DISCOURSE ANALYSIS
CONVERSATION ANALYSIS:
Conversation is one of the most prevalent uses of human language. All human beings engage in
conversational interaction, and human society depends on conversation in order to function: Social
interaction is the primordial means through which the business of the social world is transacted, the
identities of its participants are affirmed or denied, and its cultures are transmitted, renewed and
modified. (Goodwin and Heritage, 1990, p. 283) Conversation is the way in which people socialize and
develop and sustain their relationships with each other. When people converse, they engage in a form
of linguistic communication, but there is 2 An Introduction to Conversation Analysis much more going on
in a conversation than just the use of a linguistic code. Much that is important in conversation is carried
out by things other than language, including eye gaze and body posture, silences and the real-world
context in which the talk is produced.
DEVELOPMENT OF CONVERSATION ANALYSIS:
HAROLD GARFINKEL:
Conversation Analysis is an approach to the study of talk in interaction which grew out of the
ethnomethodological tradition in sociology developed by Harold Garfinkel (1964, 1967, 1988).
Ethnomethodology, as a field of sociology, studies the common-sense resources, practices and
procedures through which members of a society produce and recognize mutually intelligible objects,
events and courses of action. These main ideas for the approach were established in Studies in
Ethnomethodology (Garfinkel, 1967). . Garfinkel sought to study the social structure of everyday lived
experience and develop an understanding of ‘how the structures of everyday activities are ordinarily and
routinely produced and maintained’ (Garfinkel, 1967, pp. 35–6). For ethnomethodologists, the social
order is not a pre-existing framework, but rather it is constructed in the minds of social actors as they
engage with society. As members of a society encounter sense impressions and experiences, they must
somehow organize into a coherent pattern.
ERVING GOFFMAN:
The emphasis on studying actual instances of social interaction is further developed in the work of
Erving Goffman (Goffman, 1959, 1963, 1967, 1969, 1971, 1981, 1983), who asserted that the ordinary
activities of daily life were an important subject for study. Goffman’s work demonstrated that it was
possible to study everyday events and situations and to discover from this non-trivial information how
human beings engage in sociality. He was able to show how matters of great social significance could be
found in everyday activities. . He argued: Talk is socially organized, not merely in terms of who speaks to
whom in what language, but as a little system of mutually ratified and ritually governed face-to-face
action, a social encounter. (Goffman, 1964: 65)
HARVEY SACKS:
The work of Garfinkel and Goffman provided an impetus for the development of Conversation Analysis
by establishing a concern for investigating the orderliness of everyday life, and these were taken up by
Harvey Sacks in his lectures on conversation from the early 1960s (Sacks, 1992). In these lectures, Sacks
developed an approach to the study of social action which sought to investigate social order as it was
produced through the practices of everyday talk. By the late 1960s and early 1970s, through the work of
Harvey Sacks, and his colleagues Emmanuel A. Schegloff and Gail Jefferson, Conversation Analysis began
to emerge from sociology as an independent area of enquiry oriented towards understanding the
organizational structure of talk which has influenced a number of the social science disciplines
concerned with human communication (Lerner, 2004).
CORE ASSUMPTIONS:
The core assumptions of Conversation Analysis are listed (see Psathas, 1995):
1. Order is produced orderliness. That is, order does not occur of its own accord nor does it pre-exist
the interaction but is rather the result of the co-ordinate practices of the participants who achieve
orderliness.
2. Order is produced, situated and occasioned. That is, order is produced by the participants
themselves for the conversation in which it occurs. The participants themselves orient to the order
being produced and their behavior reflects and indexes that order. This means that in analyzing
conversation as an academic activity, orderliness being documented is not externally imposed by the
analyst but internally accomplished by the participants. This observed order is not the result of a pre-
formed conceptions of what should happen, nor is it a probabilistic generalization about frequencies.
