Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

A fundamental aspect of psychotherapy, the “talking cure,” is that it is a conversation. This is not to overlook nonverbal aspects of the psychotherapeutic process or forms of therapy that focus on the extra-discursive, such as dance therapy. But it is probably fair to say that in most psychotherapies the client-therapist dialogue is an important tool for facilitating change. Developments in psychoanalysis, such as the Lacanian theory and psychotherapy (e.g., Lacan 1968), have built on the growing significance awarded language in the creation of human subjectivity. And, with the “turn to language” in the social sciences during the 1970s, psychotherapy saw analogous developments in variant approaches such as narrative, re-authoring, and conversational methods (Hobson 1985; White and Epston 1990). Hence, in understanding the processes of psychotherapy, a focus on the language used and the transformational potential of the therapy dialogue has a long history.

Conversation analysis developed in the 1960s in North American sociology, the first publications appearing in the early 1970s (e.g., Sacks 1972), and can be defined as a rigorous approach to discovering the ways in which talk-in-interaction is choreographed. The aim of basic research is to understand how ordinary everyday conversation is organized as a self-regulating system. This is a particularly exciting field since discoveries relating to the fundamentals of conversational exchange are still to be made. Conversation analysis continues to grow in popularity and has crossed into disciplines such as psychology and linguistics. There is also a developing field of applied research in which conversation analytic discoveries and methodology are used to understand how talk functions in institutional contexts (e.g., Heritage 2005). In such contexts, interlocutors speak as incumbents of institutional identities with the obligations, responsibilities, and expectations these entail: patient–doctor, witness-policewoman, and pupil–teacher. Moreover, in institutional talk, conversations are a medium in and through which the work of an organization can be conducted, such as making a diagnosis, interrogating a suspect, or teaching a class. Hence, applied conversation analytic research seeks to understand the ways in which conversational practices are modified in order to fulfill institutional aims. Psychotherapy is, of course, one such institution (Morris and Chenail 1995).

Conversation analysis has a unique place in the cluster of methods generally considered qualitative (Madill and Gough 2008). It is avowedly empirical and has a claim to being relatively atheoretical in the sense of (a) eschewing a theory of subjectivity (e.g., not accounting for talk at the level of individual predisposition) and (b) being primarily inductive and data driven. However, arguably, like all methods, conversation analysis has theoretic elements in positing a way of approaching talk-in-interaction. But unlike some other data-driven approaches, such as grounded theory, conversation analysis is foundational (assumes that objective principles can be established) and progressive (seeks to build a corpus of knowledge from established facts). Conversation analysis is therefore not interpretivist. Conversation analysis also differs from grounded theory and other popular methods such as interpretative phenomenological analysis in that no attempt is made to categorize the content of the data with a view to theorizing how participants understand particular social processes or their own experiences. Nor is conversation analysis social constructionist, as are variants of (micro-) discourse analysis which often draw on conversation analytic methods (e.g., Edwards and Potter 1992). Unlike discourse analysis, conversation analysis, in general, is not concerned with how sociocultural meanings are mobilized to create the phenomena which furnish our world and hold individuals in place as particular kinds of subjects. In fact, in many ways, conversation analysis is closer to a natural science than human science approach through taking the stance that rigorous empirical observation and application of the correct method will reveal the (normative) rules of conversational exchange [see Lepper (2015)]. Given this, conversation analysis defies easy placement in paradigmatic schematics as outlined and examined in, for example, Madill and Gough (2008).

Ideal data for conversation analysis are naturally occurring interactions: that is, conversations which have occurred in the conduct of everyday life unaffected by the interests of researchers. Much of the earliest conversation analytic research was based was audio-recorded, over-heard telephone conversations on shared party lines. Today informed consent is required from participants but it is not impossible to collect, what is for all intents and purposes, unselfconscious conversation. For example, archives of psychotherapy interaction exist which have been collected with consent for research purposes (e.g., the Second Sheffield Psychotherapy Project: Shapiro et al. 1990; Madill et al. 2001). Moreover, clients and therapists are often willing to release audio and sometimes video recording of their interactions for conversation analytic research (see, e.g., the collection in Peräkylä et al. 2008). Conversation analysis of recorded material requires detailed transcription, particularly of aspects of interaction already demonstrated of importance such as intonation, audible intake of breath, and length of silence. Jeffersonian transcription conventions have been designed for this (see http://www.sscnet.ucla.edu/soc/faculty/schegloff/) and are under constant development (e.g., Hepburn and Potter 2007). Depending on the aims of the study, a relatively small amount of data can be enough to demonstrate the occurrence of an exchange pattern. On the other hand, a large corpus of material can be scoured for multiple examples of a phenomenon (e.g., apologies).

