L'application des concepts de la linguistique à l'amélioration des techniques d'analyse de contenu [The application of the concepts of linguistics for the improvement of the techniques of content analysis]

The paper deals first with the epistemological status of content analysis going on later to outline a concept of discourse which would allow one to dispel the ambiguity surrounding this status. Finally, the author sketches the mechanism of automatic discourse analysis in accordance with these premises. Inheriting from its linguistic and literary origins, content analysis feels the need to tackle the problem of transmitting sense effects which linguistics has been unable to deal with since its constitution as an independent science. Reactions to this state of affairs range, on the one hand, from systematic ignorance displayed by linguistics in its use of classes of equivalence worked out from statistical counts and based upon arbitrary a priori classifications and, on the other, to the reintroduction of linguistic concepts with a literary bias. The author demonstrates that the two approaches have a common epistemological basis : An explicit or implicit assumption of homogoneity, based in this instance upon the unity of the human mind and upon the existence of a form of unity implying the homogeneity of the referent, i.e. society. However, as it is difficult to uphold an assumption of this kind in view of the heterogeneous character of reality, the author suggests that one introduces into content analysis a concept of discourse which can be defined as a sequence of variable dimension referred to the conditions which determined its production. Essentially, these conditions concern the properties of discourse relative to the place and position of both speaker and receiver. By place is meant those situations which are attributable to speakers using the prevailing mode of production in which they are capable of taking positions conjecturally. A theory of discourse of this kind eventually raises the problem of the syntactical relationship between what is as yet unasserted (but capable of being asserted) and the mechanisms of assertation. It also elaborates various laws enabling one to locate utterances or minimal parts. A technique for automatic analysis based on these principles would entail three phases : One would need to establish firstly a corpus of reference discourses of a similar type on the basis of a scientific examination of the circumstances of production : secondly one would need to make allowance for the influence of linguistics by constructing a «reconnaissance grammar-system» enabling one ultimately, by using transformational rules, to plot a graph of minimal utterances interrelated in such a way as to allow one to reconstruct the original discourse ; and lastly, one would need to evolve a method of systematic comparison which would apportion the sub-sequences in accordance with classes of equivalence, the definitions of which will emerge in the course of analysis.Pêcheux Michel. L'application des concepts de la linguistique à l'amélioration des techniques d'analyse de contenu [The application of the concepts of linguistics for the improvement of the techniques of content analysis]. In: Ethnies, volume 3, 1973. Linguistique et relations interethniques. pp. 101-118