3. Order is repeatable and recurrent. The patterns of orderliness found in conversation are repeated,
not only in the talk of an individual speaker but across groups of speakers. The achieved order is
therefore the result of a shared understanding of the methods by which order is achievable. These three
formulations make it clear that Conversation Analysis assumes that there is overwhelming order in
conversation. Conversation is neither random nor unstructured; however, the order observable in
conversation does not imply an overarching uniformity in conversational structure which is generalizable
across conversations (Wooffitt, 2005). Instead, the participants themselves construct conversations in
orderly ways.
1. TURN-TAKING IN CONVERSATION:
One of the most noticeable features of conversation is that speakers change. In fact, in most cases, only
one person speaks at a time and transition from speaker to speaker occurs fluidly with few gaps and
little overlap. This is of course not to say that gaps and overlaps do not occur. However, where gaps or
overlaps occur, they can be seen as doing something of interactional significance. Speaker change is a
normative process which must be achieved by participants in the conversation. That is to say, turn-
taking behavior is socially constructed behavior, not the result of an inevitable process. For example, the
fact that overwhelmingly one person talks at a time is not the result of any physical or psycholinguistic
constraint on human beings and there are also some activities, such as laughter or responses to
greetings addressed to a group, which can overlap regularly and unproblematically, and where one
person speaking at a time would be very unusual.
THE TURN CONSTRUCTIONAL COMPONENT:
A first step towards understanding how turn-taking works in conversation involves understanding what
turns at talk actually look like. Turns at talk are made up of stretches of language, but, as has Turn-
Taking In Conversation 87 already been seen, these stretches of language can vary a lot in terms of their
structure. Sacks et al. (1974) state that turns are made up of units which they call turn constructional
units (TCUs) and that the composition of TCUs is highly context-dependent. A variety of grammatical
units may function as TCUs: words, phrases, clauses and sentences. In fact, any linguistic constituent can
potentially function as a TCU. TCUs are context-sensitive and a decision about what constitutes as TCU
can only be made in context. Importantly, it must be acknowledged that people do not just talk in
sentences but can use a range of different structures to construct their talk. In context, even linguistic
forms which are not usually considered to be able to stand alone can be found as TCUs, as in (5.5).
1 Ther: What kind of work do you do?
2 Mother: Food service
3→ Ther: At?
4 Mother: (A)/(uh) post office cafeteria downtown
5 main post office on Redwood
6 Ther: °Okay°
In (5.5) the arrowed turn contains the word At?, and in this case it functions as a complete TCU.
POSSIBLE COMPLETION:
Sacks et al. (1974) discuss a number of levels of possible completion which are relevant to TCUs. He
argues that a turn at talk can be seen as possibly complete in three main ways. First, it may be
grammatically complete: that is, it may be a syntactically complete unit. Second, it may be intonationally
complete: it may occur with an intonation contour which indicates that the unit is now ended. For
example, what with rising intonation can be hearable in context as a complete question TCU, while what
with level intonation may not be hearable as a complete question in the same context. Third, although
most importantly, the TCU needs to be complete as an action: it must count as having done what needs
to have been done at this point in the conversation.
EXAMPLE:
Having asked a question, provided an answer, issues an invitation and so on.
TRANSITION RELEVANCE PLACES:
The notion of possible completion is linked fundamentally to the idea of transition relevance places
(TRPs). It has been argued earlier that there are points where a speaker’s talk is possibly complete and
that at points of possible completion, speaker change is a possible next action. Such points are TRPs,
places at which speaker change could be appropriate. This means that TRPs are not places where
speaker change has to occur but rather places where speaker change could occur. In fact, TRPs are the
sites in conversation in which speaker change can be a ‘legitimate next action’. There are, however,
some cases in conversation where speakers do work to gain a right to produce an extended turn at talk
which is made up of more than one TCU, for example, telling a story. In these cases, the current speaker
can be seen as having some rights to produce each next TCU until the extended turn is ‘complete’.
TURN ALLOCATION:
At any TRP, there are two basic ways in which a next speaker can come to have a turn at talk: either the
current speaker can select the next speaker or a next speaker may self-select.
