This paper describes the relationship between syntactic structure and semantic interpretation for noun phrases. Specifically, I justify with new evidence Longobardi's (1994) claim that only DPs/QPs but not NPs may function as syntactic arguments. In addition, I claim that NPs, on the other hand, invariably translate as predicates at LF. These claims are prompted by evidence gathered from investigating the so-called (countable) bare singular noun phrases (BSs) as found in Balkan and Mainland Scandinavian languages, as well as bare plurals (BPs) across Germanic, Romance and Balkan languages. Crucially, I claim that as opposed to non-bare nouns which are always syntactic arguments, BSs and existential BPs are property-denoting expressions which invariably translate as predicates at LF. Consequently, they are not restricted quantifiers or variables. I follow Dobrovie-Sorin and Laca (1996) in that the basic/universal property of BPs is property-denotation. The generic reading of BPs is derived from the property-denotation by a (language specific) nominalization mechanism. This statement is prompted by a clear empirical observation: all the languages that have BPs allow them to receive an existential interpretation, but only some (notably Germanic but not Romance and Balkan languages) allow them to receive a generic interpretation. Further, the paper argues that while BSs and existential BPs are NPs lacking a D projection, generic BPs are DPs with a morphologically empty D.
