Job Name: -- /337544t

BLS 34, No 1 2008. DOI: http://dx.doi.org/10.3765/bls.v34i1.3564

(published by the Berkeley Linguistics Society and the Linguistic Society of America)

The Morphology-Phonology Connection

SHARON INKELAS
University of California, Berkeley

0. Introduction
This paper addresses several general issues in the connection between
morphology and phonology, where morphology is understood to involve
generalizations about form and meaning that relate words to one another within a
language, and phonology is understood to involve generalizations about the sound
patterns in that language. Morphology and phonology intersect insofar as the
statement of morphological generalizations includes information about sound
patterns (realizational morphology), and insofar as the statement of phonological
generalizations includes information about morphology (morphologically
conditioned phonology). This intersection is extensive, blurring the distinction
between morphology and phonology in many situations. The recent literature
features three approaches which focus squarely on the morphology-phonology
interface: Cophonology Theory (Orgun 1996, Inkelas et al. 1997, Inkelas 1998,
Anttila 2002, Inkelas and Zoll 2007), Stratal Optimality Theory (Kiparsky 2000;
2003b; a), and Indexed Constraint Theory (McCarthy and Prince 1995, Pater 2000,
Itô and Mester 1999, Alderete 2001, and Smith 1997). This paper argues that
Cophonology Theory succeeds best of the three in capturing three generalizations
that unify morphologically conditioned phonology and realizational morphology:

SUBSTANCE: Morphologically conditioned phonology and realizational

morphology involve the same operations
SCOPE: Morphologically conditioned phonology and realizational
morphology have identical scope of application within a word
LAYERING: Morphologically conditioned phonology and realizational
morphology are identical in their interactions in complex words

Sections 1 and 2 introduce examples of morphologically conditioned phonology

and realizational morphology, and Section 3 introduces the theories being
compared. SUBSTANCE, SCOPE and LAYERING are discussed in sections 4-6.

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

1. Morphologically conditioned phonology

Morphologically conditioned phonology is the situation in which a particular
phonological pattern is imposed on a proper subset of morphological
constructions (affix, reduplication, compounding) and thus is not fully general in
the lexical phonology of the language. We will see three examples here.
In Mam, suffixes partition into two classes (Willard 2004, based on England
1983). ‘Dominant’ affixes cause long root vowels to shorten (1a); ‘Recessive’
suffixes preserve root vowel length (1b). Dominant vs. recessive status is not
predictable; it must be learned individually for each affix.

(1) a. Dominant suffix: shortens long root vowel

facilitative OLLFK·- OLFK·LFK·LLQ ‘break/breakable’
resultant MXXV- MXVE
HHQ ‘burn/burned place’
locative
directional MDDZ- MDZQD[ ‘go up/up’
participial QRRM- QRMQD ‘fill/full’
b. Recessive suffix: preserves root vowel length
intransitive PXT PXTRR ‘bury (n.)/bury (v.)’
verbalizer E·LLW] E·LLW]RR>E·OLLW]D@ ‘song/sing’
instrumental OXN OXNE·LO ‘pull up/instrument
for pulling up’
remainder ZDD ZDDE·DQ ‘eat/remains of
food’

In Malayalam, gemination applies at the internal juncture of subcompounds

(compounds with head-modifier semantics) (b) but not at the internal juncture of
cocompounds (with coordinate semantics) (c) (Mohanan 1995:52):

(2) a. PHH D ‘table’ SHL ‘box’

NDVDDOD ‘chair’ ND (plural suffix)
b. >PHH DSSHL@6²ND ‘boxes made out of tables’
c. >PHH DSHL@&ND ‘tables and boxes’

In English, suffixes fall into two classes (Allen 1978, Siegel 1974, Chomsky
and Halle 1968, Kiparsky 1982a): those which shift stress and those which do not.

(3) Base Stress-shifting suffix Non-stress-shifting suffix

párent parént-al párent-ing
président prèsidént-ial présidenc-y
áctive àctív-ity áctiv-ist
démonstràte demonstrative démonstràtor

In all three of these examples, some morphological constructions in the

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

language (affixation, compounding) are associated with a pattern that other

constructions (other affixation, other compounding) are not.

2. Realizational morphology
Realizational (or process) morphology is the situation in which a morphological
category is exponed by a phonological process other than concatenation of
segmental morphemes. Three clear examples are cited below.
In Tohono O’odham, a well-known process of subtractive morphology derives
perfective verbs from imperfectives by deleting a final segment. Before a final
coronal consonant, a high vowel deletes as well. Examples come from Yu
(2000:129-30), citing Zepeda 1984, and Anderson (1992), citing Zepeda 1983:

(4) Imperfective Perfective gloss data source

VtNRQ VtNR ‘hoe object’ Yu 2000
KtZD KtZ ‘rub against object’ Yu 2000
KL QN KL Q ‘bark’ Anderson 1992

In Keley-i (Malayo-Polynesian), nonperfect aspect is marked by consonant

gemination, providing a coda to what would otherwise be the leftmost light
syllable (Samek-Lodovici 1992, citing original sources) (5a-c). In a word with all
closed (heavy) syllables (5d), gemination is blocked.

