Abstract

I survey three analyses of metaphor, offered by Goodman, Rorty and Davidson, and find the first two unsatisfactory. To explain metaphor Goodman posits a relation--metaphorical reference--which is at best only an example of the phenomenon to be explained, and which at worst fails to distinguish metaphors from non-metaphors. Rorty's analysis casts metaphor as a linguistic object operating on verbal dispositions, but fails to account for the way in the interpretation of a metaphor depends upon an interpretation of its linguistic meaning. On Davidson's theory a metaphor is a linguistic item that occurs in conversational space ; the interpretation of a metaphor thus relies on an appreciation of both its semantic features and the purposes of the speaker who is deploying it. In this way it illustrates several key Davidsonian thesis about interpretation: that theories of interpretation are designed to explain how speakers converge on shared beliefs, and how they do so using systematic descriptions of each other's utterances; that such descriptions are meant to apply to general, manifested behavior, and so evidence of a non-linguistic nature may be used to construct or evaluate interpretations; and that linguistic and non-linguistic evidence may interact in unusual ways. It is one consequence of such a theory that metaphors will lack distinctive, semantically-specifiable meanings