12h00
UPS Pouchet, salle de conférences
Gennaro Chierchia (Harvard)
Where does weak crossover come from?
Résumé
Nearly 50 years after its discovery, Weak Crossover (WC) still constitutes a formidable problem.
The phenomenology associated with it is complex, and its best theoretical characterizations remain fairly descriptive and fraught with empirical issues. In attempting a further stab at it, it is useful to start from the lucid and semantically explicit proposal put forth in Büring (2004). Büring develops a situation based approach to WC that keeps A and A’ binding rigorously separate (which he implements through two distinct operators and indices designed not to be able to directly interact with each other). He discusses the possibility of grounding his proposal in a specific semantics for pronouns (as opposed to traces), but doesn’t pursue this possibility.
I take on Büring’s challenge and develop an alternative that rides on a combination of dynamic semantics (DS) and event semantics (ES). On the one hand, DS has a natural way of distinguishing normal variables (linguistically realized as traces of movement) from dynamic ones (linguistically realized as pronouns). The latter can be bound (i.e. get an antecedent) only if such an antecedent has been properly ‘activated’ in previous discourse, typically across a conjunct. Now, in ES verbs are analyzed as properties of events and arguments are fed via theta-roles in a structurally determined fashion. As a result, simple clauses are split into a series of conjunctive statements (e.g. John kissed Bill = there is an event, John is its agent, and …). If we couple this with the standard dynamics for propositional connectives (and, if, etc.), and the ensuing ‘accessibility hierarchy’ (à la DRT) the crossover problem seems to take on a new light, and we may begin to see ways of freeing grammar from crossover specific provisions. While this basic idea might ultimately be too simple to be true, it seems worthwhile taking a close look at it.