(5) (a) (b) (c) (d)

Base: SLOL GX\DJ DJWX GXQWXN
Subject focus: XPSLOOL XPGX\\DJ PDQ DJWX XPGXQWXN
Object focus: SLOOL GX\\DJ DJWX GXQWXN
Access. focus: LSSLOL LGGX\DJ L DJWX LGGXQWXN

English provides a familiar third example: stress shift marks the conversion
from verbs to nouns in English (e.g. Kiparsky 1982b):

(6) condúct cónduct

abstráct ábstract
recórd récord

3. Three approaches to morphologically conditioned phonology

We turn next to a brief presentation of three theories designed to cover
morphologically conditioned phonology. For maximum comparative effect, and
given the limited space available, it is necessary in these sketches to portray the
strictest version of each theory, ignoring nuanced variations of each.

3.1. Cophonology Theory

In Cophonology Theory (Orgun 1996; Inkelas et al. 1997; Inkelas 1998; Anttila

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

2002), a member of the family of construction grammar theories (Goldberg, et al.),

the morphological grammar consists of a set of word-building constructions. Each
construction embodies both a meaning function, which could be inflectional,
derivational, or even the identity function, as well as a form function
(cophonology), e.g. a set of ordered phonological rules or ranked constraints.
For example, the –ify construction in English is associated with a meaning
function that takes a nominal stem as input and produces an output whose form is
predictable from the form of the input by means of a phonological mapping that
concatenates the stem with the string –ify and performs such phonological
operations as (re)syllabification, stress shift, trisyllabic laxing, and velar
softening. In (7), only the form function is denoted, as f(x), where f represents the
cophonology and x represents the phonological form of the input string(s).

(7) [Phon = f(x)] [opácify]

[x] -ify [opaque] -ify

The cophonology of the comparative –er suffix in English differs from the
cophonology of –ify in numerous ways: it is stress-preserving, not stress-shifting;
it requires roughly monosyllabic inputs; it does not trigger Trisyllabic laxing or
velar softening. In Cophonology Theory, each individual morphological
construction has its own, potentially unique, cophonology; similarities among the
cophonologies of constructions in the same language are captured with meta-
generalizations formalized as a ‘grammar lattice’ in Anttila 2002. Precedents for
cophonologies can be found in Poser 1984 and Bochner 1992.

3.2. Stratal Optimality Theory (Kiparsky 2003b)

A descendant of Lexical Morphology and Phonology (LMP; Kiparsky 1982),
Stratal OT posits that every language has three strata, each with its own
phonological system:

(8) Stem stratum

Word stratum

Postlexical stratum

In Stratal OT, the phonological differences between –ify and –er would be
modeled by assigning –ify to the Stem stratum, which imposes resyllabification,
stress shift, Trisyllabic laxing and velar softening, and -er to the Word stratum,
which imposes only resyllabification. Stratal OT thus can be characterized as a
very restrictive version of Cophonology Theory in which every morphological

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

construction is associated either with the ‘Stem’ or the ‘Word’ cophonology.

3.3. Indexed Constraint Theory

Unlike Cophonology Theory and Stratal OT, both of which assume that a
language can have multiple cophonologies, Indexed Constraint Theory assumes a
single phonological grammar for each language. Because Indexed Constraint
Theory was formulated within OT, it is always discussed with reference to OT
constraints, though it also resembles the rule-based theory of The Sound Pattern
of English (Chomsky & Halle 1968), which assumed a fixed set of general rules
for each language, plus a contingent of minor rules indexed to particular lexical or
morphological contexts. In Indexed Constraint Theory, morphologically
conditioned phonology is handled by indexing constraints to individual
morphological contexts, e.g Max-Croot, Max-Caffix, Max-CBR, etc. Proponents
include McCarthy and Prince 1995; Smith 1997, Itô and Mester 1999; Pater 2000,
2006; and Alderete 2001, among others.
With this brief introduction to the three theories being compared, we now test
them, using evidence from realizational morphology and morphologically
conditioned phonology, against the SUBSTANCE, SCOPE and LAYERING
generalizations, to be motivated in the following sections.

4. SUBSTANCE
The SUBSTANCE generalization holds that realizational morphology and
morphologically conditioned phonology overlap substantively to the point of
being essentially indistinguishable. In a brief tour below, we will see seven
different phonological effects, each instantiated once as realizational morphology
and once as morphologically conditioned phonology.

