The text discusses two major approaches to analyzing the structure of conversations: the Birmingham/Structural Approach and Conversation Analysis (CA). The Birmingham/Structural Approach focuses on the sequential organization of interaction, particularly in institutional settings, and sees interaction as being governed by a syntax of action. CA, on the other hand, is concerned with the contextual sensitivity of language use and talk as a means of social action, and was developed as a reaction to deterministic models of society. Both approaches aim to better understand the structure of conversations and how they contribute to mutual understanding.
1. 1
Conversation Structure
The two most signfiicant approaches to analysing the structure of conversation, certainly from a language
learning and teaching point of view are the structural linguisitic approach developed at the University of
Birmingham and 'Conversation Analysis'
The Birmingham/Structural Approach
A linguistic approach which focuses more directly on the sequential organisation of interaction, and which
investigated language use in institutional settings has been developed by the Birmingham school of
discourse analysis (Coulthard 1992, Coulthard 1977, Sinclair and Coulthard 1975). Interaction is described
as sequential patterns: the sequences of moves that make up these patterns of exchanges are seen to be
characteristic of specific contexts, such as the classroom or medical surgery. The analysis sees discourse
coherence as essentially rule-governed and ‘grammatical’ (Sinclair and Coulthard 1975), in the sense that
just as well-formed sentences adhere to rules of syntax, coherent and meaningful discourse is governed by a
syntax of action, that is, it is generated in terms of hierarchically organised sets of acts, moves, exchanges
and transactions. These sequences of moves which make up exchanges, are seen to be characteristic of
specific settings, such as the classroom. An analysis of interaction describes these seqential patterns and the
expectations that these patterns provoke in particular contexts. Interactional coherence is examined most
extensively at the level of exchange structures, the most prominent of which is the I-R-F (initiation-
response-feedback) exchange.
T: How many legs has an insect got ….anybody? ........................S: Six
T: Six, that's right. Good.
(T = teacher; S = student)
Teaching Exchanges
Look at the following stretch of discourse:
T: I've got some things here, too. Hands up. What's that, what is it?
P: Saw.
T: It's a saw, yes. This is a saw. What do we do with a saw? Marvelette.
P: Cut wood.
T: We cut wood with that. What do we do with that?
P: Metal.
T: We cut metal, yes we cut metal. And, er I've got this here. What's that? Trevor.
P: An axe.
T: It's an axe yes. What do I cut with an axe?
P: Wood, wood.
T: Yes I cut wood with the axe. Right. Now then I've got some more things here that cut things. that you've
seen before I think. Scissors. What do I cut with scissors?
Basically what happens time and time again in the classroom is a teacher question followed by a student response
followed by the teacher evaluating that response as correct or incorrect. We have a three part unit, a three part
exchange. There are three classes of move - Opening, Answering and Follow-up which realise the three elements of
the structure.
When we say that the structure of the exchange is IRF we do not mean that all exchanges are made up of
three parts. A fuller description of the structure would be I(R)(F). This means that all exchanges must have
an I slot and that a given exchange may have an R slot and it may have an F slot. If it has an R slot that in
turn may or may not be followed by an F slot. In other words I(R)(F) is a concise way of saying the
structure of any classroom exchange will be: I or IR or IF or IRF.
Bound Exchanges.
In addition to the exchanges we have seen there are a number of bound exchanges. Look at this sequence:
T: What's this one called? Peter.............................P: A bread knife.
T: No it's not really a bread knife. John..................P: Carving knife.
2. 2
T: Carving knife, good.
