Handout 5

Conversation Analysis


"Conversation analysis" is a popular approach to the study of discourse. It is a way of thinking about and analyzing the pragmatics of ordinary conversation, focusing on the interactive, practical construction of everyday interchanges.

1. What is Conversation Analysis?

1.1 Origins in Ethnomethodology Conversation analysis (CA) has its origins in the "ethnomethodology" of Harold Garfinkel and Harvey Sachs, which in turn built on the social phenomenology of Alfred Schutz (e.g. Schutz & Luckmann, 1974). As a result CA has some distinctly phenomenological characteristics. As Levinson puts it: "Conversation analysis... has been pioneered by a break-away group of sociologists, often known as ethnomethodologists. The relevance of the sociological background to the pragmaticist is the methodological preferences that derive from it. The movement arose in reaction to the quantitative techniques, and the arbitrary imposition on the data of supposedly objective categories (upon which such techniques generally rely), that were typical of mainstream American sociology. In contrast, it was argued cogently, the proper object of sociological study is the set of techniques that the members of a society themselves utilize to interpret and act within their own social worlds -- the sociologist’s ‘objective’ methods perhaps not really being different in kind at all. Hence the use of the term ethnomethodology, the study of ‘ethnic’ (i.e. participants’ own) methods of production and interpretation of social interaction (see Garfinkel, 1972; Turner, 1974a). Out of this background comes a healthy suspicion of premature theorizing and ad hoc analytical categories: as far as possible the categories of analysis should be those that participants themselves can be shown to utilize in making sense of interaction; unmotivated theoretical constructs and unsubstantiated intuitions are all to be avoided. In practice this results in a strict and parsimonious structuralism and a theoretical asceticism -- the emphasis is on the data and the patterns recurrently displayed therein" (Levinson, p. 295).

1.2 What is a conversation?

 A conversation is the impromptu, spontaneous, everyday exchange of talk between two or more people. "Conversation may be taken to be that familiar predominant kind of talk in which two or more participants freely alternate in speaking, which generally occurs outside specific institutional settings like religious services, law courts, classrooms and the like" (Levinson, 1983, p. 284) The participants in a conversation take ‘turns,’ and during their turn each makes a conversational ‘move’ of some kind. Conversation analysts adopt the view that when people conduct a conversation it is an interactionally managed and locally managed phenomenon. That is to say, people organize the construction of a conversation together, cooperatively, and they deal with the organization at a "local" level, one utterance at a time. "Conversation is a process in which people interact on a moment-by-moment, turn-by-turn basis. During a sequence of turns participants exchange talk with each other, but, more important, they exchange social or communicative actions. These actions are the ‘moves’ of conversation considered as a collection of games. Indeed, conversational actions are some of the most important moves of the broader ‘game of everyday life.’" (Nofsinger, p. 10) Not all kinds of verbal exchange operate this way: a formal speech is planned in advance, and is managed primarily by the person giving it (though responses from the audience play some part too). So not every kind of verbal exchange is a conversation. However, interactions that are not ‘conversations’ in this sense can still be analyzed using CA. 1.3 What is the method of CA? CA seeks to describe conversation in a way that builds upon the way it is taken up by the people who are participating in it. It does this by paying attention to the way each utterance displays an interpretation of the previous utterance, and by paying particular attention to hitches, misunderstandings, and repairs: "The methodology employed in CA requires evidence not only that some aspect of conversation can be viewed in the way suggested, but that it actually is so conceived by the participants producing it. That is, what conversation analysts are trying to model are the procedures and expectations actually employed by participants in producing and understanding conversation.... We may start with the problem of demonstrating that some conversational organization is actually oriented to (i.e. implicitly recognized) by participants, rather than being an artifact of analysis. One key source of verification here is what happens when some ‘hitch’ occurs -- i.e. when the hypothesized organization does not operate in the predicted way -- since then participants (like the analyst) should address themselves to the problem thus produced. Specifically, we may expect them either to try to repair the hitch, or alternatively, to draw strong inferences of a quite specific kind from the absence of the expected behavior, and to act accordingly" (Levinson, p. 319) For example, consider the following exchange between student (S) and teacher (T): 1 S: So I was wondering would you be in your office after class this week?

