Since our theory of metonymy interpretation relies on a full grammar and a Gricean pragmatic component tied to itm, a detailed overviews of the set-up of the paper might be of helg. We first provide an intuitive description of our data and the function of representational metonymies therein (1.1), followed by some initial methodological considerations (1.2). Then we present three arguments which show that only a pragmatic theory of representational metonymies will work (2).
In ch. 3 we develop the syntax-semantics interface for the description of metonymies following Chierchia and McConnel-Ginet (1990, 2000) where SS, LF and lf are given for a small fragment of English. It is argued that we need a context-based intensional semantics incorporating characters in the manner advocated by Kaplan (1977) and Stalnaker (1974). Arguments demonstrating the role of contexts in the resolution of mentonymies are also provided. Ch. 4. describes the model M used for semantic interpretation. In addition to traditional parameters like the interpretation function V, M contains a modal base as argued for by Kratzer and others (1981). Modal bases depend on contexts; they are used for the interpretation features of "non-classical" indexicality as exhibited by context-dependent lexical items, natural modalities or expressions for which an indirect interpretation must be given. As can be expected, ch. 4 also contains a definition of the interpretation function V for lf-expressions as well as definitions of truth, validity, and entailment. Some methodological remarks concerning the relation of truth in M and indirect interpretation lead up to the semantics-pragmatics-interface used for the interpretation of metonymies (ch. 5). Chapter 5 contains the newly developed tool for metonymy resolution, the algorithm for reconstructing false lf-formulas, and a formulation of Gricean conversational implicature as default based on M. Ch. 6 develops our account of compositional semantics for metonymies. We show that the formulation captures representational metonymies and that it ca be easily generalised to other kinds of metonymies. The discussion of more complicated examples in ch. 7 shows that developing theories of compositionality for complex metonymies presents a real challenge, despite the efficient now tools developed. We conclude with remarks concerning future research (ch. 8), several problems remain sill to be solved; we hoe to handle these on the basis of the theory developed.
(~519 k)