Here we begin with the usual IRF structure. But after Peter's failed attempt to answer the teacher does not
ask the question again. He simply nominates John to answer the question. The original question still stands
so the second exchange is bound to the first by the fact that it invokes the same question. So when an
exchange reactivates an element in another exchange instead of repeating it or rephrasing it we have a bound
exchange. This gives a structure of I Ib R F with Ib as a bound initiation:
Originally the I-R-F model was developed to describe sequences of questions and responses in formal
classroom interactions and it has since been criticised as being too restrictive in its application to such
‘traditional’ (Taylor and Cameron 1987) institutional settings as classrooms and doctor-patient interactions,
where the relatively socially-constrained context limits participants’ opportunities for action. Critics have
also claimed that because the analysts did not investigate sequential patterns in a wider range of settings they
failed to show the ways in which elements of the I-R-F sequence constitute its ‘ instructional’ character
(Taylor and Cameron 1987, Levinson 1983), although this criticism has been rebutted more recently by
Francis and Hunston’s work on everyday conversation (1992). Nevertheless, because of a preoccupation
with defining discursive rules within a context, the Birmingham School tend to ignore links between
linguistic behaviour and social relation, ie how mutual understanding is achieved by participants in
interactions.
Conversation Analysis
The contemporary analytical framework of conversation analysis or CA. originated out of the American-
based ethnomethodological school of sociology, as a reaction to deterministic models of society and owes
much to the works of Harold Garfinkel
Major concerns of CA
Concern with the contextual sensitivity of language use and with talk as a means of social action are central
issues. Essentially then, utterances are interpreted depending on the extent to which they meet or fail to meet
interactants’ expectations of what is considered appropriate in the context in which they occur. Expectations
derive not only from the general context of the interaction, but also from the social identities of the
participants and the assumptions about the scope of conduct that is conventionally attached to events such as
casual conversation, courtroom interrogation or news interviews.
What makes the analysis essentially different from other interactive frameworks is the view of the
relationship between rules and behaviour, whereby interactants, rather than being rule-governed, are seen as
being ‘accountable’; in other words, they are aware of the rules relevant to a particular situation but can
choose whether to conform to them or not. If they do not conform, interactants are aware of the interactional
consequences, that is, of the reflexive accountablity of their actions, ie an interactant’s behaviour is shaped
in the light of what reaction is expected to it.
Heritage (1984a:107) gives an example of greeting someone who is passing in a corridor; if A greets B, B is
in the position of either returning the greeting or not. If B fails to conform to the rule ‘return a greeting’, A
draw several inferences to account for the behaviour, such as B did not hear A, B did not recognise A, or B
was snubbing A. Thus, interactional behaviour is situated in a sequence of actions within which it is
interpreted, accounted for and reacted to.
Another major concern of CA analysis is the investigation of the sequential organisation of conversation
and over the last two decades they have detailed some of the main organisational features of conversation,
including; ‘adjacency pairs’ and a ‘turn-taking mechanism’ (Sacks, Schegloff and Jefferson 1974).
Adjacency pairs are seen as a basic structure of conversation: a sequence of two utterances which are usually
adjacent and which are produced by different speakers so that the first utterance requires a particular, limited
3. 3
range of second utterance(s). For instance, a question requires an answer, an invitation requires a response.
Adjacency pairs themselves often occur in sequences:
A: Are you busy tonight?
B: No I don' t think so (pre-sequence)
A: Would you like to go to the cinema?
B: I'd love to
Sometimes the first part of a presequence can allow the recipient to skip ahead to the second part of the next
pair:
eg
A: Do you know where the paper is?
B: Yes (skip)
A: Where? (skip)
B: On the table in the sitting room
This suggests a way for handling the issue of indirect speech acts whereby the second speaker does not have
to interpret a question asking for information as an indirect request but rather the first speaker's question
projects a subsequent request (where?) and the second speaker responds to this assumed request.
CA also introduces the notion of ' preference', ie that there are two types of possible responses to the first
parts in adjacency pairs; 'preferred' and 'dispreferred'. 'Preferred' responses are those which are expected or
conventional and 'dispreferred' are those which are not.
eg
A: Would you like to come round for coffee tomorrow morning?"
B: (preferred) I'd love to
B: (dispreferred) er .. well… that would be really nice but I've got to go into town tomorrow
morning.