2 (2.0)

3 S: Probably not

4 T: Hmm no

[Modified from (39) in Levinson.]


Here the two-second pause after the student’s question -- a ‘hitch’ in the conversation -- is interpreted as a negative answer to the question. (This is a much abbreviated analysis; see Levinson pp. 320-321 for more detail.) Although a silence has no features on its own, conversational significance is attributed to it on the basis of the expectations that arise from its location in the surrounding talk. (Below we shall summarize the three main kinds of interpretation of silence.)

"A fundamental methodological point can be made with respect to [this example], and indeed most examples of conversation. Conversation, as opposed to monologue, offers the analyst an invaluable analytical resource: as each turn is responded to by a second, we find displayed in that second an analysis of the first by its recipient. Such an analysis is thus provided by participants not only for each other but for analysts too" (Levinson, p. 321). When we are trying to understand a particular utterance or conversational action it is important to consider where and how that action is located in a sequence of other conversational actions. When people speak in an ongoing conversation they do so in the light of what has just been said, and in anticipation of what might take place in the future. They "design" or "construct" their own speech, and understand the talk of other people, accordingly. They also shape their utterances to take account of the identity of the speakers and what their interests are. The meaning of an utterance -- the way it is interpreted, and the way it was designed -- depends, then, on its context, both verbal and non-verbal. This construction of utterances is called recipient design.

"This ongoing judgment of each utterance against those immediately adjacent to it provides participants with a continually updated (and, if need be, corrected) understanding of the conversation" (Nofsinger, p. 66). Every person involved in the conversation has their own interpretation of what is going on, but although these interpretations are subjective in the sense that everybody has their own, they are intersubjective in the sense that every person treats the adjacent utterances in similar ways. People share a understanding of the "game" they are engaged in, and its "rules."

2. Speech Acts

Utterances are used to do things; they are actions; what John Austin called performatives. Speech acts can be grouped into several families, each containing similar types of performative:

Commissives: their point is to commit the speaker to a course of action
  promise "I’ll tell noone what you’ve said."

offer "Shall I do that?"

make a vow

take a pledge

give a guarantee
 

 Directives: their point is to get the recipient to do something
  request "Please tell me more."

command "Tell me about that"

order

suggest "Why don’t you describe what happened."

give permission "You can share that if you wish."

question "What’s your family like?"
 

 Assertives: their point is to display the speaker’s belief in the propositional content of the utterance
  assert "This rain is heavy."

describe

state

predict "It will surely rain tomorrow."

speculate "I wonder whether it will rain tomorrow"

report

announce
 

 Expressives: their point is to express the speaker’s psychological state
  compliment "Great dress!"

apologise "I’m really sorry I did that."

welcome "Nice to see you."

thank "Thanks very much!"

greet "Hi!"

acknowledge "uh huh"
 

 Declarations: done by an appropriately authorized speaker
  fire "You’re fired!"

appoint "You’re in charge."

sentence "I sentence you to thirty days in jail"

 The illustrations given here should be taken with caution. It is a central insight of Conversation Analysis that the action that participants will interpret an utterance to be will depend not just on its linguistic form, but also on its location in the sequence, on the context, on the identity of the speaker, etc..

So it’s important to examine the structure of conversation in which speech acts are produced. We consider next a central aspect of this structure.

3. Turn-taking

It is an evident fact about conversation is that it takes the form of turn-taking: two or more participants take turns to speak. But how does this happen? How does someone "get the floor"? It may seem that people simply wait for the speaker to stop, and then talk, but the gaps between turns are generally too short for this to be the case: sometimes they are just micro-seconds in length, and on average they are no longer than a few tenths of a second.