CURRENT SPEAKER CAN SELECT THE NEXT SPEAKER
If a current speaker is to select the next speaker, the talk must be designed to achieve this. Only some
forms of talk can select next speaker: for example, questions can, but answers do not. If a current
speaker produces a question, this talk makes a particular type of action a relevant next action (i.e. an
answer) and at the same time may make some recipient the relevant participant to perform that action.
NEXT SPEAKER SELF-SELECTS
Self-selection occurs when a participant becomes next speaker, but nothing in the previous talk has
selected this person to be next speaker.
OVERLAP:
Two people talk at the same time.
EXAMPLE:
Participant A: Hi, did you finish the report for the meeting tomorrow?
Participant B: Not yet. I’ve completed the data analysis, but I still need to write the conclusion.
Participant A: Okay. Do you think you’ll be able to finish it by this afternoon?
Participant B: Yes, I should be done by 4 p.m. I’ll email it to you as soon as it’s ready.
Participant A: Great, thanks. That will give me enough time to review it.
Participant B: No problem. Let me know if you want me to change anything.
2. SEQUENCE ORGANIZATION:
The notion of sequence organization is based on the premise that the central consideration for the
organization of talk is that talk is a form of social action: that is, turns at talk are places in which the
participants in a conversation perform actions through talk. Turns at talk cluster together in order for
speakers to develop a course of action. This view maintains that the action being performed is a more
important resource for understanding how conversation is organized than is topic. While it does not
deny that topic may be an organizational feature of talk (Button & Casey, 1984, 1985; Jefferson, 1984), it
does argue that action is more important as an organizational feature than topic (Schegloff, 1991, 1995).
Much talk is problematic if considered solely in terms of topic. For example, in the sentence Could you
open the window?, it is problematic to consider this as primarily an utterance about windows and much
more useful to consider it as enacting a request to do something. The legitimacy of seeing such talk in
terms of action rather than topic can be seen in the ways in which participants treat such talk in a
conversation: further talk on the topic would be considered less appropriate after such a sentence than
would the action of opening a window.
3. ADJACENCY PAIRS:
In conversation we notice that many turns at talk occur as pairs. A greeting is conventionally followed by
another greeting, a farewell by a farewell, a question by an answer. Schegloff and Sacks (1973) called
these sorts of paired utterances adjacency pairs and these adjacency pairs are the basic unit on which
sequences in conversation are built. Adjacency pairs have a number of core features which can be used
by way of a preliminary definition. They (1) consist of two turns (2) by different speakers, (3) which are
placed next to each in their basic minimal form, (4) which are ordered and (5) which are differentiated
into pair types. The first two features are rather straightforward, but the latter three require more
explanation. First, it is normally the case that where an adjacency pair occurs, the two turns occur
immediately together with no intervening talk. However, this does not mean that all adjacency pairs are
in fact immediately adjacent to each other. In some cases other talk can come between the two turns,
although what can legitimately occur in this position is actually quite constrained. Second, the two turns
which make up the adjacency pair are ordered so that one of the pair always occurs first and the other
always occurs second: for example, a question always precedes its answer. Therefore, we Adjacency
Pairs and Preference Organization 139 can say that some types of talk are designed to initiate next
actions, while other types of talk are designed to complete the action initiated. Those forms of talk
which initiate actions are called first pair parts (FPP), while those that flow from such initiations are
called second pair parts (SPP). Third, the relationship between FPPs and SPPs is constrained by the type
of FPP produced. That is, when an FPP initiates a sequence, not just any SPP can occur in the second
position: the SPP must be of the appropriate type for the action initiated by the FPP. Therefore, a
question must be followed by an answer to be heard as a completed sequence: it cannot be followed by
a greeting or a farewell even though these are also possibly SPPs in other contexts.
