| Until recently, it has been widely assumed that pragmatics is independent of the computation of syntax and semantics. However, there is a movement in the direction that at least some of the pragmatic effects should be analyzed in semantics. Some lexical items have been identified to be closely tied to implicatures. Also, language has linguistic means to indicate the nature of evidence for an uttered statement, evidentiality. The existence of these items raises a question of where exactly is the boundary of semantics and pragmatics. Japanese has a rich paradigm of morphological manifestation of implicatures and evidentiality, which also has a rigid syntax. By investigating Japanese discourse items which give rise to pragmatic effects, this dissertation sheds a new light on the issue of how these pragmatic notions are represented in language.; Contrastive-marking is analyzed as an indication of lexically specified Gricean implicature. Adopting Schlenker's (2003) notion of 'shiftable indexicals', I argue that Contrastive-marking also contains a shiftable indexical because the implicature associated with Contrastive-marking can be relativized to an attitude-holder other than the actual speaker of the sentence. This association between the implicature and the attitude-holder cannot be established in certain constructions, adjunct clauses and relative clauses. Hence, I argue that the computation of Contrastive-marking involves an island-sensitive movement of an operator.; I also present a parallelism between Contrastive-marking and Evidential-marking with respect to the distribution among adjunct clauses. I take this fact to show that both Contrastive-marking and Evidential-marking express some attitude toward a closed proposition, following Johnston's (1994) analysis that semantics of temporal and if-clauses involve an event quantification, while that of because-clauses is a relation between two particular events. Furthermore, I argue that because-operator and evidential-marking are context-shifters that can bind shiftable indexicals.; Toward the end of the dissertation, I give an analysis of a particular discourse item darou as an evidential marker that has a modal flavor, which indicates the speaker's bias toward the embedded proposition and the bias is based on non-observable reasoning that the speaker has. The data also toss a question to the discussion on levels of meaning to which discourse items commit. |