Conversation analysis requires knowledge of an increasingly large corpus of established findings and a high degree of skill operationalizing its analytic procedures. Moreover, like other specialist fields, conversation analysis utilizes a technical vocabulary which can be, at first, rather opaque: adjacency pairs, conditional relevance, repair, etc. However, once familiar with the basic concepts, a good conversation analytic paper can be, in it thoroughness and precision, a beautifully eloquent articulation of the complex, usually tacit, skills of conversational exchange—“largely seen but unnoticed” (Kozart 1996, p. 366). Its painstaking empiricism assures that analytical insights are evidenced in the data and it is often incredible how much can be gleaned from a series of relatively short extracts of dialogue. Typically, the sequence of analysis would entail identification of a conversation phenomenon of interest, the collection of a series of instances of that phenomenon from available conversational data, cross-comparison of these instances in order to determine the common practices through which the phenomenon is regulated (informed by current knowledge regarding interactional exchange), and finally the presentation of a carefully evidenced and argued case for the pattern discovered using detailed analysis of examples of real conversational data.

With its intensive microanalytic focus, conversation analysis is compatible with the change process paradigm as it has been developed in psychotherapy research. Psychotherapy process research is interested in how therapy gets done. Early process research usually made the attempt to be representative in the selection of therapy segments for analysis and hence had generally used random or systematic sampling strategies. Building on works such as those of Gurman (1973) and Rice and Greenberg (1984), more recent process research perceived that all parts of therapy are not the same and that client change is likely to occur at particular important junctures. It is therefore argued that an economical and productive research strategy is to focus research on these key points or, what became known as, significant events (Greenberg 1991). Methodologically, this has entailed intensive analysis of key therapy events using task analysis (e.g., Greenberg 1984), sequential analysis (e.g., Mahrer et al. 1984), and comprehensive process analysis (e.g., Elliott 1984), among others. Intensive process analyses tend to use qualitative language-oriented methods, view the therapeutic dialogue as a communication event, and focus on the development of meaning as it occurs between client and therapist. In recent years a research literature has developed in which conversation analysis has been applied to psychotherapy interaction. However, in the main, this research has been more clearly furthering the aims of conversation analytic research (i.e., understanding conversation employed as an institutional practice) as opposed to understanding and developing psychotherapeutic techniques.

Five main features of talk-in-interaction are identified in conversation analysis as of crucial importance in understanding how conversations work: turn-taking, sequence organization, repair, word selection, and action formation. Depending on the phenomenon of interest, one or two of these features may predominate in any particular analysis. In the following sections I provide an overview of these five features of conversational exchange. Each is illustrated with commentary on an extract of psychotherapy interaction from published conversation analytic research.

2 Turn-Taking

It is easy to overlook the importance of the seemingly trivial observation that interlocutors take turns to speak during conversation. It is usual for there to be one beat of silence between speaker turns, which translates to about the length of a spoken syllable. Speaker change is choreographed between interlocutors as a conversation progresses, and conversation analysis has identified normative rules that speakers use as a resource for managing this process.

Central to turn-taking is the phenomenon of a turn-constructional unit (TCU). Normally a speaker has the right to one TCU only before a transition relevance place opens up and another speaker can take the floor. Recipients are able to anticipate the ending of a TCU since it is hearable, in context, as adequately complete. Adequate completion can be judged using three main criteria. The first, and most important, relates to pragmatics. TCUs complete an action: that is they do something in the talk, such as make a request. The second relates to prosody. TCUs sound complete in the way in which they are intonated. The third is syntactic. TCUs are grammatically complete within the conversational context in which they are spoken. Nonverbal cues, such as gaze directed towards the next speaker, also help signal the end of a TCU. Recipients monitor ongoing talk for the projectable ending of the current speaker’s TCU and the opening of a transition relevance place in order to take turns at talk in smooth progression of the conversation.

The normative rules of turn-taking are a resource for managing and understanding conversation. For example, they allow recognition of different kinds of silence, each with a range of different interactional significance. Mid-TCU pauses belong to the current speaker and, since the TCU is not complete, do not signal the end of the speaker’s turn. Hence, if a recipient attempts to speak during a mid-TCU pause, it is likely to be treated as problematically interruptive. Such pauses may function, for instance, to secure a person’s gaze as indication that he or she is willing to converse (see Repair, Sect. 24.4). When a sequence of turns is complete in that a series of conditionally relevant actions is fulfilled (e.g., a question-answer sequence), a lapse in the conversation may occur (see Sect. 24.3). Lapses are not necessarily problematic for the interlocutors. More problematic are inter-turn silences occurring after the end of a speaker’s turn during sequences which are not yet complete (e.g., after a question and before some answer has been provided). Conversation is conducted so as to minimize such gaps since they are hearable as belonging to the next speaker who, for some reason, is not supplying their turn.

