Computer Science Colloquium, 2003-2004

Ron Artstein
Department of Computer Science, Technion
November 26th, 2003

Quantificational arguments in temporal adjunct clauses

This talk concerns a peculiarity of temporal clauses (e.g. after John resigned), which allow quantificational expressions like each executive to interact with expressions outside the clause. Sentence (1) below is true if a different secretary cries for each executive; such an interpretation is not possible for sentence (2), where the subordinate clause is not temporal.

  1. After each executive resigns, a secretary cries.
  2. If each executive resigns, a secretary cries.
I use a logical language with explicit temporal variables, in which temporal modifiers form generalized quantifiers (proposed by Pratt and Francez 2001, Linguistics and Philosophy, to account for nested temporal modifiers as in John drank before each meeting during most conferences). A flexible architecture for the semantics permits the formation of a temporal generalized quantifier in the dependent clause before the application of a quantificational argument, giving the argument scope outside its clause. The semantics derives three kinds of readings for temporal clauses: dependent-time, where the evaluation times of the main clause depend on a quantifier inside the temporal clause; single-time, where the main clause is evaluated at a single time regardless of quantifiers in the temporal clause; and aggregate-time, where the main clause is evaluated in an interval which encompasses the individual times quantified over by the temporal clause. The semantics can be extended to deal with long-distance temporal dependencies (I saw Mary in New York before she claimed that she would arrive) as well as with locative clauses (A tree grows where each car had crashed).


Shuly Wintner
Last modified: Mon Nov 17 12:05:15 IST 2003