Professional Documents
Culture Documents
Title Pages
(p.i) Paradigms in Phonological Theory
General editors
Advisory editors
Page 1 of 5
Title Pages
PUBLISHED
by Jason Merchant
by Utpal Lahiri
by Lutz Marten
IN PREPARATION
by Ray Jackendoff
Page 2 of 5
Title Pages
(p.iv)
With offices in
Page 3 of 5
Title Pages
1 3 5 7 9 10 8 6 4 2
Page 4 of 5
Title Pages
Page 5 of 5
Oxford Studies in Theoretical Linguistics General Preface
Page 1 of 2
Oxford Studies in Theoretical Linguistics General Preface
David Adger
Hagit Borer
Page 2 of 2
Abbreviations
(p.viii) Abbreviations
ACL
Association for Computational Linguistics
BLS
Proceedings, Berkeley Linguistics Society
CLS
Proceedings, Chicago Linguistics Society
DM
Distributed Morphology
ESCOL
Eastern States Conference on Linguistics
GA
Gradient Attraction
GTT
Generalized Template Theory
MA
Metrical Alignment
MC
Metrical Consistency
MWF
Metrical Well-Formedness
NDEB
Non Derived Environment Blocking
NELS
Proceedings, Northeastern Linguistics Society
Page 1 of 3
Abbreviations
OP
Optimal Paradigms
OT
Optimality Theory
PC
Paradigm Contrast
PES
Principles of English Stress (Burzio 1994a)
PU
Paradigm uniformity
REH
Representational Entailments Hypothesis
ROA
Rutgers Optimality Archive
SPE
Sound Pattern of English (Chomsky and Halle 1968)
SSG
Sonority Sequencing Generalization
SWP
Stress to Weight Principle
TCT
Transderivational Correspondence Theory
UE
Uniform Exponence
UR
Underlying Representation
VOT
Voice onset time
WCCFL
West Coast Conference on Formal Linguistics
WSCLA
Workshop on the Structure and Constituency of the
Languages of the Americas
Page 2 of 3
Abbreviations
Page 3 of 3
Contributors
(p.ix) Contributors
Adam Albright, Massachusetts Institute of Technology
Page 1 of 2
Contributors
Page 2 of 2
Introduction: The Role of Paradigms in Phonological Theory
DOI:10.1093/acprof:oso/9780199267712.003.0001
1.1 INTRODUCTION
Page 1 of 34
Introduction: The Role of Paradigms in Phonological Theory
Page 2 of 34
Introduction: The Role of Paradigms in Phonological Theory
Page 3 of 34
Introduction: The Role of Paradigms in Phonological Theory
Page 4 of 34
Introduction: The Role of Paradigms in Phonological Theory
(1)
a. The more tightly words are associated with
each other, the stronger the preference for
uniform paradigms (e.g. inflectionally related
words are more likely to be subject to leveling
than derivationally related words) (Paul 1880:
206).
b. The higher the token frequency of individual
members of a paradigm, the higher the
resistance to undergo paradigm leveling (e.g.
lack of paradigm uniformity in paradigms like
{do, does}, {am, are, is}, as opposed to {walk,
walks}, {blame, blames}) (Paul 1880: 208).
c. Alternations within paradigms which come to
be associated with grammatical functions resist
leveling (e.g. umlaut in German, which came to
serve as a plural marker).
d. The more alternations there are within a
paradigm, the higher the resistance to undergo
paradigm leveling.
a. Stage I Stage II
b. Stage I Stage II
Page 5 of 34
Introduction: The Role of Paradigms in Phonological Theory
(3)
b. Rad [ra:t]
Rades [ra:d∂s]
Page 6 of 34
Introduction: The Role of Paradigms in Phonological Theory
Page 7 of 34
Introduction: The Role of Paradigms in Phonological Theory
(4)
ab [ap] ‘off’
(5)
Page 8 of 34
Introduction: The Role of Paradigms in Phonological Theory
Stage I Stage II
Page 9 of 34
Introduction: The Role of Paradigms in Phonological Theory
(6)
a. Vowel Lengthening: V → V: / __ [-son, +voice]
b. Final Devoicing: [-son] → [-voice] / __ #
(7)
Level 1:
Dentalization — — /mætr/
Level 2:
Affixation — le1tr —
Page 10 of 34
Introduction: The Role of Paradigms in Phonological Theory
(8)
a. [t̪]rain, [d̪]rain
b. ma[t̪]er, la[d̪]er, spa[n̪]er, pi[l̪]ar
(9) Dentalization:
The data
relevant
for PU
effects are
presented
in (10).
Here we
see that Dentalization does not affect a stem-final /t d n l/ if it
is followed by a class II (derivational) suffix (in 10a) or an
inflectional suffix (in 10b). (Note, however, that the rule does
apply across class I suffixes, e.g. elemen[t̪]-ary.)
(p.8)
(10)
a. shou[t]er, ru[n]er, ki[l]er
b. la[t]er, lou[d]er, fi[n]er
Page 11 of 34
Introduction: The Role of Paradigms in Phonological Theory
(11)
Page 12 of 34
Introduction: The Role of Paradigms in Phonological Theory
a. o(rígin) *! *
b. (óri)gin *
c. (óri)gin *
d. ? (óri)gin *
Page 13 of 34
Introduction: The Role of Paradigms in Phonological Theory
Page 14 of 34
Introduction: The Role of Paradigms in Phonological Theory
2: morphology — — o(rigi)n-al-ity
3: footing — — o(rigi)(n-al-i)ty
Page 15 of 34
Introduction: The Role of Paradigms in Phonological Theory
Page 16 of 34
Introduction: The Role of Paradigms in Phonological Theory
(13)
a. NONFINALITY: Do not parse a word final
syllable into a foot.
b. ALIGN-R: AlignR(Foot, Word)
c. ALIGN-L: AlignL(Foot, Word)
d. O-O-IDENT: The foot structure of a derivative
and Base are identical.
(14) Recursion A
Page 17 of 34
Introduction: The Role of Paradigms in Phonological Theory
a′. o(rígi)nal * *
c ′. o(rígi)nal * * *
d′. ? o(rígi)nal * * *
Page 18 of 34
Introduction: The Role of Paradigms in Phonological Theory
Recursion B
Page 19 of 34
Introduction: The Role of Paradigms in Phonological Theory
a″. o(rìgi)(náli)ty * *
b″. (òri)gi(náli)ty *
d″. ? o(rìgi)(náli)ty * * *
Page 20 of 34
Introduction: The Role of Paradigms in Phonological Theory
Recursion C
Page 21 of 34
Introduction: The Role of Paradigms in Phonological Theory
Recursion A Recursion B
Page 22 of 34
Introduction: The Role of Paradigms in Phonological Theory
(15)
? a. *
(càpi[?]a)
(lístic)
b. (càpi) *!
{tha(lístic)}
c. (càpi) *!
(thà)(lístic)
Page 23 of 34
Introduction: The Role of Paradigms in Phonological Theory
(16)
Page 24 of 34
Introduction: The Role of Paradigms in Phonological Theory
a. l∂́mnámš
l[∂́]mnálxw
l∂́mnál∂
l∂́mn∂xw
b. lèmnámš
lèmnálxw
lèmnám∂ *!***
lèmnál∂
lémn∂xw
c l∂́mnámš
l∂́mnálxw
l∂́mnál∂
lémn∂xw
Page 25 of 34
Introduction: The Role of Paradigms in Phonological Theory
morphological Base for the 3rd plural. Rather, both take a verb
root as Base and independently take the appropriate
inflections. Further, the cyclic model can explain only
unexpected similarities among morphologically related words,
not unexpected differences. However, Output–Output anti-
Faithfulness correspondence constraints can account for this
sort of data, as they establish a relationship between outputs
of morphologically related words and penalize identity
between outputs. As in Base-Priority accounts of paradigm
uniformity, in paradigmatic contrast we find an asymmetrical
relationship between members of the paradigm. The
phonological Base (in the Russian case, the 3rd person
singular) conforms to the regular phonology of the language,
while a morphologically related form (the 3rd person plural)
shows exceptional phonology due to a correspondence
relationship with the Base.
Page 28 of 34
Introduction: The Role of Paradigms in Phonological Theory
a. (óri)gin, *
(óri)ginal, **!
(òri)gi(náli)ty *
b. ? (óri)gin, * √
o(rígi)nal, * * *
o(rìgi)(náli)ty * *
c. o(rígin), *! √ *
o(rígi)nal, * *
o(rìgi)(náli)ty * *
d. (óri)gin, * √
o(rígi)nal, * * *
(òri)gi(náli)ty ** *! *
Page 29 of 34
Introduction: The Role of Paradigms in Phonological Theory
(18)
Page 30 of 34
Introduction: The Role of Paradigms in Phonological Theory
‘died’
Page 31 of 34
Introduction: The Role of Paradigms in Phonological Theory
Notes:
(1) There are a wide variety of terms in use referring to
analogical changes, some indicating the author’s attitude
towards the phenomenon, including falsche Analogie ‘false
analogy’ (cf Whitney’s (1867) mistaken analogy),
Formübertragung ‘transfer of form’, Uniformierung
‘uniforming’, Parallelbildung ‘parallel formation’, Association
‘association’, Angleichung ‘similarity’, Ausgleichung ‘leveling’,
Verschleppung ‘abduction, spreading’, Contamination
‘contamination’, Stumpfsinnige Übertragung ‘mindless
transfer’, Verflechtung ‘interweaving’, Verschmelzung
‘blending’, Kreuzung ‘crossing’, Rückbildung ‘back-
formation’ (Curtius 1885: 36, 37).
Page 32 of 34
Introduction: The Role of Paradigms in Phonological Theory
(7) But see Orgun (1996,1999) for arguments against the view
that cyclicity is inherently derivational.
Page 33 of 34
Introduction: The Role of Paradigms in Phonological Theory
(11) See work like Kiparsky (2000) and Orgun (1996, 1999) for
other critiques of the Base-Priority approach to cyclicity, and
alternative theories of cyclicity within OT
Page 34 of 34
The Morphological Basis of Paradigm Leveling
DOI:10.1093/acprof:oso/9780199267712.003.0002
Page 1 of 53
The Morphological Basis of Paradigm Leveling
2.1 INTRODUCTION
In older stages of Latin, many nouns exhibited paradigmatic s
~ r alternations created by rhotacism in suffixed forms, as in
(1a). In the period immediately before Classical Latin, these s
~ r alternations were eliminated by extending the r to the
nominative form. At approximately the same time, an
independent change shortened long vowels before word-final
sonorants, resulting in the paradigm in (1b).1
(p.18)
a. Pre-leveling b. Post-leveling
etc… etc…
Page 2 of 53
The Morphological Basis of Paradigm Leveling
Page 3 of 53
The Morphological Basis of Paradigm Leveling
a. F √ ** *
[honoːs],
[honoːr-is],
[honoːr-em]
b. [honoːs], *(gen.)! √ √
[honoːs-is], *(acc.)
[honoːs-em]
Page 4 of 53
The Morphological Basis of Paradigm Leveling
Page 5 of 53
The Morphological Basis of Paradigm Leveling
a.… √ √ *(nom.)! ** √
[honoːs],
[honoːr-is],
[honoːr-em]
c. [honoːr], *! √ √ *** √
[honoːr-is],
[honoːr-em]
d.… F √ √ √ *** *
[honor],
[honoːr-is],
[honoːr-em]
Page 6 of 53
The Morphological Basis of Paradigm Leveling
Page 7 of 53
The Morphological Basis of Paradigm Leveling
Page 8 of 53
The Morphological Basis of Paradigm Leveling
Page 9 of 53
The Morphological Basis of Paradigm Leveling
(7)
a. Why was a basic, ‘unmarked’ isolation form
(the nominative) rebuilt on the basis of more
marked suffixed forms, contrary to the usual
direction of analogical change? (Lahiri and
Dresher 1984; Bybee 1985: ch. 3)
b. What role (if any) did similar words, like
[soror], play in the change? Is there a minimum
number of such words necessary to effect such a
change?
c. Why did [honoːs] change to [honor], and not
[soror] to [soroːs]?3
d. Why were monosyllables and neuters
generally not affected?
e. Why did both [oːs] and [or] variants persist for
so long, and why was no variation induced in
etymological [or] words such as agentives?
Page 10 of 53
The Morphological Basis of Paradigm Leveling
Page 11 of 53
The Morphological Basis of Paradigm Leveling
nom. gen.
[gluptus] ~ [gluptiː]
[nokus] ~ [nokiː]
[reptus] ~ [reptoris]
[kortus] ~ [kortoris]
In the case of (8), there are two classes of words: the [-us] ~ [-
iː] words and the [-us] ~ [-oris] words. The two classes are
neutralized in the nominative, so the mapping from the
nominative to the genitive is unpredictable. If a learner were
to memorize just the nominative form, then two rules would be
necessary to project genitives ([-us]→[-iː] and [-us]→[-oris]).
Each of these rules would only have 50 percent accuracy in
the existing lexicon, each covering two out of four words.
Furthermore, for a hypothetical new word [tulpus], there
would be two possible genitives ([tulpiː] and (p.24)
[tulporis]), each with a 50 percent chance of being right,
leaving a 0 percent margin to decide by.
Page 12 of 53
The Morphological Basis of Paradigm Leveling
Page 13 of 53
The Morphological Basis of Paradigm Leveling
Such pairwise
comparisons
are iterated
across the
entire lexicon,
yielding a
sizeable set of
(often Fig 2.1. Minimal generalization to
redundant) discover environments
generalized
rules for each
morphological change. When a change occurs in a diverse set
of phonological environments, then the generalization
procedure in Fig.2.1 can yield very general rules; for example,
comparing the words [reptus] ~ [reptoris] and [kraus] ~
[kraoris] would yield the very general rule [us] → [oris] /——#
(assuming that [a] and [t] have no feature values in common).
Page 14 of 53
The Morphological Basis of Paradigm Leveling
nom. gen.
[neks] ~ [negis]
[arks] ~ [argis]
[teks] ~ [tekis]
[flurs] ~ [fluris]
Page 15 of 53
The Morphological Basis of Paradigm Leveling
Page 16 of 53
The Morphological Basis of Paradigm Leveling
Page 17 of 53
The Morphological Basis of Paradigm Leveling
Page 18 of 53
The Morphological Basis of Paradigm Leveling
(p.29)
Page 19 of 53
The Morphological Basis of Paradigm Leveling
Page 20 of 53
The Morphological Basis of Paradigm Leveling
Page 21 of 53
The Morphological Basis of Paradigm Leveling
Page 22 of 53
The Morphological Basis of Paradigm Leveling
(p.32)
As can be
seen, the
criteria
proposed in s.
2.2.2
generally
agree on the
relative
effectiveness
of the various Fig. 2.3. Average rank of forms as
forms as potential bases
possible
bases. The
only exception is the number of rules in the grammar, which
yields an uninterpretably different ranking. However, we saw
above that the number of rules failed to distinguish even the
simple four-word language in (8), so it is not surprising that it
performs poorly here. The combined results from all five
remaining criteria (excluding number of rules) are shown in
Fig. 2.3.
Page 25 of 53
The Morphological Basis of Paradigm Leveling
Page 26 of 53
The Morphological Basis of Paradigm Leveling
will use the genitive singular as the base for deriving the
nominative, but the same result could be achieved using the
ablative or dative singular.9
Page 27 of 53
The Morphological Basis of Paradigm Leveling
As Fig. 2.4.
shows,
agentive
nouns, which
are all
polysyllabic
and masculine
or feminine,
strongly favor
-r. In fact, Fig. 2.4. Preference for -r or -s in four
these words distinct sets of words
contained -r
Page 28 of 53
The Morphological Basis of Paradigm Leveling
2.4 DISCUSSION
Page 30 of 53
The Morphological Basis of Paradigm Leveling
Page 31 of 53
The Morphological Basis of Paradigm Leveling
Page 32 of 53
The Morphological Basis of Paradigm Leveling
Page 33 of 53
The Morphological Basis of Paradigm Leveling
Polysyllabic, non-neuter
Monosyllabic
floːs ‘flower’ 16 25 0 56 19
oːs ‘mouth’ 65 29 9 6 55
Neuter
onus ‘burden’ 40 45 40 — 15
Masculine, agentive
senaːtor ‘senator’ 43 33 23 28 14
Page 34 of 53
The Morphological Basis of Paradigm Leveling
Note: A problem arises in counting frequencies for neuters, since the nom. and acc. forms are identical. The hypothesis
being tested here is that the frequency of [s] forms in the paradigm determines their susceptibility to leveling, so I have
counted all s forms in the nom. column, to facilitate comparison with the masc. and fem. nouns.
Page 35 of 53
The Morphological Basis of Paradigm Leveling
Page 36 of 53
The Morphological Basis of Paradigm Leveling
Page 37 of 53
The Morphological Basis of Paradigm Leveling
Page 38 of 53
The Morphological Basis of Paradigm Leveling
Page 39 of 53
The Morphological Basis of Paradigm Leveling
Page 40 of 53
The Morphological Basis of Paradigm Leveling
Page 41 of 53
The Morphological Basis of Paradigm Leveling
[t ̚] [pat ̚] [th], [t∫h], [s] [pathi ɭ], [pathi ɭ], [pasi ‘field’
ɭ]
Page 42 of 53
The Morphological Basis of Paradigm Leveling
The contrast between Latin and Korean may also help to shed
light on why there seems to be a typological bias towards
paradigm leveling; in Latin, rhotacism affects just one segment
(/s/), so the only words that exhibit the alternation are those
with stem-final [s]. In Korean, on the other hand, coda
neutralizations affect all obstruents, so almost all lexical items
exhibit alternations. I conjecture that morphophonemic
alternations typically affect only a smallish subset of the
phonemic inventory; rhotacism affects just /s/, umlaut affects
only back vowels, palatalization tends to affect just coronals
(and often just coronal stridents), and so on. The end result is
that alternations tend to be the minority pattern, and there
will be a tendency to generalize non-alternation, or in other
words, to level.
Page 43 of 53
The Morphological Basis of Paradigm Leveling
2.5 CONCLUSION
I have argued here that the Latin [hono?s] ⇒ [honor] change
was caused by more than simply a sporadic pressure for
paradigm uniformity or Uniform Exponence constraints to
assert themselves over IO-Faithfulness constraints. I have
shown that the spread of [r] to nominative forms did more
than just create uniform paradigms; it also extended a pattern
of non-alternation that was already dominant in the lexicon.
Details of the change, such as its restriction to polysyllabic
nouns and non-neuters reflect the fact that these were
especially strong contexts for [r] stems. Furthermore, the
‘backwards’ direction of the leveling, with oblique forms
influencing the nominative singular, can be explained by a
particular model of Latin noun paradigms, in which an oblique
form served as the base, and nominative forms were derived
from oblique forms by rules operating on surface forms.
