Reading Guide to: Atkinson, J and Heritage, J (eds) (1984) Structures of Social Action: Studies in Conversation Analysis, Cambridge: Cambridge University Press.

Chapter 2 Sacks, H 'Notes on methodology'

[These notes were apparently taken from lectures delivered by Sacks in the late 1970s, and transcribed and edited by Gail Jefferson].

There is a new domain of research -- ethnomethodology/Conversation Analysis -- which 'seeks to describe methods persons use in doing social life' (21). Apparently results have already shown that these social activities can be formally and methodically described, and are generalisable 'in intuitively nonapparent ways' and reproducible. These results show that natural observational scientific methods can be used in sociology [and in linguistics, according to the Orientations section by Atkinson and Heritage]

Conversation analysis offers a bottom-up way of understanding social order, rather than trying to isolate general principles of order which are then largely indifferent to content, the details of which are usually taken to be 'idiosyncratic or random'. Parsons's notion of the 'unit act' serves as the example for Atkinson and Heritage (Orientations, page 17), but Chomsky is also mentioned.

The old top-down approach tends to focus on the 'big issues' rather than on the 'mundane, occasional, local' (22), for methodological reasons as well. We can proceed instead by deciding that 'whatever humans do can be examined to discover some way they do it, and that way will be stably describable' (22). One implication is that we can examine just about any local and mundane activity, since 'there is order at all points' (22).

[Some strange assumptions are made here as justification for this approach -- Sacks recommends that 'we figure or guess or decide that whatever humans do, they are just another animal after all', and this leads him to a strangely functionalist point that 'That sort of order would be an important resource of a culture... Were it important for nature to insure that if persons are to be workable things in a society... then a culture might be well so arranged. And then, of course, research might employ the same resources: Tap into whomsoever, wheresoever and we get much the same things' (22). Sacks suggests that the 'fact of order at all points' [we have gone from a supposition to a 'fact'] can also explain why sociological or anthropological research methods get orderly results despite their dubious methods [Schutz has a much better explanation]. It all ends very happily for him -- 'given the possibility that there is overwhelming order, it would be extremely hard not to find it, no matter how or where we looked' (page 23). Further, if there is such inherent order, we can use this to do research. Tremendous circularity here!]

We must not assume that every day conversations are easily dismissed as exhibiting less important functions of communication [Sacks has Sapir in mind here]. If we do not study these mundane uses, we miss many opportunities to understand how human beings 'construct and order their affairs' (24), and if we can describe these formally, we can develop 'another grammar' (25).

Sacks then discusses another problem with conventional social science -- that it can operate only with hypothetical or typical versions of the world. This leaves out much interesting activity [and there is another hint that this sort of omission is deliberate, permitting conventional social science to proceed, enabling conversations to take place with other sociologists]. Sacks claims to be 'using observation as a basis for theorising' instead (25) [the old empiricist claim?]. This helps us focus on things that have happened but which may not be typical. We should be dealing with 'transcriptions of actual occurrences in their actual sequence' (25).

Sacks says that this interest led him to work with tape recorded conversations, which are available for anyone else to examine. Tape-recordings were capable of being replayed and transcribed for extensive study. Of course they omit aspects of conversation, but they are 'a "good enough" record of what happened' (26) [good enough to permit research? But why do this research?].

The intention is to see 'how conversation works... whether actual single events are studiable [sic], and how they might be studiable and then what an explanation of them would look like' (26). We want to transform descriptions into explanations, 'to find the machinery' (27). 'We sit down with a piece of data, make a bunch of observations, and see where they will go... in an unmotivated way' (27). We can thus gain novel findings [The old promise of 'surprises', just as ethnographers claim -- whether you are surprised by this sort of stuff depends very much on how straight you were in the first place, I reckon. If you live in an ivory tower, I suppose it is surprising how resourceful and creative 'ordinary' people can be. If  'surprise' relates to conventional sociological findings, again it all depends on how conventional your training has been, I suppose. Rhetorically, the whole thing seems to work off straw men, I find -- paraphrase Parsons in such a way as to make him look terribly formalist, then surprise yourself by discovering complexity in everyday life -- etc  Atkinson and Heritage in their Orientations section describe the approach as wholly appropriate for a 'fledgling discipline', indeed as 'a hallmark of scientific endeavour and a precondition of genuine discovery' (18), and go on to claim that the approach is now fully justified by the 'independent  adoption of the methodological a range of domains where interaction is the focus of interest' (18) . The whole thing is a testimony to 'the pioneering quality of [Sacks'] work...and [his]..."lateral thinking"' (17). I still have doubts though -- as soon as the 'fledgling discipline' became well known, of course, its methodological orientations would be pursued quite knowingly. As for the quality of Sack's thinking -- judge for yourselves? ].

