Formal Method in Discourse Analysis

Developing a workable method of discourse analysis matters, presumably, and if it does, because it would allow us to demonstrate the play of power in language, the play of ideology. To demonstrate means being able to substantiate arguments in the public arena – to politicians, in courts of law, in hospitals or prisons or museums. What we envisage is something like the possibility of proof. We envisage, not a political process of judgement, but procedures of demonstration which could be mechanically repeated to produce identical results, and so bear witness to the underlying structure which causes them. Discourse analysis, like other uncertain knowledges, aspires to a state of hardness. My tone of irony covers a real ambivalence about this project, these projects, and certainly I intend no easy dismissal of them. Here I want to raise a series of questions about some of the work being done in and around the Dutch Konteksten series, especially by Tony Hak and Brian Torode;1 and to see whether it might be possible to define some of the limits touched by this raw, exploratory, interesting work. This is not, even in the case of individual writers, a unified body of research, but it has certain common points of reference. One of them is the ethnomethodological conversation analysis of Sacks and Schegloff; the other, more properly linguistic, is the work of Michel Pêcheux, and to a certain extent the analytic methods of Zellig Harris. The attraction of Pêcheux’s work is, I think, that it promises a rigorous and quasigrammatical account of presupposition: that is, the concept of the preconstructed promises direct access to ideology insofar as it appears to be equivalent to the Gramscian categories of 'common sense' – the obvious, the unthought, the taken-for-granted; and insofar as it both embodies these in a particular grammatical construction and relates them to a particular process of subject formation