(p.43)
Page 44 of 53
The Morphological Basis of Paradigm Leveling
Page 45 of 53
The Morphological Basis of Paradigm Leveling
Page 46 of 53
The Morphological Basis of Paradigm Leveling
Page 47 of 53
The Morphological Basis of Paradigm Leveling
Page 48 of 53
The Morphological Basis of Paradigm Leveling
Notes:
(1) There is some evidence for an intermediate stage in the
late pre-Classical stage in which [honoːr] with long [oː] was at
least one possible variant of the nominative form. Early poets
such as Plautus and Enni-us still allowed final long [oːr], using
it to satisfy metrical requirements for heavy syllables, and also
sometimes used [-o(ː)r] instead of [-oːs], e.g. Ennius 545
(Skutsch 1985) Clām[ōː]r ād cǣlūm vōlvēndŭs pěr ǣthěră
vāgît; also Ennius 409, 428. As far as I know, there is no
evidence concerning how ‘clean’ this intermediate stage was—
the shortened variant [honor] may also have been used from
the very beginning of the [s] ⇒ [r] change. The analysis that I
propose here is compatible with the existence of an
intermediate [honoːr] stage, but does not rely on it. It would,
however, make a crucial difference for some other analyses,
such as the Uniform Exponence analysis sketched below.
Page 49 of 53
The Morphological Basis of Paradigm Leveling
(6) All the training input sets and results files for the
simulations discussed here, as well as the original database of
nouns, can be downloaded from http://ling.ucsc.edu/~albright/
papers/latin.html, accessed 25 Feb. 2004.
Page 50 of 53
The Morphological Basis of Paradigm Leveling
Page 51 of 53
The Morphological Basis of Paradigm Leveling
Page 52 of 53
The Morphological Basis of Paradigm Leveling
Page 53 of 53
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
DOI:10.1093/acprof:oso/9780199267712.003.0003
Page 1 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
3.1 INTRODUCTION
Crystal (1985: 220) defines a paradigm as ‘a set of forms
having a common root or stem’, where each form appears ‘in a
certain grammatical environment’. A paradigm thus expresses,
according to van Marle (1985: 225), ‘the ways in which
linguistic entities may be mutually connected’. These
definitions set the basis for the principles of paradigm
uniformity proposed in s. 3.1.2.
(1)
Page 2 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 3 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(2)
Page 4 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 5 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(3)
↓ tiftax-tiftexi
Page 6 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
That is, in
both
paradigms,
Structural
Identity is
respected.
However, in the main paradigm, Existence is violated, while
Structural Relation is respected, and in the sub-paradigm it is
the other way around.
Page 7 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(6)
Page 8 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Imperative truncation
Masculine Feminine
Future TI Future TI
Page 9 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
Page 10 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(p.50)
(7)
a. Markedness constraints:
• ONSET (Prince and Smolensky
1993) A syllable has an onset.
• *[σCCC (a member of *COMPLEX;
Prince and Smolensky 1993) A
triconsonantal onset is prohibited.
b. Morphological constraint:
• IMPERATIVE TRUNCATION
(IMPTRUNC) (anti-MAX; see
discussion below) Not every segment
in the base has a correspondent in
the output.
c. Faithfulness constraint:
• MAX SEGMENT (MAXSEG)
(McCarthy and Prince 1995) Every
segment in the input has a
correspondent in the output.
Page 11 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(8)
Page 12 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
CV truncation—in bases with an initial CVCσ]: Future tiftax–TI ftax ‘to open’
ti-ftax IMP ONSET *[σCCC IMP TRUNC MAX SEG
a. tiftax *!
b. tiftax *! *
c. tiftax *! *
d. ### tiftax **
e. tiftax ***!
Page 13 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(p.51)
(9)
Page 14 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
V truncation—in bases with an initial CVσ]: Future tevakeš–TI tvakeš ‘to request’
te-vakeš IMP ONSET *[σCCC IMP TRUNC MAX SEG
a. tevakeš *!
b. tevakeš *! *
c. #### tevakeš *
d. tevakeš **!
e. tevakeš *! ***
Page 15 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(10)
Page 16 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 17 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(p.52)
(11)
Page 18 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
ti-k.xí kxí
Page 19 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(12)
a. FAITH STRESSED SYLLABLE (FAITH[ineq])
Corresponding stressed syllables are identical
b. MAX SEGMENT STEM (MaxSegS) (Beckman
1997) Every segment in the stem of the base has
a correspondent in the output
(13)
Page 20 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. tit.fór *!
b. <t>it.fór *! *
c. titfór *! *
d. <t>itfór **
e. <tit>fór *! * * ***
Page 21 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(p.53)
(14)
Page 22 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. tit.fór *!
b. <t>it.fór *! *
c. t<i>tfór *! * *
d. <ti>tfór * **
e. <tit>fór *! ***
Page 23 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
Page 24 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
When the base begins with a CV syllable and the stem begins
with a sonorant consonant, the TI should have an initial
sonorant-consonant cluster in the onset (cf. (9)). Such a cluster
violates the Sonority Sequencing Generalization (SSG), which
is an undominated markedness constraint in the language.
(16)
Page 25 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 26 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(17)
Page 27 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. tirkod *
b. <t>irkod *!
c. t<i>fór *!
d. <ti>kod *
e. <tir>kod *! *!
f <ti>rekod *!
g t<i>erkod *!
Page 28 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
The paradigm thus ends up with quite a few gaps, as not every
future base has a truncated imperative form (see more gaps in
s. 3.4.2). These gaps, however, are not accidental; they are
predicted to appear for every base beginning with a CVC
syllable, where the second C (the stem initial consonant) is a
sonorant. Bases beginning with a CV syllable truncate the V
and the stem initial sonorant is thus the second member of the
cluster (e.g. telamed, tlamed ‘to teach Future, TI’). However,
as will be shown in the ensuing section, there is one case
where such verbs do have a TI.
When the stem of such verbs begins with a sonorant, only the
masculine TI can be derived (e.g. tered, red ‘to descend
Future, TI’); the feminine form would have an impermissible
sonorant-consonant cluster (e.g. terdi, *rdi ), and its formation
should thus be blocked (as in (15)). However, this is not the
case; as noted in s. 3.1.2, there are no gaps in the sub-
paradigm. As can be seen from the examples in (18), the
feminine TIs, when compared to the feminine future forms,
seem to have undergone enthesis (DEPviolation) and stress
shift (FAITHHEAD violation).
Page 29 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(18)
Page 30 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 31 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(19)
Page 32 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 33 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
Page 34 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(20)
Page 35 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. Imperatives
without a prefix
b. Imperatives
with a prefix
Page 36 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(21)
Page 37 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 38 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(22)
Page 39 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. tivrah *! *
b. t(i)vrah *! *
c. (ti)vrah *! *
d. tibrah *! * *
e. tivərah *! *
f. ə tibərah * * *
Page 40 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(23)
Page 41 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 42 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(24)
Page 43 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 44 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(25)
Page 45 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
Page 46 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(26)
Page 47 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. ə tazkir *
b. tazkir *! *
c. Tazkir *! (a) *
d. tazkir *! (a) **
e. thazkir *! (h) *
Page 48 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
Page 49 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(27)
Page 50 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. tiədal *! *
b. <t>hiədal *! * *(h)
c. ə <tig>gəðal ** *(ə)
Page 51 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
b.
Page 52 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. təəaddel *! *
b. <t>həəaddel *! * *(h)
c. <t>gaddel **
Page 53 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
(p.64)
Page 54 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. titqaddeš *!
b. <ti>təqaddeš ** *!
c. ə <t>hitqaddeš * *
Page 55 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
b.
Page 56 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew Imperative Paradigm
a. tikkaanes *!
b. <ti-k>kkaanes ***!
c. ə <t>thikkaanes * *
Page 57 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
3.5 CONCLUSION
Paradigms can be non-uniform in different ways. In this
chapter, we have seen differences within languages and across
two stages of a language.
Notes:
(1) A paradigm is usually named after the grammatical
category specifying all members in the paradigm (e.g. tense
paradigm, gender paradigm, etc.). Here I am concerned with
two categories only, future and imperative, and thus refer to
the paradigm using the derived form, the imperative.
Page 58 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
Page 59 of 60
Competing Principles of Paradigm Uniformity: Evidence from the Hebrew
Imperative Paradigm
Page 60 of 60
Sources of Paradigm Uniformity
DOI:10.1093/acprof:oso/9780199267712.003.0004
Page 1 of 80
Sources of Paradigm Uniformity
4.1 INTRODUCTION
Generative Phonology was only in its teens, when J. W. Harris
(1973) argued on the basis of facts like those in (1) that
Paradigm Uniformity (henceforth ‘PU’) must be recognized as
a principle of grammar.
(1)
Page 2 of 80
Sources of Paradigm Uniformity
(2)
a. √ n/a √
an(tágonis)tø
a.′ a(mérica) √ √ *
(nìstø)
b.(ònoma) √ n/a √
(tólogy)
Page 3 of 80
Sources of Paradigm Uniformity
b.′. √ √ *
phe(nòme)
(nólogy)
Page 4 of 80
Sources of Paradigm Uniformity
(3)
a. Exactly which other output is the source of
Output–Output Faithfulness for any form under
calculation?
b. How may the existence of Output–Output
Faithfulness (OO-F) constraints affect our view
of morphology? Specifically, what must we make
of the fact that the type of surface-to-surface
resemblances established by traditional
morphological relations are insufficient,
requiring the addition of OO-F?
c. May ‘cyclic’ derivations, once employed to
capture some transfer of phonological properties
(such as the one in (2b′)) still be viable within OT
after all as an alternative to OO-F, should
questions (3a, b) prove troublesome?
Page 6 of 80
Sources of Paradigm Uniformity
(4)
a.rémedyremé:di- * *
able
b.lévy lévi-able √ √
Page 7 of 80
Sources of Paradigm Uniformity
Page 8 of 80
Sources of Paradigm Uniformity
Page 9 of 80
Sources of Paradigm Uniformity
b. pʁɔ∫εn √ * √ √
Page 10 of 80
Sources of Paradigm Uniformity
Page 12 of 80
Sources of Paradigm Uniformity
Page 14 of 80
Sources of Paradigm Uniformity
(8)
Page 15 of 80
Sources of Paradigm Uniformity
b. nuvεl √ * √ √
Page 16 of 80
Sources of Paradigm Uniformity
The data in (8) are laid out identically to those in (5), but here
there is no dialectal variation, as the form that takes the final
C from the feminine and the preceding vowel from the
masculine:*nuvol, is not found. To make sense of this
divergence, we must revise the above discussion slightly. We
must assume that there are two simultaneous attraction
effects from the masculine allomorph (Base 1): one at the level
of the whole word, and one at the level of individual segments,
exactly along the lines of the Ganong effect. The word-level
effect is evidently by itself insufficient to impose the vowel of
the masculine, whence (8a) *nuvol, while the segment-level
attraction is absent in (8) because [o] is not a perceptual
neighbor of [ε]. FAITH-1(V-quality) is thus voided, as indicated
in (8) by the question mark, and consequently so is the
clustering of (8a). In contrast, in (5a) above the vowel [ε] of
feminine [pʁɔ∫εn] is a perceptual neighbor of the [ε̃] of the
masculine [pʁɔ∫ε̃n]. The resulting attraction by the masculine
form thus results in satisfaction of FAITH-1(V-quality) and the
clustering of (5a) above. Segment-level attraction cannot be
sufficient to turn [ε] to [ε̃], however, since it is not a general
property of French that [ε]s merge with their nasalized
counterparts. Hence the word-level attraction must therefore
also be at work, drawing the liaison form to the masculine
allomorph under gender identity. Thus it appears that the
purely distributional data of the French liaison effectively
reproduce the properties of the Ganong effect: a segment-
sized representation is influenced simultaneously by two types
of neighbors, segment-level and word-level.
Page 17 of 80
Sources of Paradigm Uniformity
Page 18 of 80
Sources of Paradigm Uniformity
The REH yields the attraction over distance effect of (7) in the
way illustrated in (10).
Page 19 of 80
Sources of Paradigm Uniformity
1 2 3 4
Varying R2: A A A ¬A
B B ¬B ¬B
C ¬C ¬C ¬C
¬D ¬D ¬D ¬D
Page 20 of 80
Sources of Paradigm Uniformity
In (11),
the R1 of
(10) is a
point in
Page 21 of 80
Sources of Paradigm Uniformity
(12)
A Basic meaning A A
B and segmentism B B
C Stress C ¬C
D Length of stressed V ¬D ¬D
(A, B, (A, B ⇒ D)
C ⇒ D)
Page 22 of 80
Sources of Paradigm Uniformity
(13)
Page 23 of 80
Sources of Paradigm Uniformity
a. rémediable * *
b. remédiable * *
c. ? remé:diable * *
Page 24 of 80
Sources of Paradigm Uniformity
(14)
Page 25 of 80
Sources of Paradigm Uniformity
a. ????? léviable *
b. lé:viable *
Page 26 of 80
Sources of Paradigm Uniformity
In sum, the REH (9) directly yields the attraction over distance
effect that we earlier (p.77) found to be pervasively present.
The latter is directly relatable to principles of neural
computation and in fact coincides with the claim that
representations are ‘distributed’. The reason is that taking an
entailment A ⇒ B due to a representation R1 to be relevant
also to some other representation R2 presupposes that the A
and B of R1 and those of R2 are the same entities, and
therefore that representations such as R1 and R2 are
‘distributed’ over a general pool of units that include A and B,
rather than being autonomous or local. The foregoing
discussion has provided a partial answer to question (3a)
above on the sources of OO-F effects: the latter effects obtain
between representations that are independently similar.
However, in the case of allomorphy, the independent similarity
has a specific source that we still need to deal with. For
example, in the case of remediable, the independent similarity
with remedy is partly a semantic fact: the intended meaning is
‘able to be remedied’, not ‘able to be read’, or other. But it is
also a morphological fact. It is a general property of the affix -
able that it attaches to verbs; hence the presence of -able
enforces some identity relation between its stem and some
verb in the lexicon. Our residual question is then how could
this be expressed within the framework of the REH (9), or to
put it differently, what is the role of morphology in a system
that sees words as related to one another ‘surface to surface’
in the ways prescribed by the REH. This is in fact question
(3b) above, and the topic of s. 4.6. Section 4.5 next takes us on
a brief detour to address some of the more pressing questions
for the REH.
Page 27 of 80
Sources of Paradigm Uniformity
A w, x A⇒B w ⇒ y; w ⇒
z
B y, z x ⇒ y; x ⇒ z
Page 29 of 80
Sources of Paradigm Uniformity
w⇒xy⇒z x ⇒ y; (x ⇒ z)2
Page 30 of 80
Sources of Paradigm Uniformity
Page 31 of 80
Sources of Paradigm Uniformity
Phonemes i n e p t
Time elapsed 0 1 2 3 4
Time Vector ¬1 1 1 1 1
¬2 ¬2 2 2 2
¬3 ¬3 ¬3 3 3
¬4 ¬4 ¬4 ¬4 4
Page 32 of 80
Sources of Paradigm Uniformity
Phonemes i n t
Page 33 of 80
Sources of Paradigm Uniformity
Page 34 of 80
Sources of Paradigm Uniformity
Phonemes k a t a b a
Major Class [+ +C -C +C -C +C -C
conson.]
Page 35 of 80
Sources of Paradigm Uniformity
(22)
Phonological alternations
a. compUlS-ive, der[i]vAT-ive
derivAT-ive
b. regrett-able, adm[i]r-able
INEVIT-able
Page 36 of 80
Sources of Paradigm Uniformity
Phonological alternations
d. permi[?]-ing
In (22), the suffixes -ive, -able, -ing all select verbs as their
stems, but rather than the forms in (22), the actual verbs are:
compel (a), der[ay] ve (a), adm[ay] re (b), and permi[t] (d). In
addition, there is no verb (in)evit (b), and regrettable (b) does
not have the expected meaning ‘able to be regretted’, but
rather refers to things that must be regretted instead.
Similarly, -ic selects nouns, but the actual nouns are rather
symptom and paras[ay]te instead of the forms in (22c). Such
distortions can be divided into the three categories of (22).
There are simply irregularities, both of form and meaning, as
in column (I), although irregularities may in turn correspond
to various degrees of sub-regularity, e.g.: symptomAt-ic,
problemAT-ic, axiomAT-ic, etc. Then there are regular
phonological effects, further subdivided into contrastive
(column (II)), and allophonic (column (III)). English vowel
shortening is a contrastive type of variation because it
neutralizes a contrast, that between short vowels and their
long or diphthongal counterparts. These contrast because they
have non-complementary distributions: [bayt]/ [bIt], etc.
Instead, American English flapping is allophonic as the
alveolar flap [?] is in complementary distribution with regular
alveolar stops [t/d]. Each individual suffix appears to control
the general degree to which its stem can fail to match an
independent word in the lexicon. The affixes that were once
classified as ‘Level 1’ in ‘Lexical Phonology’ (Kiparsky 1982a,
b) and exemplified in (22a, b, c) tolerate mismatches rather
well, hence featuring both sheer irregularities and regular
phonological alternations. In contrast, those that were
classified as Level 2, such as -less, -ness, -ful, and -ing of (22d)
tolerate few mismatches, essentially only those coming from
allophonic variation, such as the one shown in (22).
(23)
Page 37 of 80
Sources of Paradigm Uniformity
Page 39 of 80
Sources of Paradigm Uniformity
(24)
Page 40 of 80
Sources of Paradigm Uniformity
a. compell-ive *
b. ☞ compuls-ive *
a. ☞ compell-ing *
b. compuls-ing *
(25)
a. Phonological irregularity (Marked value,
English)
a.crime *
b.☞ cri:me *
a.cri:mine *
b. ☞ crimine *
Page 41 of 80
Sources of Paradigm Uniformity
That is to
say, in the
single
surface
form [?
mérIk?], some properties, like the specific choice of segments,
will result from input information. Let us call these ‘PI’. Other
properties on the other hand, like the stress and
syllabification, will result from Mark-edness constraints. Let us
call these properties ‘PM’. In the Lexicon Optimization
approach, the PI/PM distinction is not reflected by respective
distributions over different levels, but rather by constraint
ranking. For any aspect of the representation α, PI α obtains iff
IO-F(α) ≫ Markedness(α), while PM α obtains iff the opposite
ranking holds. Hence, the long vowel of cri:me (25) above is a
type of PI, while the short one of Italian crimine (25b) is a type
of PM. So, while under Lexicon Optimization the input to [?
mérIk?] is the same full-fledged [?mérIk?], only some if this
input—what I have called PI, is effectively ‘active’ in
determining the output. The rest of it is passive, under the
sway of the Markedness constraints—what I have called PM.
Page 42 of 80
Sources of Paradigm Uniformity
The new
set of
properties
PO will
represent
the aspects of the representation that are determined by OO-F
constraints by virtue of the constraint ranking. As in the case
of [ƏmérIkƏ], the input to compulsive can be taken to be the
exact same form, but only (p.87) some of it will be effectively
active. In (24) above this part of the input was given in caps.
The rest of the input conforms with OO-F, just as the stress of
[?mérIk?] conforms with Markedness. I return below to the
question of why only a fragment of the input /compUlS/ (the
portions in caps), is active in this fashion (i.e. corresponds to
PI of (27)). Now in (24b) above the same input compUlS- was
given for the form compelling, but in that case it was only as a
hypothetical input, to show that it would have no effect.