Chapter 8 Button, G and Casey, N 'Generating topic: the use of topic initial elicitors'

Conversations can be organised in a number of ways, including by 'stepwise progression' (as studied by Sacks) [showing how conversations can take new directions gradually, via topics which bridge old and new], but they can also be divided into limited segments addressing different topics. The segments are initiated by 'topic initial elicitors' (TIE), which allow a topic to be developed 'interactionally and mutually' [This is the normal course of events? If so, this is functionalism?]. A common pattern consists of three 'speaker and turn-alternating parts' (167): (a) an inquiry about the 'possibility of presenting a report of a newsworthy event' (b) a positive response to the first part involving a 'newsworthy-event-report that has the status of a possible topic initial' (c) a topicaliser, which 'topicalises the prior topic initial and provides for talk on the reported event' (167).

[A number of extracts of transcribed (telephone?) conversations are provided here and throughout, carefully transcribed using the most elaborate transcription protocols,  which indicate pronunciation, stress, tone, the size of pauses and so on. Apart from the first example , I have not followed these faithfully, and they hardly seem to be warranted here, since these matters are not relevant to the argument, as far as I can see. A full explanation of the transcription protocols is provided in the Introduction to the whole book]

The sort of remarks that are relevant include utterances like 'Whaddiyu kn : ow', or 'Anything else to report?'. The first speaker can respond to a reply, and encourage with remarks such as 'Oh really?' or 'Yeah?'. The first reply can be a positive one [reporting a newsworthy event] or a negative one 'Nothing', or 'Oh nothing doing'. Positive and negative responses have different consequences for the interaction, and only positives produce the full three stages as outlined above. Nevertheless the three-part structure is 'organisationally "preferred"' [another special term, it seems]. It seems to offer a constraint system on what might be said in future turns. Further examination of these turns leads to 'organisational grounds for a preferred sequential development' (169) [an empirical claim is being made here?, or is this a purely formal set of organisational grounds?]. Actual sequences may be sensitive to topics.

Topic initial elicitors [TIE]. These have a number of key features: (a) they segment talk; (b) they inquire but do not present 'newsworthy events' themselves (170); (c) they provide an 'open, though bounded, domain'.

Segmenting 'regularly occurs' ['soft quantification' here?] 'following closing components, following opening components and following topic-bounding issues' (170) [that is whenever conversation branches, begins or ends]. They can take the form of offering 'further' or 'else' options -- for example as in 'OK. Is there anything else you - happen today of any interest?' or 'What else'. They refer back to the inquiry rather than to previous talk, that is they are relevant to topic organisation. They show how conversations can be closed too, although they can also replace a 'closing component' (171), and permit further conversations, the introduction of new newsworthy items. They are also found in openings, where they ask about 'immediately current events... as opposed to further newsworthy events' (172) (original emphasis). Transcript examples include 'What's doing?', 'Oh hello there, what are you doing?'. They can operate as a bridge between the opening pleasantries and the introduction of the topic, to move from 'inquiries and returned greetings... to... reason-for-call and/or first topic' (173). They help move between segments of introduction and more topic-based conversation. They can be used by both caller or called. Callers sometimes use them 'to delay the reason-for-call presentation' [Why anyone should want to do this remains unexplored, however. I can only think of strategic reasons myself, which involve feigning interest in pleasant conversation with someone for its own sake, while thinking how to get round to asking for information or a favour or whatever. This takes us beyond the mere analysis of conversation, of course, into something approaching the banned 'social structure']. Sometimes [for equally unclarified reasons] callers want to invite the called to initiate a topic of their own.

