The only way to access those grids of interpretation is to attempt to ‘talk nuclear’ with local inhabitants and with people who work at the plant. Accordingly, I was led in the course of my researches to pay most attention to the act of speech. That is to say, I conducted the study mainly through live interviews, either with individuals or with groups, always making due allowance for the context in which the enunciative act took place. Indeed, that act must never be considered in isolation from the circumstances in which it is performed, from the place where it occurs, or from the social and professional identity of the speaker or speakers. A further characteristic is that it sets up an interaction with the interlocutor (the interviewer), locating each party within a network of relationships that itself requires decoding. Adopting this approach involves as it were taking all the material available to you and incorporating it in your analysis of a person's speech: the combinative aspect of discourse sequences, the use of certain words rather than others, the position and weight of silences, the proportions of narratable and memorisable elements in the conduct of the account, the inflections introduced by the reflexive nature of narrative exchange.

There is quite rightly a big question mark over the soundness of such an approach when the area under investigation covers everyday practices to which people resort unthinkingly and in connection with which their speech is necessarily forgetful, or when the researcher is trying to capture thoughts, feelings, and private, secret areas of behaviour that people likewise find difficult to express in speech.