Page 43 of 80
Sources of Paradigm Uniformity
Page 44 of 80
Sources of Paradigm Uniformity
Page 45 of 80
Sources of Paradigm Uniformity
a. cri:minal cri:min-al *
b. ☞ criminal crimin-al *
Note that the case in (29) is not just an instance of the ‘tri-
syllabic’ shortening, which would naturally not be applicable
to cri:me. As argued in Burzio (1993, 2000a) and PES, tri-
syllabic shortening is a purely descriptive category, of no
theoretical significance. Shortening applies to all positions
within Level 1 items in principle (including final syllables:
infinite), the constraint at work being the maximally general
one of (29), which is independently motivated by the
difference between English and Italian of (25) above. The
English Level 1 lexicon is in fact very much like the whole of
Italian, favoring short vowels. Vowels may turn up long under
certain circumstances, however, in stressed open penultimates
in particular, as in desí:rous. This is because of competition
with the Metrical Consistency (MC) of (2) above. Given a
prohibition on stressed light penultimates (the foot *(L σ),
banned by the Metrical Well-Formedness of (2)), shortening
would force a violation of MC, as in desí:re/*désirous. Even this
effect is paralleled by Italian, where stressed open
Page 46 of 80
Sources of Paradigm Uniformity
(30)
Page 47 of 80
Sources of Paradigm Uniformity
Page 48 of 80
Sources of Paradigm Uniformity
(32)
Representation: A, Generalization:
¬B, C, D
Page 49 of 80
Sources of Paradigm Uniformity
Our discussion has thus far not accounted for the fact that
morphological idiosyncrasy within the Level 1 lexicon is not
totally unrestrained. If the dominance relation ‘IO-F ≫ OO-F’
postulated above were categorical, with an input
representation always prevailing over the demands of
morphology, then OO-F should have no effect at all, and any
resemblance between affixed stems and independent lexical
items should arise only by sheer accident, which is surely
incorrect. The remedy to this is in the fact that the basis to
OO-F constraints is not only in the SELECT entailments, which
may have a fixed rank, but also in the GA effect, which
modulates the overall OO-F effect in accordance with
representational distance. The revised prediction is, then, that
for regions that are very close to an attractor, OO-F may well
dominate IO-F even in the Level 1 domain, while for more
Page 50 of 80
Sources of Paradigm Uniformity
Page 51 of 80
Sources of Paradigm Uniformity
Page 52 of 80
Sources of Paradigm Uniformity
Page 53 of 80
Sources of Paradigm Uniformity
where X=V
X-A2 X ⇒ /___ A2
where X= /___
A2
am am á ⇒ / ___bamos
x00E1;-
bamos
Page 54 of 80
Sources of Paradigm Uniformity
Page 55 of 80
Sources of Paradigm Uniformity
Page 56 of 80
Sources of Paradigm Uniformity
Page 57 of 80
Sources of Paradigm Uniformity
Page 58 of 80
Sources of Paradigm Uniformity
(36)
a. -ist: Stem consistency prevails a(mérica)(n-
ìstø); (pròpa)(gánd-is)tø
b. -ic: ffix consistency prevails ti(tán-icø); *(títan-
i)cø
Page 60 of 80
Sources of Paradigm Uniformity
dv ènd-it-óre‘seller’ d′ v ìnc-it-óre‘winner’
Page 61 of 80
Sources of Paradigm Uniformity
Page 62 of 80
Sources of Paradigm Uniformity
(39)
Page 63 of 80
Sources of Paradigm Uniformity
Page 64 of 80
Sources of Paradigm Uniformity
Page 65 of 80
Sources of Paradigm Uniformity
1pl
2pl
Page 66 of 80
Sources of Paradigm Uniformity
(41)
Page 67 of 80
Sources of Paradigm Uniformity
Fewer Subjunctive
contrasts
English, OE,
Romance
Passive
Gothic
English, OE Past
Russian
adjectives,
OE nouns,
adjectives
Obliques
OE pronouns,
demonstrativ
es, nouns,
adjectives
Page 68 of 80
Sources of Paradigm Uniformity
…that he x00F8;
walk-
x00F8;
Page 69 of 80
Sources of Paradigm Uniformity
Page 70 of 80
Sources of Paradigm Uniformity
Indicative Subjunctive
B B B B
b. Added component: C C
[Subjunctive]
c. Alternating S S S S
component:
C⇒S C ⇒¬S
e. Leveling pressure 2 3
(contradictory
entailments)
Page 71 of 80
Sources of Paradigm Uniformity
Page 72 of 80
Sources of Paradigm Uniformity
Page 73 of 80
Sources of Paradigm Uniformity
4.8 CONCLUSION
Paradigm Uniformity can be dealt with in terms of the OT
notion of Output–Output faithfulness: members of a paradigm
are required to be faithful to other members of the same
paradigm. Yet this view raises a number of fundamental
questions, listed in (3) above, which the present chapter has
sought to answer.
Page 74 of 80
Sources of Paradigm Uniformity
Notes:
(1) To the limited extent that nouns also form inflectional
paradigms, these are also uniform, as Harris had noted: señór/
señóres ‘gentleman/gentlemen’, etc.
Page 75 of 80
Sources of Paradigm Uniformity
(2) It also seems less than an accident from this point of view
that the only sequentially discontinuous morpheme in English
is constituted of segmentally identical parts as in en-light-en,
etc.
((i))
Page 76 of 80
Sources of Paradigm Uniformity
Page 77 of 80
Sources of Paradigm Uniformity
(6) This abstracts away from the possibility that either affixed
form in (34) may be more prominent than the other perhaps
for semantic reasons, or for reasons related to frequency of
use (Bybee 1995).
Page 78 of 80
Sources of Paradigm Uniformity
Page 79 of 80
Sources of Paradigm Uniformity
Page 80 of 80
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
DOI:10.1093/acprof:oso/9780199267712.003.0005
Page 1 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
5.1 INTRODUCTION
In an influential paper, Steriade (2000) argues that
noncontrastive details are relevant for the evaluation of
paradigm uniformity, thus showing the role of noncontrastive
phonetic properties on phonological patterning. As Steriade
(2000: 314) says, her argument is part of ‘a larger agenda’
that claims that ‘the distinction between phonetics and
phonological features is not conducive to progress and cannot
be coherently enforced’. This is reflective of a functional
approach to phonology developed in the works of the UCLA
group (Flemming 1995; Jun 1995; Kirchner 1997; Silverman
1997; Steriade 1999b; Zhang 2000; etc.) which maintains that
phonology is reducible to general properties of human motor
behavior and perception, and this entails explic- (p.108) itly
incorporating low-level phonetic machinery into the
synchronic phonological grammar. Such a position can be
contrasted with the one espoused by Ladefoged (1983) who
specifically emphasizes the importance of abstract units in
phonology such as the phoneme commenting that ‘… some of
the really interesting linguistic phenomena are precisely those
for which there are no general phonetic explanation’ (ibid. 8).
Page 3 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
a. [t Mèditerránean Nàvratilóva
h
]
Page 4 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
Page 5 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
Further, van Dam and Weaver report that the bolded stops in
(2) are also more aspirated than the bolded stops in (5) where
the stop is at the beginning of the last syllable of a word-final
dactyl.
Page 6 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
Page 7 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
Page 8 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
The first two columns in (6) show that voiceless stops are
aspirated word-initially whether at the beginning of a stressed
syllable or a stressless one. The last two columns in (6) show
that a voiceless stop is aspirated at the beginning of a non-
initial stressed syllable, be it a syllable with primary stress or
secondary stress. The voiceless stop which begins the third
syllable of a word like Mèditerránean in (1) does not seem to
match any of the environments in (6), since it is neither word-
initial nor at the onset of a stressed syllable. In fact, it appears
to be at the onset of a syllable with a reduced vowel.
(p.112)
In (7), the
initial
dactylic
sequence
comprises
a
superfoot. This reflects the analysis of English found in Hayes
(1981) and McCarthy (1982) searchers assumed a similarity in
which a superfoot consists of a binary foot followed by an
adjoined single syllable. These researchers assumed a
similarity of environments between a word-final dactylic
sequence as exemplified by the trisyllabic words cápital and
párity and nonfinal dactylic sequences as in the first three
syllables of Wìnnepesáukee. The last syllable of a word-final
dactylic sequence behaves as foot-internal under the view that
flapping occurs in a foot-internal environment, as opposed to
aspiration, which is foot-initial (cf. Kiparksy 1979). The fact
that the /t/ in cápital and párity is flapped constitutes evidence
of a foot-internal environment. Researchers such as Hayes
Page 9 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
An alternative foot structure that I will argue for and that does
account for the aspiration in the third syllable of
Wìnnepesáukee is shown in (8). This involves a superfoot as
well, but different from that in (7). Here, the third syllable of
the initial dactylic sequence is adjoined as the first syllable of
a superfoot and does not group with the preceding syllables.
That is, the third syllable of the initial (nonfinal) dactylic
sequence forms a constituent with what follows rather than
with what precedes as in (7).
(8)
The foot
structure
in (8) for
nonfinal
dactylic
sequences
was
originally proposed by Withgott (1982) but was also suggested
in a footnote by McCarthy (1982). More recently, this foot
structure has been adopted by Jensen (2000) and Davis and
Cho (2003), and is suggested by Pater (2000). One immediate
advantage of this footing is that it makes for a unified
statement regarding American English aspiration (as noted by
Jensen 2000): voiceless stops are aspirated in foot-initial
position. This is what unifies the aspiration environments in
(6) with that shown in (1). We can illustrate this with the
convenient word pèripatétic in (9), where each foot-initial
position consists of a voiceless stop.
Page 10 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
(p.113)
(9) Peripatetic
Further
evidence
for the
superfoot
structure
comes
from
words like potáto and other words such as those in the second
column of (6) which begin with an initial stressless syllable.
When such syllables have an initial voiceless stop they surface
as aspirated. This can be accounted for by the foot structure
shown in (10) where the initial syllable forms the first syllable
of a superfoot.
(10) potáto
(11)
Page 11 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
(12)
The /t/ at
the
beginning
of the
third
syllable is
aspirated,
as expected, because it is foot-initial (that is, at the beginning
of the superfoot). Moreover, as indicated in (13), expletive
infixation shows variation with this word.
Page 12 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
(14) cápital
Page 13 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
This foot
structure
readily
accounts
for the
flapping of
the /t/ on the view that flapping occurs foot internally (in an
intervocalic environment). Even if one maintains that final
syllables in English nouns are extrametrical, the foot structure
in (14) would reflect the surface foot structure of the word
(and not necessarily the initial foot structure under a
derivational view of phonology). As mentioned earlier, an
alternative to the foot structure in (14), found in Hayes (1981),
is to assume that the final syllable in (14) is adjoined to the
end of a superfoot in a way identical to what is shown for the
first three syllables in (7). While such a structure would still
allow for the expression of the flapping environment to be
foot-internal, the foot structure would not be independently
motivated since there do not seem to be other patterns that
would require an adjoined syllable at the end of a superfoot.
This should be contrasted with the superfoot type proposed in
this chapter in which the adjoined syllable is at the beginning
of the superfoot as seen in (8) and (9). As argued, the same
superfoot structure occurs in words like that in (10).
Consequently, I would maintain that the only superfoot
structure allowed in English is one that begins with an
adjoined syllable as illustrated in (8)–(10).
(15) cápitalistic
Page 14 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
(p.116)
The
flapping of
the /t/ in
the third
syllable of
càpitalístic is a consequence of the uniformity of foot structure
with cápital. As we have seen from data like those in (1), the
beginning of the third syllable of càpitalístic is not an
environment where one would otherwise expect flapping.
(16)
a. capi[?]a-fucking-listic
b. *capi-fucking-[?]alistic
Page 15 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
Page 16 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
a. pay *!
b. ? *
h
p ay
a. *! *
h
b ay
b. ? *
bay
Page 17 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
a. * *!
h
læp s
b. ? *
læps
c. *! *
h
l æps
b. (phàe.r1){ p∂ *! **
(thέ. ?1k)}
Page 18 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
this criticism does not hold. First, Jensen’s statement that ‘we
cannot require that all feet begin with an aspirated stop, since
other segment types must be permitted in that position’
ignores the fact that the constraint requiring all feet to begin
with the feature [sg] can be dominated. This is shown by the
ranking of *[SG, +VOICE] over ALIGNL(FOOT, [SG]), as
exemplified in the tableau in (20). With this ranking a voiced
sound cannot be aspirated even if it is in foot-initial position.
a. pay
b. phay
Page 19 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
a. (mìli) *! *
h
(t à)
(rístic)
Page 20 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
b. ? (mìli) *
h
{t a(rístic)}
c. (mìlita) * *!
(rístic)
(26) cápitalistic
Page 21 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
a. ? *
(càpita)
(lístic)
b. (càpi) *!
h
{t a(lístic)}
c. (càpi) *! * *
h
(t à)
(lístic)
Page 22 of 23
Capitalistic v. Militaristic: The Paradigm Uniformity Effect Reconsidered
Page 23 of 23
Jita Causative Doubling Provides Optimal Paradigms
DOI:10.1093/acprof:oso/9780199267712.003.0006
6.1 Introduction
Page 1 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 2 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 3 of 46
Jita Causative Doubling Provides Optimal Paradigms
(1) Jita Causative (Downing 1996, 2001, and unpublished field notes; causative is bolded; ‘=’ indicates the
stem boundary)
Page 4 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 5 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 6 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 7 of 46
Jita Causative Doubling Provides Optimal Paradigms
(3) Bemba causative doubling (Hyman 1994, figs. (9–10), pp. 85–6)
Page 8 of 46
Jita Causative Doubling Provides Optimal Paradigms
(4) Kinande causative displacement (Mutaka and Hyman 1990; Mutaka 1994)
Page 9 of 46
Jita Causative Doubling Provides Optimal Paradigms
OR
b.
Page 10 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 11 of 46
Jita Causative Doubling Provides Optimal Paradigms
Cycle 1
Page 12 of 46
Jita Causative Doubling Provides Optimal Paradigms
morphology gur-i
phonology
displacement/copy N/A
spirantization + gus-i
Cycle 2
morphology gus-i-ir
phonology
displacement/copy gus-i-ir-i
spirantization + gus-i:s-i
Cycle 3
morphology gus-i:s-i-an-
phonology
displacement/copy gus-i:s-i-an-i
spirantization + gus-i:s-i-an-i-
gliding gus-i:s-y-a:n-y-
Page 13 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 14 of 46
Jita Causative Doubling Provides Optimal Paradigms
b.
Page 15 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 16 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 17 of 46
Jita Causative Doubling Provides Optimal Paradigms
a. F =gus-]y]-a *
b. =gur-]y]-a *!
(15)
Page 18 of 46
Jita Causative Doubling Provides Optimal Paradigms
(16) *Jita
Page 19 of 46
Jita Causative Doubling Provides Optimal Paradigms
a. F =gur-an]y]-a *
b. =gus-y-an]y]-a * *! *
c. =gus-y-an]a *! *
d. =gur-y-an]y]-a *! * *
Crucial rankings:
*ry >> MAX/DEP [CONT] optimizes spirantization to satisfy markedness
ALIGN /y/ >> CONTIGUITY optimizes causative /y/ displacement
Page 20 of 46
Jita Causative Doubling Provides Optimal Paradigms
(p.133)
(18)
Page 21 of 46
Jita Causative Doubling Provides Optimal Paradigms
a. F *
oku=gus]y]-a,
oku=gus-y- * * *
an]y]-a
b. *
oku=gus]y]-a,
oku=gusy-an]-a *! *
c. oku=gur] *!
y]-a,
oku=gury-an]y]- * * *
a
d. *! *
oku=gus]y]-a,
oku=gur-an]y]-a *
Crucial constraint rankings, same as (16), plus:
OP-CS >> DEP /y/ optimizes doubling to satisfy paradigm uniformity
Page 22 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 23 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 24 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 26 of 46
Jita Causative Doubling Provides Optimal Paradigms
(21)
Page 27 of 46
Jita Causative Doubling Provides Optimal Paradigms
a. oku=gur-a, *
oku=gus]y]-a, *
oku=gus-y-an] *! * *
y] -a
b. oku=gur-a,
oku=gur]y]-a, *
oku=gur-y-an] - *! *
a
c. oku=gur-a,
oku=gur]y]-a, *!
oku=gur-an]y]-a *
d. F oku=gur-
a,
oku=gus]y]-a, * *
oku=gur-an]y]-a *
Page 28 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 29 of 46
Jita Causative Doubling Provides Optimal Paradigms
BUT
Page 30 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 32 of 46
Jita Causative Doubling Provides Optimal Paradigms
(23)
Page 34 of 46
Jita Causative Doubling Provides Optimal Paradigms
a. F oku=gur- *
a,
oku=GUS]y]- *
a,
oku= GUS-y- * * *
an]y]-a
b.
oku=GUR-
a,
oku=GUR]y]- *!
a,
oku= GUR-y- * * *
an]y]-a
c.
oku=GUR-
a,
oku=GUS]y]- * *
a,
Page 35 of 46
Jita Causative Doubling Provides Optimal Paradigms
oku= GUR- *! *
an]y]-a
d. *!
oku=GUS-
a,
oku=GUS]y]- *
a,
oku= GUS-y- * * *
an]y]-a
Page 36 of 46
Jita Causative Doubling Provides Optimal Paradigms
Crucial constraint
rankings:
Page 37 of 46
Jita Causative Doubling Provides Optimal Paradigms
eni=βon-y-es-y-a ‘I am showing’
Page 38 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 39 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 40 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 41 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 42 of 46
Jita Causative Doubling Provides Optimal Paradigms
6.4 Conclusion
To sum up, I have shown that Causative Doubling is best
accounted for as a form of Output–Output correspondence
enforcing Optimal Paradigms (McCarthy, ch. 8 below). The
analysis supports the constellation approach to paradigms that
defines bases as the phonological strings shared by a set of
morphologically related words. As we have seen, the shared
string for doubling is not a well-formed morphological
subconstituent of the words in the paradigm, instead it is a
morpho-prosodic constituent, the Causative Stem. It is also not
an independent word, but instead a bound stem. Even though
McCarthy (Ch. 8 below) has suggested that TCT, rather than
OP, (p.144) better accounts for derivational paradigms, these
two properties make doubling, a derivational process,
unformalizable in the TCT theory of Output–Output
correspondence. As I have shown, though, there are good
reasons also to adopt OP instead of TCT to account for
derivational paradigms. The priority of the base requirement
characteristic of derivational paradigms can fall out from
appropriate constraint rankings in OP. The TCT formalism is
not needed to express this restriction. Further, when
derivational paradigms are embedded in a highly inflectional
system, like that found in Bantu languages, TCT cannot
maintain its claim that each paradigm member is both a fully
inflected word and separated by only one morphological
operation from its base. Finally, I have shown that OP provides
Page 43 of 46
Jita Causative Doubling Provides Optimal Paradigms
Notes:
(1) Notice that the causative is realized as half a long vowel
before an [i]-initial morpheme like the applicative /-ir-/ and as
a glide (triggering palatalization or spirantization of the
preceding consonant) before other vowels. This falls out from
the standard assumption that [i] and [y] have identical
features, and the realization as a vowel or a glide is
determined by syllable structure. Note, too, that vowels are
regularly (compensatorily) lengthened following glides in Jita,
except word-finally. The orthographic sequence ‘ny’ in the Jita
data is realized as a palatal nasal segment, and compensatory
lengthening is not found following this nasal. See Downing
(1996) for discussion and analysis.
