Whereas English bare plurals exhibit kind, generic, and existential readings, Spanish bare plurals only allow existential readings. Relatedly, English bare plurals can occur in virtually any syntactic position, whereas Spanish bare plurals are restricted to (essentially) VP-internal positions.

It has been proposed (e.g., Contreras 1986, Longobardi 1994, Chierchia 1998) that both the syntactic and semantic restrictions in Spanish fall out from a single (purely) syntactic source: Spanish bare plurals contain a null determiner which is only licensed inside VP. Given a theory of syntax-semantics mapping in which VP-internal elements map to a domain of existential closure, the syntactic restriction then accounts for the limited range of interpretations. Although it is appealing, I show that this account is empirically inadequate. Instead, I argue that the syntactic restrictions fall out from the semantic representation of Spanish bare plurals. In particular, I show that Spanish bare plurals are unambiguously property-denoting and, due to properties of the syntax-information structure-semantics interface, these NPs can only be accomodated in certain syntactic positions. In English, on the other hand, bare plurals are ambiguously kind-denoting or property-denoting. I discuss the morpho-syntactic spell-out rules in the two languages that account for the difference in possible denotations for bare plurals.