[As an example of empirical work? Empiricism?] 'It was observed that in closings the segmenting operation of TIE was related to their orientation to topic generation' (174) [This gives you an idea of the laboured and technical style as well]. They can also be used in 'following prior topic-bounding turns' too. Variations are found according to whether they are used in openings or closings, such as whether they refer to 'immediate' or 'further' events. Examples of topic - bounding from the transcripts include 'What's new?', 'What's new with you?' and so on. It is noticeable that topics can be organised into sequences over the course of that conversation, as when topics seem to head progressively towards a closing [some notion of narrative closure here?].

It is clear that TIE only elicit and do not actually initiate topics. They make available the possibilities for topic initiation, which is important for turns after the early ones. They offer both open and bounded 'domain of relevancies' (176). They vary according to whether they are found in openings or closings, whether the topic is to be extended, or new topics introduced. Topics can also be bounded by imposing time limits on questions, such as '... anything else you -- happen today of any interest?'. Note that this does not specify the event itself, but asks for anything of interest: there is no 'commitment to talk on the particular event' (177).

Next turns. Turns are constrained by TIE although there are options available -- for example the called can say there is no news to report. Taking positive responses first, though, these can be obvious responses to a prior turn, or offer possible 'topic initials' of one's own. For example, a response can be the result of a 'reach', as in '... let me see. Last night I had the girls over?', or even 'Uh::::::m::'. These indicate that 'even in the absence of an immediately available topic... the co-participant is still prepared to search for one' (178). Shifts to a new type of talk might also be indicated, as in use of the terms like 'Oh' [sic -- maybe I need the transcription conventions here to indicate a doubting and about-to-dissent type tone?]. These have been called 'change-of-state' tokens, and they indicate that the response would not have been made without a prior turn, such as the initiating question. All these types of elicited responses mean that the response loses 'newsworthiness', however, and are usually offered only in preference to saying that there is no news at all [another empirical observation, or some functional requirement?]. They contain possible topic initials [this is claimed to have been 'observed', page 179].

The response to the question 'What's New' [usually] leads to a 'newsworthy-event-report' as a 'topic initial... [but also]... as a possible topic initial' (179). [Is this a clumsy way of saying both or either an actual and possible topic initial, or is some technical term being offered here?]. This indicates the existence of both open but bounded domains as before. After a report, the first speaker can confirm the event as discussable -- until then it is only a possible topic initial.

Turning to negative responses, these can include a 'no-news report' such as 'Oh nothing doing'. There are implications for the next turn, since such reports can 'display sensitivity to the operation of a preference produced by TIE within the constraint system' (180). What happens is an initial denial of any news can then produce a report of news nevertheless. It can indicate a 'preference for next turns' [that is a desire to keep the conversation going?], and can permit a positive response next time, as in the 'searching' remarks [the transcript here shows 'Oh nothing doing' being followed by 'Oh, let's see...'].

Turns according to negative or positive responses. The caller can re-inquire after meeting a negative response, although the first response is [often?] positive, producing either possible topics, or downgraded news with a further chance to topicalise. [Full-blown literal] news reports are unknown, and this allows further talk, as in follow-up such as 'Oh really?' and so on. 'It can be recurrently observed that [my emphasis! ] a reported topicalised event is talked to by both participants as a topic for conversation' (183) [I rest my case].

Early turns establish that both participants are available to talk on such a topic, but later turns are different. [The example on page 183 shows, maybe, that permission to continue with the topic can be withheld, for example by making a new inquiry]. But 'it is observed that routinely possible topic initials are topicalised' (184) [In other words, conversations work functionally most of the time?]. '... in two turns at talk, both speakers can display a mutual orientation and sequential commitment to the generation of a topic' (184), and this is the 'preferred response'. And 'In the case of three turns at talk speakers interactionally and mutually generate a topic for talk'. [Here, we would need to know some sociological details of the speakers, though, before getting too excited about the rapidity and ease with which social order emerges. Were these conversationalists friends already? Were they from similar social backgrounds? What was the motivational context for the conversation -- were they ringing each other just for a chat? How practised were these people before being able to do this?].