Page 44 of 46
Jita Causative Doubling Provides Optimal Paradigms
(6) Space does not permit reiterating all the arguments for
PStem as an important domain for phonological processes and
phonotactic generalizations here. The interested reader should
consult Downing (2001), Hyman (1998), and Hyman and
Inkelas (1997) for discussion.
(11) The Jita tone patterns cited in (25) and (26) abstract away
from processes of final contouring and tone shift for ease of
exposition. See Downing (1996) for detailed discussion of Jita
tone. Enclitics like /-mó/ are not included in MWord. For this
reason, High tone sequences are tolerated over the MWord-
Enclitic boundary. However, they are included in PWord and
count as the final element in PWord.
Page 45 of 46
Jita Causative Doubling Provides Optimal Paradigms
Page 46 of 46
Paradigmatic Uniformity and Contrast
DOI:10.1093/acprof:oso/9780199267712.003.0007
Page 1 of 45
Paradigmatic Uniformity and Contrast
Page 2 of 45
Paradigmatic Uniformity and Contrast
7.1 INTRODUCTION
The general research question that this chapter addresses is
the proper treatment of cases of opacity in which the
triggering or blocking context for a phonological process is
found in a paradigmatically related word. Chomsky and Halle’s
(1968) discussion of the minimal pair comp[∂]nsation v.
cond[ε]nsation is a classic example of the problem. In general,
the contrast between a full vowel v. schwa is predictable in
Eng- (p.146) lish as a function of stress; but comp[∂]nsation
v. cond[ε]nsation have the same σ̀σσ̀σ stress contour and thus
raise the question whether English schwa is phonemic after
all. SPE’s insight was that the morphological bases from which
these words are derived provide a solution to the problem:
comp[∂]nsate has a schwa while cond[ε]nse has a full stressed
vowel. Chomsky and Halle’s suggestion is that such
paradigmatic relations among words can be described by
embedding the derivation of one inside the derivation of the
other. In other words, the derivation of complex words
proceeds cyclically with the morphological base forming an
intermediate stage /kómpensé:t/ v. /kóndéns/ over which stress
and vowel reduction are computed and relies on the model’s
serial architecture for subsequent destressing processes to
obscure the context for vowel reduction. The cycle became the
basic (and essentially only) tool for describing paradigmatic
phonological relations in the generative framework. It
received considerable theoretical development in the Lexical
Phonology model Kiparsky (1982a, 1985). While many
languages have been successfully described in cyclic terms
(see Cole 1995 for a recent review), cases were discovered
where the cycle does not provide the proper coverage or
elucidation (e.g. S. Chung 1983; see also Crosswhite (1997).
Page 3 of 45
Paradigmatic Uniformity and Contrast
Page 4 of 45
Paradigmatic Uniformity and Contrast
(1)
Page 5 of 45
Paradigmatic Uniformity and Contrast
(p.148)
(2)
a. /amor-sit≈-it-o/
CORR-Σ-ROLE
a. ?a.mor.-si.t-o
b.a.mo.r-i.t-o *!
cf. a.mor
b. /koron-sit≈-it-a/
CORR-Σ-ROLE
a.koron-sit-a *!
b. ?koron-it-a
cf. ko.ro.n-a
(3)
(4)
Page 6 of 45
Paradigmatic Uniformity and Contrast
(5)
Page 7 of 45
Paradigmatic Uniformity and Contrast
a. ? {ra.ton.- *
si.t-o, ra.ton.-
si.t-a}
b. {ra.to.n-i.t- * *!*
o, ra.to.n-i.t-
a}
c. {ra.ton.- *!
si.t-o, ra.to.n-
i.t-a}
Page 8 of 45
Paradigmatic Uniformity and Contrast
Page 9 of 45
Paradigmatic Uniformity and Contrast
(6)
Page 10 of 45
Paradigmatic Uniformity and Contrast
Page 11 of 45
Paradigmatic Uniformity and Contrast
(7)
Page 12 of 45
Paradigmatic Uniformity and Contrast
Page 13 of 45
Paradigmatic Uniformity and Contrast
(8)
Page 14 of 45
Paradigmatic Uniformity and Contrast
a.kúr’at *!
b. kur’át *!
c ?kúr’ut *
d.kúr’it *!
Page 15 of 45
Paradigmatic Uniformity and Contrast
(9)
(p.153)
Page 16 of 45
Paradigmatic Uniformity and Contrast
(10)
? zórn-o *
zórn-a *!
kapít-o *!
? kapít-a
(11)
infinitive perfect
Page 17 of 45
Paradigmatic Uniformity and Contrast
(12)
ll *!
ll *! *
ll *! * *
? ll **
(13)
infinitive perfect
Page 18 of 45
Paradigmatic Uniformity and Contrast
Thus in the perfect form kokos-el-e from /kokot-i:l -e/ the /t/
has mutated to /s/ and the vowel of the perfect suffix /-i:l/ is
shortened and lowered to mid and thus appears as /-el/. The
perfect form is still distinct from the applied because the latter
does not mutate the final consonant of the stem; also the final
vowels are distinct. In the applied perfect /kokot-il-i:l-e/ kokot-
el-e:l-e the vowel length of the perfect suffix is retained since it
does not immediately follow a mutation site. Both suffixes
lower their vowels as a function of the mid vowel of the root.
Finally, the lateral of the applied suffix switches to plain /l/
before the perfect suffix whose lateral in turn harmonizes with
the preceding base.4
(14)
infinitive perfect
la:l-il-e
(15)
infinitive perfect
Page 19 of 45
Paradigmatic Uniformity and Contrast
(16)
infinitive perfect
(17)
? sulili:le *
Page 20 of 45
Paradigmatic Uniformity and Contrast
suli:le *!
molele:le *!
? mole:le
(18)
Infinitive perfect
They also fail to ablaut in the applied perfect since if they did
the result would be identical to the basic perfect. Thus, while
the regular verbs in (15) such as ku-mo:l-a mutate in the
perfect (mo:z-el-e) and ablaut in the applied perfect (mol-e:l-e/
mol-el-e:l-e/), the irregular verbs of (18) such as x-sajil-a are
exceptions to mutation (cf. perfect sajil-i:l-e); the latter fact
explains the corresponding absence of ablaut in the applied
perfect form sajil-il-i:l-e(*sajil-i:l-e).
Page 21 of 45
Paradigmatic Uniformity and Contrast
Page 22 of 45
Paradigmatic Uniformity and Contrast
(19)
Page 23 of 45
Paradigmatic Uniformity and Contrast
starts with a vowel, stress appears on the first syllable and the
second stem vowel is elided instead: /seme?-u/ → sém?-u. Both
schwas surface in the 3 sg. masc. /seme?/ → séme?:, the first
because it is stressed and the second because the syllable is
closed.
(20)
a. FOOT-FORM: metrical feet are bimoraic
trochees (Hayes 1985)
b. NON-FINALITY: penalize a foot aligned with
the right edge of the Prosodic Word (Prince and
Smolensky 1993)
(p.158)
c. LAPSE: penalize a sequence of two unparsed
syllables (Selkirk 1984)
d. *e. : penalize an unstressed syllable-final
schwa
(21)
Page 24 of 45
Paradigmatic Uniformity and Contrast
? (?ál)lamu *
(?álla)mu *!
?al(lámu) *!
(séme)?u *!
? (sém)?u
se(mé?u) *! *
S(mé?u) *!
se(mé?)na *!
? s(mé?)na
Page 25 of 45
Paradigmatic Uniformity and Contrast
The paradigms in (22) show the 3 sg. bases ?állam ‘he taught’
and ?állam-et ‘she taught’ and the 1 sg. ?allám-t‘I taught’
amplified with the suffixes marking object inflection.
(22)
Page 26 of 45
Paradigmatic Uniformity and Contrast
fem. form remains distinct from the 1 sg. form at the cost of an
otherwise anomalous violation of antepenultimate stress.
(23)
?al(lámet)o *!
?al(lám)to *!
?? *
alla(méto)
(24)
Page 27 of 45
Paradigmatic Uniformity and Contrast
‘hit’ ‘see’
(25)
Page 28 of 45
Paradigmatic Uniformity and Contrast
(26)
(Dár)beto *!
? (Dáreb)to
Dar(béto) *!
(∫áa)feto *!
? (∫áaf)to
∫aa(féto) *!
(27)
Page 29 of 45
Paradigmatic Uniformity and Contrast
The singular feminine suffix /-a/ takes the allomorph /-et/ when
followed by a pronominal suffix or a governed NP—the so-
called iDafa or Construct State construction: Dáarb-a ‘her
hitting’ but Daarb-ét#ni ‘her hitting me’, Dáarb-it Faríid ‘her
hitting Fariid’. This construct suffix is equivalent in underlying
shape to the 3 sg. feminine suffix of the perfect: /-et/. Both
share the formal features [+fem, –pl]. But the phonological
behavior of the construct suffix is quite different—it is
essentially regular. Before consonant-initial object suffixes the
construct state suffix is stressed while before the vowel-initial
ones it syncopates—regardless of the shape of the preceding
stem: cf. /Daareb-et#o/ → Dáar e b-t#o ‘her hitting him’ v. /?
allam-et#o/ → ?allam-ét#o ‘she taught him’.7 Why this
difference in the behavior of the same suffix in essentially the
same phonological context? The notion of paradigmatic
contrast provides an attractive explanation. There is no other
member of the participial paradigm competing for the same
phonetic output. Hence the regular phonology can have its
way.
Page 30 of 45
Paradigmatic Uniformity and Contrast
Page 31 of 45
Paradigmatic Uniformity and Contrast
7.3.5.1 Oujda
The perception of stress in Moroccan Arabic is quite subtle
and largely remains a question for future research.
Fortunately, the location of accent does not bear on the gen-
(p.163) eral points we wish to make and so we do not
attempt to transcribe it here. In our transcriptions schwa is
indicated by e; we abstract away from the coloring induced by
neighboring consonants. In (28) we see the perfect tense
paradigm for the reflex of a Measure 1 verb. An exceptionless
ban on open-syllable schwa accounts for the realization of
CeCeC as CCeC. This constraint is also responsible for the
syllable reshuffling found in the stem when the following suffix
starts with a vowel.
(28)
2 sg. m. rfed-t
2sg. f. rfed-ti
3sg. m. rfed
3 sg. f. refd-et
1pl. rfed-na
2pl. rfed-tu
Page 32 of 45
Paradigmatic Uniformity and Contrast
3pl. refd-u
(29)
Page 33 of 45
Paradigmatic Uniformity and Contrast
(p.164)
(30)
(31)
(32)
rfedetu *!*
Page 34 of 45
Paradigmatic Uniformity and Contrast
refdetu *!
rfedtu *!
? rfedatu *
∫afetu *!
? ∫aftu
∫afatu *!
(33)
Page 35 of 45
Paradigmatic Uniformity and Contrast
2. sg. f. ðreb-ti
3 sg. m. ðreb
3 sg. f. ðerb-et
1 pl. ðreb-na
2 pl. ðreb-tu
3 pl. ðerb-u
3 sg. f. ?ellm-et
1 sg. ?ellem-t
3 sg. f. ∫af-et
1 sg. ∫ef-t
Page 36 of 45
Paradigmatic Uniformity and Contrast
(34)
2 sg. m. fihím-t
2sg. f. fihím-ti
3sg. m. fíhim
3sg. f. fíhm-it
1pl. fihím-na
2pl. fihím-tu
3pl. fíhm-u
(35)
Page 37 of 45
Paradigmatic Uniformity and Contrast
(36)
Page 38 of 45
Paradigmatic Uniformity and Contrast
(37)
Page 39 of 45
Paradigmatic Uniformity and Contrast
Page 40 of 45
Paradigmatic Uniformity and Contrast
Page 41 of 45
Paradigmatic Uniformity and Contrast
Page 42 of 45
Paradigmatic Uniformity and Contrast
Notes:
(1) There are also cases where the competition is resolved in
favor of syllable matching: cf. aleman, aleman-a ‘German’ and
the corresponding diminutives aleman-sit-o, aleman-it-a; also
the personal names Ramon, Ramon-a and Ramon-sit-o, Ramon-
it-a. These appear to be in the minority.
Page 43 of 45
Paradigmatic Uniformity and Contrast
(9) The survey was conducted in the spring of 2001 while the
author was a Visiting Scholar at the Institut de Linguistique et
Phonétique Générales et Appliquées in Paris. I wish to thank
Fatiha Abdulaziz, Nassira Betar, Samira Boumelassa,
Mohamed Elmedlaoui, Abdi Fassi-Fehri, Lilia Ould-Younes,
Karim Shoul, Abderrazzak Tourabi, Mohamed Yeou, and
Chakir Zeroual.
Page 44 of 45
Paradigmatic Uniformity and Contrast
Page 45 of 45
Optimal Paradigms
Optimal Paradigms
John J. McCarthy
DOI:10.1093/acprof:oso/9780199267712.003.0008
Page 1 of 71
Optimal Paradigms
8.1 Introduction
Traditional ideas of analogy, SPE’s phonological cycle, and
recent proposals about Output–Output faithfulness all have the
same goal: to account for surface resemblances among
morphologically related words. For instance, the trisyllabic
participial form lightening has a syllabic n because of its
relationship to the verb lighten—a relationship not shared with
the disyllabic noun lightning. Phenomena like this have figured
prominently in discussions of analogy and its successors.
Page 2 of 71
Optimal Paradigms
Page 3 of 71
Optimal Paradigms
(p.172)
(1)
a. UNIFORM EXPONENCE (Kenstowicz 1996)
Minimize the differences in the realization of a lexical
item (morpheme, stem, affix, word).
b. METRICAL CONSISTENCY (Burzio 1994a: 228)
Every morpheme must be as metrically consistent as
possible.
Page 4 of 71
Optimal Paradigms
Page 5 of 71
Optimal Paradigms
(2) OP in Outline
a. Candidates consist of entire inflectional
paradigms, where an inflectional paradigm
contains all and only the words based on a single
lexeme (for similar ideas, see Bonet and Lloret
2001; Kenstowicz 1996: 385; McCarthy 1998;
Raffelsiefen 1995, 1999c ; Tesar and Smolensky
2000).2
b. Markedness and Input–Output faithfulness
constraints evaluate all members of the
candidate paradigm. The violation-marks
incurred by each paradigm member are added
to those incurred by all the others.3
c. The stem (output form of the shared lexeme)
in each paradigm member is in a
correspondence relation ℜop with the stem in
every other paradigm member. (That is, for
every candidate paradigm P there is a relation
ℜop on P×P.) There is no distinctive base—rather,
every member of a paradigm is a base of sorts
with respect to every other member.4
(p.174)
d. There is a set of Output–Output faithfulness
constraints on the ℜop correspondence relation.
Page 6 of 71
Optimal Paradigms
As we will see in greater detail in ss. 8.4 and 8.5, this model
predicts certain interactional patterns that set it apart from
other approaches, particularly TCT. One pattern is
overapplication-only. Given a language with a general process
of coronal palatalization, there are two ways to level
alternations within the paradigm of /mat/: <mat∫, mat∫i> or
<mat, mati>. The first of these paradigms shows
overapplication of the palatalization process—there is
palatalization of /t/ even in the unsuffixed form, where the
conditioning i is absent. The second paradigm shows
underapplication of palatalization: the process is blocked in
the suffixed form mati because there is no palatalization in the
unsuffixed form mat. OP-IDENT(high) is satisfied either way,
but the paradigm with underapplication cannot be obtained in
the OP model. (Some care is required in defining what over-
and underapplication mean in the context of a (p.175)
constraint-based theory like OT, so this statement should not
be applied indiscriminately. See s. 8.5.2.)
instead, see ss. 8.4.2 and 8.5.2. For a general evaluation of the
overapplication-only hypothesis, see s. 8.5.2. And for the
reduplicative parallel, upon which this argument is based, see
McCarthy and Prince (1995, 1999).
Page 8 of 71
Optimal Paradigms
Page 9 of 71
Optimal Paradigms
Over the years, there has been gradual progress toward this
goal. Work started with the CV-template, which was applied to
root-and-pattern morphology (McCarthy 1981) and to
reduplication (Marantz 1982). This was later generalized to
incorporate syllabic information (Levin 1983) and prosodic
structure generally (McCarthy and Prince 1986/1996), leading
to the hypotheses in (3).
Page 10 of 71
Optimal Paradigms
Page 11 of 71
Optimal Paradigms
Page 12 of 71
Optimal Paradigms
The template of templates also says that verb stems can begin
with [CV or [CCV: faʕal, faʕʕal v. ftaʕal, stafʕal, etc. But the
stems of Arabic nouns (except for obviously deverbal nouns)
always begin with a single consonant: faʕal, faʕi:l, faʕl, etc. In
this case, it is the nouns, rather than the verbs, that are
subject to the more stringent requirement.
Singular -u ‘nominative’
-i ‘genitive’
-a ‘accusative’
-aj ‘gen./acc.’
Page 13 of 71
Optimal Paradigms
a. Perfective
C-initial suffixes
V-initial suffixes
b. Imperfective indicative10
CV prefixes
C-initial suffix
Page 14 of 71
Optimal Paradigms
V-initial suffixes
-u ‘1st and 3rd sg. com., 2nd sg. Masc., 1st pl.
com.’
-a:ni ‘du.’
Page 15 of 71
Optimal Paradigms
Page 16 of 71
Optimal Paradigms
Page 17 of 71
Optimal Paradigms
fin.na:.si a.bul.wa.zi:.ri
b. Epenthesis
qa:.la.tis.mat⊕ mu.ham.ma.du.nin.na.bij.ju
Page 18 of 71
Optimal Paradigms
a. ☞ a.bul.wa.zi:.ri *
b. a.builµ.wa.zi:.ri *!
c. a.bu:lσ.wa.zi:.ri *!
(p.184)
a. ☞ a.bul.wa.zi:.ri *
b. a.bu:.li.wa.zi:.ri *!
b. a.bu:.wa.zi:.ri *!
Page 19 of 71
Optimal Paradigms
a. ☞ qa:.la.tis.matf *
b. qa:.latsµ.matf *!
c. qa:.latsσ.matf *!
d. qa:.lat.matf *!
Page 20 of 71
Optimal Paradigms
(p.185)
Page 21 of 71
Optimal Paradigms
a. ☞ <faʕala, faʕaltu, **
…>
b. <faʕa:la, faʕa:lσtu, *!
…>
c. <faʕa:la, faʕa:lµtu, *!
