Case Syncretism, Categories, and the Morphosyntax of Polish Numeral Phrases
Gilbert C. Rappaport
The University of Texas at Austin

    This paper proposes a new approach to a familiar problem. The problem 
is providing a satisfactory linguistic analysis of the complex morphosyntax 
of Polish numeral phrases. The proposal will be embedded in a synthesis of 
recent work in both the Minimalist Program and Distributed Morphology.
    In Babby's terminology, (1a,b) exhibit heterogeneous morphosyntax (the 
numeral is not in the same case as the noun it quantifies), and (1c) 
exhibits homogeneous morphosyntax:
(1)a.   Tam spi piec studentek.
                 `Five students[female] are sleeping there.'
         b.      Widze piec studentek.
                 'I see five students[female].'
         c.      z piecioma studentkami
                 `with five students[female]'
Heterogeneous morphosyntax is found in direct case positions (nominative 
and accusative), and homogeneous morphosyntax is in the remaining, oblique 
case positions.
     Following recent work in Minimalism we take structural case marking to 
be the result of a syntactic operation Agree which applies to pairs of 
categories under appropriate (if elaborate) conditions and copies a feature 
from one category to the other. We will develop an approach wherein the 
contrast between heterogeneous and homogeneous morphosyntax lies not in the 
syntax, but in the morphology. This approach is implemented by assuming an 
`abstract' case we called Quantitative case, which is invariably spelled 
out for nouns and adjectives as a syncretism with another case (as the 
Accusative case is throughout most of its paradigm). Heterogeneous 
morphosyntax is the result of the numeral bearing the Quantitative case, 
spelled out on nouns and adjectives as the Genitive, and on the numeral (in 
this context) as the attested form (traditionally, if incorrectly on our 
view, called the Nominative). Heterogeneous morphosyntax is found only in 
direct case positions because these are positions of structural case 
assignment, which in the assumed framework are positions in which the noun 
is inserted in syntactic structure without case (to be assigned it in the 
syntax). These are precisely the syntactic contexts permitting various 
quantified expressions, which entail, as does a Quantitative numeral, that 
structural case need not be realized. Homogeneous morphosyntax is found in 
the remaining positions, because they are positions of inherent case 
assignment, in which case is licensed by selection, i.e., the noun is 
inserted with case (and the numeral, without case, is assigned case by 
agreement); case in these positions must be realized (ruling out quantified 
expressions and heterogeneous numeral phrase morphosyntax). We take the 
noun to be the head of the numeral phrase in both instances.
    Numerals are a hybrid category (unlike collective nouns, on one hand, 
and modifiers, on the other) defined by having the lexical option of being 
associated either with a case feature valued as Quantitative or as 
unvalued. The architecture of grammar automatically accounts for the fact 
that a valued case feature invokes heterogeneous morphosyntax, and an 
unvalued case feature invokes homogeneous morphosyntax in the appropriate 
contexts. The impossible combinations are easily ruled out by morphological 
considerations. Polish contrasts with Serbian/Croatian, in which numerals 
always exhibit heterogeneous morphosyntax, indicating the possibility in 
this language of only the former of the two lexical options available in 
Polish.
    While the above facts and analysis apply to certain other Slavic 
languages, Polish seems to exhibit a twist in the virile category:
(2)a.   Szesciu pilotow weszlo do kawiarni.
         `Six pilots[masculine] went into the coffeehouse.'
     b.  Zastalem szesciu pilotow w kawiarni.
         `(I) encountered six pilots[masculine] in the coffeehouse.'
On our analysis, while in both these direct case positions the numeral 
phrases may appear to exhibit homogeneous morphosyntax, in fact they are 
structurally identical to (1a,b). The difference is a superficial, 
morphological one: the spell-out of the Quantitative case gives the same 
result (Genitive) in a virile numeral as in nouns and adjectives (cf. the 
dedicated Quantitative form in (1a,b).
    The paper proceeds to discuss the somewhat more complicated situation 
of the `lower numerals', in which Polish differs (in its simplicity) from, 
say, Russian (and the other East Slavic languages). Aside from the virile 
forms, lower numerals in Polish are simple modifiers, with no government 
properties (assigning case or number). In the virile forms, both 
possibilities for the case feature are possible, as in the higher numerals: 
Quantitative and no value. However, unlike the higher numerals, lower 
numerals enjoy an additional spell-out option, the dedicated virile 
Nominative forms dwaj, trzej, czterej:
         (3)a.   Dwie studentki weszly do kawiarni.
         `Two students[female] went into the coffeehouse.'
  b.     Dwoch studentow weszlo do kawiarni.
         `Two students[male] went into the coffeehouse.'
  c.     Dwaj studenci weszli do kawiarni.
         `Two students[male] went into the coffeehouse.'

                                                       -the end-