On the lexical semantics of lexical categories (Joint work with Itamar Francez, U of Chicago)

Grammatical categories, distributionally defined classes of expressions, exist in all languages and form an essential part of any theory of grammar.  A persistent question since antiquity has been whether such distributional classes correlate with, or are motivated by, semantic categories, and the traditional enterprise, found also in the functional-typological literature, is to seek a universal notional core underlying the major categories. For example, it is often proposed that verbs (prototypically) name (transient) actions, nouns name (time-stable) things, etc. (see e.g., Givon 1984, Langacker 1987, etc.).  Such ideas have come under criticism (e.g. Newmeyer 1998, Baker 2003), partly due to the allusion to prototypes and the lack of formal articulation of key notions, both of which make it difficult to identify falsifiable predictions.

 In this talk, we propose to shift attention away from the search for notional universals, focusing instead on a search for systematic ways in which category membership constrains denotation. First, we show that so called property concept lexemes can have two kinds of denotations, the first of which is available freely to both nouns and adjectives, but the second of which is universally unavailable to adjectives. We then propose a derivation of this restriction on adjectival denotation from a precise understanding of the intuitive idea that the essence of the adjective category is adnominal modification. Putting forth a prima facie naive formal definition of the semantics of adnominal modification as subsective strengthening, we then describe and motivate some interesting and controversial theoretical commitments that follow.