In this essay I respond to criticisms of semantic externalism that Noam Chomsky has presented in several recent publications. In the first section of the essay I present reasons to think that there are virtues to both semantic externalism and to semantic internalism, and that these two views are compatible. Linguistic items and intentional phenomena might have semantic content which is determined by factors external to their possessor, and also have semantic content which is determined by factors that are internal. In the second section I respond to Chomsky's argument that the cognitive sciences are internalist, and hence semantic reference has no place in naturalistic inquiry. I argue that the cognitive sciences are not exclusively internalist, that they are significantly externalist. In my discussion I attend mainly to the case of scientific studies of vision. In the third section I deal with issues specifically concerning language. I counter Chomsky's arguments that semantic reference can play no role in the scientific study of language; I argue that it in fact does play a role in significant developments that should be regarded as belonging to the scientific study of language. I also present considerations in defense of semantic externalism with regard to proper names and natural kind terms, and in defense of there being a social factor to semantic content.

According to externalism, reference is a relation between uses of an expression and features of the environment. Moreover, the reference relation is normative (constitutive of correct semantic use), and the referential relata of our expressions are explanatory of successful language use. This paper largely agrees with the broad conception underlying externalism: it is what people do with words that makes them have the references they have, and the world constrains what people can successfully do with words. However, the paper strongly disagrees with the details (at least as usually presented). A centrally important feature of what people do with words is how they use them in inferential contexts. When due attention is given to the reference-determining role played by inferential properties of expressions, I argue, we arrive at a more satisfactory account of semantic norms and explanations. Much of the argument is based on a detailed look at the language of chemical classification used in the late 19th century.

In recent years Donald Davidson has outlined main features of a “unified theory” of language and action. The article tries to lay open the central theoretical steps one has to take from his “radical interpretation” to his theory of rationality and his triangulation model of externalism. It is argued that Davidson's reinterpretation of Tarski's T - sentences can be used to show a fundamental symmetry between representation and expression of propositional contents. Yet, his theoretical framework has to be enriched to deal with the problem of contextualism that arises from his redescription of utterance meanings. The paper shows in order to elaborate Davidson's claim that rationality is a normative concept one has to address the question of an internal relationship between radical interpretation, rationality and externalism.