4.1. Segment deletion

As seen earlier, in Tohono O’odham, final segment deletion marks the perfective
category in verbs. Along similar lines, final vowel deletion marks nominative case
in Lardil (9) (Blevins 1997:249, citing original sources):

(9) NonFuture Accusative Nominative gloss

NHQWDSDOLQ NHQWDSDO ‘dugong’
QJDOXNLQ QJDOX ‘storey’
PD\DUUDQ PD\DUU ‘rainbow’
PHODQ PHOD ‘sea’

Segment deletion commonly occurs as a morphologically conditioned

phonological process, as well. In Turkish, vowel hiatus arising at morpheme
boundaries is repaired in most cases by glide epenthesis, but in one case – that of
the progressive suffix –Iyor –by vowel deletion:

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

(10) C-final root V-final root

‘do’ ‘come’ ‘understand’ ‘say’
\DS JHO DQOD V|\OH
Facilitative/-Iver/: \DS YHU JHOLYHU DQOD\ YHU V|\OH\LYHU
Progressive/-Iyor/: \DS \RU JHOL\RU DQO \RU V|\O\RU

4.2. Gemination
In section 1 we saw gemination serving as the sole mark of nonperfect aspect in
Keley-i, and as morphologically conditioned phonology in Malayalam, where it
served as a phonological accompaniment to subordinate compounding. Here we
see two additional examples. In Woleaian, denotatives are formed by geminating
the stem-initial consonant (Kennedy 2003:174). This is realizational morphology:

 ILOL  IILOL ‘choose it/to choose’

XJD  EEXJD ‘boil it/to boil’
WDEHH\  WWDEH ‘follow it/to follow’

In Hausa, prefixing pluractional verb reduplication includes a process of stem-
initial gemination that other prefixing reduplication constructions to not exhibit
(Newman 2000:235, 425). This is morphologically conditioned phonology:1

 E~Jj   E~EE~Jj ‘beat’

GiQQq   GiGGjQQp  ‘press down, oppress’
J\j U~  J\jJJ\j U~ ‘be well repaired’
   
4.3. Truncation to a prosodic constituent
Truncation can serve as realizational morphology, e.g. Spanish nickname
formation (13) (Pineros 2000:71); it also commonly accompanies affixation, e.g.
in Swedish nicknames (14) (Weeda 1992:121, citing original sources):

(13) Ricardo Rica

Armando Arma
Jesus Jesu
Concepción Conce

(14) a. DONRKROLVW  DONLV ‘alcoholic’

ODERUDWRUL XP  ODEELV ‘lab’
b. PDWV  PDWWH (proper name)
IDELDQ  IDEEH (proper name)

4.4. Dissimilation and ‘exchange’ rules

Both realizational morphology and morphologically conditioned phonology
1
‘ ’ represents trilled r, written in Hausa as an r-tilde. Plain ‘r’ is a rhotic approximant.

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

include effects where one segment surfaces with a value opposite either to its own
input value (‘Exchange rules’, ‘toggles’) or to the output value of another segment
in the same word (‘dissimilation’). For a survey, see Kurisu 2001.
In Nuer (Frank 1999), input/ouput vowel length dissimilation marks the
singular/plural distinction in nouns. (The language has multiple ways of marking
the number distinction, of which this ‘exchange’ process is just one.)

(15) Nominative singular Nominative plural gloss

a. OH\ OHH\ ‘animal(s)’
ZX N ZX N ‘(upper) arm(s)’
b. NDDW NDW ‘vulture(s)’
\LHHU \LsU ‘river(s)’

In Hausa, ‘stabilizer’ clitics have a fixed segmental component (QH  for

masculine, FH  for feminine) but exhibit tone polarity. The stabilizer surfaces with
tone opposite from that of the preceding syllable (Newman 2000:160ff., 598):

 …L-H …H-L
 JZjGzQp  ‘it’s a blanket’ Nq NpQq ‘it’s a bicycle’
 ]y Eq Qp  ‘it’s a ring’ QiQQq ‘it’s there (by you)’
 Py Wj Fp ‘it’s a car’ iNZiOi Fq ‘it’s a piece of junk’
 Jy Qk Fp ‘it’s the farm’ Uu Ji Fq ‘it’s a gown’

4.5. Stress/pitch-accent (re)assignment

Stress and accent shift commonly expone morphological categories on their own,
as seen in the example of English verb-to-noun conversion in (6), and are also
very frequently morphologically conditioned concomitants of affixation and other
overt morphological processes, as in the example of English stress-shifting
suffixes in (3).

4.6. Review
The phonological operations used to realize morphological constructions are
essentially the same operations that can accompany overt affixation, reduplication
and compounding. There is no clear basis for distinguishing the two (cf. Anderson
1975). A more comprehensive survey might well find that certain types of
phonological effects are much more rarely found as the sole markers of
morphological categories than others are, and that certain types of phonological
effects are more likely to be morphologically restricted (in any way) than others
are. The reasons for this would be interesting to explore. However, for present
purposes the overlap in type is more significant. It creates a problem of
discriminability. Theories which offer separate treatments of realizational
morphology and morphologically conditioned phonology require some criteria for
telling the two part, even when they resemble one another in form.

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

The practical criterion seems to be that a phonological alternation is classified

as ‘realizational morphology’ if it is the sole exponent of a morphological
construction, whereas it is classified as ‘morphologically conditioned phonology’
if it accompanies something else which is judged to be the primary exponent of a
morphological construction (affixation, reduplication, compounding). All of the
examples discussed in Section 4 were tacitly classified according to this criterion.
The problem is that in many cases it is difficult or impossible to determine which
phonological effect is the primary marker of a morphological construction (i.e.
morphology), and which is the secondary phonological correlate (i.e.
morphologically conditioned phonology).
In Hausa (Newman 2000), the dimensions of whether a morphological
construction is tone-replacing and/or has overt affixation are independent, so that
the same tone-replacement phenomenon in some cases is classified as
realizational morphology (17a) and in others as morphologically conditioned
phonology (17c).