4. PREFERENCE ORGANIZATION:
In a conversation, a participant may be able to choose among alternatives to design a particular
contribution to the talk and, as we have seen earlier, these choices may have different interactional
import. The term preference has been developed to characterize these non-equivalent alternatives
(Atkinson & Heritage, 1984b; Pomerantz & Heritage, 2013). The concept of preference deals with the
possible ways in which some conversational action may be accomplished. It does not refer to the
personal desires of the speakers but rather to the recurrent patterns of talk in which actions are carried
out. The basic distinction made in preference organization is that in a particular context, certain actions
may be avoided, delayed in their production or produced in mitigated form, while other actions are
normally performed directly and with little delay. it can be seen that the acceptance was done without
delay, while the declining is delayed in its turn, and, in fact, the invitation is not even explicitly declined.
The immediate and direct acceptance of an invitation is usually treated as unremarkable, as nothing
unusual being done; however, an equally immediate and direct refusal of an invitation would not
Adjacency Pairs and Preference Organization 143 normally be treated as unremarkable but would more
likely be interpreted as rude or hostile (Goodwin & Heritage, 1990; Pomerantz, 1984). Actions which are
routinely performed immediately, and whose immediate production is unremarkable, are termed
preferred actions, while those which would not normally be performed in this way are called
dispreferred actions. For example, for invitations we have the following possibilities: Invitation: Accept –
preferred Decline – dispreferred. These two concepts – preferred and dispreferred – are essentially
social in nature. They express the fact that some responses are problematic for social relationships,
while others are not. If speakers need to produce a next turn which is dispreferred, then they need to
design the turn in different ways in order to do extra conversational work. This extra conversational
work orients to the need for the contribution not to be disruptive of the relationship the speaker has
with the recipient. As a general statement it may be said that actions such as agreement, acceptances
and so on are preferred, while disagreements, rejections and declinings are dispreferred. This
distinction, however, depends on what precedes the turn. For example, next speaker’s agreement with a
negative self-assessment (I’m just no good at anything) would be dispreferred rather than preferred
(Pomerantz, 1984).
EXAMPLE:
Case 1: Invitation → Acceptance (Preferred response)
Speaker A: Do you want to come to my birthday dinner on Saturday?
Speaker B: Yeah, I’d love to! What time should I be there?
→ Adjacency pair:
First pair part: Invitation
Second pair part: Acceptance
Case 2: Invitation → Implicit Rejection (Dispreferred response)
Speaker A: Do you want to come to my birthday dinner on Saturday?
Speaker B: Saturday’s a bit tricky for me—I’m already tied up most of the evening.
→ Adjacency pair:
First pair part: Invitation
Second pair part: *Rejection (implicit, mitigated)
5. SEQUENCE EXPANSION:
While the adjacency pair structure is the basis of sequences of talk, it is possible for these sequences to
be expanded in various places in their production. Sequence expansion allows talk which is made up of
more than a single adjacency pair to be constructed and understood as performing the same basic
action and the various additional elements are seen as doing interactional work related to the basic
action under way. Sequence expansion is constructed in relation to a base sequence of an FPP and an
SPP in which the core action under way is achieved. Expansions may occur prior to the articulation of the
base FPP (pre-expansion), between the base FPP and the base SPP (insert expansion) and following the
base SPP (post-expansion). Most examples of expansion are also sequences in their own right made up
of FPPs and SPPs and so may also be called pre-sequences, insert sequences and post-sequences in
order to focus more on the ways in which the expansions are constructed.
Sequence expansion and pre-sequences (overview)
Sequence expansion in Conversation Analysis refers to how a basic adjacency pair (e.g. question–
answer, invitation–acceptance) can be extended with extra turns before, between, or after the core
pair. These expansions help participants manage interaction smoothly.
One common type of sequence expansion is the pre-sequence.
Pre-sequences occur before a base action and are used to test the conditions for that action (availability,
willingness, relevance) before committing to it.
Example (sequence expansion with a pre-sequence):
A: Are you free later? (Pre-sequence)
B: Yeah, I think so.
A: Do you want to go for coffee? (Base action: invitation)
B: Sure!
Here, the invitation is expanded by a pre-invitation that checks availability first.