Dispreferred responses usually contain a delay component (initial pause + 'well') and hedging (I'd love to
but). Also, they are likely to contain 'accounts', ie excuses or explanations ('Ive got to go into town'). The
person who refuses an invitation to coffee simply by saying 'no' or even 'no sorry' or by offering an insulting
account such as 'I'd love to but you make such awful coffee', is unlikely to get another invitation!
Conversation management
Turn Taking
Another central feature of an analysis of conversation and one of particular concern in CA is the
management of turn taking in conversation.
A turn , in a conversation-analytic tradition, has generally been described on the basis of speaker exchange,
being seen as a ‘slot in an adjacency pair’ (Schegloff 1972), with any change of speaker constituting a trade
of turns (Sacks, Schegloff and Jefferson 1974). However, while such mechanistic, behaviour-oriented
descriptions of turns may facilitate data analysis they are frequently counter-intuitive and fail to take account
of the turn-taker’s intentions as part of the definition. It is argued, following Yngve (1970) and Edelsky
(1993), that a more ‘meaning-oriented’ rather than ‘exchange-structure’ definition is needed, where a turn is
a complete or incomplete utterance which is intended (or seen to be intended) to convey an on-record
message.
Speakership (ie the allocation of speaking turns in verbal interactions) can be allocated in one of two ways:
either the current speaker selects the next speaker or a listener self-selects as the next speaker, at a
‘transitional relevance place’ or TRP (Sacks, Schegloff and Jefferson 1974:704). As stated earlier, in more
restricted speech exchange systems, such as meetings, speakership may be allocated by the chairperson
acting as gatekeeper.
4. 4
Research into speakership in conversational settings has suggested that group members who gain more
speaking turns or speak more are accorded a higher influence ranking than others who have fewer turns or
speak less (Brooke and Ng 1986). Swann (1988) reported a similar result in his study of classroom talk in
which boys as a social category dominated talk more than girls and supported the view that volume of talk is
determined by the number of turns rather than number of words.
Interruptions
Recent evidence has shed doubt on the predominantly negative view of interruptions previously held in
much research (Zimmerman and West 1975, Jacob 1975 Mishler and Waxler 1968) suggesting that the
relationship between interruptions and dominance is much more complex than had previously been assumed
(Coates 1989, Tannen 1983, Beattie 1981).
Types of interruptions
It is clear that while interruptions may function to prevent others from completing a turn and to allow the
interrupter to take over the floor, this ‘disruptive’ role is only one of various functions they can perform.
For instance, there are many examples of cases of overlapping or simultaneous speech, including what are
technically successful interruptions, which are clearly both intended and perceived as collaborative and
rapport-building (Edelsky 1993:196, Coates 1989:112, Tannen 1989:271-8).
‘Successful’ interruptions refer to one participant's ability to interrupt another speaker's turn and cause that
speaker to yield the floor before the completion of his or her turn.
‘Unsuccessful’ interruptions, or ‘rebuttals’, conversely, occur when the original speaker manages to rebut
the attempted turn claim and keep the floor.
Smooth Speaker Switches
Another point which is of interest with regard to the management of turns and interruptions is the occurrence
of smooth speaker switches, where turn changes are negotiated smoothly in that there is no simultaneous
speech present and the first speaker’s utterance is complete.
Topic Control
Topic control refers to a speaker’s ability to control the flow of topics in a discussion or conversation by
either initiating, developing or closing a topic. Topic control is part of the normal flow of conversational
activity but beyond that it also provides a resource for executing conversational influence and dominance.
Topic control can be a key factor in determining subcategories of talk or conversation. Some research
(Fishman 1983) suggests that in mixed gender conversation, topics initiated by men are more likely to be
developed than those initiated by women.
Conversational management skills
Students need an awareness of what is appropriate in conversation in terms of topics and of strategies to
maintain conversation. ie the need for participants to check understanding and adapt to other speakers as a
conversation unfolds.
Some conversational management skills include:
effective turn-taking
initiating, developing and closing topics
finding common ground
eliciting/giving feedback
checking/ giving clarification
indicating friendliness
asking for/giving opinions and information (disclosure)
avoiding taboo topics.