With negative responses, the first speaker can try again, for example by 'recycling' [the examples here refer to that odd American convention, whereby a second speaker repeats a statement made by the first speaker but as a question, as in 'I live in Plymouth' 'You live in Plymouth?'-- I have made this example up. The actual example is 'Nothing doing, huh?'.] Such responses invite talk about the 'no-news report' (184). Alternatively, a speaker can reassess, a second speaker can repeat or rephrase, or even turn back the question to the first speaker --'No, how is it with you?' (185). The second speaker is agreeing on the topic, but returning it. Negative responses can also lead to topic nomination, which is not elicitation but the 'presentation of a possible topic initial to be talked to' (185). [There seems to be some naive belief here that what looks like an elicitation actually is an elicitation, and not a presentation in disguise?]. For example in response to a no-news report, a speaker can offer a prompt [as we would ordinarily say]. A  negative response can risk terminating a conversation, because the first speaker is only providing an opportunity for the topic to emerge, and the second speaker seems to be refusing the chance to initiate a topic. Such responses therefore can lead to closure -- or attempts to renew the conversation. [There is some assumption here that mostly participants will attempt to renew conversations? Why should they do this? Is there some functional imperative to do so, or is it just that these particular telephone conversations were made specifically in order just to have a conversation with somebody? Were these American examples, where the cheap call has prompted a peculiar kind of recreational phone conversation?].

Generating a topic using TIE. The above does describe 'regular' uses of TIE, but they can also be used to move to or delay a 'reason-to-call' [as in the tactical examples I have described above?]. It can be difficult to generate new topics 'simply, because there has not been a prior topic form after which a new topic can emerge' (187) [This seems conspicuously obvious to me]. TIE can be used where there is no existing flow of conversation. They can also be used in closings: closing components are 'bereft of topical features' (188), and if they crop up in conversations they seem to provide no basis to continue [quite so!]. The normal techniques to continue are not available, including 'stepwise progression', because there is no topic to reference. Closing involves 'topic-bounding turns', however, which means conversations can be re-opened with further topicalisations. TIE are very useful here, because no retopicalisation is required [you don't have to go back over previous ground]. This is why TIE are so often found in openings, closings and topic-bounding turns, all cases where movement into or out of topics is required. [There seems to be functionalism here again -- conversations must take place, continue, and 'interactionally and mutually generate and determine a first topic' (188). Or perhaps this is a tautology? Conversations are defined as events that interactionally and mutually generate and so on, and so of course, by definition, they must persist, or they're not conversations?]. So 'It has been observed (Schegloff and Sacks 1997) that closings are not appropriate places for new materials' [!] (188) -- so again some kind of TIE is to be preferred. TIE are also useful where the topic has been exhausted, but the conversation needs to continue (189).


Chapter 15 Atkinson, J 'Public speaking and audience responses: some techniques for inviting applause'

This is a description of interaction taking place at public meetings. Such meetings largely feature one speaker, and the audience has to indicate responses by collective displays of 'affiliation (such as applause, cheers, and laughter) or disaffiliation (such as boos, jeers, and heckles) (371). Atkinson wants to argue that applause dominates the affiliation techniques 'clapping is an activity not substantially dependent on or constrained by breathing considerations' (371) [naively biologistic -- and what of prolonged boos?]. These responses are patterned:

(1) They occur at specific moments in a speech ( more below)

(2) They last about the same amount of time ( 8 seconds plus or minus 1) and have a characteristic aural pattern (a peak is reached quickly, the applause dies away). Speakers also seem to agree that this is about right for applause, and Atkinson also suggests that reporters work with this 'normal' period too (later in this piece he suggests that applause can be prolonged deliberately to make the remark or speech seem to be of unusual merit).

(3) These remarks apply to applause generated during a speech.