3.1 Turn-construction Turns can be made up of a single word, a phrase, a clause, or a full-sentence. They are not syntactic or semantic units, but genuinely pragmatic units. The recognizable potential end of a turn is called in CA a "transition relevance place" (TRP). A TRP may be identified by "a change in the pitch or volume of the voice, the end of a syntactic unit of language, a momentary silence, or some sort of body motion" (Nofsinger, p. 81). Transition between speakers usually occurs at such a point, and it is at a TRP that speakers employ the conversational techniques that CA aims to discover. 3.2 Turn-allocation Sacks et al. (1974) suggest a handfull of techniques that assign the rights and responsibilities of the participants in a conversation. In simplified form, these techniques are the following: 1. The current speaker (C) can select the next speaker (N) while still talking, but must then stop talking at the next TRP. (Current speaker selects next)

2. If N is not selected, anyone can jump in, and the first to do so gains rights to the floor. (Self -selection)

3. If neither (1) nor (2) occurs, C may (but need not) continue talking. (Speaker continuation)

4. If (3) happens, rules (1)-(3) apply again at the next TRP.

 For example, technique 1 can be employed by pointing, using a name, making eye contact, etc.. Another way the current speaker can select the nextr speaker is to use the first part of an "adjacency pair," as described in the next paragraph. 3.3 Adjacency Pairs Conversational actions tend to occur in pairs. We speak of an "exchange of opinions" and "an exchange of greetings" because many conversational actions call for a particular kind of conversational response in return. Greetings and farewells typically call for another utterance of the same type. Other actions call for a different type of action: invitations with acceptances (or rejections); congratulations with thanks; offers with acceptances (or refusals). Such pairs of conventionally linked conversational actions are said to have two "parts": a "first part" and a "second part." The pairs are said to have "conditional relevance."

More formally stated, adjacency pairs are sequences of two utterances that are:

(i) adjacent (unless separated by an "insertion sequence"; see below)

(ii) produced by different speakers

(iii) ordered as a first part and a second part

(iv) typed, so that a particular first part requires a particular second (or range of second parts) - e.g. offers require acceptances or rejections, greetings require greetings, and so on.

 ...and there is a "rule" governing the use of adjacency pairs, namely: Having produced a first part of some pair, current speaker must stop speaking, and next speaker must produce at that point a second part to the same pair. Adjacency pairs are often found linked together in closely integrated ways, and the next two sections describe two of these. One pair may follow another (question, answer; question, answer), or one pair may be embedded inside another pair . A "presequence" is an example of the former, an "insertion sequence" is an example of the latter. 3.4 Presequences A presequence occurs when some preliminary action is taken before initiating the first part of an adjacency pair, and the preliminary action itself involves an adjacency pair. Before making a request, for instance, it often makes sense to check whether the other person has the item one wants. Here a question-answer pair (turns 1 and 2) prepares for a request-agreement (or request-rejection) pair (initiated in turn 3).
1 A: Do you have the spanner? ) presequence

2 B: Yes. )

3 C: Can I have it please? ) R-A pair

4 B: [...] )

Another example:
1 Teacher: Mike, do you think you know

the answer to question four? )presequence

2 Mike: Yes. )

3 Teacher: Can you tell the class, then, please? )R-A pair

4 Mike: [...] )

3.5 Insertion sequences

The person towards whom the first part of an adjacency pair has been directed may want to undertake some preliminary action before responding with the second part. A request for clarification by the recipient will take place after the first pair part, but before the second pair part. This is an insertion sequence. Here turns 1 and 4 make up one adjacency pair, and turns 2 and 3 make up a second adjacency pair inserted between the two parts of the first pair:
 
1 P: Martin, would you like to dance? )

2 M: Is the floor slippery? )

3 P: No, it’s fine. )

4 M: Then I’d be happy to. )

Another example of this:
1 Teacher: Will you tell us the answer to question four? )

2 Mike: Is that one page six or seven? )

3 Teacher: Six. )

4 Mike: Oh, okay. The answer is factorial two. )

3.6 Silence

Depending on where silence occurs in a conversation, and its location in the conversational structure, it will be interpreted as a gap between turns, a lapse in the conversation, or a pause that is attributed to the designated speaker. A gap is silence at the TRP when the current speaker has stopped talking without selecting the next speaker, and there is a brief silence before the next speaker self-selects. A gap does not "belong" to anyone.

