In the following chapters we will be examining two related but somewhat different kinds of data. The first kind is the recordings and transcripts of ordinary members of the public calling the British Airways flight information service. The second kind of data consist of recordings and transcripts of people talking through a telephone system to what they believed was a prototype speech-based computer information service. Our examination of these materials will focus on the systematic properties of the ways that speakers verbally interact with each other, or with what they think is a computer. With this goal in mind we are adopting a broadly conversation analytic approach to empirical investigation. There have been other attempts to adopt a conversation analytic approach to inform the development or analysis of interactive systems (for example, see the collection of papers in Luff et al., 1990). However, the use of a conversation analytic approach is still sufficiently novel to require us first to address some methodological issues, in particular, some of the assumptions which underpin empirical work in conversation analysis. Our emphasis upon explaining the principles which inform empirical work means that we will not try to review here the major findings from conversation analysis. In subsequent chapters, however, we do draw upon and discuss the research literature on turn-taking, the organisation of overlap, preference, closing sequences and repair in conversation.