Particular moments in speeches seem to be followed by applause -- when 'a speaker announces the name of the prize winner and congratulates or commends someone... at political meetings... after a speaker criticises or insults the opposition, praises or boasts about his or her own position, or does both at once' (376 - 7). There are also 'terminating declarations... commendations, congratulations' (377). Audiences seem to expect or predict these opportunities, and some start to clap either just before or just after they occur. Speakers also predict audience responses, and often wait until applause has finished before resuming. There is thus co-ordination and possibly even a form of  turn-taking between audiences and speakers: 'In other words,... audiences may be said to come in on cue' (378). There may be a number of these cues in order to indicate that applause is required [and different sections of the audience can respond to different sets of cues] :

(1) Using names, often after an introductory sequence, so the audience can applaud when they hear the name. The precise name may not be necessary, and a delay may be used to build tension. Atkinson notices some examples where the speaker has not managed this kind of cueing and delaying [but, later, he suggests that such mismanagement can also be tactical].

(2) Using lists. An extract from a speech by Mrs Thatcher is analysed to show that three- item lists are especially useful 'given two items so far, a recipient can see that a third will occur, and that upon its occurrence utterance completion can have occurred whereupon it will be [time to respond collectively]' (386). As an example, Mrs Thatcher declared Soviet Marxism to be bankrupt on three counts, earning applause on uttering the third item. Such lists can be used to expand a point that has been made, or indeed to specify points to come. This list technique is so well-established that the audience sometimes applauds after the third list item even where the speaker does not intend this to be the last one. Atkinson goes on to analyse non-verbal cues which can supplement this list structure [shifting dazes, pursing lips, moving head downwards, moving arms so as literally to conduct the audience]. Sometimes audiences responded differently to these different cues [to explain unusual extracts] .

(3) Contrastive devices, to compare us with them, for example, especially suitable for adversarial politics, although with a long history in rhetoric. Contrasts may both complete projects and indicate expected continuation. In practice, they may involve a punchline or an element of surprise, or be involved in a '"puzzle - solution" format' (393), for example where the first part of the contrast sets up the puzzle, and the last part solves it. The most effective contrasts seemed to be ones where the second part resembles the first closely, especially  'in the details of its construction and duration' (395), which cues in the audience specifically and avoids distraction.

(4) Where this still produces confusion, speakers may continue with 'recompletion, with the point just made being referred back to and reiterated in summary form' (396). Recompletions can also be used tactically, to hint that the audience applause has somehow drowned out a speech before it has been completed. There are some disadvantages however in that recompletions can draw 'attention to the absence of a response where one had been expected and/or ineffectively projected' (398).

Prosodic features are also important ('shifts in volume, intonation, emphasis', page 398). Here too, there are some regularities. Indeed, listening to speeches in foreign languages can still provide clues about cueing applause. Speakers may increase volume, alter gestures, or speak 'with a more emphatic beat or rhythm' or different intonation (399). It is a way that speakers have of adding emphasis, and indicating that response is appropriate. These features usually work in conjunction with some of the features mentioned above.

The same points can be made for non-vocal activities such as bodily movements -- raising an arm, making short forward movements co-ordinated with the talk, shooting the hand down again and so on. Again these help the audience to focus and get ready to respond, to hold the attention of the audience.

In conclusion, Atkinson suggests that further observation will suggest even more identifiable techniques or procedures, and promises more attention to examples where the techniques go wrong, or where individual orators use them to become 'outstandingly effective (or ineffective) public speakers' (404). It might even be possible to use this research to train effective speakers. One thing to avoid, it seems, has been noted since the first century AD -- 'that applause should not "be too eagerly sought"' (405). Such a device produced a derogatory term -- 'claptrap'. [Atkinson has also hinted at techniques which might be used to make the applause seem spontaneous, or particularly enthusiastic, as above].

[There seem to be a few confusions about agency here. Sometimes these regularities are generated 'from below', as in classic ethno, but Atkinson also calls them 'techniques', and allows for particular individual virtuosos. Is the audience 'genuinely' participating, or merely playing its part in the occasion? -- Atkinson offers both, maybe suggesting that the audience is itself divided. Is the [virtuoso?] speaker manipulating the audience?].