(17) base tone replaced base tone preserved

zero derivation
overt affixation

a. No affixation; tone replacement (imperative formation)

 Ni Pj  Nj Pi ‘catch (!)’
EtQFuNp  EuQFuNp ‘investigate (!)’
QiQQp Py  QjQQq Py ‘seek repeatedly (!)’ (< né mó ‘seek’)
b. No affixation, no tone replacement (Grade 2 verbal noun formation)
IjQVi  IjQVi  ‘redeem/redeeming’
WjPEi\j  WjPEi\j ‘ask/asking’
c. Overt affixation, tone replacement (various plural classes)
Pi OjP  Pj OjPiL ‘teacher-pl’ -LH
Uu Ji   Ut J~Qj ‘gown-pl’ -HL
WjPEi\j   WiPEi\y \t ‘question-pl’ -H
d. Overt suffixation, no tone replacement (various)
GiIj   GiIj Zi ‘cook-ppl’ -LH
JjMp Up   JjMp Uu\i ‘short-fem’ -LH
 K Oi   K Ok ‘hat-def’ -L

For theories making any kind of analytical distinction between the two effects,
treating exactly the same process, tone replacement, as morphology in (17a) but
phonology in (17c) poses a duplication problem.
In Barasana, a paradox is actually created. A number of Barasana suffixes
exert effects on stem tone. The Non3rdSubj suffix E causes H tone to align all
the way to the right in words containing it, while the Interrogative suffix -ri
causes H to align all the way to the left (Pycha 2005, citing Gomez-Imbert and
Kenstowicz 2000):

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

(18) EDDE  ‘swim-non3rdSubj = I/you/we swim’

+++
EDDUL ‘swim-Interr = did he/she/they swim?’
+

These suffixes exhibit what Pycha (2005) calls mutual partial blocking. Their
segmental components cannot co-occur (18a), nor can their mutually incompatible
effects on tone both be realized. In words where both meanings are desired, we
find the segments of the Interrogative—and the tones of the Non3rdSubj (18b):

 D EDDULE  EDDE UL ¶GLG,\RXZHVZLP"·

 E EDDUL ¶GLG,\RXZHVZLP"·
  +++

Pycha’s interpretation of the facts in (18) is that both categories (Non3rdSubj,

Interrogative) achieve exponence, by using the segments of one and the
cophonology of the other. The paradox this poses for a theory that distinguishes
realizational morphology from morphologically conditioned phonology is that the
tone pattern of the Non3rdSubject must, by the criterion used above, be analyzed
as morphologically conditioned phonology based on the fact that it co-occurs with
a ‘primary’ exponent, namely the suffix E ; yet its ability to expone the
Non3rdSubject even when E  is absent identifies it as realizational morphology.
This is a paradox.
One possible way to avoid the problems illustrated in Hausa and Barasana
would be to reduce everything to morphologically conditioned phonology,
reanalyzing apparent cases of realizational morphology as zero derivation
accompanied by morphologically conditioned phonology. This would, however,
pose a problem in Barasana, where the morphologically specific tonal effects of
the Non3rdSubj are present even when the affiliated suffix is not. Alternatively,
we could try to reduce all morphologically specific phonological effects to
realizational morphology, reanalyzing apparent cases of morphologically
conditioned phonology as instances of ‘extended exponence’, the multiple
marking of a morphological category (e.g. Matthews 1972; Stump 1991).
Multiple exponence of overt morphology is a common enough phenomenon; in
Hausa, for example, the formation of class 13 noun plurals involves suffixation
and reduplication, as well as tone replacement (Newman 2000:458):

(20) WVtUz  WVuUpWVuUp ‘shoot, sprout(s)’

NZiQj  NZjQpNVjQp ‘corner, curve(s)’
KiEiLFu  KjEjLFpKjEjLFp ‘innuendo(s)’

The challenge for any theory of morphologically specific phonological effects

is in accounting for their overlap in substance, which makes them difficult to

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

distinguish from one another and creating a potential duplication problem.

Observations like these have been made before in the literature, leading to
proposals that realizational morphology and morphologically conditioned
phonology should be analyzed in the same way (Ford and Singh 1983; Poser 1984;
Dressler 1985; Ford and Singh 1985; Singh 1987; Anderson 1992; Bochner 1992;
Singh 1996). We turn next to a discussion of how the three theories compared in
this paper do in this regard.

4.7. Theoretical discussion

Cophonology Theory is naturally suited to capturing the overlap in substance
between realizational morphology and morphologically conditioned phonology,
since it uses exactly the same mechanism – a cophonology – to account for both.
For example, truncation is modeled by a cophonology which maps an input to an
output of a certain size. In the English examples below, the output of the
truncating cophonology, g(x), is two syllables. In the construction on the left, in
which truncation is the sole mark of the construction, the input is the long stem
Rebecca and the truncating cophonology produces the disyllabic output Becca. In
the construction on the right, in which truncation to two syllables accompanies
overt suffixation of -y, the input is Becky, with material from the stem and the
suffix both competing for a spot in the disyllabic output. The inputs differ,
because the constructions differ, but the cophonologies are the same.