…>
d. <faʕa:la, faʕaltu, … *! *
>
Page 22 of 71
Optimal Paradigms
The tableau at (13) makes the same point for the matched pair
of faithfulness constraints OP-DEP-V and IO-DEP-V. Candidates
(13b) and (13c) have the same markedness problems that
afflict (12b) and (12c). In candidate (13d), there is vowel
epenthesis to relieve the forbidden triconsonantal cluster. But
this leads to an intraparadigmatic (p.186) vowel/zero
alternation: faʕla ℜop faʕiltu. This alternation violates OP-DEP-
V (or, symmetrically, OP-MAX-V). In (13a), epenthesis
metastasizes throughout the paradigm, even in forms where it
Page 23 of 71
Optimal Paradigms
Page 24 of 71
Optimal Paradigms
a. ☞ <faʕila, faʕiltu, … **
>
Page 25 of 71
Optimal Paradigms
Page 26 of 71
Optimal Paradigms
Page 27 of 71
Optimal Paradigms
Page 28 of 71
Optimal Paradigms
a. After pause
ʔif.ʕa:.lu
…Cif.ʕa:lu
…Vf.ʕa:lu
ALIGN-L(Stem, σ) IO-DEP-
V
a. ☞ <.fiʕa:lu, .fiʕa:la, **
…>
Page 29 of 71
Optimal Paradigms
ALIGN-L(Stem, σ) IO-DEP-
V
Page 30 of 71
Optimal Paradigms
a. ☞<s.taf.ʕa.la, **
jas.tafʕilu, … >
b. <.sitafʕala, jas.taf *! *
ʕilu, … >
Page 31 of 71
Optimal Paradigms
IO-MAX-V SWP
Page 32 of 71
Optimal Paradigms
Page 33 of 71
Optimal Paradigms
(19)
Page 34 of 71
Optimal Paradigms
(20)
a. ☞ <.(fíʕa)lu, .(fíʕa)la, … **
>
Page 35 of 71
Optimal Paradigms
(p.194)
(21)
The lines
indicate
proven
constraint
Page 36 of 71
Optimal Paradigms
Page 37 of 71
Optimal Paradigms
a. Ⅾ <(fáʕi)la, fa(ʕìl)tu, *
…>
b. <(fáʕ)la, (fáʕlσ)tu, *!
…>
c. <(fáʕ)la, (fáʕlµ)tu, *!
…>
d. <(fáʕ)la, fa(ʕíl)tu, *!
…>
Page 38 of 71
Optimal Paradigms
Page 39 of 71
Optimal Paradigms
a. ☞ <(fáʕa)la, *
fa(ʕál)tu, … >
b. <fa(ʕái)la, *!
fa(ʕá:lσ)tu,…>
c. <fa(ʕá:)la, *!
fa(ʕá:lµ)tu,…>
d. <fa(ʕá:)la, *!
fa(ʕál)tu,…>
Page 40 of 71
Optimal Paradigms
Now that all the elements of the analysis are in hand, we are
in position to bring them together and see how it works, with
our eyes on the goal of reconciling the facts of Arabic with the
tenets of Generalized Template Theory. A [CCV noun stem
like /fʕal/ undergoes epenthesis to improve alignment, and
epenthesis is not blocked by SWP (24).
(24)
a. ☞ <.(fíʕa)lu, . ** **
(fíʕa)la, … >
b. <f .(ʕálu), f . ** **
(ʕála), …>
(25)
nouns because only verbs must deal with both -V and -CV
suffixes. At the left side of the stem, verbs are more diverse
than nouns because only verbs take prefixes. These
differences in inflectional morphology, combined with
independently motivated markedness and OP faithfulness
constraints, explain the templatic differences between nouns
and verbs. There is no need for a template per se nor for
special templatic constraints or similar mechanisms. Deriving
templatic effects from independently motivated constraints, as
in this analysis, is in accordance with the reductionist goals of
Generalized Template Theory.
Page 42 of 71
Optimal Paradigms
(26)
SWP ALIGN-L(Stem, σ)
Page 43 of 71
Optimal Paradigms
Page 44 of 71
Optimal Paradigms
(27)
Page 45 of 71
Optimal Paradigms
a. ☞ <s.
(táf )ʕala, s.
(tàf )(ʕál)tu,
(jàs).(táf )ʕilu, 4 4
(jàs).(tàf )
(ʕíl)na, …>
b.
<.si(táf )ʕala, .si
(tàf )(ʕál)tu,
(jà.si)(táf )ʕilu, 2! 4 4
(jà.si)(tàf )
(ʕíl)na, …>
c. <s.(tàf )
(ʕá:)la, s.(tàf )
(ʕál)tu,
(jàs).(tàf )(ʕí:)lu, 4! 4 2
(jàs).(tàf )
(ʕíl)na, …>
Page 46 of 71
Optimal Paradigms
d.
<.si(táf )ʕala, .si
(tàf )(ʕál)tu,
(jàs).(táf )ʕilu, 4! 2 2 4
(jàs).(tàf )
(ʕíl)na, …>
e. <.si(tàf )
(ʕá:)la, .si(tàf )
(ʕál)tu,
(jàs).(tàf )(ʕí:)lu, 4! 4! 2 2 2
(jàs).(tàf )
(ʕíl)na, …>
Page 47 of 71
Optimal Paradigms
Page 48 of 71
Optimal Paradigms
Page 49 of 71
Optimal Paradigms
/samam/
samamtu ‘I poisoned’
/ja-smum/
/ħmarar/
ħmarartu ‘I reddened’
Page 50 of 71
Optimal Paradigms
Page 51 of 71
Optimal Paradigms
Page 52 of 71
Optimal Paradigms
kǝlb ‘dog’
bǝrd ‘wind’
dǝnb ‘sin’
∫ǝm∫ ‘sun’
lǝʕb ‘game’
mwǝxx ‘brain’
b. C1C2ǝC3 if C2 ≤ C3 in
sonority
r3ǝl ‘leg’
ktǝf ‘shoulder’
ħbǝl ‘rope’
bƔǝl ‘mule’
wtǝd ‘peg’
kfǝn ‘shroud’
Page 53 of 71
Optimal Paradigms
Page 54 of 71
Optimal Paradigms
Page 55 of 71
Optimal Paradigms
(32)
Page 56 of 71
Optimal Paradigms
/∫ǝrb/ + {t, na, *ǝ]σ *CCC OP-MAX-V SON CON IO-MAX-V IO-DEP-V
ti, tu, u, ǝt}
d. <∫ǝrb, ****!
∫ǝrbt, ∫ǝrbna,
∫ǝrbti, ∫ǝrbtu,
∫ǝrbu, ∫ǝrbǝt>
Page 57 of 71
Optimal Paradigms
Page 58 of 71
Optimal Paradigms
Page 59 of 71
Optimal Paradigms
Page 60 of 71
Optimal Paradigms
Page 61 of 71
Optimal Paradigms
(33)
Page 62 of 71
Optimal Paradigms
exist. The remarks here certainly do not address them all, but
rather they suggest the overall approach that can be taken
within the strictures of OP.
8.6 Conclusion
In this chapter, I have introduced the Optimal Paradigms
model of the interaction of phonology with inflectional
morphology. Candidates in OP consist of entire inflectional
paradigms. Within each candidate paradigm, there is a
correspondence relation from every paradigm member to
every other paradigm member. Faithfulness constraints on this
intraparadigmatic correspondence relation resist alternation
within the paradigm.
1 faʕala jafʕalu
7 nfaʕala janfaʕilu
8 ftaʕala jaftaʕilu
Page 63 of 71
Optimal Paradigms
10 stafʕala jastafʕilu
Note: The vowel a is just a stand-in for any of the three vowels
a, i, and u. So /faʔl/, /fiʔl/, and /fuʔl/ are all licit noun stems.
Arabic also has non-templatic nouns. Templatic nouns are by
far the majority and include not only native words but also
many (p.210) loans. Non-templatic nouns are rare and are
nearly all loans. There is an independent criterion for
determining whether a noun is templatic: with few exceptions,
all and only templatic nouns form their plural by internal
change (‘broken’ plurals— McCarthy and Prince 1990a).
Notes:
Page 64 of 71
Optimal Paradigms
Page 65 of 71
Optimal Paradigms
(8) This is properly true only for ‘sound’ verbs, those without
glides or double consonants in the root. See s. 8.4.4.
Page 66 of 71
Optimal Paradigms
(16) This statement does not hold for words whose final root
consonant is a high glide. See s. 8.4.4.
Page 67 of 71
Optimal Paradigms
Page 68 of 71
Optimal Paradigms
(26) For simplicity, I present the analysis here using only the
perfective verb. The full paradigm includes the imperfective as
well, as I have argued for Classical Arabic. The result still goes
through when the imper-fective is considered, though some
additional analysis is required to account for prefixed forms
like nǝ∫rǝb ‘I drink’.
Page 69 of 71
Optimal Paradigms
Page 70 of 71
Optimal Paradigms
Page 71 of 71
Paradigm Uniformity Effects Versus Boundary Effects
DOI:10.1093/acprof:oso/9780199267712.003.0009
9.1 Introduction
Morphologically complex structures sometimes exhibit sound
properties that differ from those occurring in simplexes. The
relevant properties concern both distinctive features as in (1a)
and arguably non-distinctive features as in (1b). In (1a) the
occurrence of the diphthong before a stressless syllable closed
Page 1 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(1)
‘gnomic’ ‘cómic’ ‘shyness’ ‘minus’
Page 2 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(2)
nɑ́m1k *!
ünó℧mik *
(p.213)
Steriade
(2000)
emphasizes
the
theoretical implications of a PU-based description for the
phonology–phonetics interface. To be affected by PU
constraints, phonetic features must be represented in the
lexicon. If universally non-distinctive features could be shown
to be affected by PU the distinction between phonology and
linguistic phonetics would break down and the notion of
distinctiveness would lose its relevance.
Page 3 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(p.214)
Given the
Page 4 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 5 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 6 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(5)
Constraint violated:
wáηal#gimbáːl#du * LAPSE
bigúːn#gímbal * CLASH
(p.216)
(7)
a. [σσσσ]WORD → (σσσσ)ω
b. [[σσσ]STEM[σ]SUFFIX]WORD → (σσσσ)ω
c. [[σσ]STEM[σσ]SUFFIX]WORD → (σσ)ω(σσ)ω
Page 7 of 114
Paradigm Uniformity Effects Versus Boundary Effects
d. → (σσ)ω(σσσ)ω
[[σσ]STEM[σσ]SUFFIX[σ]SUFFIX]WORD
e. [[σσ]STEM[σσ]SUFFIX[σ]MOD]WORD → (σσ)ω(σσ)ω(σ)ω
The prosodic
hierarchy is
subject to the
Strict Layer
Hypothesis (cf
Selkirk 1981),
which
consists of two constraints, here referred to as HEADEDNESS
(cf. Selkirk 1995) and CONTAINMENT. The definitions given
below are adopted from Nespor and Vogel (1986: 7)7
(9)
a. HEADEDNESS
Page 8 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 9 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(10)
a. ALIGN (GCat, E; PCat, E)
b. ALIGN (PCat, E; GCat, E)
(p.218)
Page 10 of 114
Paradigm Uniformity Effects Versus Boundary Effects
9.2.2.4 Summary
It is proposed here to recognize boundary effects if and only if
all of the following criteria are satisfied:
(13)
a. Both the morphosyntactic and the
phonological structure of the input forms are
independently motivated.
b. The relation between the morphosyntactic
structure and the prosodic structure is
describable in terms of alignment constraints,
such that ‘cohesion’follows from constraint
domination.
c. Inviolability of the constraint ALIGN (ω,E;
GCat,E): pword boundariesnecessarily align with
morphosyntactic boundaries.
d. Inviolability of the constraints HEADEDNESS
and CONTAINMENT.
Page 11 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(15)
rádical+ity → rádical+ness →
ràdicálity rádicalness
phónème+ic → devélop+ment →
phonémic devélopment
(16)
(ràdicálity)ω (rádical)ωness
(phonémic)ω (devélop)ωment
(solídify)ω (hóspital)ωìze
Page 12 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(Sùdanése)ω (selèct)ωée
(17)
a. [radical]STEM+ity b. [radical]WORD+ness
[phoneme]STEM+ic [develop]WORD+ment
[solid]STEM+ify [hospital]WORD+ize
[Sudan]STEM+ese [select]WORD+ee
(18)
a. [[radical] ity] → (radicality) ω
b. [[radical][+ness] → (radical)ω(ness)ω
c. [[alcohol][+ize] → (alcohol)ω(ize)ω
Page 13 of 114
Paradigm Uniformity Effects Versus Boundary Effects
In the
descriptions
reviewed here
Page 14 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Violations of
the patterns
in (20)
abound.
Specifically,
lenition in
(21a, b) indicates coda association, although exclusive onset
syllabification is expected. Similarly, fortition in (21c) indicates
exclusive onset syllabification, even though (ambisyllabic)
coda association would be expected.
The choice of
syllable-
structure
related
properties
such as
fortition or
lenition as decisive diagnostics for recognizing pword
structure is supported by the generalization in (22), which is
free of exceptions.
Page 15 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Similarly, the
t in pro-
Tibetan is
prevented
from closing
the preceding
stressed
syllable
because of
undominated
CONTAINMENT. The occurrence of lenition in late ex is less
clear because flapping possibly indicates ambisyllabicity. Here
I follow Kahn who separates the obligatory syllabification of
word-final consonants as codas (his rule II) from the ‘later’ and
optional rule of associating such consonants with following
vowel-initial words (his rule V). That is, coda association of
prevocalic t in (24b) is necessary to satisfy CONTAINMENT
and the relevant structure is preserved in registers where
perhaps ONSET is satisfied as well.
The analysis
of the
syllabification
patterns in
(23b), (24b)
as boundary
effects
correlates
with the
satisfaction of
HEADEDNESS, manifested by the stability of the weak foot in
Page 16 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(p.223)
While
accounting
for
‘unexpected’
lenition in
(23b), (24b),
the
syllabification
of the pword-
final
consonant as a coda raises the question of how to describe the
exceptional rhyme structure in those words. The fact that
‘superheavy’ rhymes such as [a1t] occur word-finally, but not
word-internally (cf [æ´t.] las, but not *[á1t.] las) has led some
Page 17 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Given the
Page 18 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(28)
a. [[la1m]WORD[la1t]WORD]WORD → b. (la1m)ω
(la1t)ω
c. ALIGN (Word,L; ω,L), ALIGN (Word,R;ω,R)
Unmistakable
9.3.1.2 Cohesion
Page 19 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(31)
Phonetic Prosodic
transcriptions representations:
(Wells 1990a):
Page 20 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(32)
Page 21 of 114
Paradigm Uniformity Effects Versus Boundary Effects
a. ωW ωS b. ωS ωW
Page 22 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(33)
a.
I [sedúce]WORD → sedúce)ω
II [[re]HEAD → ((re)Σ
PREF[dúce]STEM]WORD dúce)ω
[[be]HEAD → ((be)Σ
PREF[ráte]STEM]WORD ráte)ω
Page 23 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(34)
(35)
a. [[be]
[little]ADJ]VERB
Page 24 of 114
Paradigm Uniformity Effects Versus Boundary Effects
[[be]
[devil]VERB]VERB
b. [[be]
[low]ADJ]ADVERB
(38)
Page 25 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 26 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(39)
Page 27 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(40)
9.3.2.2 Cohesion
Category I prefixes cohere by definition whereas category III
prefixes exhibit only HFF illustrated in (41a) (compared to
regular prosody in (41b)):
(41)
a. (disintegrate)ω b. (dis)ω(organize)ω
(mistake)ω (mis)ω(print)ω
(extraordinary)ω (extra)ω(marital)ω
(42)
Page 28 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(43)
Page 29 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 30 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(45)
Page 31 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 32 of 114
Paradigm Uniformity Effects Versus Boundary Effects
9.3.3 Suffixation
9.3.3.1 Morphosyntactic structure and alignment
Unlike prefixes, suffixes never function as modifiers but
always as heads.26 Analogous to the representation of head
prefixes, suffixed words are consequently uniquely
represented as [[X]STEM[Y]HEAD SUFFIX] WORD. The
morphosyntactic and corresponding pro-sodic structures are
illustrated in (46).
(46)
(47)
Page 33 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(48)
Page 34 of 114
Paradigm Uniformity Effects Versus Boundary Effects
a [[goal]NOUN[less]]A b.
DJ
[[time]NOUN[less]]A f [[X][less]]ADJ,
DJ
[[tooth]NOUN[less]]A then
DJ [[X]STEM[less]HEAD
SUFFIX]WORD
Page 35 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(50)
(51)
Page 36 of 114
Paradigm Uniformity Effects Versus Boundary Effects
[[sánit]STEM[íze]HEAD → (sáni[th]ìze)ω
SUFFIX]WORD
[[límit]STEM[átion]HEAD → (lìmi[th]átion)ω
SUFFIX]WORD
[[góat] →(gòa[th]ée)ω
STEM[ée]HEAD
SUFFIX]
WORD
(52)
Page 37 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 38 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(53)
Page 39 of 114
Paradigm Uniformity Effects Versus Boundary Effects
‘playlet’
Page 40 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(54)
(55)
Page 41 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(56)
(p.235)
(58)
Page 42 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(59)
9.3.5 Summary
In this section I have presented evidence that the pword
structure of English words can be described on the basis of
independently motivated morphosyntactic structure with no
need for diacritic marking of affixes. Some representative
examples are repeated in (60):
(60)
Page 43 of 114
Paradigm Uniformity Effects Versus Boundary Effects
[[góal]STEM[íe]HEAD → (góalie)ω‘goalie’
SUFFIX]WORD
(61)
a. Vietnám+ése → ballóon+líke →
(Vìetnamése)ω (ballóon)ω(lìke)ω
b. émphas-is+íze → cúri-ous+ness →
(émphasìze)ω ((cúrious)ωness)ω
ámput-àte+ée → shóv-el+ful →
(àmputée)ω ((shovelful)ω
c. revéal+al → Ø goal+less →
((goal)ωless)ω
Page 44 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 45 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(63)
Page 46 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(64)
Page 47 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(66)
Page 48 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 49 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(67)
[[rádical]STEM[ness]HEAD → ((rádical)ωness)ω
SUFFIX]WORD
[[frólic]STEM[some]HEAD → ((frólic)ωsome)ω
SUFFIX]WORD
Page 50 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(68)
Page 51 of 114
Paradigm Uniformity Effects Versus Boundary Effects
[úrban]STEM[ity]HEAD → (urbánity)ω
SUFFIX]WORD
Page 52 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(p.240)
(69)
Page 53 of 114
Paradigm Uniformity Effects Versus Boundary Effects
a. [[vámpìre]STEM[ic]HEAD → (vampíric)ω
SUFFIX]WORD
Page 54 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Consider next the evidence from the velar nasal, which on the
domain analysis cannot be followed by g before non-cohering
suffixes. The relevant words include suffixes with initial liquids
or glides (cf the data in (64b)), which in fact, are never
preceded by [Ŋg]:
(70)
[[ríng]STEM[let]HEAD → ((ri[Ŋ])ωlet)ω
SUFFIX]WORD ‘ringlet’
Page 55 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(p.241)
(71)
a. No PU (stró[Ŋg]er)ω ‘stronger’
effect
(diphthó[Ŋg]al)ω ‘díphthongal’
(díphtho[Ŋg]ìze)ω, ‘díphthongìze’
Page 56 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(72)
(73)
Page 57 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(74)
Page 58 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 59 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(75)
Page 60 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 61 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 62 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(78)
Page 63 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 64 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(79)
Page 65 of 114
Paradigm Uniformity Effects Versus Boundary Effects
atmosphere’
(80)
norm’
Page 66 of 114
Paradigm Uniformity Effects Versus Boundary Effects
instance of consoling’
Page 67 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Given a
domain
analysis as
described in
9.2.2 (based
on [[re]HEAD
FW[cogníze]STEM]WORD and [[re]MOD FW[combíne]WORD]WORD)
a PU analysis of aspiration would be ruled out by the Prosodic
Consistency Criterion in (81). This is because aspiration
correlates systematically with other sound properties entailed
by the pword structure in (83).