1. Pre-invitation
What it is:
A pre-invitation checks availability or interest before actually issuing an invitation. It helps the speaker
avoid rejection.
Example:
A: Are you doing anything tonight?
B: Not really.
A: Then do you want to grab dinner together?
2. Pre-request
What it is:
A pre-request checks whether the conditions are right for making a request (ability, willingness, or
resources).
Example:
A: Do you have a charger with you?
B: Yeah, I do.
A: Could I borrow it for a few minutes?
3. Pre-announcement
What it is:
A pre-announcement prepares the listener for news, often to secure attention or assess readiness
before delivering it.
Example:
A: I have something to tell you.
B: Okay, what is it?
A: I got the job.
4. Pre-offer
What it is:
A pre-offer checks whether help or a resource is needed before actually offering it.
Example:
A: Are you still working on that report?
B: Yeah, I’m behind.
A: I can help you finish it if you want.
5. Pre-telling
What it is:
A pre-telling checks whether the recipient already knows something or wants to hear the information.
Example:
A: Have you heard about what happened at the meeting?
B: No, what happened?
A: The deadline got moved up.
6. Pre-closing
What it is:
A pre-closing signals that a conversation may be coming to an end and checks whether the other
participant agrees.
Example:
A: Anyway, I should probably get going.
B: Yeah, same here.
A: Okay, talk to you later.
REPAIR MECHANISM IN CONVERSATION:
Repair refers to the processes available to speakers through which they can deal with the problems
which arise in talk. All levels of conversation are potentially subject to difficulties and conversation as a
self-regulating system needs to have available practices for dealing these. Repair is itself a mechanism of
conversation: a set of practices designed for dealing with the sorts of difficulties which emerge in talk.
Like other aspects of the conversational system, the practices of repair are independent of the nature of
the thing which needs to be repaired. 198 An Introduction to Conversation Analysis Repair is a broader
concept than simply the correction of errors in talk by replacing an incorrect form with a correct one,
although such corrections are a part of repair (Jefferson, 1987, 2018c; Kitzinger, 2013; Schegloff et al.,
1977). In order to emphasize the broad nature of repair as a conversational phenomenon, Conversation
Analysis uses the term repair rather than correction to indicate the overall phenomenon of dealing with
problems in talk and the terms repairable or trouble source to indicate the thing in talk which needs to
be repaired (Schegloff et al., 1977, p. 363). Repairs may relate to problems of production of talk or to
problems of reception: that is, speakers may have trouble in producing talk and recipients may have
trouble in receiving talk (House, Kasper & Ross, 2003; Schegloff, 1987c).
TYPES OF REPAIR:
Schegloff, Jefferson and Sacks (1977) have proposed a model of the mechanism for repair in
conversation which makes a central distinction between who initiates repair and who makes the repair.
Repair can be initiated by the speaker of the repairable (self-initiated repair) or it may be initiated by its
recipient (other-initiated repair). In addition, a repair can be made by the speaker of the repairable (self-
repair) or it may be made by the recipient of the item (other-repair). In combination, these possibilities
allow for four types of repair:
1. Self-initiated self-repair: in which the speaker of the repairable item both indicates a problem in the
talk and resolves the problem.
Example:
A: I met her on Thursday—uh, sorry—on Friday afternoon.
2. Self-initiated other-repair: in which in which the speaker of the repairable item indicates a problem in
the talk, but the recipient resolves the problem.
Example:
A: I talked to… what’s his name… the guy from marketing.
B: You mean Alex?
A: Yeah, Alex.
3. Other-initiated self-repair: in which the recipient of the repairable item indicates a problem in the
talk and the speaker resolves the problem.
Example:
A: I’ll meet you at the library at six.
B: At which library?
A: Oh—at the main campus library.
4. Other-initiated other-repair: in which the recipient of the repairable item both indicates a problem in
the talk and resolves the problem.
Example:
A: The meeting is on Tuesday the 12th.
B: It’s actually Wednesday the 13th.
A: Right, thanks.