(21) g(x): a cophonology limiting the output to two syllables ( >> Max)

g(Rebecca) = Becca g(Rebecca, -y) = Becky

/Rebecca/ /X/Stem /-i/

(Realizational morphology) (Morphologically conditioned phonology)

By collapsing the formal treatments of realizational morphology and

morphologically conditioned phonology, Cophonology Theory eliminates the
analytical ambiguity of cases of the type discussed in Section 4.6.
In Indexed Constraint Theory, all phonological alternations are accomplished
by the ranking of phonological constraints, and thus the expectation is that
Indexed Constraint Theory should make essentially the same predictions as in
Cophonology Theory regarding the substance of realizational morphology and
morphologically conditioned phonology, even though the mechanism of relating
phonological subpatterns to particular constructions is different from the
mechanism used in Cophonology Theory. (We will come back to this issue in later
sections.)
For example, Kurisu (2001) has proposed that the constraint REALIZE-MORPH

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

(RM) could be responsible for many effects classified here as realizational

morphology; RM essentially requires that the phonological output of a
morphological construction be non-homophonous with the input, such that a
construction with no overt affix or other morphological exponent would be
required through RM to undergo some phonological change. The resulting change
is predicted to be the least expensive one, as determined by the ranking of
markedness and faithfulness constraints of the grammar. In Icelandic, deverbal
nouns are formed by deleting the final vowel from the infinitive:

(22) NOLIUD  NOLIU ‘climb/climbing’

JUHQMD  JUHQM ‘cry/crying’
V||WUD  V||WU ‘sip/sipping’
SXXNUD  SXXNU ‘conceal/concealment’

Kurisu derives this outcome by ranking Dep and RM above Max, such that the
need to satisfy RM compels a Max violation.

(23) NOLIUD RM DEP MAX

a. NOLIUD !*
b. NOLIU *
c. NOLIUDWD *!

RM is a type of indexed constraint; it is an anti-faithfulness constraint indexed

to a morphological constituent, in this case deverbal noun. Indexed Constraint
Theory commonly indexes faithfulness constraints as well: Base-Reduplicant
Correspondence Theory indexes faithfulness constraints to Base and Reduplicant
constituents (McCarthy & Prince 1995), Smith (1999) indexes faithfulness
constraints to nouns vs. verbs, etc. The same approach ought to be able to capture
the morphologically conditioned phonological effects we have seen thus far. For
example, Ito & Mester (1997) analyze a case of truncation in German comparable
to the Rebecca Becky example, above, exploiting an abstract morpheme TRUNC
which is compelled, by indexed constraints, to be faithful segmentally to the full
stem and to fit, with the German equivalent of –y, into two syllables. This
constraints that participate in this analysis are very similar to the ones in a
cophonology account, and while the theories differ in other ways they make
similar predictions about substance.
In contrast to Cophonology Theory and Indexed Constraint Theory, Stratal OT
has little to say about realizational morphology or its relation to morphologically
conditioned phonology, making it hard to evaluate any predictions Stratal OT
might make about substance. Like LMP, Stratal OT focuses on generalizations
holding over stems and words, but ignores alternations that are construction-
specific. Since not all stem morphology in English is truncating, Stratal OT
cannot accomplish the truncation seen in Rebecca Becca through Stem

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

phonology. Instead it would require some constraint or constraint ranking specific

to nickname formation – i.e. indexed constraints or cophonologies, merging
Stratal OT with one or the other of the two approaches with which it contrasts.

6. SCOPE
With both morphologically conditioned phonology and realizational morphology,
the SCOPE p of the phonological effect(s) is the stem produced by the word
formation process in question. By associating cophonologies with morphological
constructions, Cophonology Theory predicts that the scope of each cophonology
will be the morphological subconstituent built by the associated construction.
For example, in a word with three suffixes, Cophonology Theory predicts that
the cophonology of Stem2 can affect the surface form of Stem1 and Suffix2, but
that the cophonology of Stem2 cannot affect the surface form of Suffix3:

(24) word

stem2

stem1

root suffix1 suffix2 suffix3

A case study from Hausa illuminates the significance of this type of prediction.
Cophonology Theory predicts that if a tone-replacing construction is embedded
within a tone-preserving construction, it will not replace the tones of any affixes
introduced by the outer construction; these are outside its scope. The ventive
construction is tone-replacing (Newman 2000:663): IuWi  (LH) ‘go out’ ItWy  (H)
‘come out’,JiQJjUi +/+‘roll down’ JiQJiUy  (H) ‘roll down here’, etc. As
seen in (25), a ventive stem can be converted to a verbal noun through the
suffixation of -`wá the tone-preserving verbal noun-forming suffix

(25) ItW{ Zi (H-LH)

Tone preserving
cophonology
ItWy (H)
Tone-replacing cophonology
replaces LH with H melody
IuWi (LH) y (H) CZi (LH)
‘go out’ VENTIVE VERBAL NOUN FORMER
‘coming out’