Page 68 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 69 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(85)
(86)
Page 70 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 71 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(87)
The
German
data show
that
boundary
signals
can be
more
salient in casual speech than in careful speech. That is, [έsiç]
‘ess ich, a more careful pronunciation, is at least near-
homophonous to the simplex [έsiç] ‘Essig’, whereas the more
casual pronunciation [Ezέç] ‘ess ich differs markedly.
(88)
Page 72 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 73 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(90)
(91)
Page 74 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 75 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(92)
Page 76 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(b[æ´]nner)ω *!
✓(b[É]nner)ω *
Page 77 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(93)
a. No PU (cl[æ]ssic)ω ‘classic’
effect
(cl[æ]ssify)ω ‘classify’
(m[æ]ssive)ω ‘massive’
(p[æ]ssage)ω ‘passage’
(cl[E]sses)ω ‘classes’
(cf. cl[E]ss)
(m[E]ssing)ω ‘massing’
(cf. m[E]ss)
Page 78 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(94)
Page 79 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(auxiliary)
Page 80 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(95)
(96)
Page 81 of 114
Paradigm Uniformity Effects Versus Boundary Effects
sw[E]m]PAST **!
✓ sw[æ]m]PAST * *
sw[1]m]PAST *!
Page 82 of 114
Paradigm Uniformity Effects Versus Boundary Effects
clusters (e.g. cild > cild ‘child’, climban > cliimban ‘climb’, cf.
Luick (19641 242 ff ). Luick (1964: 243) notes that the rule did
not apply to function words such as and, under, should, would.
(97)
(98)
a. ✓ pæm *
b. pEm *!
Page 83 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Consider next
the question
of why pword-
final coda
phonetics (as
opposed to
the phonetics
of the
nucleus)
seems to be immune to undergoing paradigmatic leveling (e.g.
Pe[t⌝]e—Pe[ɾ]er, not *Pe[t⌝]er, wri[t⌝]e ‘write’—wri[ɾ]er, not
*wri[t⌝]er ‘writer’). A possible explanation here is that the
lexicalization of coda-specific phonetic features is blocked if
the relevant consonant is associated with onset position in the
lexicon. The relevant hypothesis is stated in (101).
Page 84 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(102)
(103)
Page 85 of 114
Paradigm Uniformity Effects Versus Boundary Effects
⌗(pæt)ω *
(pət)ω *!
b. LR: /pæmələ/
‘Pamela’
⌗(pæm)ω *
(pEm)ω *!
Page 86 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(104)
Page 87 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(105)
Page 88 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 89 of 114
Paradigm Uniformity Effects Versus Boundary Effects
The ratings for the two candidates in both exclusive onset and
exclusive coda position are fairly clear. For the ambisyllabic
context the ratings are interpreted as indicative of ‘free
variation’ with a ‘slight preference for [l]’ (Hayes 2000: 95,
96).56
(106)
Page 90 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 91 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(p.255)
(107)
c. [[mail]STEM[er]HEAD → (mailer)ω
SUFFIX]WORD
Page 92 of 114
Paradigm Uniformity Effects Versus Boundary Effects
The parallel
treatment of
the cases in
(108) raises
the question of why the preference for light [l] in grayling is
stronger than the preference for dark [ł] in mailer. This
contrast (p.256) could not be caused by the pronunciation of
the alleged attractors: grey is clearly pronounced with [éI] as
mail is clearly pronounced with [éə]. On my view this contrast
indicates that two separate phenomena are involved. Light [l]
in (108a) is a boundary effect and the alleged ‘match-up’ with
the preceding vowel is simply a matter of ‘pro-sodic
consistency’ (cf. s. 9.4.4). Internal boundary effects are
Page 93 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(110)
[[ców]STEM[less]HEAD ((cow)ωless)‘cowless’
SUFFIX]WORD
Page 94 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Page 95 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(p.257)
(111)
Steriade
(2000) rejects
Rialland’s
analysis
because both
d’rôle and
drôle count as monosyllabic in metrical scansion. The
association of [ʁ] with both nucleus and onset position in
(112b), presumably a novel type of syllable structure resulting
Page 96 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(114)
a. incorrect predictions regarding syllable count
(cf Steriade 2000: 328);
b. questionable type of ambisyllabicity in (112b)/
introduction of novel,non-structure preserving
syllable structure;
c. inadequate representation of phonetic
correspondence; cf. (112a) vs (112b)
(p.258)
d. lack of discrimination between distinct types
of schwa loss (cf. (113));
e. the question of acquisition.
Page 97 of 114
Paradigm Uniformity Effects Versus Boundary Effects
Invoking PU
to account for
the effect
observed in
(111) is
questionable
since non-
distinct word
forms
belonging to a single paradigm but distinct registers are
related.59 Before discussing Steriade’s analysis further I will
demonstrate that reference to PU constraints is unnecessary
in that none of the criticisms in (114) apply to a boundary
analysis based on the syntactic distinctions specified in (117).
The crucial and, I believe, uncontroversial distinction is that
r(e) trouvé, but not d(e)rôle, is a word. Note the use of a single
phoneme /ʁ/ in the left column and the predictability of the
three allophones in the right column on the basis of position
within the parenthesized syllabification domains (i.e. [ʁi]:
initial, [ʁi: prenuclear, non-initial, [ʁk]: coda).
(117)
Page 98 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(118)
a. NUC: VO C (‘a nucleus must be filled by a
vowel’)
b. ALIGN (Word, L; σ,L), ALIGN (Word, R; σ,R)
c. ALIGN (Mod, L;σ,L), ALIGN (Mod, R; σ,R)
d. ALIGN (Head FW, L; σ,L), ALIGN (Head FW,
R; σ,R)
(119)
Page 99 of 114
Paradigm Uniformity Effects Versus Boundary Effects
(pa)σ(d)σ( Kol)σ *!
(pa)σ (dirol)σ *!
√ (pad)σ (Kol)σ
b. [ba]WORD[[ʁə]MOD[
tʁuve]WORD]WORD
(ba)σ (ʁə)σ(tʁuve)σ *!
(ba)σ (ʁ)σ(tʁuve)σ *!
(baʁ)σ(tʁuve)σ * *!
√ (ba)σ (ʁtʁuve)σ *
c. [fəʁε]WORD
(fəʁε)σ *!
√ (fʁε)σ
Unlike
Rialland’s
analysis, the
domain
analysis
relates the
distribution of
r-allophones in (111) to subphonemic syllabification effects
which do not involve distinct registers. Consider now the
boldfaced contrasts for (121a, a'), (121b, b') and (121c, c')
respectively, which are determined by contrasts in
morphosyntactic structure.
(121)
‘destabilize’
(122)
a. [inisjal]WORD → ([i.ni]sjal) initial ‘initial’
b. [[in]MOD[imitabl]WORD]WORD → ([i.ni]mitabl)
inimitable‘inimitable’
9.6 Conclusions
In this chapter 1 have argued for the necessity of properly
distinguishing PU effects from domain effects and have
proposed various criteria for doing so. Proper distinction of
the relevant phenomena is a crucial prerequisite for all further
inquiry. PU effects, once they have been properly delimited,
shed light on many aspects of lexical structure of major
concern to linguists, including the organization of paradigms
in the mental lexicon and the question of which phonetic
features are represented in the lexicon. Mostly false
conclusions will be drawn on the basis of misclassified PU
effects.
Notes:
(1) The rule has exceptions, especially among proper nouns
(cf. Jacob, Josef).
(3) The conclusion follows from the premise provided that the
relevant features are indeed affected by PUconstraints directly
(cf Davis (Ch. 5 above), see also n. 61).
(6) The citation marks are used here because the precise
meaning of ‘surface’ is the bone of contention here.
(15) cf. Hall (2001, 2002) for an alternative analysis, where the
special status of pword-final consonants isdescribed by a
constraint which requires trimoraic rhymes to align with the
right edge of a pword.
(19) The verb rerate is not listed in Wells but the prefix re- in
many comparable words including rerun,rewrite, redo is
transcribed with a stress mark and vowel length.
((i))
(33) The sort of stress shift resulting from the ‘rhythm rule’
across pword boundaries in cases like thìrteen mén differs
(34) This claim is consistent with the fact that bound stems
can arise historically if the base of an established affixed word
becomes obsolete as in ruthless, †ruth, doleful, †dole.
(39) For instance the suffixes -ity and -acy would presumably
both be assigned to Class I based on theirassociation with
stem-allomorphy, which raises the question of how to account
for their distinct stressbehavior. In fact, there is not a single
pair of cohesive suffixes In English whose systematic
morphophono-logical properties could be described
exhaustively by some unique ranking of constraints (cf
Raffelsiefen1999b, 2004).
((i))
(45) The claim is that for instance fa[ŋ] less will remain fa[ŋ]
less, even if fang becomes obsolete.
(48) The fact that the final foot is strong if it ends in a cluster
(còndescé[nd], cìrcumvé[nt], rèprehé[nd]) indicates that stress
is based on syllable structure. The last syllable in réconcìle is
open (recall that the pword-final consonant is syllabified as an
onset in the lexicon) and therefore has a non-branching rhyme.
The last syllable in còndescénd is closed, that is, with a
branching rhyme, which attracts main stress.
(59) Steriade (2000: 331) notes that her analysis agrees with
Rialland’s in that a characteristic property of the citation or
careful form is inherited by the schwa-less variant.
DOI:10.1093/acprof:oso/9780199267712.003.0010
Page 1 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 2 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 3 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
(1) PAR:
All surface realizations of µ, where µ is the morpheme
shared by members of paradigm x, must be identical.
(2) CON:
The surface realizations of morphologically distinct
members m i … m n of a paradigm x must be
phonetically distinct.
Page 4 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 5 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
c. IPl.Pres.Indic d. IPl.Cond
Page 6 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 7 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 8 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 9 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 10 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 11 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 12 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
INDEFINITE DEFINITE
Page 13 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 14 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
(8) Template: in one word-form tense/mood and definiteness markers are mutually exclusive, e.g. (relevant values
emboldened)
1Pl.def.Ind akar – j uk
1Pl.def.Past akar t – uk
Page 15 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
dimensions: values:
(p.270)
Page 16 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 17 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 18 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
akar
Page 19 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 20 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a.PAST b. CONDITIONAL
*akar-t-ak
‖‖ ‖‖
*akar-ná-k
Page 21 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 22 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 23 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 24 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. ☞ akar-ok
b.akar-om !*
(1Sg.indef: -ok, -j-
ak, -né-k)
c.akar-op !* *
Page 25 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 26 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. akar-t-ak !*
(=3Pl ‘they wanted
sg.’)
b. ☞ akar-t-am * *
(=def ‘I wanted it‘) (1Sg.indef:
-ok,-j-ak, -né-k)
c.akar-t-ap !* *
(1Sg.indef: 1Sg.indef:
-ok,-j-ak, -né-k 1Sg.def: -ok, -j-ak, -né-k)
-om,-j-am, -ná-m)
Page 27 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
(20) PAR(TENSE/MOOD)
The situation is very similar in the 1Pl case. The regular 1Pl
forms can be analyzed as the same way as the 1Sg forms in
(18).
Page 28 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
akar + PAR(PERSON/PAR(PERSON /
1Pl.def.Past NUMBER) NUMBER × ×
‘We wanted it’ DEFINITENESS)
a. ☞ akar-t-uk
b. akar-t-unk !*
(1Pl.def: -j-uk, -j-uk, -
ná-nk)
c. akar-t-up !* !*
(1Pl.indef: - (1Pl.def: -j-uk, -j-uk, -
unk, -j-unk, - ná-nk)
ná-nk
1Pl.def: -j-
uk, -j-uk, -
ná-nk)
Page 29 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. akar-ná-k !*
(=3Pl ‘they would
want it’)
b.☞akar-ná-nk * *
(=indef: ‘we would (1Pl.def: see (21))
want sg.’)
Page 30 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Note that the forms (22 a) and (22b) are not completely
concatenative: in both cases the initial high vowel of the
Person /number suffix is deleted for phonological reasons
(akar+na+unk→ akarnánk and akar+na+uk → akarnák). It is
clear that this deletion is due to a higher-ranking phonological
constraint, which bans this type of hiatus at morphological
boundaries (i.e. *akar-ná-uk, *akar-ná-unk) and thus enforces
phonological allomorphy. (Note that the form *akar-ná-uk
satisfies all the paradigmatic constraints in (22), i.e. it does
not show homophony and is paradigmatically uniform.
Therefore the phonological constraint must be higher up than
CON(Definiteness).)11
10.5 Anti-Harmony
The second type of lexical allomorphy we discuss is a case
when a phonological regularity is in interaction with
paradigmatic uniformity/contrast. The phonological regularity
involved is backness harmony, which we briefly describe in the
following section.
Page 31 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 32 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
FRONT BACK
[aː]<á> LONG
Page 33 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
(p.279)
Page 34 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 35 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 36 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
etc.
Page 37 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 38 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 39 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 40 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
1Sg.indef 3Pl.def
*akar-ná-k
Page 41 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 42 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
b. Ⅾ akar-né-k *…a…+…é…
Page 43 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 44 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 45 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. teker-ná-k !*…e…+…á…
Page 46 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 47 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Note that if the most harmonic forms in (31) and (34) were the
actual grammatical forms, then we would get lexical
allomorphy in 1Sg.indef.Cond. This lexical allomorphy of the
Person /number marker k~m in akar-né-k ~ *teker-né-m would
be of a (p.284) completely new kind. Up to this point we saw
cases of ‘lexical allomorphy’ where the allomorphs occur at
different points of the paradigm. If we think of a paradigm as a
matrix where a row or a column lists forms that belong to a
given value of a dimension, then in the cases we have
examined so far, lexical allomorphy holds between forms in
different cells14 of the matrix, i.e. lexical allomorph α occurs in
some cell or cells, while allomorph a occurs in another cell or
other cells. Figure (35) shows this state of affairs for
Definiteness Neutralization.
Page 48 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. teker-ná-k !*…e…+…á…
Page 49 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
The PAR
Page 50 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
… … …
1Sg. indef
… … …
Page 51 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
(p.286)
Page 52 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. teker-ná-k ! * … e… + … á
…
Page 53 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 54 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 55 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. teker-t-ak !*…e…+…a…
Page 56 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 57 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 58 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. akar-t-ek !*…e…+…a…
Page 59 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 60 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. akar-ná-k !*
(= 3Pl
‘they would
want it’)
b. akar-né-k !* * *
c.☞ akar-ná-nk * *
(1Pl.indef: (=indef
d. akar-né-nk * !*
(1Pl.indef: …a…+
-j-uk,-t-uk, …é…
-j-uk)
Page 61 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
(p.290)
Page 62 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a. teker-ná-k !*
Page 63 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 64 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
b. akar-né-tok *…a…+…é…
Page 65 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
It can be
seen in
(45) that
some
avoidance
strategy
Page 66 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 68 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 69 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Con(Definiteness)
PAr(PeRSON /number x
Definiteness)
Par(Tense/mood)
WeakPar
Page 70 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
a
*akartak
b
*akarnák
c
*akarál /*akarol, *hozál
d
*tekerje, *tekerjétek, *tekerjék
e
*akarnák
f
*akartanak
(p.295)
Page 71 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
PRESENTSUBJUNCTI
VE
INDEF. DEFINITE INDEF. DEFINITE INDEF. DEFINITE INDEF. DEFINITE INDEF. DEF.
2Sg. akarsz akarod akartál akartad akarj ál, akarjad, akarná l akarnád -Á1 −d
c akarj akard
hozol
3Sg. akar akarja akart akarta akarjon akarja akarna akarná ∅ -(j)−a
d tekerte tekerje tekerné
tekeri
1Pl. akarunk akarj uk akart unk akart uk akarj unk akarj uk akarná nk akarná nk −unk -(j)−uk
tekerj ük tekerj ük
2Pl. akartok akarjátok akart atok akartátok akarjatok akarjátok akarná akarná −tok -(j)Á-tok
tekeri tek tok tok
d
3PI. akarnak akarják akartak f akarták akarjanak akarják akarná akarná k −nAk −(j)Á-k
tekerikd nak
Page 72 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Notes:
(1) See also Kiparsky (1972, 1978) for a discussion of related
issues in the framework of classical generative phonology
(6) For a more radical case that involves consonants too, see
Rebrus (2000b).
Page 73 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
(9) In this tableau and the following ones we have included the
‘bases’ of the violations in parentheses forease of reference,
i.e. we list the specifications and glosses of the forms that the
candidate examined wouldbe identical with in the case of CON
constraints, and the (sub)strings that the candidate (sub)string
wouldbe different from in the case of PAR constraints. The
crucially relevant parts of the strings and glosses
areemboldened.
Page 74 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 75 of 76
Uniformity and Contrast in the Hungarian Verbal Paradigm
Page 76 of 76
A Note on Paradigm Uniformity and Priority of the Root
DOI:10.1093/acprof:oso/9780199267712.003.0011
Page 1 of 33
A Note on Paradigm Uniformity and Priority of the Root
11.1 Introduction
The role of the paradigm in explaining phonological
(ir)regularities has recently reemerged in phonological and
morphological theory. Generally speaking there is a tendency
toward phonological uniformity within a paradigm, but under
some conditions uniformity is not maintained. In order to
understand the role of the paradigm in the grammar it is
important to learn more about when identity is and is not
maintained. Approaches to paradigm uniformity include cyclic
application of phonological rules (Kiparsky 1982 b) or positing
a relationship between fully formed words (see Benua (1997a);
Burzio (1994a); Flemming and Kenstowicz (1995) and other
chapters in this volume). Two issues that have emerged
recently are determining the base for word-formation and
understanding the conditions under which uniformity does or
does not (p.297) occur. What unites these two issues is a
central goal to understand which words stand in a
paradigmatic relation and how phonological processes interact
with morphological structure. In investigating these issues, a
common assumption is that the base is the stem, minus the
formative, which is being added. McCarthy (Ch. 8 above)
refers to this perspective as illustrating base-priority. In this
chapter base-priority refers explicitly to the output word to
which some morphological process applies. It is interesting to
examine cases where base-priority seems to be violated. For
example, in the formation of imperfective stems in Upriver
Halkomelem (Central Salish), the base would be the stems in
(1a.i) and (1b.i).1
Page 2 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 3 of 33
A Note on Paradigm Uniformity and Priority of the Root
a. [q’é:l] ‘believe’
Page 4 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 5 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 6 of 33
A Note on Paradigm Uniformity and Priority of the Root
b. [šəmén] ‘enemy’
c. [txWə́məcən] ‘September’
(5) Halq’eméylem
a. [qwé:l] ‘talk’2
Page 7 of 33
A Note on Paradigm Uniformity and Priority of the Root
(6) Halq’eméylem
(7) Halq’eméylem
a. /-stəxw/ CAUSATIVE
Page 8 of 33
A Note on Paradigm Uniformity and Priority of the Root
(p.301)
Page 9 of 33
A Note on Paradigm Uniformity and Priority of the Root
a. ‘control’
b. ‘limited control’
Page 10 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 11 of 33
A Note on Paradigm Uniformity and Priority of the Root
While
candidate (9a)
illustrates
perfect
faithfulness to
the segmental
identity of the
segments in
Page 12 of 33
A Note on Paradigm Uniformity and Priority of the Root
(10) Constraints
c. IO-MAX-VFEATWD:
d. OP-MAX-VFEAT-OBJ:
(p.304)
Page 14 of 33
A Note on Paradigm Uniformity and Priority of the Root
b. lèmnám̓š *!***
w
lèmnál̓x
lèmnàmə
lèmnálə
lémnəxw
Page 15 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 16 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 17 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 18 of 33
A Note on Paradigm Uniformity and Priority of the Root
/CV-c̓qw
̓ / c̓əc̓ə́q̓w ‘getting pierced, shot’
/ABLAUT- c̓qw
̓ -n-əxw/ c̓aq̓wnəxw ‘shooting it, piercing it
(manage to)’
d. /c̓ət̓q̓w-ətə́/ c̓ə́t̓qw
̓ t ‘grind, pulverize, smash
it’
/ABLAUT-c̓ət̓qw
̓ ət/ c̓at̓q̓wt ‘grinding, pulverizing,
smashing it’
e. /c̓ət̓q̓w-els/ c̓ət̓qw
̓ els ‘grind up’
/ABLAUT-c̓ət̓qw
̓ els/ c̓at̓q̓wəl̓s ‘grinding something up’
Page 19 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 20 of 33
A Note on Paradigm Uniformity and Priority of the Root
(p.306)
There is a
pervasive
process of
unstressed
vowel
reduction that
drives the
reduction of
the root vowel
to schwa. In
order to
achieve this, a
markedness
constraint
that penalizes
unstressed full vowels is required, represented in (15a) below.