Speakers can, of course, continue speaking beyond the end of their TCU. The right to do so is achieved interactionally. At the end of a TCU, a speaker may produce a rush through which provides no time for the next speaker to start their turn at the projected turn-transition point. On the other hand, a speaker may indicate during their TCU that a multi-TCU turn is required. This can be done, for example, through structuring the turn around a list of items, all of which need to be expanded upon before the work of the turn is complete. Hearers collaborate with multi-TCU turns through indicating that they are giving up the right to take the floor through using continuers such as “mm” or “uh huh” at turn relevant points. These are not turns in and of themselves but indicators to the current speaker that the recipient is still listening.

A common strategy, which has been the subject of particular analytic interest, is where a speaker indicates the need for multi-TCU in order to tell a story. Storytelling is likely important to psychotherapy interaction as extended sequences of troubles telling are often elicited from clients by therapists (Pain 2009). When a story has reached a point of recognizable completion, one way in which a recipient may indicate cognizance of this is through offering some kind of assessment. In therapy, this may be provided by the therapist by way of a therapeutic formulation or interpretation.

Analysis of the following extract from brief eclectic psychotherapy makes use of the observation that silence following the end of one speaker’s turn, when allocatable to the next speaker—that is, an inter-turn gap—is accountable and highly problematic. Wynn and Wynn (2006) provide this extract within a more extended analysis of failures of empathy in psychotherapy. They argue that such turn-withholds are an important way in which failure to bring off empathy is achieved interactionally through interruption of the smooth progressivity of the conversation.

Example 1 [Excerpt 8 from Wynn and Wynn (2006)]

  1. 1.

    T it must be quite…hurtful not to…not to feel the will to live…

  2. 2.

    T that wi[ll]

  3. 3.

    P [hmm] ((patient looks away from the therapist))

  4. 4.

    T that lies in…lacking that will? I think that must do something with you?

  5. 5.

    (7 s)

  6. 6.

    T do you know what it does with you?

  7. 7.

    (2 s)

  8. 8.

    P no I don’t know

  9. 9.

    (5 s)

  10. 10.

    P then I become insecure with this work with assistance and insecure

  11. 11.

    P if I will manage working from eight to four

At line 4, the therapist has, in context, adequately completed her turn. Moreover, she has asked a question so the sequence is still underway as she has yet to receive some kind of answer. Hence a transition relevance space has opened up and the silence at line 5 belongs to the client who can therefore be heard as withholding her turn. Given that the norm is one beat of silence between turns, a 7-s gap is extremely long. This inter-turn gap can therefore be analyzed as signaling a problem in the interaction. This problematic interactional pattern is repeated over lines 6–7. The therapist issues a reformulated version of her prior question in a further hearably complete TCU which is, once again, followed by an interactionally long silence belonging to the client. The client eventually does provide a relevant, although blocking, response in a hearably complete TCU. The client’s response provides some kind of answer to the therapist’s question and so completes the series of conditionally relevant actions underway. The following 5-s silence is therefore a lapse in the conversation, in which the therapist could but is not obliged to speak, after which the client continues with a change of topic and new sequence.

There are many more observations that could be made of extract 1. The above commentary, however, allows us to see how some of the silences in this extract, i.e., on lines 5 and 7, are in places in which the next speaker, the client, is expected to talk. These are therefore analyzable as turn-withholds and, hence, indicators of interactional problems. In a more extended analysis, Wynn and Wynn argue that the specific interaction problem here is a failure to bring off empathy.

3 Sequence Organization

As touched on in the previous section, turns at talk come in series which build up coherent sequences. Interestingly, conversation analysis draws attention to action, as opposed to topic per se, as the most analytically useful resource for identifying how conversations are organized into sequences (see, e.g., Bercelli et al. 2008). That is, a sequence is recognizable as such through having completed a series of conditionally relevant actions, such as making and receiving a response to an invitation.

The most basic and minimal kind of sequence consists of two turns at talk in the form of an adjacency pair. An adjacency pair consists of a first pair part (FPP) issued by one speaker initiating an action followed immediately by an appropriate second pair part (SPP) issued by a second speaker completing that action. Hence, a particular kind of FPP makes a particular kind of SPP conditionally relevant and the second speaker is accountable if she or he fails to offer an appropriate type of response. The technology of the adjacency pair is therefore linked to the turn-taking systems because an FPP makes the issues of an SPP by a second speaker a relevant next action.

The actions performed in conversation are not all the same social valence. This is oriented to by speakers and captured in conversation analysis by the concept of preference organization. Preference is a continuum linked to the degree of social delicacy associated with performing particular actions: that is, the relative potential of the action to create interactional difficulties. Dispreferreds tend to block, rather than to progress, the action of the sequence and can make vulnerable the interlocutors’ relationship: for example, an SPP declining rather than accepting an invitation. Dispreferreds are avoided, if possible, or delayed in their production. Hence, they tend to appear towards the end of a turn, can be presaged with a silence, hesitation, or lexical marker such as “well” and may include a warrant or explanation. In contrast, preferred actions tend to be performed immediately and directly.

