I address the assumption that communicative interaction is made possible by knowledge of a language. I argue that this assumption as it is usually expressed depends on an unjustified reification of language, and on an unsatisfactory understanding of "knowledge'. I propose instead that communicative interaction is made possible by (Rylean) know-how and by the development of (Davidsonian) passing theories. We then come to see that our focus ought to be, not on propositional knowledge of a language which we internally represent, but on the practical application of know-how in our understanding and interpretation of others.