Semantics Research Group Meeting, November 27, 2009

I propose to reconsider the semantic analysis of event coercion as well as part-whole metonymy, often treated as a case of coercion, through an ontological analysis of functional objects and their partition into functional parts. More precisely, I will argue that the telic view of functionality as conceived by Aristotle and underlying Pustejovsky's analysis of coercion in the frame of the Generative Lexicon does not fit the linguistic data. I will defend that the kind of functionality underlying event coercion is an intentional one, where the event or function attributed to an object depends on the physical characteristics of the object itself and the agent intentionally using the object for a given purpose. If I am right, the intentional analysis of functionality supports Asher's analysis of event coercion in terms of dependent types, defining events introduced by coercion as dependent on the object designated by the coerced term as well as on the agent designated by the subject term.

The link between event coercion and metonymic coercion then rests on the notion of functional part. The different views of functionality adopted in the analysis of event coercion will indeed render different principles of partitioning of an object into functional parts, and therefore different accounts of metonymic coercion. Here again I will argue that the notion of functional part conceived in a telic way appears unsatisfactory. What we need to adequately account for metonymic coercion is a kind of context-dependent notion of functional part, the contextually salient principle of functional partitioning being prescribed by the object itself as well as the property expressed by the coercive predicate and which acts as an integrating property. I will therefore defend an account of metonymic coercion in terms of dependent types, capturing the functional dependence of parts with respect to the functionally integrated whole they belong to.