It would be odd to conceptualize conversation as consisting only of short sequences of adjacency pairs. And observation of talk-in-interaction shows that, although sequences are organized around a base adjacency pair, longer sequences relevant to performing this core action are produced. Expansions around a base adjacency pair can consist of pre-sequences, insert sequences, and/or post-expansion sequences.

Some kinds of action initiated in FPPs are potentially problematic for social relations and are, themselves, dispreferred. These include, for example, requests since they can put an imposition on the recipient. Hence, dispreferred FPPs often involve a pre-sequence which checks out the likely response of the recipient to the projected action. Pre-sequences can be responded to by the recipient with a go-ahead (intimating that a preferred response to the projected action may be forthcoming), block (which stops the action progressing), or hedge (e.g., seeking further information before an SPP is supplied). The recipient’s response to the pre-sequence influences the trajectory of the subsequent talk. For example, a pre-request projecting the possibility that a request is about to be made will be heard as such by an interlocutor. The preferred response is one that heads off any potential interactional difficulty. Hence, the most preferred response to a pre-request is that the recipient makes the relevant offer so that the request itself does not have to be made. A blocking or, possibly, hedging response to the pre-request allows a speaker to avoid progressing with the request as there are indications that it may be refused. Interestingly, then, conversation analysis allows analysis of where a base adjacency pair integral to the action of a sequence is, in actual fact, never performed.

Preference organization calls for the production of a relevant SPP as soon as possible following the FPP. However, sequences can be inserted between the FPP and SPP interrupting, but recognizably related to, the action underway. Insert sequences come in two main types: those that orient to the FPP and those that orient to the SPP. Post-first insert expansions most commonly consist of repair initiations in which a second speaker requires some clarification of the FPP before being able to produce an appropriate SPP (see Sect. 24.4). Pre-second insert expansions project forward in the conversation and consist commonly of the second speaker requesting additional information necessary to producing the SPP.

The final place at which a base adjacency pair can be expanded is after the issue of the SPP. Minimal post-expansions consist of what is known as sequence closing thirds. The speaker of the FPP receives an SPP and then finishes the sequence in the third turn with a minimal response such as an acknowledgement token like “okay,” change of state token like “oh,” or brief assessment of the sequence. Longer post-expansions occur, in particular, when a dispreferred SPP is given and often account for and soften their impact. These can consist of extended post-sequence self-talk-like musings (which appear to be an opportunity for a speaker to “have the last word”), a new FPP which continues the sequence (indicating that the action has not yet been completed, e.g., a question topicalizing the SPP as worthy of further discussion), a repair initiation (indicating a problem in the talk), or a rejection of the SPP.

To bring post-expansions to a close, speakers collaborate typically in the following series of turns. The first speaker provides an assessment, summary, or aphoristic formulation of the upshot of the sequence (which projects its closure). A second speaker agrees (which provides the go-ahead for closure). A third turn (sequence closing third) is produced consisting of a closing token or brief assessment following which there is, possibly, the initiation of a new sequence. This kind of sequence closing sequence and initiation of a new one is illustrated in the following extract (which is from either cognitive or relational-systemic therapy).

Bercelli et al. (2008) observe that therapists make both formulations and reinterpretations in response to client-narrated events and examine the organization of sequences in which these two different kinds of action occur. In formulations, therapists offer a candidate understanding of what the client has meant in her or his previous talk. Hence, in terms of sequence, formulations are contingent on the prior talk, which usually consists of sequences of questions and answers providing the information on which the therapist’s formulation is based. Formulations are also the FPP of a sequence closing sequence and, as such, make conditionally relevant an SPP. Bercelli et al. note that a preferred SPP response to a formulation is a confirmation. Alternative, but less preferred, SPPs are disconfirmations or reformulations. Example 2 illustrates an FPP formulation, followed by an SPP confirmation, and followed by a sequence closing third and initiation of a new sequence.

Example 2 [Extract 2 in Bercelli et al. (2008)]

  1. 1

    Cl: [no, there] and then I saw my father who::: who

  2. 2

    hm:: (.) was protecting my mother.

  3. 3

    (0.3)

  4. 4

    Cl: and n[ot

  5. 5

    Th: [you read this thing as (0.3)

  6. 6

    protecting your mum=

  7. 7

    Cl: =yes

  8. 8

    Th: okay explain to me how.

