A Semantics for Temporally Dependent Referring Expressions

this paper, we propose a unified treatment of these cases of temporal reference by appealing to a different ontological framework. Musan [Musan, 1995] makes the hypothesis that some of these sentences with quantified temporal NPs actually express something about stages of objects and not objects themselves. The distinction between entities and stages is inspired from Carlson's dissertation [Carlson, 1980] , but it can actually be traced back to works by philosophers such as [Whitehead, 1929] or [Quine, 1960] , who hold the view that every material object reduces to a spatio-temporal process. In a similar perpsective, more recent studies consider entities we speak about as essentially temporal [Noonan, 1976, Heller, 1990] . As far as we know, these ideas and their implications for formal semantics have been entirely overlooked. They seem to suggest a very natural revision of classical semantic intepretations, one that allows temporal relations on predicate arguments to be expressed. We shall investigate such a revision, here. For instance, we propose that example (2) be given the following logical form: #t, e, y(stage(t, y) hostage(t) attend_the_party(e, y) 1 In words, the hostage "episode" t of some entity y is temporally independant from the event e described by the verbal predicate (we could have t < e or t e). Alternatively, one might posit that verbs too are only predicated of stages of individuals (so stages are all there is): #t 1 , t 2 , t 3 (stage(t 1 , t 3 ) stage(t 2 , t 3 ) hostage(t 1 ) attend_the_party(t 2 ) t 2 < now) (6) This representation makes it clearer that the intepretation of any element of our domain of interpretation is temporally extended. This amounts to considering everything as a kind of event; this is line with ..