The strategic concern of the discourse theory in Foucault

In order to fully understand Michel Foucault's archeology of knowledge it is necessary to put it not only in the context of the discourse theories of the French school, but also in that of the linguistic theories of science. On the one hand, Foucault rejects the representational theory of scientific language (developed by Carnap and others). In his analysis of the discourse of the human sciences, with its complicated definition of the statement (enonce), he aims to show that we cannot limit the language of knowledge to a vocabulary. On the other hand, Foucault's approach is a serious challenge to structuralism because it is impossible to analyze the language of savoirs without taking into account the statement's correlate (the referent issue). If the discourse of sciences is an artificial vocabulary created by means of the purification of the natural language, Saussure's semiology could not compete against the conceptual dynamics of representational semiotics, and should admit that, in the cognitive usage, signs are representations. Therefore, Foucault admits that the statement has a correlate, that this correlate is not a physical object and that it functions as a historical structure of the discourse. This structure (episteme) can be seen as a norm; the changing of this norm cannot be identified but in the archive and its description is always retrospective, structural and archaeological.