The therapist issues an FPP formulation in lines 5–6 in which he offers a candidate understanding of the client’s prior talk. She responds immediately and directly in line 7 with a preferred SPP confirmation. Bercelli et al. note that this kind of minimal confirmation token “yes” is the most common such response to therapist formulations in their corpus. The minimal post-expansion, and sequence closing third (line 8 “okay”), is also one of the most common ways in which therapists close such formulation sequences. Then, with the issue of an FPP question, the therapist goes on to initiate the next action and, hence, new sequence. Bercelli et al. contrast formulation sequences, such as this, to reinterpretation sequences in which therapists are observed to offer more of their own perspective on client-narrated events. They argue that reinterpretation FPPs make relevant a much wider range of client SPPs than do formulation FPPs and provide an analysis of the implications of this for therapy interaction.

4 Repair

One common type of insert or post-expansion sequence performs the action of repair. In repair attention is drawn to some trouble source in the talk and so occurs when a speaker indicates an (ostensible) problem in speaking, hearing, or understanding. Repair is designed to be heard as correcting, covers a very broad range of phenomena, and is integral to conversation as a self-regulating system. The mechanisms involved are highly organized and sensitive to both linguistic and social considerations.

There are four types of repair. In self-initiated self-repair, the speaker of the trouble source both indicates a problem in her or his own talk and resolves that problem. In self-initiated other-repair, the speaker of the trouble source indicates a problem in her or his own talk but an interlocutor resolves that problem. The pattern is reiterated for the final two types of repair: other-initiated self-repair and other-initiated other-repair.

The preference is for self-repair. In order to clear up misunderstandings quickly, the norm is to issue repair as close as possible to the trouble source. Hence, repairs are often done during the turn in which the trouble source occurs or quickly following in the turn-transition space before another speaker takes the floor. Self-repairs can correct errors but can also be used to reformulate a turn in order to express something more clearly or fine-tune, or redesign, it to perform a particular action (e.g., to mitigate further a dispreferred). Hence, self-repair includes operations such as redoing part of the turn with the insertion of an additional word or phrase. Repair in the form of a recycled turn beginning may have an interactional function such as securing a person’s gaze as a sign of their willingness to act as recipient to the talk. Another common place for self-repair is in the turn immediately following that of the next speaker, even if this next speaker has indicated no ostensible trouble in hearing or understanding.

Following the norm of doing repair as close as possible to the trouble source, other-repair is by far most commonly issued in the turn following the trouble source. The recipient may just make a straightforward correction. However, this may not be possible and repairing someone else’s talk can be risky socially. So, even if the repair is other-initiated, it is usually designed to allow self-completion. This can be achieved, for example, by the recipient drawing attention to the problem in a mitigated way (such as asking a question) so that resolution is passed back to the original speaker. Hence, where two speakers are involved in a repair, they can make use of the adjacency pair system consisting of an FPP repair initiation followed by an SPP repair solution. Conversation analysis provides detailed analysis of further positions in which repair can be performed in relation to the trouble source and the features commonly associated with such repairs.

Rae (2008) provides an analysis of repair sequences in psychotherapy in which he argues that therapist-initiated repairs containing lexical substitutions are a resource that can be used to prompt clients to describe their feelings in a more explicit or freer way. The following extract illustrating this process is from a person-centered counseling session.

Example 3 [Extract 5 in Rae (2008)]

  1. 07

    Cl: I am surviving and I am

  2. 08.

    Th: But it feels (.) doesn’t feel right

  3. 09.

    Cl: It feels a little uncomfortable

  4. 10.

    Th: Or a lot uncomfortable.

  5. 11.

    Cl: It feels a l(hoh)ot unc(huh)omfortable actually

In line 10, the therapist initiates a repair through proposing a correction to how the client feels. The lexical substitution he uses is to change the client’s word a “little” to the word a “lot.” He therefore suggests a refinement, specifically an upgrade, to the way in which the client has described her feelings. As a proposed correction, this therapist-initiated repair orients to the preference for self-repair and the repair completion itself is bounced back to the client. Hence, in the turn immediately following (line 11), the client responds with the repair completion in which she accepts the therapist’s corrective upgrade. Interestingly, Rae notes that individuals are usually considered to have privileged access to their own feelings. In Example 3 we can see how this often taken-for-granted theory of mind can be problematized in therapy interaction. Antaki et al. (2007) suggest that such claim of expertise about the experience of others may be part of what “doing therapy” is about. However, we still might be surprised to see such therapist recasting of the client’s feelings within a client-centered session. Conversation analysis shows how this process can be performed through the everyday practices of repair.

5 Word Selection

Word selection is an important aspect of conversational exchange. For example, word selection as an aspect of repair was illustrated in Example 3 above in which the therapist substituted the word a “lot” for the client’s word a “little” and in doing so suggested an alternative, more expansive, characterization of her feelings. Such observations reveal the importance of word choice for interlocutors and how attention to the words used, when alternatives are available, is a useful analytic resource for understanding what is being achieved through talk.