The outer suffix retains its lexical LH tone pattern; it is immune to the tone

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

replacement pattern which is imposed on the ventive stem subconstituent by the

ventive cophonology. Scope effects of this kind are an intrinsic prediction of
Cophonology Theory.
Stratal OT can handle some but not all scope effects. Like Cophonology
Theory, Stratal OT assumes a layered structure in which the cophonology of an
higher (e.g. Word) level applies to the output of the cophonology of an earlier (e.g.
Stem) level. Stratal OT thus predicts that the Word cophonology will have scope
over Stems (and the suffixes combining with them to form Words), but that the
Stem cophonology will not have scope over Word-forming suffixes.
The challenge faced by Stratal OT is describing the scope of morphologically
specific phonological effects that are not general within Stems or Words. To
model the division between tone-preserving and tone-replacing morphology in
Hausa, for example, Stratal OT must assign one effect, e.g. tone replacement, to
Stems, and the other, e.g. tone-preservation, to Words. However, tone-replacing
and tone-preserving morphological constructions can be embedded in either order.
In (26), the tone-replacing ventive construction is embedded within the tone-
preserving pluractional, which in turn is embedded within the tone-replacing
imperative (represented with a dummy suffix for graphical clarity). If Words and
Stems are strictly ordered, Stratal OT cannot handle this case:

(26) QqQQq Py

QpQQp Py

Qp Py

CVC- Qq Pi (LH) y (H) -Ø (LH)

PLURACT.- ‘seek’ -VENTIVE -IMPERATIVE
‘seek repeatedly!’

Indexed Constraint Theory faces two challenges in describing and predicting

SCOPE effects. One, addressed here, is the question of what constraints are indexed
to. (The other is layering, discussed in Section 6). The indexation issue can be
illustrated in Hausa with the tone-replacing cophonology associated with the
Ventive (and several other morphological constructions) and the tone-preserving
cophonology associated with verbal noun-forming CZi (and many other affixes).
Cophonology Theory would posit the constraint rankings in (27):

(27) Ventive cophonology Tone=H » Ident-tone, Tone = LH

CZi verbal noun cophonology Ident-tone » Tone=H

Indexed Constraint Theory has one constraint ranking for the entire language.
The cophonologies in (27) could translate into indexed constraints as follows:

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

(28) a. Ident-tone-`wá » Tone=H » Ident-tone

b. Tone=HVentive » Ident-tone » Tone=H

The ranking in (28a) indexes faithfulness: the verbal noun-former -`wá  is

specially faithful, outranking the general markedness constraint Tone=H to which
other stems are subject. The alternative ranking in (28b) indexes markedness
constraints. All-H is the unmarked pattern for the Ventive, but faithfulness trumps
markedness for other constructions, which are therefore are tone-preserving. It
does not matter here which method is chosen; the literature on Indexed Constraint
Theory favors indexing faithfulness, rather than markedness constraints (see e.g.
Alderete 2001, though cf. Inkelas & Zoll 2007).
On either option, capturing SCOPE requires indexed constraints to refer not to
morphemes, but to complex stems. The H tone mandate of the ventive, captured
in the constraint Tone=HVentive, must refer to the entire ventive stem, not just the
ventive suffix y in order to generate, for input nè má , the correct output Qp Py 
For this reason, recent work in Indexed Constraint Theory has moved in the
direction of Cophonology Theory by indexing constraints to subconstituents of
words, not individual morphemes (e.g. Alderete 2001).

6. Layering
A corollary of the scopal prediction of cophonologies is layering, the effect in
which, given a structure where X is a daughter of Y, the output of the
cophonology associated with X is the input to the cophonology of Y. This
prediction holds for both realizational morphology and morphologically
conditioned phonology. A good illustration of this prediction can be found in
example (26), repeated below, which contains two tone-replacing morphological
constructions. The inner one (ventive) imposes all-H; the outer one (imperative)
imposes LH. The word surfaces LH, as Cophonology Theory predicts.

(29) QqQQq Py

QpQQp Py

Qp Py

CVC- Qq Pi (LH) y (H) -Ø (LH)

PLURACT.- ‘seek’ -VENTIVE -IMPERATIVE

The way two cophonologies in the same word interact depends intrinsically on
the hierarchical structure of the word. The outer construction has the last say.
Stratal OT also predicts layering, to which Kiparsky 2000 has pointed as a

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

possible explanation for phonological opacity. The problem for Stratal OT, as
mentioned above, is simply that it does not provide enough layers to capture the
richness of morphologically conditioned phonology and realizational morphology
within a language.
In contrast to Cophonology Theory and Stratal OT, in which the interaction
between morphologically conditioned phonological patterns follows from the
hierarchical structure of a given word, in Indexed Constraint Theory interactions
of these types follow from constraint ranking, which is fixed in the language. To
illustrate this, consider the constraint ranking needed to generate the imperative
ventive word in (32), in which the LH imperative tone melody takes precedence
over the all-H melody associated with the ventive:2

(30) Tone=LHimperative » Tone=HVentive » Ident-tone » Tone=H, Tone=LH

 >>Qq Pi y @9HQWLYH²‘@,PSHU TONE=LHImper. TONE=HVent IDENT

 D Qp Py   
 E Qq Py   

The outcome of this tableau is determined by the highest ranked morphologically

indexed constraint, not by the hierarchical structure of the word.
If the morphological constructions involved always occur in a fixed order,
then layering of cophonologies and ranking of indexed constraints make
essentially the same predictions. However, there are good examples of languages
in which the same constructions can occur in either order, with different
phonological results. This was an important result of Mohanan 1986, in which it
was demonstrated that the two types of compounds in Malayalam could embed
inside each other; a similar freedom of combination occurs in Turkish, as pointed
out in detail in Inkelas and Orgun 1998, and in Cibemba, as pointed out by
Hyman 1994. Indexed Constraint Theory does not capture the overarching
generalization that scope is related to hierarchical position.

