in this and the following sub-
sections, are shared by the two dialect groups. In other words, we ignore the effect of
Spreading.
1 9
A formal version of the rule is provided in Section 4.2.
2 0
T h e situation is actually somewhat more complex. Just like stems, affixes may either
be toneless or have tone of their own. The suffix -ov is obviously toneless. If a toneless
stem combines with an affix with its own tone, the derived form will be associated with
tone. For example, the derived form oblacic 'cloud (dim.)' has a High on its final syllable;
this High comes from the suffix -ic, which is affiliated with tone in its underlying form.
When a toneless simplex stem is combined with a toneless suffix, the derived form remains
toneless.
2 1
It is worth noting that, at the earlier stages of the language, before the fall of yers, the
nominative ending had exactly this form.
2 2
Metatony is triggered by any process that delinks a High tone. A High may be de-
linked either by rule (see the case discussed in Section 4.3), or by virtue of vowel loss. This
latter case could be motivated as follows: once the vowel is delinked from its mora, the
mora in its turn delinks from the higher structure, which then leads to the delinking of tone.
Tone delinking caused by yer deletion is only one such case; other cases are found in the
verbal paradigm where a vowel other than yer is deleted when followed by another vowel
(Zee, 1988).
2 3
Zee (1988) posits two cyclic, that is, lexical levels, and associates Tone Linking with
Level 2. The failure of this rule to apply at Level 1 is attributed to the prosodic setup of this
level. Several phonological phenomena receive a unified account under the assumption that
Level 1 includes only moras while Level 2 includes both moras and syllables. Since con-
straint (59) is operative throughout the cyclic component, it will preclude the linking of
tone to a mora that does not bear the s marking, and such moras are not available before
Level 2. Absence of relevant structure at some point in the grammar is the standard type of
interference with the strict domain hypothesis (cf. Kiparsky, 1984).
2 4
T h i s type of accent shift is found only in the Stokavian dialects (Leskien, 1914:
2 2 1 - 2 2 2 ) . The Cakavian dialects, which in many ways preserve an older accentual situ-
ation than that in the Stokavian, lack entirely the accent shift in the genitive plural form.
Moreover, only the genitive plural forms taking the -a desinence are subject to this type of
accent shift; no accent shift is exhibited by forms taking the -i desinence.
2 5
This ending has another peculiar property: it lengthens the vowel of the immediately
preceding syllable. The lengthening process, however, appears not to be responsible for the
genitive accent, as the discussion below will make evident. This lengthening process is
ignored in the derivations given below.
Rule Domains and Phonological Change 403
1 focus here only on those forms which do not contain yer vowels in the stem. Forms
2 6
with yers, like komaarac (nom.sg.), komaarca (gen.sg.), komaraaca (gen.pl.) 'mosquito'
exhibit a shift in the genitive plural which I believe is of a different sort. In this class of
forms, tone is linked to the desinence in all case forms other than the genitive plural. Thus
tone is delinked here from the genitive plural desinence rather than from the stem-final
syllable. While I do not have an account for this set of forms, I will assume that the pattern
they follow is distinct from that discussed here.
2 7
We will have to assume that yer delinking operates before Spreading, in order to
account for the fact that this process is not sensitive to syllable weight.
2 8
Evidence for the postlexical version of the phonological word in NS2 comes from
postlexical applications of Initial High Insertion. One class of forms, indefinite pronouns,
is lexically marked as exceptional to Initial High Insertion and can thus undergo this rule
only postlexically; as a result, the High is assigned to a proclitic. Thus the negated prepo-
sitional phrase ni sa kim 'not with anyone' has a High tone on the negative particle ni,
which is a proclitic, rather than on the pronominal stem kim. Note that ni gets no tone in ni
sa Petrom, since Petrom possesses a High tone on its initial syllable.
1 have no evidence for the application of this rule within any of the levels larger than
2 9
the phonological phrase; however, it is reasonable to assume that the rule has bled itself, so
to speak, beyond the phonological phrase level.
3 0
This situation is not restricted to Serbo-Croatian. Hyman (1990) discusses several
tonal rules of the domain limit type which apply postlexically. Two such rules in Kinande,
for example, operate on the domains of the phonological and the intonational phrase, but
not on any of the smaller domains.
3 1
Kaisse (this volume) argues that a relatively new rule may be applying within larger
(most likely, postlexical) domains while still failing to descend to the lexical domains; this
most likely captures the progression of Spreading in the Neo-Stokavian dialects. However,
while I agree with Kaisse that a relatively new rule may be applying in the larger but not in
the smaller domains, I also believe that this kind of situation may become stable only in
the case of domain limit rules. Initial High Insertion, for example, is a very old rule with
deep Indo-European roots. As for the Old Stokavian version of Spreading, it is not clear
whether it dates as far back as its Neo-Stokavian counterpart; if it does, then its failure to
descend to smaller domains is due to its domain limit properties.
REFERENCES
Belie, A. (1956). Osnovi istorije srpskohrvatskog jezika, vol. 1, Fonetika. Naucna knjiga,
Beograd.
Booij, G., and Rubach, J. (1987). Postcyclic versus postlexical rules in lexical phonology.
Linguistic Inquiry 18, 1 - 4 4 .
Browne, W. E., and McCawley, J. (1965). Srpskohrvatski akcenat. Zbornik za filologiju i
lingvistiku 8, 1 4 7 - 1 5 1 .
