From Heritage and Greatbatch, 1991: 100

3.9 Child: Have to cut the:se Mummy 1.3 Child: Won’t we Mummy 1.5 Child: Won’t we Mother: Yes It is clear that the child’s second and third versions of the initial question dis- play that she has ‘noticed’ the absence of the mother’s answer. Moreover, her repeated attempts to solicit an answer display her orientation to the norma- tive expectation that an answer should follow a question. Thus what seems on first inspection to provide evidence which undermines claims about the prop- erties of paired action sequences for example, that second parts should follow first parts, actually displays the participants’ orientation to the nor- mative relevance of those properties. So too in extract 3.8, we can see that the IR’s withholding of a further question and the subsequent elaboration by the IE demonstrated that they were orienting the expectation albeit belatedly in the case of the IE that answers should be extended. However, participants’ orientation to normative expectations is not only exposed in those circumstances in which a specific kind of turn is noticeably absent: it is also revealed by the way in which participants engage in non- normative activities. One of the overriding norms of news interview interaction is that it is the IRs who ask questions, and thereby guide the interview. It is not expected that the participants will initiate their own topical agenda. But there are circum- stances in which IEs do offer comment or opinion which is not directly solicited by the IR. Some news interviews are conducted with more than one participant; and to generate a lively exchange, or to ensure that a wide range of perspectives will be represented, it is often the case that the IEs will hold different opinions. In such interviews it is not unusual to find disagreement between participants. This disagreement can result in attempts to by-pass the IR and address directly what the other IE has said. Yet when this happens, the IE asking the question will seek permission from the IR, thus displaying that they are aware that their question deviates from normative conventions. Extracts 3.10 and 3.11 provide illustration. The two participants in this interview hold differing opinions about imprisonment as a deterrent to crime.

3.10 From Heritage and Greatbatch, 1991: 103

IE 1: … and therefore I’m not going to accept the criticism that I haven’t tried to help victims = = I’ve . been trying to help them 0.2 off and on for twenty-five years. = 62 CONVERSATION ANALYSIS AND DISCOURSE ANALYSIS ?: = ·hhhh = IE 2: = Can I- can I say something abou t this IR: Yes in deed. 0.5 IE 2: e:r 0.7 As 0.5 Frank . Longford knows so well ·hh er my views … continues

3.11 From Heritage and Greatbatch, 1991: 104

IE 1: … there was no evidence whatever that stiffer penalties diminish crime. = IE 2: = Can I make a point about that. = hhh, Which is that if only this country … continues In these cases IE2 responds to the point made in IE1’s answer to a prior question. However, instead of simply embarking on a response, IE2 seeks per- mission from the IR. In extract 3.10, this is treated as a literal request in that the IR gives permission; in extract 3.11, IE2 makes a token request, in that he begins to address IE1’s point before being authorised by the IR. We can see, then, that the normative expectations which underpin news interview interaction are of a markedly different kind to those which inform more con- versational interaction. These normative expectations are most clearly visible in the way that participants restrict themselves to – and exhibit an expecta- tion of – a limited range of primarily question–answer activities. Context and interaction In this section I want to make some points about the relationship between sequences and possible characterisations of participants. A common sense assumption would be that prior to analysis, it is necessary, or at least useful, to have some understanding of the participants: their history, their relationship to each other, their personality, the topic of their talk or the context in which they are speaking. It seems intuitively sensible to assume that all these things must cohere together and influence the way people speak, and therefore, that it is appropriate to take account of these factors when analysing talk. In CA, however, there is a principled reluctance to draw on ethnographic characteri- sations of the setting and its participants in the analysis. The first reason for this is that there is a very real problem in formulating what the context might be Schegloff, 1991, 1997. This is because, as we have seen in earlier chapters, the description of any event, situation, place, or person can be done in a variety of ways Heritage, 1978; Schegloff, 1972b. The point is that any description or reference is produced from a potentially inexhaustible list of possible utterances, each of which is ‘logically’ correct or ‘true’ by any test of correspondence. How is the analyst to decide which particular version of the context is most appropriate? METHOD AND CRITIQUE 63 关 兴 This is not to say the context of interaction is analytically irrelevant. Context is a relevant issue for the participants. During interaction speakers orient to, and display to each other in the design of their turns, what they understand to be the salient features of their context. And in the same way that we can discover speakers’ own interpretations by examining the design of their turns, so we can discover what they take to be the relevant features of the context of their interaction. We can investigate if the participants’ turns are designed to display, for example, that they are orienting to each other’s work or gender identities; or we can explore how the relevance of the relationship between the participants is invoked, if at all, and so on. However, features of the interaction may themselves be the relevant context for any subsequent contributions. For example, what might be relevant to the way an utterance is produced is the activity performed by the prior turn: a ques- tion, an excuse, a repair, an instruction, and so on. Consider extract 3.9 again. Child: Have to cut the:se Mummy 1.3 Child: Won’t we Mummy 1.5 Child: Won’t we Mother: Yes This child’s first turn is the first part of an adjacency pair. However, after 1.3 seconds, the child produces two further turns, both of which display the under- standing that the conditionally relevant second turn has not been produced. Her turns suggest that the salient context, on this occasion, is an accountably absent turn. Because participants’ turns will exhibit their analysis of relevant features of the context, the analyst is provided with a significant methodological advan- tage. We do not have to speculate what might be relevant, we can see directly what is relevant to the participants. And as the ‘relevant context’ may be as immediate and transitory as the prior turn, CA treats context as a fluid and contingent achievement: a notion like ‘context’ will have to remain substantively contentless, and uncommitted to any prespecified referent and be instead ‘programmatically relevant’ [that is] relevant in principle, but with a sense always to-be- discovered rather than given-to-be-applied. Schegloff, 1987b: 112 The second reason that conversation analysts do not rely in their research on a characterisation of the context is because interaction is viewed as a domain of activity in its own right, and not a reflection of individual personalities or social or cultural constraints. Following Goffman, who was the first to focus on the organisation of mundane, everyday activities, interactional practices are regarded as exhibiting an order which is not reducible to the personality, intentions or mood of the speakers, nor the social or cultural context in which 64 CONVERSATION ANALYSIS AND DISCOURSE ANALYSIS