7. Conclusion
Cophonology Theory has clear advantages over Indexed Constraint Theory and
Stratal OT in capturing SUBSTANCE, SCOPE, and LAYERING. Yet cophonologies
have been viewed with concern, principally over the issue of cophonology
proliferation: without a lid on cophonology variability, a language might vary as
much internally as unrelated languages can vary (see e.g. Benua). This concern
has been addressed in two ways in the literature (Inkelas and Zoll 2007). On the
formal side, Anttila (2002) has proposed that cophonologies in the same language
must conform to a master ranking of constraints; only constraints left unranked in
this master ranking are allowed to vary in their ranking across individual

2
Note that this ranking indexes markedness constraints, rather than faithfulness constraints. An
indexed faithfulness account would be much more challenging to develop.

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

cophonologies. More substantively, researchers such as Bermudez-Otero and

McMahon (2006) have observed that cophonological diversity arises from
diachronic change, and that languages change too slowly and in too systematic a
fashion to permit the kind of wildly divergent cophonologies that have been cited
as a reason to avoid Cophonology Theory.
We have also seen in this study, however, that Cophonology Theory, Stratal
OT and Indexed Constraint Theory have many properties in common, and
whatever successor to these theories ultimately ends up being adopted will share
their common goal of tying morphologically conditioned phonological effects to
morphological subconstituents of complex words.

References

Alderete, John. 2001. Dominance effects as transderivational anti-faithfulness.

Phonology 18: 201-253.
Allen, Margaret. 1978. Morphological investigations. Ph.D. diss., Massachusetts
Institute of Technology.
Anderson, Stephen R. 1975. On the interaction of phonological rules of various
types. Journal of Linguistics 11: 39-63.
Anderson, Stephen R. 1992. A-morphous morphology. Cambridge: Cambridge
University Press.
Anttila, Arto. 2002. Morphologically conditioned phonological alternations.
Natural Language and Linguistic Theory 20: 1-42.
Bermudez-Otero, Ricardo, and April Mcmahon. 2006. English phonology and
morphology. In B. Aarts and A. McMahon, eds., The Handbook of English
Linguistics, Oxford: Blackwell.
Blevins, Juliette. 1997. Rules in Optimality Theory: two case studies. In I. Roca,
ed., Derivations and Constraints in Phonology, 227-260. Oxford: Clarendon
Press.
Bochner, Harry. 1992. Simplicity in generative morphology: Publications in
Linguistic Sciences 37. Berlin: Walter de Gruyter.
Chomsky, Noam, and Morris Halle. 1968. The sound pattern of English. New
York: Harper and Row.
Dressler, Wolfgang. 1985. Morphonology, the dynamics of derivation. Ann Arbor:
Karoma Publishers.
England, Nora. 1983. A Grammar of Mam. Austin: University of Texas Press.
Ford, Alan, and Rajendra Singh. 1983. On the status of morphophonology. In J. F.
Richardson, M. Marks and A. Chukerman, eds., Papers from the parasession
on the interplay of phonology, morphology and syntax, 63-78. Chicago:
Chicago Linguistic Society.
Ford, Alan, and Rajendra Singh. 1985. Towards a non-parametric morphology. In
M. Niepokuj, M. VanClay, N. Nikiforidou and D. Feder, eds., Papers from
BLS 11, 87-95. Berkeley: Berkeley Linguistic Society.

