This paper continues earlier work developing a formal theory of morphological categories within a feature-based model of morphology (Anderson 1992, Rappaport 1998) in which rules can manipulate abstract grammatical information divorced from concrete morphemes. The project aims to better understand the range of complex, even idiosyncratic, morphological rules possible within a tightly constrained model of grammar.
Animacy, a lexical feature of nominal stems spread to adjectival modifiers by Concord, has no distinctive morphological expression. It is a parameter defining case syncretism: in Russian the syntactically assigned case feature [accusative] is replaced by [nominative] or [genitive] before morphological spell-out in the absence vs. presence of the feature [animate]. The interaction of animacy and numerals in Russian generates known paradigmatic puzzles. Among them is the following. When a quantified animate noun phrase is in the nominative case, numerals stand in the nominative and appear to head the phrase, governing the noun in the genitive case (Babby's 'heterogeneous case marking'). When the same phrase is in the accusative, the numerals 'two' through 'four' agree with the following (apparently head) noun standing in the genitive case ('homogeneous case marking' typical of quantified phrases in oblique cases); the remaining numerals behave as in the nominative case.
In contrast to a syntactic approach to such facts (cf. Babby 1987, 1988), we pursue a morphological approach (cf. Halle 1990, 1993). We attribute the case contrast (genitive vs. nominative) in the numerals in the accusative (znat' dvux [Gen] vs. pjat' [Nom] studentov) to a morphological fact: Russian numerals of the latter type do not distinguish animacy in their morphological spell-out. This is an idiosyncratic fact of Russian, as can be shown on the basis of historical, dialectal, and contemporary data in other Slavic languages. In contemporary Polish, for example, the default pattern for all numerals outside the nominative (modulo exceptions for compounds ending in -oneí) is homogeneous case marking.
We argue here that the number contrast in the noun following 'two' through 'four' (singular in the nominative vs. plural in the accusative; cf. dva studenta [Sg] znajut dvux professorov [Pl]) results from a third value for the feature [number] in Russian, which we call paucal (cf. Corbett 1978). While historically derived from the dual number (and expressed directly to some extent in Belorusian and Ukrainian in the form of a stress contrast), in contemporary Russian this number is completely analogous to animacy in that its function is to define a syncretism: paucal nouns in the nominative take the form of the genitive singular. Further nuances (e.g., alternative case forms of modifiers of feminine nouns and, in earlier stages of the language, for masculines as well) are treated as variation in rules of morphological detail in the spell-out of [number: paucal]. In Polish, for example, there is no paucal (dwa polskie [Nom Pl] paszporty [Nom Pl] ).
Two historical developments in progress demonstrate that only a [nominative] quantifier assigns paucal number. First, animate accusative nouns of the second declension (ending in {a} in the nom. sg.) in combination with numerals 'two' through 'four' admit the heterogenous case pattern characterizing syntactically nominative phrases, although this is currently being replaced by the expected homogenous pattern (videt' tri utki --> trex utok). Second, while the syntax of phrases quantified by compound numerals is generally determined by the last numeral, compounds ending in 'two' through 'four' combined with animate accusative nouns tend to move from a homogenous to a heterogenous pattern (videt' sto dvadcatí dvux klientov --> dva klienta). Generalizing to the puzzle addressed here at the outset, we conclude (pace Halle) that animacy is a parametric variable and that Russian quantifiers can undergo a rule of deanimacy (under specific conditions defined here), which makes it possible for the quantifier to take the nominative case feature required to assign to the noun the paucal number, spelled out as the genitive singular.