Several years ago, after publishing one or two small-scale studies relating to children's requests, I sat down to think through the relationship between these studies and the spectrum of research carried out on children. The approach I had been taking was a minority interest, one which placed its methodological emphasis on the rigorous analysis of small numbers of sequences. Here, ‘rigour’ meant identifying the details within these exchanges which documented the understandings of the participants involved. It seemed fairly clear that by proceeding in this way one could tap into forms of interactional organization which seemed quite powerful, but the question arose as to how these findings meshed in with the large amount of other knowledge about young children's behaviour which had been generated by alternative and more conventional modes of research. This was the issue I sat down to address.

I found this task very difficult. Much of this other research was heavily quantitative, and thus generated through the application of various kinds of pre-specified taxonomy to the flow of what took place in children's interaction. These taxonomies usually came with some evidence of high reliability, in the technical sense, but there was little compelling basis for selecting one taxonomy rather than another. And, more importantly, in spite of pioneering work by people such as Carter (1975, 1978), there was little systematic attention paid to finding ways of figuring out the significance which these different forms of speech act had for the children themselves.