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

The Morphology-Phonology Connection

Frank, Wright. J. 1999. Nuer noun morphology. MA thesis, University of New

York at Buffalo.
Gomez-Imbert, Elsa, and Michael Kenstowicz. 2000. Barasana tone and accent.
International Journal of American Linguistics 66: 419-463.
Hyman, Larry M. 1994. Cyclic phonology and morphology in Cibemba. In J.
Cole and C. Kisseberth, eds., Perspectives in Phonology, 81-112. Stanford:
Center for the Study of Language and Information.
Inkelas, Sharon. 1998. The theoretical status of morphologically conditioned
phonology: a case study from dominance. Yearbook of Morphology 1997,
121-155.
Inkelas, Sharon, and Cemil Orhan Orgun. 1998. Level (non)ordering in recursive
morphology: evidence from Turkish. In S. Lapointe, D. Brentari and P. Farrell,
eds., Morphology and its relation to phonology and syntax, 360-392. Stanford:
Center for the Study of Language and Information.
Inkelas, Sharon, Cemil Orhan Orgun, and Cheryl Zoll. 1997. Implications of
lexical exceptions for the nature of grammar. In I. Roca, ed., Constraints and
Derivations in Phonology, 393-418. Oxford: Clarendon Press.
Inkelas, Sharon, and Cheryl Zoll. 2007. Is Grammar Dependence Real? A
comparison between cophonological and indexed constraint approaches to
morphologically conditioned phonology. Linguistics 45.133-171.
Itô, Junko, and Armin Mester. 1999. The phonological lexicon. In N. Tsujimura,
ed., The handbook of Japanese linguistics, 62-100. Malden, MA: Blackwell.
Kennedy, Robert. 2003. Confluence in phonology: evidence from Micronesian
reduplication. Ph.D. diss., University of Arizona.
Kiparsky, Paul. 1982a. Lexical morphology and phonology. In I.-S. Yang, ed.,
Linguistics in the Morning Calm, 3-91. Seoul: Hanshin.
Kiparsky, Paul. 1982b. Word-formation and the lexicon. In F. Ingemann, ed., 1982
Mid-America linguistics conference papers. Lawrence, KS: University of
Kansas.
Kiparsky, Paul. 2000. Opacity and cyclicity. The Linguistic Review 17: 351-367.
Kiparsky, Paul. 2003a. Fenno-Swedish Quantity: Contrast in Stratal OT. Stanford
University, ms.
Kiparsky, Paul. 2003b. Finnish noun inflection. In D. Nelson and S. Manninen,
eds., Generative approaches to Finnic and Saami linguistics, Stanford: Center
for the Study of Language and Information.
Kurisu, Kazutaka. 2001. The phonology of morpheme realization. Ph.D. diss.,
University of California, Santa Cruz.
Matthews, P. H. 1972. Inflectional Morphology: A Theoretical Study Based on
Aspects of Latin Verb Conjugation. Cambridge: Cambridge University Press.
Mccarthy, John, and Alan Prince. 1995. Faithfulness and reduplicative identity. In
J. Beckman, L. Dickey and S. Urbanczyk, eds., University of Massachusetts
Occasional Papers in Linguistics 18: Papers in Optimality Theory, 249-384.
Amherst, MA: Graduate Linguistic Student Association.
Mohanan, K. P. 1995. The organization of the grammar. In J. Goldsmith, ed., The

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.
 Job Name: -- /337544t

Sharon Inkelas

handbook of phonological theory, 24-69. Cambridge, MA: Blackwell

Publishers.
Newman, Paul. 2000. The Hausa language: an encyclopedic reference grammar.
New Haven: Yale University Press.
Orgun, Cemil Orhan. 1996. Sign-based morphology and phonology: with special
attention to Optimality Theory. Ph.D. diss., University of California, Berkeley.
Pater, Joe. 2000. Non-uniformity in English secondary Stress: the role of ranked
and lexically specific constraints. Phonology 17.237-274.
Pineros, Carlos-Eduardo. 2000. Prosodic and segmental unmarkedness in Spanish
truncation. Linguistics 38.63-98.
Poser, William J. 1984. The phonetics and phonology of tone and intonation in
Japanese. Ph.D. diss., Massachusetts Institute of Technology.
Pycha, Anne. 2005. Partial Blocking. Paper presented at 41st Regional Meeting of
the Chicago Linguistic Society.
Samek-Lodovici, Vieri. 1992. A unified analysis of crosslinguistic morphological
gemination. Proceedings of CONSOLE-1, Utrecht.
Siegel, Dorothy. 1974. Topics in English morphology. Ph.D. diss., Massachusetts
Institute of Technology.
Singh, Rajendra. 1987. Well-formedness conditions and phonological theory. In W.
Dressler, et al. eds., Phonological 1984. Cambridge: Cambridge University
Press.
Singh, Rajendra. 1996. Natural phono(morpho)logy: a view from the outside. In B.
Hurch and R. Rhodes, eds., Natural phonology: the state of the art. Papers
from the Bern Workshop, Berlin: Mouton.
Smith, Jennifer. 1997. Noun faithfulness: on the privileged status of nouns in
phonology. ms. 242-1098.
Stump, Gregory. 1991. A paradigm-based theory of morphosemantic mismatches.
Language 67: 675-725.
Weeda, Donald Stanton. 1992. Word truncation in prosodic morphology. Ph.D.
diss., University of Texas, Austin.
Willard, E. Rainbow. 2004. Dominance effects in a dialect of Mam Maya. Paper
presented at Linguistic Society of America annual meeting, Boston.
Yu, Alan C.-L. 2000. Stress Assignment in Tohono O'odham. Phonology 17: 117-
135.
Zepeda, Ofelia. 1983. A Papago grammar. Tucson: University of Arizona Press.
Zepeda, Ofelia. 1984. Topics in Papago morphology. Ph.D. diss., University of
Arizona.

Sharon Inkelas
Department of Linguistics
University of California
Berkeley, CA 94720-2650

[email protected]

In order to view this proof accurately, the Overprint Preview Option must be checked
in Acrobat Professional or Adobe Reader. Please contact your Customer Service Rep-
resentative if you have questions about finding the option.