Cupic, D. (1977). Govor Bjelopavlica. Srpski dijalektoloski zbornik 23, 1-226.
404 Draga Zee
A C
407
408 Language Index
Manam Serbo-Croatian
free element condition, 302 genitive plural delinking, 3 9 1 - 3 9 4
stress and clitics, 101-103 high tone spreading, 14, 365, 3 6 8 - 4 0 0
Margi, tone assignment and suffixation, 91 initial high insertion, 373, 3 7 8 - 3 8 1
Mende, interaction of mutation and reduplica- metatony, 3 8 7 - 3 9 0
tion, 48, 53 Neo-Stokavian dialects, 365, 3 6 8 - 3 7 1 ,
373-400
Old Stokavian dialects, 368, 3 7 1 - 3 7 3 , 382,
N 394, 3 9 7 - 4 0 0
postcyclic lexical level, 100
Navajo tone linking, 3 8 3 - 3 9 0
functional element order, 157 Shanghai, phonological domains, 105
Shi, apparent evidence for interactionism, 137
morphological structure, 1 4 8 - 1 4 9
Slave
Nganhcara, pronominal clitics, 3 8 - 3 9
morphological structure, 1 4 6 - 1 5 1 , 153-161
Norse, Old, scattered rule, 361
phonological domains, 9, 153, 1 6 2 - 1 6 6
Spanish
interaction of la/el allomorphy and stress,
P
4 - 5 , 69
interaction of stress, e raising, and adverb for-
Polish mation, 6
comparative suffix allomorphy, 26, 42 spirantization, 351, 362
imperative suffix allomorphy, 2 6 - 2 7 Sundanese, nasal harmony and plural infixation,
non-isomorphic phonological and morpho- 48, 5 4 - 5 5
logical structure, 6 - 7
postcyclic lexical level, 100
T
Sanskrit W
compounds, 83
infixation and ruki rule, 48, 54 Warlpiri, stress, 106
ruki rule, 2 8 4 - 2 8 5
Vedic, 2 8 4 - 2 8 5
Sekani Y
morphological structure, 1 4 8 - 1 4 9
s deletion, 3 4 6 - 3 4 7 Yawelmani Yokuts, vowel harmony, 272
subject prefix allomorphy, 361 Yidiny reduplication, 3 0 - 3 2 , 42
tone, 361 Yoruba, vowel harmony, 305
SUBJECT INDEX
A 9 4 - 9 6 , 9 8 - 1 0 3 , 2 1 9 , 331
and nonderived environment blocking, 278
Absolute neutralization, see Phonological rule, postcyclic level, 200, 3 6 6 - 3 6 7 , 3 7 3 - 3 7 4
neutralizing precyclic rule application, 1 4 1 - 1 4 2
Across-the-board application, see Postlexical as property of affix or level, 88, 1 0 2 - 1 0 3 ,
rule, characteristics of 200, 201
Allophonic rule, see Phonological rule, of rule application, 5, 16, 7 5 - 1 0 4 , 112,
allophonic 1 1 5 - 1 1 , 2 8 0 - 2 8 2 , 382, 3 8 5 - 3 9 3
Allomorphy, 4 - 5 , 2 5 - 2 7 , 4 8 - 5 2 , 69, 126-127,
1 3 3 - 1 3 5 , 1 3 7 - 1 3 8 , 361
Analogical change, 3 2 7 - 3 2 8 , 329 D
Anaphoric island, see Lexical integrity
Derived environment condition
blocking in non-derived environment, 9,
277-307
B
as characteristic of lexical rules, 16, 255, 267,
Bound root, see Morphological constituent 270
Boundary strength, 1 7 5 - 1 7 9 as characteristic of structure-preserving rules,
Bracketing, morphological 255,263-265,267
erasure, 81, 173, 188 explanation via strict cycle condition, 315,
paradoxes, 3 4 - 3 7 , 40, 105 319
visibility to phonology, 2 - 3 , 16, 29, 77, 81, explanation via underspecification, 319
88, 90, 173, 180 phonological versus morphological, 9, 320
in vacuously derived environment, 2 9 1 - 2 9 5
Domain, of rule
C and historical change, 1 4 - 1 5 , 3 4 4 - 3 4 8 ,
353-361,365-366
Clitic, 3 7 - 4 0 , 8 6 - 8 7 , 101, 367, 396 noncyclic, see Cyclicity
Clitic group, see Domain, prosodic, clitic group predictability of, see Strong domain
Constraint, persistent, 326, 338, 339, see also hypothesis
specific constraint as property of individual rule, 15, 255, 258.
Cyclicity 262, 265
versus noncyclic rule application, 8 9 - 9 0 , word-internal, see Level, word
411
412 Subject Index
V o l u m e 1: R h y t h m a n d M e t e r
edited by Paul Kiparsky and Gilbert Y o u m a n s
V o l u m e 2 : T h e Special S t a t u s of C o r o n a l s : Internal a n d E x t e r n a l E v i d e n c e
edited by C a r o l e Paradis and J e a n - F r a n c o i s Prunet
V o l u m e 3 : C u r r e n t I s s u e s in A S L P h o n o l o g y
edited by Geoffrey R. C o u l t e r
V o l u m e 4 : S t u d i e s in L e x i c a l P h o n o l o g y
edited by Sharon H a r g u s and Ellen M . Kaisse