Heidi Klockmann (Utrecht University)

On sharing number: The English kind-construction

Previous mentions (Carlson 1977, Lehrer 1986) and studies (Zamparelli 1998) of English kind-words (kind, type, sort) have noted the kind-generalization, an identity condition on the expression of number on a kind-word and its accompanying noun (N2). If a kind word is singular, N2 is singular ((1) vs. (3)), and if a kind-word is plural, N2 is plural ((2) vs. (4)).

(1)            This kind of rabbit, this type of car, this sort of rug                    sg-sg

(2)            These kinds of rabbits, these types of cars, these sorts of rugs     pl-pl

(3)            *This kind of rabbits, *this type of cars, *this sort of rugs           *sg-pl

(4)            *These kinds of rabbit, *these types of car, *these sorts of rug    *pl-sg

((4) improves if N2 is interpreted as massified)

Examples from the Corpus of Contemporary American English (COCA) (Davies 2008-) show there to be systematic exceptions to this generalization: if either the kind-word or the N2 lacks a canonical count syntax, the kind-generalization fails to apply. The identity requirement seems to be restricted to those examples where both the kind-word and N2 are marked canonically for number, suggesting that number plays a crucial role. In this talk, I propose that the effect arises from a sharing of number between the kind-word and the N2. This sharing has a morphological consequence, namely the expression of the same number value on both the kind-word and the N2. I further pursue the claim that the “sharing” is a case of restructuring in the nominal domain, and illustrate how differences in the amount of structure projected above the N2 create the limited matching effect of the kind-generalization.