Numeral induced agreement mismatches in Serbo-Croatian

Abstract

Certain numeral phrases in the subject position in Serbo-Croatian trigger apparent mismatches in terms of features between the verb and the quantified noun. The behavior of numerals in Serbo-Croatian follows predictable trends observed in other Slavic languages in forming natural groups of cardinal numbers that impose distinct forms on the nominal elements and trigger different verbal agreement. Apart from numeral 1, which behaves like a true adjective and has no effect on the verbal morphology, cardinal numerals 2, 3 and 4 on the one hand, and 5 and above on the other, give rise to seemingly unexpected verbal forms. Similarly, the two sets of collective numerals – masculine collectives and mixed collectives, also exhibit agreement mismatches. In this thesis I focus on cardinal numbers and collective numerals in Serbo-Croatian, and I discuss agreement patterns that these numerals impose on the elements in both the nominal and the verbal domain. To account for the behavior of different numerals, I propose that they are endowed with distinct feature specifications. Whereas numeral 1 one enters the derivation with all its features unvalued, numerals 2, 3 and 4 have inherently valued number feature and unvalued gender feature. 5+ numerals are defective in that they lack the gender feature altogether, but have an unvalued number feature. Similarly, masculine collective numerals behave on a par with 2, 3, 4 cardinals in having inherently valued number, but unvalued gender features. Mixed collectives, on the other hand, behave like 5+ cardinals and have only unvalued number features. I further argue that the observed verbal patterns are the result of the agreement or in some instances the lack of agreement with the numeral. Additionally, the behavior of adjectival elements with respect to scopal relations and agreement patterns also depends on the type of the numeral (i.e. cardinal or collective) as well as the feature specification of the numeral. By assigning the proposed feature specification to different groups of Serbo-Croatian numerals, I implicitly suggest that they indeed are in-between categories, as previously suggested by Klockmann (2012) for Polish. If the categories of nouns and adjectives are defined by their feature specification as in Baker (2003), numerals seem to fall somewhere in between the two categories.