A lapse is silence when no next speaker is selected, and no-one self-selects: the conversation comes to an end for at least a moment. (N.b., a gap and a lapse can be distinguished from one another only in retrospect.)

A pause is silence when the current speaker has selected the next speaker and stopped talking, but the next speaker is silent. A pause is also silence that occurs within a participant’s turn (i.e., before a TRP is reached). A pause "belongs" to the person currently designated speaker.
 

  3.8 Preference When speakers have a choice between two conversational actions, one will typically be considered more usual, more normal, than the other. This phenomenon is called "preference." The term doesn’t refer to the psychological desires of a speaker, but the norms of the intersubjective conversational system. These shared norms mean that "any of the conversational tendencies and orientations that we commonly attribute to participants’ personalities or interpersonal relationships derive (at least in part) from the turn system" (Nofsinger, p. 89).

For example, in response to the first part of an adjacency pair some second part responses are preferred, while others are dispreferred. Refusals of requests or invitations are nearly always dispreferred, while acceptances are preferred. (See the table below.)

Conversationalists’ grasp of preference will influence their interpretation of conversational actions. For instance, a silence in response to a request may be taken as evidence of a likely upcoming dispreferred response (a refusal), so that further inducements may be added (this was the case with the first example in this handout). Conversational actions that are not the preferred response are often conducted in a manner that displays this: "dispreferred seconds are typically delivered: (a) after some significant delay; (b) with some preface marking their dispreferred status, often the particle well; (c) with some account of why the preferred second cannot be performed" (Levinson, p. 307).

4. Mutual Understanding as ‘Alignment’

For a conversation to run smoothly and effectively the organization of turns must be managed, but in addition the conversation must also be kept ‘on track.’

How do participants in a conversation get a sense of understanding and being understood? The response to an utterance often provides some kind of interpretation of the prior utterance, and so indicates the alignment. Assessments ("That’s good"), newsmarks ("Oh, wow!"), continuers ("uh huh"), formulations (giving the gist of what’s been said), collaborative completions (finishing the speaker’s sentence), all provide evidence to the speaker of how their talk is being understood.

Repairs are the things done to fix a conversational breakdown and restore alignment. Breakdowns can be misunderstandings ("What did you say?"; "What do you mean?") as well as disagreements ("I think you’re wrong"), rejections ("No, I won’t") and other difficulties. Revisions may occur when the speaker can anticipate that trouble is likely and reformulates talk accordingly.

Alignment is displayed and adjusted not only in responses to an utterance but also in advance. Preventatives such as disclaimers ("I really don’t know much about this, but...") are examples of such "pre-positioned alignment devices." Pre-sequences (see above) do this too.

Alignment is especially important at the openings and closings of conversation.

5. Extended Discourse: Narratives and Arguments

Our examples have been of conversations where the turns are brief, but the lengthy stories and substantial arguments that sometimes occur in conversations can be examined with the focus and methods of conversation analysis, too. It turns out that, counter to what we might expect, "Stories and other extended-turn structures in conversation are not simply produced by a primary speaker, but are jointly or interactively produced by a primary speaker together with other cooperating participants" (Nofsinger, p. 94).