This constraint must be ranked above the OP-Faith constraint
in (10d), because there is non-uniformity among
corresponding vowels. In order to arrive at the correct output,
it must be the case that the paradigmatic faithfulness
constraint evaluates only words in the aspectual paradigm.
This constraint is represented in (15b) below with the
additional -ASP ending.
(15) a. v˘:
b. OP-MAX-VFEAT-ASP:
Page 21 of 33
A Note on Paradigm Uniformity and Priority of the Root
This paradigm
would be
optimal
because IO-
MAX-VFEAT is
ranked below
the OP-Faith
constraint. The ranking established thus far is repeated below.
Only those candidates with unstressed vowel reduction are
included. The symbol indicates that the incorrect candidate is
chosen as optimal.
(17)
Unstressed
vowel
reduction
occurs in
each of the
candidates. If it is assumed that there is only one OP-Faith
constraint evaluating all the words in a paradigm, then
candidate (17a) is ruled out because it violates it. Candidate
(17b), in which vowel reduction (p.307) overapplies, would
be incorrectly selected as optimal. Candidate paradigms
containing words like * [ləl̓ém̓ət]are not included because all
imperfectives have initial stress (Urbanczyk 2000). The
constraint which demands initial stress on imperfective words
would be violated fatally by all such candidates.
Page 22 of 33
A Note on Paradigm Uniformity and Priority of the Root
(18)
If Broad-Faith
were assumed
for the
candidates
above,
candidate
(18b) would
be optimal.
Candidate
(18a) would
incur a
violation because there is a vowel feature in the imperfective
that is not parsed in the perfective. (I am considering only
candidates with schwa in the perfective words, because that
can be derived by pressure from the majority-rules effect of
the other members of the object paradigm. The other members
of the paradigm are excluded to save space.)
Page 23 of 33
A Note on Paradigm Uniformity and Priority of the Root
(19)
Candidate
(19a) is
optimal, as it
does not
violate any of
the
markedness
constraints
and the only
faithfulness
constraint it
violates is one on paradigm uniformity for aspect. Candidate
(19b) is entirely faithful, but it contains an unstressed full
vowel. Candidate (19c), which illustrates identical vowels in
the correspondent bases, is ruled out (p.309) because it has
a stressed schwa. Therefore, the constraint ranking correctly
eliminates ill-formed paradigms.
(20)
Tableau (20)
illustrates
that this same
ranking will
derive the
correct
results for the
paradigm in
which
stressed
schwa is
derived. In this case, all candidates have at least one stressed
schwa. Recall that this is compelled by high-ranking *CÀC.CÁC
in the object agreement paradigm. In the interests of saving
(possibly three-dimensional) space, I leave aside how the
entire system works. All candidates in (20) have the stressed
schwa derived by the ranking established above and so each
Page 24 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 25 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 26 of 33
A Note on Paradigm Uniformity and Priority of the Root
A rule reducing the full vowel to schwa would then apply to all
3rd-person object forms. The fact that the reduction applies to
[lə́mnəxw] is an accident, and is not related to the observation
that it forms part of the object agreement paradigm. One
might just as easily expect it not to occur, or to occur in the
absence of any other reduction rule (such as that found for the
other object agreement).
(22) Hul’q’umi’num’
Page 27 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 28 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 29 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 30 of 33
A Note on Paradigm Uniformity and Priority of the Root
11.5 Conclusion
One of the implications of this short study is that base-priority
does not straightforwardly result in predicting the correct
surface form. The analysis presented here offers support to
McCarthy (Ch. 8 above) who proposes that there is no base
priority in inflectional paradigms. It is not surprising
therefore, that root-priority can emerge from time to time,
supplanting derived properties in paradigmatically related
word forms. Root priority takes precedence over paradigmatic
uniformity, precisely when certain phonotactic considerations
are not relevant in the expression of some inflectional
morpheme. Reference to the input root is clearly a requisite in
deriving the correct imperfective stems. This follows naturally
from OP, where IO-Faith and OP-Faith interact. When OP-Faith
fails, IO-Faith emerges, thus deriving priority of the root.
Furthermore, it was shown that paradigmatic identity is
sensitive to the morphosyntactic information that is being
conveyed. When object agreement and aspectual marking are
morphologically distinct, there are two distinct OP-Faith
constraints: one demanding identity in the object agreement
paradigm, the other forgoing identity in the aspectual
paradigm.
Notes:
(1) I have regularized the transcriptions of data from various
sources to be consistent. They are presented phonemically
using the APA, in which the symbols [c, ƛ, č, š, x̌, y] correspond
Page 31 of 33
A Note on Paradigm Uniformity and Priority of the Root
to IPA [ts, tɬ, t∫, ∫, χ, j]. Other symbols and diacritics are the
same in both systems. The surface forms abstract away from
the differing qualities of schwa, representing them as [ə].
(5) The initial /s/ is a prefix, and is not considered part of the
root. Even if it is considered part of the stem, the result is an
initial cluster and may not be problematic.
(6) As such the pattern here does not provide evidence for or
against extending Broad-Faith to Output-Output
correspondence.
Page 32 of 33
A Note on Paradigm Uniformity and Priority of the Root
Page 33 of 33
References
(p.313) References
Bibliography references:
Page 1 of 33
References
Page 2 of 33
References
Page 3 of 33
References
Page 4 of 33
References
Page 5 of 33
References
Page 6 of 33
References
Page 7 of 33
References
Page 8 of 33
References
Page 9 of 33
References
Page 10 of 33
References
Page 11 of 33
References
Page 12 of 33
References
Page 13 of 33
References
Page 14 of 33
References
Page 15 of 33
References
Page 16 of 33
References
Page 17 of 33
References
Page 18 of 33
References
Page 19 of 33
References
Page 20 of 33
References
Page 21 of 33
References
Page 22 of 33
References
Page 23 of 33
References
Page 24 of 33
References
Page 25 of 33
References
Page 26 of 33
References
Page 27 of 33
References
Page 28 of 33
References
Page 29 of 33
References
Page 30 of 33
References
Page 31 of 33
References
Page 32 of 33
References
Page 33 of 33
Index
(p.331) Index
Figures and notes are indexed as ‘f’ and ‘n’ in bold.
Page 1 of 43
Index
Page 2 of 43
Index
Page 3 of 43
Index
Page 4 of 43
Index
dialects 190
Belfast English 7, 8
consonants in 8
dentalization in 7
Bemba 124n, 128, 134
causative doubling 123, 125, 139
suffixes 124
Benua, L. 8, 9, 10, 11, 11n, 46, 58, 61n, 67, 116, 122, 129, 130, 134,
139, 146, 168, 169, 171, 172, 206, 207, 207n, 212, 240, 251, 296
Bermúdez-Otero, R. 197n
Bianco, V. 299, 311
(p.333) binary foot see foot structure
Black, H. A. 197n
Bloomfield, L. 167
Bobaljik, J. 199, 205n
Boersma, P. 77
B. Hayes 21, 255
Bolozky, S. 49
Bonet, E. 102
M.-R. Lloret 173
Booij, G. 146, 220, 233n
R. Lieber 134, 135
J. Rubach 226
Borowsky, T. 197
bound stems 232 n, 236n, 241, 242, 242n, 256;
see also free stems; stems
in English 242
boundary 256
analysis 258, 262
effects:
noncohesion, affixation in 244;
noncohesion,
suffixation in 241, 242;
versus paradigm uniformity effects 211–62
phenomena 215
signals 213, 226n, 230
Brame, M. 126, 134, 191n
broad-IO Faith see constraints, broad-IO Faith
Broselow, E. 166
Buckley, E. 123, 129, 173, 188n
Bulgarian 147, 150, 152–3, 156, 168, 169
homophony in 292n
reduction in 292n
Bullock, B. 95
Burzio, L. 12, 16, 62, 66, 67, 69, 71, 73, 84, 87, 88, 89, 92, 93, 95,
96n, 100, 101, 114, 123, 129, 134, 137, 138, 139, 146, 172, 172–3n,
264n, 292, 296
Bybee, J. 22, 84, 95n, 138, 293, 293n
Page 5 of 43
Index
Bynon, T. 263
Cairene Arabic 166, 167
Calabrese, A. 150
canonical phonology 214; see also phonology
violations of 237
Cantineau, J. 160
careful speech see speech, careful
Carlson, K. 178n
Carstairs, A. 162
Casablanca dialect 166–7
casual speech see speech, casual
Casual Speech Criterion 246, 261
causative 124n, 128, 131, 135, 137, 138
doubling 121–44
Bantu languages 123;
Bemba 123, 126, 135, 139, 141;
cyclicity, analysis of 134–9;
inflectional paradigms in 139–43;
Jita 121–44;
paradigm optimization 130–4;
spirantization in 144
glides 127, 130, 131, 132, 134
languages 134, 136
morphemes 123
palatal (causative) glides 123, 124
optimal paradigms (OP), analysis of 123
stems 124, 127, 132–4, 136, 138, 139–40, 142, 143
non-causative 136;
of verbs 134;
pronunciation of 138
suffixes 126
verbs 131
cells 264, 267; see also Hungarian, cells; paradigms
Central Hessian dialect 169
children and languages 26n
Chi-Mwi:ni 147, 153–7, 168
Chomsky, N. 131, 178
M. Halle 6, 119, 145, 146, 172, 173n, 197, 228n, 263
Chung, S. 146, 178n
Cicero 21n, 38
clash avoidance constraints see Salish languages,
constraints
Classical Arabic 162
see also Arabic; various dialects
conjugations 209
grammar of 200
majority rules in 206
noun templates in 209–10
Page 6 of 43
Index
Page 7 of 43
Index
dentalized 8
in Belfast English 7
intervocalic 251
paradigm uniformity (PU) and 134, 150, 151, 160, 161, 163
phonetic contrasts of 221
pword-final 223, 224n, 233
spirantization of 124
stem-final 232,
aspiration of 233
stressless vowels and 235
syllable-initial 214
tautosyllabic 248
voiced constellation theories 139, 142, 143
constraints 9, 41, 117–18, 140, 157, 205, 206, 208, 238, 243n,
247, 251, 255, 259, 260, 262; see also under alignment;
Hebrew; Hungarian; I-O Faithfuness; markedness; Salish
languages;
against voiced obstruents 247
broad-IO Faith 307
correspondence 9, 130, 173, 173n, 174, 181, 197, 251
relations 301, 303;
theory 171, 172, 182
domination 218
exceptionless 11
hierarchy 9, 281, 289
in morphology 51, 87
in phonology 87
in rhotacism 21
paradigms:
leveling of 200;
uniformity (PU) 18, 22, 53, 215, 217, 218, 232, 234,
phonotactic 19
ranking 1, 5, 9, 11n, 15, 19–20, 21, 139–40, 173, 184, 194, 199,
200, 207, 208, 218, 240
containment, paradigm uniformity (PU) and 216, 217n, 218,
220, 222–3, 225, 226, 236, 238, 239, 240, 242, 246, 248, 251,
253, 254, 261
superordinate units in 217
coronal:
obstruents see obstruents, coronal
palatalization 174, 176, 206
sonorants 126
stops 89–90, 154
correspondence constraints 9, 130, 173, 173n, 174, 181, 197, 251
relations 301
theory 171, 172, 182
Côté, M.-H. 150
Cowell, M. 161
Page 8 of 43
Index
Page 9 of 43
Index
Page 10 of 43
Index
Page 11 of 43
Index
Page 12 of 43
Index
head 249
fusion 225, 229, 234
Gafos, A. 178n, 181n, 200, 301, 308n
Galloway, B. 297, 298, 299, 300n, 304
Ganong, W. F. 72, 73
Garr, W. R. 207, 208n
geminates 231n, 234
verbs 200; see also verbs
gemination 157, 162, 168
generalizations 4, 5, 69–70, 80, 90, 162; see also Hebrew,
generalizations; phonology
‘concatenative’ morphology of 80
entailments and 91
faithfulness 70
minimal 33
learners of 31;
morphology, rule-induction of 22
Sonority Sequencing Generalization (SSG) see Hebrew
surface 13
Generalized Template Theory (GTT) 171, 178, 182, 187, 189n, 195,
209
generative 150
grammar 169
linguistics 73
phonology 65, 212; see also phonology
classical 263n
genitive 23–4, 26, 27
German 3, 6, 227n, 231n, 233n, 234n, 242n, 247
boundary effects in 241n
paradigm leveling in 4
vowels in 6, 247
lengthening of 6
words, morphology of 6
gerunds 244, 244n
Gesenius, W. 207
Giegerich, H. 223
glides 123, 124, 128, 130, 178n, 187n, 238n, 240
verbs with high 200; see also verbs
glottalization 221, 246, 252; see also lenition
in Salish languages 304
Goldsmith, J. 146
Gougenheim, G. 260
Gouskova, M. 190
government phonology 264; see also phonology
gradient attraction (GA) 16, 74, 83–4, 91, 94, 97, 99, 100, 102, 105
affixes 98
entailments in 106
grammar 22
Page 13 of 43
Index
morphology, rules of 4
neutralizations and 100
grammar 25–6, 30n, 33, 74, 146, 152, 172n, 178, 186, 263
allomorphy and 146
and forms 35, 37
and lexicon 35
constraints 146
rankings in 12
early generative 6
functions in 3
gender in 231n
paradigms 296
uniformity (PU) and 64
phonology and 145–69
rules 146
in Latin 32
Greek 242n
groups 3, 5
formal 3, 4
leveling 5
material 3, 4
(p.337)
words in 5n
Grundzüge der Lautphysiologie (Sievers) 2
Gussenhoven, C. 169
Gussmann, E. 188n
Haddad, G. 161
Hale, M., and M. Kissock 197n
C. Reiss 18, 19, 197n, 251, 252, 252n, 253
Halkomelem see Salish languages
Hall, T. A. 130n, 217n, 224, 224n, 232n, 247n
Halle, M. and A. Marantz 102
Halq’eméylem see Salish languages
Hammond, M. 51, 215, 219
Hannah, S. J. 217n
Hargus, S. 197
E. M. Kaisse 7, 197
Harris, J. 7, 224, 252
Harris, J. W. 65, 66, 66n, 147
Harris, Z. D. 212
Hayes, B. 26n, 40, 41, 64n, 112, 115, 157, 190, 214, 253, 254, 255,
256
head:
affixations 233, 241n
boundary effects in 241
function words 249, 250n, 251, 259
prefixes 230, 231, 231n, 232, 233, 241, 244n
suffixation 232, 242
Page 14 of 43
Index
headedness 216, 217n, 218, 222–3, 225, 226, 229, 236, 250, 253,
259
Heath, J. 157
Hebb, D. O. 73
Hebb’s rules 73–7
Hebrew 12, 207, 207n;
see also Colloquial Hebrew; Tiberian Hebrew
alternations 45, 49
base selections 54–6
constraints 50, 52, 53, 55, 58, 62, 64
correspondences, multiple 48
forms:
future 58;
imperative 58;
syllables in 62
fricatives 49, 49n, 59, 60, 60n
phonemes 49;
spirantization 49, 60
generalizations in 52
imperative paradigms 44–5, 45n, 46–64
future in 48;
inflectional 45;
person– number–gender 44–5;
sub-paradigm 45–8;
suffixes in 45, 46n;
tense 45;
words 46
markedness constraints 45, 51, 53, 54, 58, 62
Optimality Theory (OT) 45, 46, 48
constraint-based 48;
Correspondence Theory 46
paradigm uniformity (PU) in 19, 39, 41, 44–64,
Sonority Sequencing Generalization (SSG) in 54
structure 57, 57n
and Existence 53, 54;
identity of 53, 54, 56, 58;
relations 57
verbs 55–6, 57n, 62
with prefixes 60–2
vowels 62
Hendricks, S. 178n
heterosyllabic clusters, in French 260, 261
Hetzron, R. 205n
hiatus 187, 260
‘High Frequency Fusion’ (HFF) 225, 226n, 233, 238n, 239n, 241
Hock, H. H. 17, 21, 102
hollow verbs 159, 160, 161, 162, 164, 164n, 165, 167;
see also verbs
Page 15 of 43
Index
Page 16 of 43
Index
Page 17 of 43
Index
Page 18 of 43
Index
Kisseberth, C. 184
M. I. Abasheikh 153, 154, 155n, 168, 184
Klausenburger, J. 30n
(p.339) Korean 39, 41, 162
alternations in 40–1
base forms in 15
markedness constraint 40
coronal obstruents in 40
faithfulness constraint 40
obstruents in 41
paradigms:
inflectional in 15;
leveling of 40;
nouns 40
phonotactics of 40
Kosati, plurals in 48n
Kraska-Szlenk, I. 146, 148, 172, 212
Kuhl, P. and P. Iverson 72
Kühner, R. 28, 36
Kurisu, K. 48n
Kurylowicz, J. 5, 18, 95, 172
labiodental fricatives see fricatives, nasal
Ladefoged, P. 108
Lahiri, A. 167n
B. E. Dresher 22, 41, 146
languages 174, 176, 177, 186, 189, 192, 199, 200, 205, 206, 223n,
231n, 237, 252, 265
learners 23, 26, 27, 30, 33–4, 37, 232, 252
minimal generalization 31;
of Latin nouns 28, 33
phonology of 49
Triebe 5
universals of 78
LaSor, W. S. 60
Latin 22n, 33n, 41, 45, 65, 159, 172, 242n, 286
alternations in 17, 30, 35, 41
o-o 19; s-r 17–18, 35
analogy 22
notation in 18
forms:
ablative 28, 30, 32, 33;
