Documente Academic
Documente Profesional
Documente Cultură
The constraint is satisfied even if the input affix is not realized faithfully:
(2) /kæt/ + /s/Pl. *[khætz]Pl. (RM satisfied)
A more complex case is the one represented by plural formation is Romanian consonant-
final masculines, in which the plural morpheme /i/Pl. is not realized as a segment, but as
palatalization on the final consonant of the stem:
In (3) the final consonant of the stem and the plural formative segment /i/ coalesce, at the
expense of UNIFORMITY-IO and the consonant acquires a secondary [coronal] articulation
originally present on the plural marker:
[coronal]
(5)
/pom/ + /i/Pl. *PK/ i, REALIZE- UNIFORMITY- MAX-IO
u MORPH(EME) IO
a. pomi *!
b. pom *! *
c. pomj *
A non-trivial question here is whether we really need REALIZE-MORPH(EME). Couldn’t
MAX-IO just be used to the same effect, possibly indexed for the morpho-syntactic
category represented by the affix, for instance MAX-IOPl.? What MAX-IO says is that it is
necessary for every segment in the input to have a correspondent in the output. From
Tableau (5) it appears that REALIZE-MORPH(EME) duplicates MAX-IO and in general
every time a morpheme is specified in the input, it will be realized in the output to the
extent to which is MAX-IO satisfied with respect to that morpheme. [I will return to this
question later.]
The English and Romanian cases above clearly represent instances of concatenative (or
additive) morphology, in which the output for a morphosyntactic category (plural) is
obtained by essentially ‘adding’ phonological material to a bare stem taken from the
lexicon which gets realized according to the grammar of the language.
There are however numerous cases in which no addition seems to be taking place and the
morphosyntactic category is realized by modifying the stem in terms of segment
alternation, truncation or reduplication etc. This is the case of nonconcatenative
morphology. A natural question that arises at this point is to what extent a theory of
morpheme realization, with its pivotal REALIZE-MORPH(EME) constraint is capable of
unifying the two paradigms. At first blush, there is hardly a common denominator for the
concatenative and nonconcatenative morphology: the former can be captured in terms of
item-and-arrangement, while the former is best described as item-and-process. In the
item-and-arrangement model both roots/ stems and affixes are specified in the lexicon,
whereas in the item-and-process model roots/ stems are part of the lexicon, but
morphemes are viewed simply as rules.
A first pass is to try to reduce nonconcatenative morphology to concatenative
morphology and assume that processes that affect the stem are the consequence of adding
phonological morphologically-specified information to the bare stem. Consider German
plural formation with Umlaut. Instead of positing a rule which changes the backness
specification of the bear stem in order to form the plural, one may instead assume the
existence of a plural morpheme for this class of nouns is represented by a [-back]
autosegmental feature that docks onto the nucleus of the stressed syllable turning it into a
[-back] vowel:
On this view, REALIZE-MORPH(EME) will be satisfied in all output forms whose stressed
vowel has been fronted as compared to the corresponding input vowel:
(7)
a.
/Vater/ + ØSingular IDENT-IO-[+back] REALIZE-SINGULAR
a. Vater
b. Väter *!
2
b.
/Vater/ + [-back]Plural REALIZE-PLURAL IDENT-IO-[+back]
a. Väter *
b. Vater *!
(9)
Singular Plural
dege ma-tege ‘piercing tool’
gala ma-gala ‘game of imitation’
(10)
Singular Plural
tgu ma-tgu ‘navel’
kuga ma-kuga ‘body hair’
Assuming that the singular morpheme is [+voice] for Class 5 nouns and null for Class 9
nouns and the plural morpheme is ma for all the nouns in (9) and (10), REALIZE-
MORPHEME will be satisfied whenever the respective specification is phonologically
realized on the output, and otherwise violated.
(11)
/tege/ + [voice]Sg. REALIZE- ID-[voi]
SINGULAR
a. tege *!
b. dege *
3
(12)
/ma/Pl. + /tege/ REALIZE- ID-[voi]
PLURAL
a. ma-tege
b. ma-dege *!
c. tege *!
d. dege *!
(13)
/gala/ + [voice]Sg. REALIZE- ID-[voi]
SINGULAR
a. gala
b. kala *! *
(14)
/ma/Pl. + /gala/ REALIZE- ID-[voi]
PLURAL
a. ma-gala
b. ma-kala *!
c. kala *! *
d. gala *!
(15)
/tgu/+ ØSg. REALIZE- ID-[voi]
SINGULAR
a. tgu
b. dgu *!
(16)
/ma/Pl. + /tgu/ REALIZE- ID-[voi]
PLURAL
a. ma-tgu
b. ma-dgu *!
c. tgu *!
d. dgu *! *
4
Luo plurals revisited
1. bare consonantal root + -e/-be/- (no change in voicing) - reported to be the largest
group of plurals
lak lake ‘tooth/teeth’
kalatas kalatase ‘paper(s)’
mabuus mabuuse ‘slave(s)’
glas glase ‘glass(es)’
stima stimbe ‘lamp(s)’
prfjum prfjumbe ‘perfume(s)’
san sande ‘plate(s)’
2. consonant final devoicing in the singular; in the plural the root final consonant appears
as voiced, plus the plural affix -e
ut ude ‘neck(s)’
4. vowel-final stems
- epenthesize [k] in the plural
obo oboke ‘lung(s)’
- delete the final vowel of the stem
gweno gwen ‘window(s)’
2. + 3. represent cases of polarity, but as it seems, polarity is just one of several ways of
forming the plural; it may be plural formation for stems that end in voiced obstruents or
are lexically marked
Luo polarity
Singular Plural
a. alot alode ‘vegetable(s)’
bat bade ‘arm(s)’
lu lu ‘stick(s)’
ruo ruoi ‘chief(s)’
guok guogi ‘dog(s)’
5
db dp ‘debbi(s)’
A Sympathy analysis
Selector constraint: ID-[voi]
Sympathy candidate: alod
ID-[voi]
/alod/ + /e/Pl. ID-[voi] REALIZE- *VoiObs ID-[voi]
PLURAL
a. alode *
b. alote *! *
c. alod *!
d. alot *(!) *(!) *
In b. assume that stems end in voiced obstruents; both numbers, singular and plural, are
marked by vocalic suffixes
A Sympathy analysis
Selector constraint: ID-[voi]
Sympathy candidate: kid
6
ID-[voi]
Plural
/kid/ + /e/Pl. REALIZE- *VoiObs ID-[voi]
PLURAL
a. kite *
b. kide *!
c. kid * *
d. kit * *