A particularly interesting aspect of word selection and focus of conversation analytic work is the way in which persons (and objects and places) are referred to in talk. English-type languages provide dedicated terms, pronouns such as “you” and “I,” to allow reference to persons. Different languages provide alternative possibilities, such as the lexically indexed singular-plural and formality differences in French between “tu” and “vous” (although there are colloquial ways of referring to groups of recipients in English such as “yous-all”). Pronouns, in the main, can be considered to be the reference simpliciter: that is, the simple solution allowing speakers to refer to each other and to third parties (she, he, they) during conversation.

When interlocutors use anything other than the reference simpliciter, it suggests that something over-and-above simple reference is being done. Alternatives to the reference simpliciter to refer to the current speaker or recipient include use of third person (e.g., “she” rather than “I” or “you”), use of one’s own or the recipient’s proper name, but also using the reference simpliciter “I” or “you” with attention-drawing prosody. Such person referencers invite analysis of what is being achieved by this word selection at this point in the conversation.

There are numerous options for referring to non-present third parties, each with different kinds of interactional implication. Conversation analysis identifies two useful sets of overlapping differentiations. First, is the use of locally initial or locally subsequent reference. Locally initial forms tend to be used on first mention of a particular third party. These include proper names, descriptors (e.g., my son), or categories (e.g., one of my colleagues). Once this has been established, reference to this person can be made using a locally subsequent reference such as “she,” “he,” or “they.” Second is the use of recognitional or non-recognitional reference. Recognitionals indicate to the recipient that he or she knows that third party and can, through the description offered, figure out who that person is (e.g., use of the person’s name). Non-recognitional person reference indicates to the recipient that he or she does not know that particular third party. These include descriptions such as “someone,” “this person I met,” and “a guy.”

Person reference, particularly non-recognitionals, can be done using a membership categorization device (MCDs). MCDs go beyond person reference and display culture through additional actions such as description since they make use of categories of person found in particular social settings. Categories of person cluster together into types (e.g., youth subcultures), teams (such as family members mother, father, son, etc.), and sequences (e.g., young, middle aged, old). Categories bring with them a stock of cultural associations and assumptions and so can be a shorthand for typifying an individual more broadly. Moreover, given that a person can be categorized in many different ways, it becomes relevant to analyze what a particular MCD is achieving at the point it is used in a conversation.

The following example illustrates use of a zero-person reference by a therapist during group therapy for addicts. Central to the Minnesota model used in this therapy is the idea that for successful outcome the client must both admit to being an addict and identify with the other group members. Halonen (2008) provides an analysis of how zero-person reference can help facilitate both aims.

Example 4 [Extract 3 in Halonen (2008)]

  1. 01.

    Th2: well how about in the morning when let’s say 0 has

  2. 02.

    drunk more in the evening and in the morning when

  3. 03.

    you wake up so, like when there is a hangover and,

  4. 04.

    like you go

Prior to this extract, the client addressed had not yet produced a description of himself as an alcoholic. The therapist then produces a story in which aspects of this client’s own account of his drinking are typified as addict-like. In producing her description, the therapist uses first a zero-person construction (line 1). This is not a reference simpliciter and so invites analysis of what it achieves over-and-above simple person reference. A zero-person construction leaves the person reference open. Hence, the recounted behavior is framed as something familiar to addicts in general. In her next two person references, the therapist seemingly addresses the client directly: “you wake up” (line 3) and “you go” (line 4). Use of the recipient reference simpliciter “you” closely following the open zero-person construction implies that the client’s behavior, too, is typical of the established addict-like pattern. In English, the pronoun “you” can also be used as an open category of people in general (where the word “one” would be more formally correct). The person reference “you” in lines 3 and 4 therefore may open the possibility of both linking the specific client’s behavior with problematic drinking and describing problematic drinking in a way in which the group members in general can identify. This extract, however, is a translation and Halonen does not comment on whether this ambiguity is also present in the original Finnish. However, using further examples of zero-person construction, she argues that this kind of zero-person reference in the context of group therapy does allow individuals to talk about their own addictions in a way recognized as not unique to the speaker but general to members of the group.

6 Action Formation

It has been made clear throughout that talk-in-interaction is a form of social action in that conversationalists do things with their utterances. This section highlights and pulls together action formation as glossed in previous sections and then considers conversational closings as an important “doing” of relevance to psychotherapy interaction.

The fundamental building block for performing social actions in conversation is the adjacency pair. That is, after a first pair part (FPP) initiating an action, the turn-taking system makes conditionally relevant from a recipient a certain kind of second pair part (SPP) which is responsive to that action. Action is important for the turn-taking system also in that, along with syntactic and intonational features, a turn-constructional unit (TCU) is hearably complete when, in context, it completes an action (e.g., makes a question). Speaker change then becomes the relevant next action.

