This thesis consists in a compared and thorough epistemological analysis of the whole of Saussure's published work and of a significant part of Hjelmslev's, Jakobson's, Martinet's and Benveniste's. Its aim is to show that although it has acknowledged Saussure as an influence, European structuralism is not Saussurean. In doing so, it aims at bringing out the specificity of Saussure's problematics and its stakes for linguistics and more widely for the sciences of the human being. Saussure's problematics had made a theoretical apprehension of language (la langue) possible for the first time in the history of linguistics. On the other hand, the problematics of structuralism is entirely empirical, so that its scientificity is in fact a product of a scientific ideology (idéologie scientifique), in Georges Canguilhem's sense of the word. The very core of this radical divergence in the problematics lies in the lack of structuralist theorization of the sound/meaning relation, and correlatively, the misunderstanding of Saussure's concept of system. This system then becomes a structure, that is to say, as we try to show, the structural apprehension of an object whose commonly held and evident definition (language as an instrument of communication) is not questioned. Thus, Saussure's etiological problematics, which is constitutive of the concept of language (langue), is echoed by an analytical problematics, which itself leads to the construction of an object (form or structure), instead of a concept. More precisely, the structuralist problematics is idiomological. Thus, it misses the difference there is between language (langue) and idiom, a distinction we thus try to demonstrate as necessary as well as constitutive for the theorization of language in the narrow sense of the word (la langue), and beyond it, of language in its larger sense (le langage), particularly in the context of an articulation between linguistics and psychoanalysis.
