In a series of early essays, Derrida examines theories of language in the philosophical tradition. He finds a persistent privileging of speech at the expense of writing, which he attributes to the conception of meaning as presence, as an ideal object present to inner perception, and of the sign as a transparent unity of signifier and signified. I trace the notion of "presence" to Heidegger's 'destruction of the history of ontology'. Heidegger criticizes the tradition for privileging perception as the only means of access to being. I accept Heidegger's criticisms of the tradition, but offer an argument to show that he is wrong to deny the priority of scientific ontology. Derrida denies three theses in traditional theories of language: (i) Language has a relation to the world (the reference relation). (ii) The intermediaries in this relation are meanings, which are ideal objects present to inner perception. (iii) Complex sentences derive their relation to the world from simple names. By a close analysis of Derrida's arguments, I show that Derrida is correct to reject (ii), but that once that is done, (i) and (iii) can be retained.