In conversation there is a preference for progressivity: that is, for moving forward with, or accomplishing, the action projected by the FPP (e.g., invitation-acceptance). However, turns at talk may do, or respond to, more than one action at a time. For example, membership categorization devices (MCDs) may do description as well as person reference. Sequences of adjacency pairs, along with their pre-, insert-, and post-expansions build and close around the performance of an action. And conversation analysis demonstrates that the completion of an action is a more useful way of conceptualizing sequence than is topic: that is, considering what the talk is doing rather than what it is purportedly about.

Where in a sequence a turn is positioned and how it is designed are resources for analyzing what the turn is doing. For example, insert sequences between an FPP and SPP may be doing repair and/or possibly presage a non-preferred SPP through breaking the contiguity between the adjacency pairs—as do design features such as mitigations, hesitations, and weak agreements. In self-repair, the action performed in a turn can be refined or modified through, for example, substituting or adding a word. Moreover, in other-initiated repair, a recipient may indicate a problem in understanding the action being performed in prior talk.

One important action relevant to talk-in-interaction which has been studied in conversation analysis is the act of closing a conversation. Closing requires interlocutors to disengage from the turn-taking system. In practice they need to do two things. First, interlocutors need to check that nothing more needs to be talked about in this conversation. Second, they need to design turns that occasion no further talk but to do so without making their relationship vulnerable. So, interestingly, the act of closing a conversation is achieved by passing up opportunities to do something. This is performed through a particular set of sequences.

The first observation is that closings cannot occur at any point in a conversation. They must occur within closing implicative environments. These environments can be created when a topic is closed down in such a way that it appears that nothing more is to be said on it. This includes making of arrangements for future interactions, providing an overarching summary or assessment which implies that the talk on that topic is complete, offering an appreciation for the opportunity to have interacted, back references to prior topics which suggest that conversational topics have been exhausted, and announcements of closure in which external circumstances are invoked to account for the need to move into terminating the conversation.

Closing the conversation then becomes a possible next action and the interlocutors may move into a pre-closing sequence. Pre-closing sequences consist of an adjacency pair which performs the action of checking if anything more needs to be talked about. In successful closings, the pre-closing FPP and SPP consist of each participant passing up the opportunity to raise further matters for discussion. A second action then follows in which the participants agree to end the conversation. In English, this terminal sequence usually consists of an adjacency pair exchange of goodbyes.

As in ordinary conversations, therapists face the problem of ending sessions in a way that does not damage their relationship with clients. This can be particularly difficult in group therapy, as in the following extract, when clients are engaged in multiparty talk. This example is taken from the ninth of 53 sessions of group psychodynamic psychotherapy for seven women diagnosed with an eating disorder and illustrates interactional achievement of a closing implicative environment.

Example 5 [Final Extract in Lepper and Mergenthaler (2005)]

  1. 1.

    P3: Sure but my mum knows about it—hers

    doesn’t.

  2. 2.

    P5: To have a problem + if you have a problem

    or—if you fail like she said—

  3. 3.

    and you stay there—that’s not right but if you have a problem and you try to

  4. 4.

    find a solution like you did by coming here searching for help—why would

  5. 5.

    your mum feel that everything goes wrong? On the contrary okay, she looks

  6. 6.

    for help.

  7. 7.

    T: Well—very + well.

  8. 8.

    P9: I must go to see the doctor at ten.

Clients P3 and P5 are exchanging turns at talk towards the end of the session. Then, in line 7, the therapist produces a brief assessment “well very well.” We know that turns projecting the end of a sequence often take the form of a brief assessment, summary, or aphoristic formulations of the gist so far since their production implies that the previous talk has been adequately concluded. In line 7 the therapist also passes up the opportunity to continue the sequence or to introduce a new topic. She therefore prepares the way for another interlocutor, client P9, to move the conversation more formally into a closing implicative environment. P9 does so through an announcement of closure in which she makes reference to external circumstances, an appointment with “the doctor at ten,” as imposing an end to the conversation. Hence, in sequence terms, P9 provides the go-ahead for closure. In their more extensive analysis of extracts, Lepper and Mergenthaler argue that therapists fail to tie their turns strongly to the pervious talk as a way to end sequences and, in the above example sessions, in a deliberate manner.

7 Discussion

In outlining the main features of talk-in-interaction of interest in conversation analysis—turn-taking, sequence organization, repair, word selection, and action formation—it appears that the analytic lens and conceptualization of key phenomena is different in many respects to that of traditional psychotherapy research. Moreover, as mentioned earlier, even though a literature has developed on conversation analysis of therapy interaction, selection of material has been, in the main, in accordance with conversation analytically-informed, as opposed to therapy-informed, observations. It is therefore pertinent to ask if the projects of conversation analysis and of psychotherapy research can cohere? And one of the most experienced and methodologically informed psychotherapy researchers, Bill Stiles,Footnote 1 does caution that “CA concepts cannot be inserted unchanged into gaps in therapy theory. Therapists and conversation analysts must learn each other’s theories and make adjustments if the product is to be mutually useful” (Stiles 2008, p. 2).