The same is true of argument. Researchers distinguish making an argument from having an argument. "Roughly, the first involves using reasons, evidence, claims, and the like to ‘make a case.’ The latter involves interactive disagreement (for example, "You can’t," "I can," "Oh no you can’t," "Well I certainly can"). Conversational argument often consists of participants making arguments in the process of having one" (Nofsinger, p. 146).

The dynamics of both narratives and arguments "can be understood by applying our knowledge of ordinary conversational practices and structures such as the turn-taking system, adjacency pair sequencies, the preference system, and repair" (Nofsinger, p. 154).

6. Conversation Analysis and the Ontological Blueprint

The ontological blueprint summarizes Heidegger’s analysis of human being. To be human is to understand and interpret: to have an understanding, albeit tacit and unarticulated, of the being of entities, of our own being, and of the world we are in. We encounter entities with a kind of concern that grasps them and puts them to use, not with a bare perceptual cognition. To understand an entity (an artifact) like a hammer is to grasp it in practical activity: to project it into and onto the world that is the situation or context, like the workshop.

What shows up is the being of the entity, and this shows up upon the ground that is the meaning of that being. When action is going smoothly what we are aware of is a towards-which: the point of the activity, the project, in terms of which the entity is grasped.

A tool functions as a "prosthetic device": when functioning smoothly and ready-to-hand it is like an extension to the body, providing the tool-user with a feel for the material being worked on, and a sense of how they are doing, as well as a way of making something.

An utterance is like a tool, except it is made on the spot, off hand, in the moment: it is an improvised artifact (literally: sudden; unforeseen). An utterance is a "conversational device"; its production makes a point, provides the speaker with a feel for the other person (through their response), gives her a sense of herself (we discover ourselves in our words), as well as a way of accomplishing something socially.

To understand an entity like an utterance is to be aware of its point, by grasping the utterance and projecting it. The way the utterance is projected depends on (1) the ongoing conversation of which it is a part, (2) the context: the here-&-now, and (3) familiarity with the public conventions of language. These three form the fore-structure of the conversational participant. The conversation is an ongoing project, a way of being involved: the fore-view. The here-&-now provides a fore-grounding. And the conversationalist knows the language (including its conversational maxims), and thus has a preliminary sense of how to interpret: a fore-grasp.

To understand a speech act is to recognize its point: where the speaker is coming from; where they are going; what they are getting at?what their concern is, not what their beliefs are. We can talk of alignment among participants in a conversation in terms of the ontological blueprint. When participants project an utterance in ways that line up, they are aligned. Their projections are not necessarily identical, they may also be reciprocal: the speaker’s good news may be the recipient’s bad news.

Our goal as interpretive researchers is to understand and articulate the way the participants understand their interaction. We want to understand the constituting, the construction, that produces the phenomena of social reality. It is in social interaction that human being is remade. In addition, settings are continually reconstructed, at the same time as they are used as grounds for human activity. Both the world and persons in at are socially constructed. It is this social construction that is the broad object of interpretive inquiry. (Nofsinger points out that conversational utterances are both context-shaped and context-renewing: "Context, in this immediate and narrow sense, is composed not just of what people know, but of what participants do to show each other which items of their shared knowledge should be used in making interpretations. The conversational actions produced by participants create an interpretive resource that is used to align conversational understanding" Nofsinger, p. 143).

Table: Preferred and Dispreferred Second Parts to various First Parts:
 
SECOND PARTS:
FIRST PARTS:
Request Offer/Invite Assessment Question Blame
Preferred: acceptance acceptance agreement expected answer  denial
Dispreferred: refusal refusal disagreement unexpected answer or non-answer admission

[From Levinson, p. 336]

References

Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall.

Levinson, S. C. (1983). Pragmatics. Cambridge: Cambridge University Press.

Nofsinger, R. E. (1991). Everyday conversation. Newbury Park: Sage.

Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of turn-taking in conversation. Language, 50, 696-735.

Schutz, A. and T. Luckmann (1974). The structures of the life world. London, Heinemann.

  © Martin Packer 1999