ABSTRACT
Alec McHoul has persuasively argued that there is no principled reason why conversation analysis (CA) cannot be used on fictional texts.2 As Hammersley and Atkinson have demonstrated, a preference to work with ‘naturally occurring data’ is just that.3 No data can ever be said to be more ‘natural’ than any other-no data are untouched by human hands. So telephone conversations, for instance, are not intrinsically better sources of data than, say, research interviews, official statistics or fictional texts. In principle, it would seem, there are only good and bad methods of analysisnot good and bad data.
