Abstract

In Section 1, we questioned the evidence for iteration of tenses, even with abstraction. To permit abstraction would in any case risk neutralizing our distinction between tensed and untensed sentences. Sequence of tense phenomena, far from supporting iteration, were incompatible with it. Instead, we argued, tense always retains its full deictic character; tenses never have scope over each other. The future modal WILL is exceptional (Section 2), but abstraction is not required to deal with this.An important suggestion, first made in Section 1 and developed further in Section 3, would restrict quantification induced by the interpretation of Past to a domain determined by pragmatic factors. When Past serves as a reference time, e.g. when it forms part of would, as inSam believed that he would be in BostonJames bought a stamp that would be priceless then a previous occurrence of Past serves to define the domain relevant to the quantificational element of Past. Without semantic binding or reference, the two cases of Past in such sentences will thus be satisfied by the same interval in the past.The extension of this idea in Section 3, permits an account of the supposed incompleteness of the past tense without invoking a hidden THEN in e.g. I turned the lights off. In this case, restriction of the relevant domain does not proceed via an earlier occurrence of Past but depends solely on discourse factors.In Section 4, the analysis of have given in Part I is modified to introduce an element of ‘pastness’. The relevant sub-interval must be non-final-and perhaps need not even be a sub-interval of the evaluation interval itself. Empirically necessary, this change makes the incompatibility of Pres,have and a definite adverb like at 6 o'clock in a single sentence even harder to explain semantically. However, the vague restrictions on domain imposed by the present tense and relevant to the sub-interval required by an embedded have below it seem clearly pragmatically incompatible with the explicit restriction of that same interval to a very specific interval in the past — a restriction which, we argue, is imposed by a definite time adverb.Thus, Part II represents a consistent, though entirely informal plea for the embedding of the semantics of Part I in a framework in which pragmatic considerations can interact freely with the semantics, to restrict the domains within which quantification is permitted. It is in effect a plea for the introduction of partial models along the lines suggested by Barwise and Perry (1981). The advantages of the basic distinctions between tense, aspect and adverbs introduced in Part I would, we have suggested in the second part, be yet more obvious in such a framework. Formalization in a system of partial models must await further work