Compositional routes to (non)intersectivity

Abstract

This dissertation is concerned with a particular instance of ambiguity in natural language, between intersective and nonintersective interpretations of certain adjectives. Adjectives like good display a consistent alternation across many languages between a noun-dependent reading and a noun-independent reading, and this alternation is sensitive to various syntactic factors concerning word order and locality. In this dissertation, I develop a compositional account of the (non)intersective ambiguity. Central to this account is the idea that ambiguity is derived in the course of syntactic derivation, from uniform underlying adjective denotations, where different interpretations result from modification occurring in different domains of the nominal phrase. I argue that these interpretive domains are divided by the introduction of grammatical number to the noun, in the form of number marking or a classifier, and correspond to the ontological distinction between nouns-as-kinds and nouns-as-objects. Empirical motivation for this analysis comes from word order alternations in Italian, the interaction of plurality and kind-readings in Turkish, and focus-driven movement in Bangla, with a particular focus on the previously underemphasized behavior of privative adjectives. Maintaining uniform adjective denotations is enabled by positing covert structure in intersective readings of adjectives, motivated by visible morphophonological effects of that structure influencing suppletion across languages. I also extend the account by offering experimental evidence in favor of a fine-grained, semantically active lexical-conceptual structure for kinds, in order to capture compositionally the nonintersective readings of privative adjectives.