![]()
This talk investigates the semantics of sentences that express numerical averages, such as (1)-(3).
We begin with a discussion of the use of 'average' in definite descriptions, as in (1), which has been argued by Chomsky, Hornstein and others to provide an argument against the hypothesis that natural language semantics includes a reference relation mapping words to objects in the world. We then develop an analysis in which the definite construction in (1), the adverbial construction in (2) and the nominal construction in (3) all involve truth conditions that are about amounts rather than individuals, and involve interpretive operations that are independently necessary for other, related constructions (in particular comparatives and prenominal 'same/different'). Finally, we show both how our analysis defuses the arguments against a reference relation that have been presented on the basis of (1), and that an empirically adequate account of the full range of sentences that describe numerical averages supports two more general conclusions about the syntax-semantics interface. First, numerals must be able to take scope independently of the nouns they modify. Second, it is not always the case that the function created by abstracting over the base position of a scope-bearing element composes with the scopal term (as in e.g. Heim and Kratzer 1998); instead, this function may compose with a third element intervening between the scopal term and its base position.
|
Return to Colloquia Page || LSO Home || MSU Home
....