Abstract

The basic claim of conceptual-role semantics is that the meaning of a sentence is its conceptual, or functional, role within the language. Roughly, this role consists in the inferential relations to other sentences, the causal relations to perception and behavior, and expressive relations to beliefs and belief-states. CRS is a relatively new approach in the philosophy of language that has not yet been sufficiently explored. I first explore ways to organize competing theories of meaning in order to find a place for CRS within theories of meaning. I then critically examine two approaches to developing CRS. These approaches are found in the work of Harman and Brandom. From here, the major points and arguments in the debate over CRS can be distilled. There are two arguments against CRS that are the most significant, The first argument is that CRS cannot account for the compositionality of language. Compositionality is needed to explain both the incredible creativity of language and the understanding of the meaning of these novel utterances. The explanation is that one understands the meaning of a novel utterance because its meaning depends upon the meaning of its parts. The second argument is that CRS must be committed to semantic holism, the idea that the meaning of any one expression depends on the meaning of all other expressions. This idea is appealing, but raises many problems including the problem of meaning instability. Lormand has attempted to resolve this problem by appealing to a multiple-meaning account. I argue that the problem of instability returns in Lormand's account via the inferential links of participating representations. The resolution of both the problem of compositionality and meaning instability requires a re-examination of idea that some inferences can be justified by meaning alone. I follow Boghossian in arguing against Quine that a notion of analytic inference can be maintained. However, analytic inferences are not sufficient to account for meaning, Analyticity is neither a problem nor a solution for CRS. With this re-examination, I argue that neither counter-argument succeeds, and the possibility remains open for developing a successful approach to CRS