Abstract

Disputes between realists and non-realists about the nature of truth and objectivity have appeared intractable when stated in metaphysical terms. Michael Dummett has proposed that one can work one's way into these issues by considering the theoretical demands of an adequate account of meaning in a natural language. Specifically, he argues that questions about the nature of truth can be profitably recast as questions about the role of the notion of truth in a full explanation of human linguistic behavior. Dummett develops a specific conception of a meaning-theory for natural language--a conception driven by his view of language as a rational practice--and argues on the basis of this that a realist, i.e. non-epistemic, notion of truth is unacceptable. This thesis is an attempt to motivate, rationally reconstruct, and defend Dummett's conception. ;I begin by defending Dummett's characterization of realism in terms of commitment to classical logic. I then outline the formal adequacy requirements Dummett lays on a meaning-theory. These formal constraints follow, on Dummett's view, from more substantive requirements, namely that a meaning-theory be genuinely molecular and that a meaning-theory embody a substantive conception of truth. and derive, in turn, from Dummett's assumption that speakers possess a body of semantic knowledge which can be isolated and studied apart from the speaker's general empirical knowledge. I argue that it is the issue of the psychological reality of semantic knowledge that divides Dummett from truth-conditional theorists like Donald Davidson. The two constraints are discussed and defended separately. I conclude with a sketch of the sort of meaning-theory I believe Dummett to be envisioning, and in a brief coda, connect his meaning-theoretic conclusions with his metaphysical views.