base 15, 32;
dative 28, 30, 32, 33;
genitive 28, 30, 32–6;
nominative 29, 31, 32, 33, 33n, 38, 34–6, 41;
oblique 32, 33, 33n, 34–6, 36n, 38, 41
genders and word lengths 37
grammar 2, 34, 35
Page 19 of 43
Index
learners of 33, 35
non-alternations 36
nouns 28–33, 30n, 38, 41
abstract 38–9; agentives 18, 34, 34f, 35, 39; analysis of
36–41; base status 42–3; declensions of 28, 36; i-stem
32, 32n, 33, 33n; monosyllables 35, 37; neuter 37;
nominative 30, 36, 36n, 37– 9; paradigms of 28–3, 33,
36, 173n; polysyllables, neuters 35; polysyllables, non-
neuters 21, 33, 35–7, 38–9; preference category 34f;
token frequencies in 38
paradigms, inflectional in 15
rhotacism in 17, 30, 30n, 36n, 41
in grammar 34; nominative case 19
stress rule 65–6, 157
subgrammars in 31
voicing in 30
vowels, long 17
word-final sonorants 17
lax vowels 244, 245, 251; see also vowels
learners, of languages 23, 26, 27, 30, 33–4, 37, 232, 252n, 257
minimal generalization 31
lenition 211, 221–3, 253
Leslie, A. 298, 300, 305
Leumann, M. 21, 28, 36n
leveling:
of alternations 18
of paradigms 3, 5, 18, 21, 37, 41, 214, 252
allomorphy in 37;
in German 4, 5;
majority rules in 37;
morphology of 17–43
Levin, J. 177
lexemes see stems
lexical allomorphy 267, 268, 271–2; see also Hungarian
lexical items 151, 167, 181, 251
lexical phonology 197, 199, 261, 263
Lexical phonology and morphology (Kiparsky) 6, 7, 68, 82, 93, 146
lexical representation 213, 224, 252, 253, 261
lexical strata see paradigm uniformity (PU)
lexical suffixes, in Salish languages 310
‘Lexicon Optimization’ 85, 87, 251
lexicon 72, 84, 87, 90, 94–5, 98, 101, 102, 180, 224, 232, 246n, 251,
252, 252n, 261
and grammar 35
conservatism, analysis of 22n
items in 138
non-alternation in the 39
paradigmatic relations in the 214
Page 20 of 43
Index
constraints 214
phonetics 213
Spanish 149
stress 67
‘limited control’ object paradigm 14
Lindblom, B. 70, 77
linguistics 44, 147, 171, 173, 262, 263, 301; see also generative
linguistics
European pre-Structuralist 2
pre-generative 1–5
liquids 238n, 240; see also glides
Lowenstamm, J. 260
Luick, K. 250n
l-velarization, in English 253–6
McCarthy, J. J. 11, 13, 14, 15, 21, 37, 53, 60n, 71, 81, 112, 113, 123,
130, 132, 133, 134, 139, 141, 142, 143, 146, 157, 161, 173, 177, 178,
180, 181, 186, 191n, 196, 197n, 206, 206n, 207, 209, 212n, 214,
231n, 250n, 298, 301, 303, 305, 308n, 312 (p.340)
A. Prince 9, 46, 53, 63, 135, 175, 177, 178, 178n, 187, 189,
189n, 190, 197n, 207n, 210, 213, 264
‘majority rules’ hypothesis 4, 37, 176, 201, 202, 205, 205n, 206,
209, 307
Makan 167
Malone, J. L. 207
Manczak, W. 5, 38
mapping, from syntactic to prosodic structure 217
Marantz, A. 177
Marcus, G. et al. 35
markedness constraints 14, 15, 72, 86, 87, 89–90, 130, 139, 224,
231, 234, 236, 259, 301
and alignment constraints 230
local, and semantics 39
of attractors 175
of superheavy syllables 182
of verbal inflections 293
optimal paradigms (OP) and 177, 178, 181, 182, 183, 184, 185,
186, 187, 188, 189, 194, 195, 196, 197, 198, 200–8, 208n
paradigm uniformity (PU) and 150, 173, 176,
Marshallese 150
Martin, J. 48n
Martin, S. E. 39, 40
Mascaró, J. 147
material groups see groups, material
Mayerthaler, W. 6n
Meiser, G. 36n
mental:
learning 15
lexicon 5, 78, 83
Page 21 of 43
Index
Page 22 of 43
Index
Page 23 of 43
Index
Page 24 of 43
Index
leveling of 21
paradigms 18
in Latin 31
stems of 179, 186, 192
and vowels 187
templates 179, 189, 193
Noyer, R. 102, 104
nucleus 259
position 257
numerals, sound changes in 5n
object agreement paradigms 300, 303, 307
and aspectual markings 312
oblique forms, in Latin see Latin, forms obstruents:
clusters in 26, 29
homorganic 72
Korean 41
coronal 40
voiced 7
ambisyllabic 246
Yucatec Maya 72
Old English 102, 243, 249n–50n; see also English Old French 242n;
see also French
onset 224, 232, 235, 239, 240, 243, 252, 254, 254, 257, 260, 261,
261n, 262, 312
position 255, 257
syllabification of 221, 222, 223, 223n, 230, 238, 245
o–o alternations, in Latin 19
OP-Faith constraints 312
opacity 4, 11, 12n, 71, 145–69
optimal paradigms (OP) 9–16, 18, 123, 143, 170–3, 173n, 174–207,
207n, 208–10 see also under morphology; prefixes; Salish
languages; suffixes; syllables; verbs
causative:
doubling 123, 129–30, 133, 135–6, 143–4; languages
134, 136
constraints 171, 173, 175, 177, 182, 184, 187, 191, 192, 193–4,
200, 201, 202, 203,
cyclicity analysis in 134–43
(p.342)
faithfulness constraints 1174n, 178, 181, 182, 183, 184, 185,
187, 188, 189, 190, 194, 195, 196, 197, 199, 201, 204, 205
formalism 132
phonology with inflectional morphology 170–210
representations in 129
Optimality Theory (OT) 1, 16, 67, 122, 138, 146, 152, 22, see also
under Hebrew; Hungarian; Salish languages
alignment constraints in 213
analyses 116–21
Page 25 of 43
Index
of cyclicity 129
constraints in 87, 199, 171, 279
cyclicity in 12n
faithfulness constraints in 71, 132, 146
in morphology 51
Input–Output faithfulness (IO-F) in 84, 90, 146
markedness constraints in 75
neutralizations in 71
‘opacity’ in 71
optimal paradigms (OP) and 173, 175, 180, 199–201, 205, 206
Output–Output Faithfulness (OO-F) 9, 12, 83–4, 90, 105
correspondence 263
paradigm uniformity (PU) and 66, 67, 83–4, 85, 89, 93
reduplication in 206
serial 12n
Oran dialect 164
Orgun, C. O. 9n, 12n
Osthoff, H. and K. Brugman 5n
Oujda 162–4, 165
Output-Output Faithfulness (OO-F): 9–11, 11n, 12–14, 16, 21,
46, 85, 86–92, 92n, 93, 96n, 100–1, 106, 116, 168, 170, 171,
174, 206, 302, 302
see also constraints; Optimality Theory (OT)
and morphology 67
anti-faithfulness correspondence 13
base-priority approach to 13–15
constraints 123, 135, 212, 263–5, 267
correspondence 122, 140, 143, 144, 171
in optimality theory 264;
of causative doubling 130, 132;
relations 240
foot structure (OO-foot structure) 119–21
symmetrical model 14–15
outputs 6, 9, 11, 12
identity relationships 6, 13
of words in morphology 16
optimal 11
strings 14
overapplication v. underapplication:
in inflectional paradigms 206
of vowel reduction 309–12
optimal paradigms (OP) and 174, 174n, 175–6, 181,
187, 188, 188n, 197, 201, 202, 206–7, 207n, 208, 209
paradigm uniformity (PU) in 249, 250, 251
Padgett, J. 70, 81n
palatal harmony 126, 127, 128, 151
Palestinian dialects 162
par constraints see Hungarian, par constraints
Page 26 of 43
Index
Page 27 of 43
Index
Page 28 of 43
Index
Page 29 of 43
Index
prefixed words:
aspiration in 245
prosody of 226, 228
prefixes:
in English 224
inflections in 188–97
monosyllables, in 230
optimal paradigms (OP) of 172n, 179, 188, 193, 199
(p.344)
paradigm uniformity (PU) and 226, 227, 228, 242, 245, 259
syntactic properties of 227
pretonic foot 227; see also foot structure
pretonic position see onset, position
Prince, A. 61n, 191, 207, 208n
P. Smolensky 9, 11, 66, 85, 86, 87, 157, 171, 180, 181, 196, 251,
301
Principles of English stress (PES) (Burzio) 66, 67, 68, 76, 84, 87, 88,
91, 92, 93, 95, 97, 98
privative meaning 232
pronunciation 1, 13, 138;
see also speech
careful 261
casual 247
in US English 11
of attractors 255
of consonants 255
of English 11n, 242
dialects 253
paradigm uniformity (PU) and 226–7, 228, 233–4, 238n, 246,
248 n, 250n, 254
Russian 13, 151
‘proper nesting’ constraint 216n
prosodic categories 217
phonetic grounds of 227
Prosodic Consistency Criterion 261
prosodic contrasts 227, 233
prosodic domination 236
prosodic fusion 231, 242n
prosodic hierarchy 216, 217
prosodic properties 215, 241
prosodic representations 226, 255
prosodic structures 177, 189, 192, 215, 218, 219, 225, 231, 232n,
238, 241, 246, 248, 249, 254, 256
and syntactic structures 219
mapping in 217
of words 225–6
stems in 241
prosody 190, 228, 229, 236, 240
Page 30 of 43
Index
Page 31 of 43
Index
Page 32 of 43
Index
Rose, S. 200
Rosenthall, S. 200
Rubach, J. 9, 146, 197, 197n
rule 187
based phonolgy 1, 8, 206
generative phonology see phonology, rule-based
generative
in Latin grammar 32, 34
induction 24–7
ordering 1, 6, 7
and the cycle 1, 6
phonology 201
Russell, K. 51
Russian 13, 147, 150–2, 156, 168, 169
cyclic model in 13
phonology of 13
pronunciation 13
verbs in 13
Sagey, E. 81n
Salish languages (BC, Canada) 264, 298–312
ablaut in 304, 305
allomorphy in 300, 304, 305
alternations in 297–8
constraints 303, 307, 308
clash avoidance 303, 311; rankings 309
derivational theory 309–12
faithfulness constraints 302–3, 306, 308
glottalization in 304
Halkomelem 299, 302–4, 308n, 311
Upriver 297–8
Halq’eméylem 298–300, 305
Hul’q’umi’num’ 298, 299, 300n, 304, 310–n
schwa reduction 301; stress in 301
markedness constraints 302–3, 306, 308
optimal paradigms in 309–12
OP-Faith 303, 306, 308, 308n Optimality Theory (OT) in 311–12
Output–Output correspondence 308n
paradigm uniformity (PU) in 303–5, 309
paradigms 302–3, 305, 307, 308, 309, 311–12
aspectual 308; object agreement 300, 302, 307, 308,
310
reduplication in 304–5, 307–8, 311
roots/stems in 298–302, 304, 305, 312
root priority 309
schwa in 298–306, 310
reduction 310;
stress 304, 309, 310
suffixes in 300, 311
Page 33 of 43
Index
lexical 310
vowels 304–6, 309–12
reduction 303, 306, 310, 311;
root 308;
stress in 299, 310
words in 304–5
Sanders, N. 188n
Sanskrit 69
Schane, S. 260
Scherer, J.
Schmidt, J. 3
Schuchardt, H. 5
schwa 14, 151, 157, 158, 160, 162, 163
see also under Salish languages in English 145–6
loss 256, 257, 257n, 258, 258n, 259, 260, 261
paradigm uniformity (PU) and 203, 208, 236n, 243n, 244n,
246, 252n, 256, 257, 257n, 259–60
violations of 244
‘second law of analogy’ (Kurylowicz) 95–6
segmental neutralizations see neutralizations,
segmental
Selkirk, E. O. 109, 158, 172, 213, 216, 229, 235, 250n
semantics 5n, 84, 128n, 243, 244, 256
affinity in 244, 244n
and local markedness 39
dissociation in 256
drifts in 244, 249
naturalness in 37
representations in 98
shortening, in Latin 19
Semitic languages 81, 178
morphology in 45, 81
sequences, illegal 26, 26n
in nouns 30
word boundaries in 30
Shaw, P. et al. 299, 303
Shere, T. 183
Sievers, E. 2
Silverman, D. 107
(p.346) simplexes 8, 171
paradigm uniformity (PU) and 214, 215, 217, 218, 221, 227,
227n, 230, 241, 241n, 242, 245, 246, 247, 248, 255, 261
phonetics of 258
polysyllabic 245
Siptár, P. and M. Törkenczy 268
Skutsch, O. 17n
Smolensky, P, L. Davidson and P Jusczyk 26n
Soh, H.-L. 150
Page 34 of 43
Index
Page 35 of 43
Index
Stampe, D. 181
‘Stampean occultation’ 181, 189
stem-final consonants see consonants, stem-final
stems 7, 95, 97n, 98, 99, 100, 138
see also under bound stems; free stems; Hungarian; pstems;
verbs Salish languages, roots/stems; words
base and 297
bound 91–2, 93
causative see causative, stems
metrical consistency of 101
morphology of 130
mwords and 143
of nouns 179, 180
of verbs 180
of words 242, 265
optimal paradigms (OP) in 172, 173, 173n, 174, 176, 178, 200,
203, 205, 207
paradigm uniformity (PU) 228, 232, 233 n, 236, 240, 241n, 244,
261
stress 101, 230n
templates 184n
uniformity of 94
Steriade, D. 9, 12, 16, 18, 22n, 52, 62, 68, 69, 71, 102, 107, 108, 109,
110, 114, 116, 121, 123, 137, 147, 150, 213, 214, 257, 258, 258n,
260, 261–2, 262n, 263, 264, 265, 267n
Stojkove, S. 152
stops 25, 107–21, 245, 246, 247, 260
aspiration in 110–12110–11, 121
and non-final 110–11; and word-final dactyls 110–11, 121
foot-initial 112–13
unaspirated 119
voiceless 110–12, 117
Strata OT 197, 197n
Strauss, S. 197
stress 68, 70, 100, 153, 157, 159, 161–2, 166, 168, 173n, 227–8,
237–8, 239n, 246, 259, 267n
see also English, stress in; Italian, stress in
constraint 121, 190
in allomorphy 101
analysis of 10
identity 11
in suffixation 243
neutrality 218
of verbs 95
patterns 190, 215
schwa 14
in English 145–6
shift 218
Page 36 of 43
Index
stems 101
violations 219
(p.347)
stress-behavior 220, 240
stress-neutrality 219, 237, 239
vowel-initial suffixes in 220
stress-related diagnostics 222
stress shift 236, 236n, 239 –40
Stress to Weight Principle (SWP) 190–2, 192n, 193, 193n, 194, 195,
196, 197, 199
stressed vowels 233, 246; see also vowels
stressless:
syllables 211, 236n, 239n; see also syllables
stems of 230
vowels 234
Strict Layer Hypothesis 216, 216n
structure:
identity of, in Hebrew see Hebrew, structure, identity of
Struijke, C. 178n, 303, 307
sub-paradigm see paradigm, sub
subgrammars 27–8, 31, 33
forms in 27
in Latin 31
rules with confidence values 31
rules in 27–8
metrics of 2
subphonemes 260; see also phonemes
Sudanese Arabic 71
suffixation 231–2, 233, 236,
stress in 243
suffixed words 212, 231, 247, 254n
suffixes 8, 82, 97, 100, 127, 300 see also under affixes;
dentalization; Hungarian; Jita; prefixes; Salish languages
causative 126
stems 124
class 8
derivational 7, 123
English 224
forms, rhotacism in Latin 17, 18
Hungarian 266, 268
Latin 29
nouns 33
inflectional 7
junction characteristics of 232n
optimal paradigms (OP) in 176, 179, 180n, 181, 182, 184, 185,
187, 191, 195, 199, 201, 203, 217, 218, 230n, 233, 234, 234n,
235, 237, 239, 239n, 173n, 174, 174n, 240, 240n, 242n, 243–4,
249, 256, 265, 279n
Page 37 of 43
Index
paradigm uniformity (PU) 153, 154, 155, 158, 214, 160, 161,
162, 163, 166
phonology of 240
in Jita 137
Spanish 147
stress of 99
syntactic representation of 219
suffixing inflection 182
superfoot see dactyls, sequences in
superheavy syllables 180–1, 181n, 183, 184; see also syllables
markedness of 182
surface:
exceptions 9
forms 85, 87, 172, 212, 264
base-priority and 312;
outputs 5
prosody 216
resemblance 170–2, 173, 187, 197–9, 214
shapes, of words 263
stem-finals 186
Suttles, W. 299, 305
Swedish 234n
syllabification 189, 267n
domains of 254, 258
in French 259, 261n
lexical versus phonetic 251
onset, of consonants 238
paradigm uniformity (PU) and 220, 221, 222, 224n, 228-22,
231n, 234, 260
syllable-conditioned:
allophony 252
phonetics 252
syllable-structures 257, 258, 259, 260
diagnostics 221
syllables 8, 111, 157, 215, 220, 220n, 223, 224, 234, 248, 257, 259
count of 257
faith-stressed 52–3
flapping in 121
in Latin 17n
intitial consonants 214
loss of 257
onset 221
optimal paradigms (OP) and 160, 162, 168, 170, 177, 181–3,
187, 188–92, 194, 197, 199, 200, 202, 202n, 204, 261
positions, final 246
Spanish 147–8
stressed 119, 221
structure of 246n
Page 38 of 43
Index
Page 39 of 43
Index
Page 40 of 43
Index
Page 41 of 43
Index
vowels 4, 77–8, 89–90, 150, 150–1, 154–7, 159, 161, 166, 176, 267,
271
see also under English; epenthesis; Hungarian; lax; Salish
languages
allomorphy in 98–9
allophony in 254
Bulgarian reduction in 152–3
epenthesis in 185
German 6, 247
lengthening of 7
Hebrew 62
Italian 88
lengthening of 181, 185n, 216, 245, 249n, 280n long 186, 198–
9
optimal paradigms (OP) 179, 180n, 183, 185, 191, 201, 202,
207, 208n, 209
paradigm uniformity (PU) 220, 235, 245, 248, 253, 255, 256,
257
shortening of 184
stems 187
stressed 221, 246
voiceless 7
Walker, R. 178n
Webster’s ninth new collegiate dictionary
(Webster) 227, 228, 231n
Welden, A. 166
Wells, J. C. 212, 214n, 226, 226n, 227, 228, 231n, 240, 242n
Wetzels, L. 30n
Whitney, W. D. 2, 2n
Wilbur, R. 206
Wilson, C. 71
Wiltschko, M. 301
Withgott, M. 108, 109, 112, 262n
Word Faithfulness (Struijke) 303
word-final:
consonants 222, 223n, 224
sonorants see sonorants, word-final
word-formation, base for see base, for word-formation
word-sized units 72–3
words 1, 3, 45, 113–14, 146
see also under English; function; head, function; Jita; Latin,
nouns; morphology; mwords; phonology; pwords; Salish
languages
affixes 219n
analogy of 3
and clitics 234–5
embedding 146
in groups 5n
Page 42 of 43
Index
Page 43 of 43