Abstract:

This dissertation investigates the semantics and pragmatics of the English and Japanese perfect. It first argues that the perfect is a stativizer and provides a monosemous analysis of the meaning of the English perfect by modifying the standard Discourse Representation Theory's approach to the perfect (Kamp and Reyle 1993, De Swart 1998) and incorporating into it some semantic underspecification. In particular, it is argued that the perfect introduces a state whose category is semantically an underspecified free variable and that the underspecification of that category triggers inferences that lead to the various interpretations of the perfect. Secondly, an analysis of the meaning of the Japanese aspect marker -te-i- , which corresponds to both the English progressive and the English perfect, is presented. It is shown that the progressive and perfect readings of -te-i- derive from the single meaning of the two morphemes that compose -te-i- , the imperfective marker -te- and the stativizer -i- . Third, a model of the inferential process that leads to the distinct interpretations of the perfect in both English and Japanese is provided. This model is validated by two corpus studies of English present perfect and Japanese te-i-ru (nonpast perfect) examples. These studies bring out two further results. First, the kinds of inference rules which addressees must use to arrive at the interpretations of perfect sentences are very few in number. Second, by invoking these inference rules, the use of the perfect helps establish discourse coherence. This thesis finally compares the uses of the present perfect and the past tense. I show that the Japanese marker -ta is a past tense marker, not a present perfect marker and that the difference in invited inferences between the English and Japanese past tenses can be explained in neo-Gricean terms. Second, I revise current definitions of reference time and show that differences in use between the past tense and the perfect follows from the fact that the eventuality described by the main verb and its arguments is anchored to reference time only when a past tense is used.