One important issue is that conversation analysis is agnostic to the psychological theories that inform psychotherapy and its interventions and which frame most traditional psychotherapy research. Conversation analysis approaches dialogue as intersubjectivity in action: that is, how interlocutors produce a shared understanding of the matter at hand in their talk, and true it its ethnomethodological roots, just as it is the talk and associated observations that is available to speakers, conversation analysis limits its observations strictly to this material. What are in other fields considered “internal” (emotions, motivations, the unconscious, etc.) are analyzed as they are constituted in and through the interaction and conversation analysis refuses to comment on the experience of interlocutors (Forrester and Reason 2006). This may seem psychologically shallow to the psychotherapy community: too removed from basic assumptions about human subjectivity and mute on questions of experiential change which are likely of interest to therapists. Moreover, conversation analysis could be considered naïve methodologically if it ignores the vast corpus of psychotherapy theory and research which could help target episodes of therapy for microanalysis.

This therapy-neutral orientation may, however, be a significant strength in allowing conversation analysis to complement and enhance process research through revealing what psychotherapy may not notice about itself (Antaki et al. 2007). In particular, Georgaca and Avdi (2009) argue that conversation analysis is useful for illustrating “the intricate micro-processes through which psychotherapeutic technique is pursued (…), throw light on unacknowledged therapist competencies and illuminate the elements that differentiate successful from unsuccessful implementation of therapeutic techniques” (p. 241). In fact, conversation analysis may be the method par excellence Footnote 2 for raising to awareness tacit skills of both therapist and client in progressing the therapeutic project: fine-grained, moment-by-moment, making the ordinary appear extraordinary (Kozart 1996). At the very least, the orientation and observations of conversation analysis may have potential to hone therapists’ skills of attention during their training (Forrester and Reason 2006).

For example, as outlined above, Rae (2008) noticed that lexical substitutions (e.g., exchanging the word a “lot” for a “little”) can be used by therapists to prompt clients to describe their feelings in a more expansive manner. Traditional therapy researchers may too make this, seemingly mundane, observation. However, as a conversation analyst, Rae recognizes it as an important conversational phenomenon—a repair—and to bring into play an extensive literature on how repair works in conversation which sheds light on how this therapy technique is used and responded to by clients. Moreover, person-neutral reference was identified by Halonen (2008) as a potentially nonthreatening way of helping clients to acknowledge problematic behavior and to facilitate identification with other clients during group therapy. Such attention to word selection and its importance in bringing off social actions is central to conversation analysis and, again, connects an easy-to-overlook therapy intervention to a relevant, and extensive, research literature.

Kozart (1996) reminds us that conversation analysis of institutional talk is not new and that the growing conversation analytic literature on psychotherapy is positioned within a wider literature on medical discourse more generally. The specific focus on therapy interaction, however, has developed to the extent that it has warranted review. Georgaca and Avdi (2009) provide a useful overview highlighting how conversation analysis has contributed to understanding the processes through which therapy is accomplished in practice and in assessing the role of the therapist. In particular, studies are noted to have focused on therapist formulations of client’s talk, use of specific interactional formats, and of idiomatic expressions as important aspects of the therapeutic process. Although published too recently to be included in Georgaca and Avdi’s review, Bercelli et al.’s (2008) analysis of the contrast between therapist formulation sequences and reinterpretation sequences, described above, contributes further to this theme, and Vehviläinen et al. (2008) draw attention to sequence organization as a particularly important “site at which many therapy-relevant phenomena happen” (p. 188).

Conclusion

Conversation analysis of therapy talk has, as yet, hardly scratched the surface and the possibilities for future research are exciting. Vehviläinen et al. (2008) suggest that, in particular, there is potential for more analysis of client, as opposed to therapist, actions, and examination of the extent to which the phenomena identified in conversation analytic research are general or specific to different types of therapy. My own suggestion is that there is potential also for conversation analysis to be informed to a greater extent by traditional psychotherapy research without losing its relatively atheoretical stance. This may be accomplished, for example, through accepting for analysis episodes of therapy interaction deemed significant from the viewpoint of psychotherapeutic theory. The link to what therapists find important is therefore strengthened while the usefully distinct orientation to the material provided by conversation analysis can be maintained. For example, as described above, Wynn and Wynn (2006) acknowledge the theoretical and psychological importance of empathy in therapy while exploring what different types of empathy and, indeed, empathy failures look like interactionally through, for instance, the technology of turn-taking.

This chapter has offered an introduction to conversation analysis and the ways in which it has contributed and is likely to continue to contribute to psychotherapy research. It has been suggested that, although their projects may differ, the “noticings” facilitated by a conversation analytic approach are being demonstrated to offer something unique and interesting to psychotherapy researchers and practicing therapists.