The Japanese postpositional particle wa has probably received more attention within the linguistic literature dealing with that language than any other single element of Japanese grammar. The present paper is a further contribution to the long and honorable series of attempts to come to grips with the function of this highly frequent, but still poorly understood, linchpin of Japanese grammar.

The particle wa may occur with subject nouns, as in (1), with other parts of speech, and may even be attached to entire sentences. Because of the variability of the elements with which wa can be associated, the element to which wa is attached is referred to throughout the present paper simply as X.

Any theory of semantics must account for the role of inference in the interpretation of a sentence, however, it is not clear at which level of the grammar this inferencing capability must be represented. In Sperber and Wilson (1986) for example, inference plays an important role in the interpretation of an utterance in discourse, yet their theory does not integrate lexical representations with discourse and inferencing. Jackendoff (1987, 1990) on the other hand, while giving explicit representations for sentences and lexical items, does not address the question of discourse and inference. In this paper I will argue that it is essential to take account of discourse-level factors when analyzing sentence meaning, and indeed word meaning. In particular I will show that the meaning of the auxiliary verb fεla. in Bengali can be properly represented only if such discourse-level factors are included.

One proposal which has played and continues to play an important role in the syntactic analysis of Romance prepositional phrases is that not everything which looks like a preposition is in fact a true preposition. First proposed by Vergnaud (1974) for French, the idea is that some apparent Ps actually act as simple Case-marking on an NP; in Vergnaud’s case, this was the à associated with Goal arguments. As a result, the syntactic representation of Goal arguments involves no PP-node, but rather an NP with an inflectional Case-marking à. Vergnaud’s proposal was developed further for French à by Kayne (1975) and Jaeggli (1981), and is assumed to be valid for French de by Elliott (1986). Chomsky (1986) also suggests that English of is in many cases the realization of genitive Case, and Demonte (1989) argues that many instances of Spanish a or de phrases selected for by verbs are not PPs but rather NPs marked with Case.