An introduction to conversation analysis in social work research

What is conversation analysis?

Conversation analysis (CA) is an approach to the study of talk (and embodied conduct) in interaction (Schegloff, 2007). Its name is somewhat of a misnomer since ‘conversation’ is not conceived in terms of ‘just words’ but encapsulating all the resources used to interact, including prosody, gesture and gaze, across any and all settings and often in aggregate. Conversation analysts have, for example, analysed encounters on the telephone with friends (Sacks, 1995) or with service providers on helplines (Baker et al., 2005), in the operating theatre during surgery (Mondada, 2014), in police stations during interviews with vulnerable witnesses (Richardson et al., 2019), during remote (e.g. Seuren et al., 2020) and in-person GP appointments (Albury et al., 2022), and while playing board games or using maps for orienteering (Hofstetter, 2020; Pehkonen et al., 2021). CA emerged from the sociological field of inquiry called ethnomethodology (Garfinkel, 1967), or the study of people’s methods, which is concerned with investigating the everyday, common-sense practices people use to make sense of and construct their social worlds (Clayman, 2001). Developed by Harvey Sacks, and his colleagues Emmanuel Schegloff and Gail Jefferson in the 1960s and 70s, CA built from these ethnomethodological roots to develop an empirical approach for studying the structure of talk-in-interaction. The agenda of CA is to make visible and analyse the ‘common-sense’ of social interaction, centrally examining ‘how [people] themselves make sense of what is said’ Psathas (1995: 52) by looking at what they orient to in interaction and how they orient to it (Drew and Heritage, 2006; Liddicoat, 2021; Sacks, 1995).

The core assumptions of CA are that 1) interaction is a form of social action; 2) interaction is orderly; 3) interaction creates and maintains intersubjectivity (Wilkinson and Kitzinger, 2017) and 4) how people make sense of any given moment in interaction can be seen on the conversational surface (Sacks, 1995). Let us unpack each a little further. First, and similar to other discursive approaches, people do things with talk. Social actions include things like asking, answering, inviting, persuading, advising, complaining and listening. By analysing ‘naturally occurring’ interaction, conversation analysts show how conversational actions are done ‘in the wild’ across and within different contexts. CA has examined everyday interactions, unpicking actions such as greetings (Sacks, 1995) or compliments (Pomerantz, 1978), as well as actions in institutional interactions, for example, advice giving in healthcare (Heritage and Sefi, 1992) or active listening in child counselling (Hutchby, 2005).

The second core assumption, that social interaction is orderly, has been demonstrated and explicated through the identification of fundamental structures or ‘rules’ that are maintained and oriented to by people themselves. These rules of conversation are considered social ‘common-sense’ to people (Goffman, 1983; Liddicoat, 2021). For example, that people take turns at talking relies on the fact that turns are contingent on what came before, which sets conditions on what can come next, for example, a greeting makes relevant a greeting in return; an invitation makes relevant an acceptance or declination. The system of turn taking is observable in unfolding sequences of reciprocal actions in talk, revealing a further structure of conversation called sequence organisation. For conversation analysts, understanding where a turn comes at a specific point in a sequence is essential to understanding what action that turn is ‘doing’. Two further important structures, which are foundational to a conversation analysis approach, are turn design and preference structures. Turn design refers to how people shape what they say, based on the context and recipient, to do the desired action and preference structures refer to how the shape of these turns places certain constraints on how people can respond. For example, accepting an invitation is interactionally more straightforward than declining it since acceptance does not typically require an account and declinations typically involve and reveal the management of social relations. Understanding the structures of talk helps us to identify what people are doing in interaction with others and how they are doing it. These fundamental structures are foundational to the analyses you will see across the papers in this special issue (for further information on the structures of talk, see Sidnell and Stivers, 2012).

The third assumption is that through talk people create and maintain intersubjectivity. Here, intersubjectivity is the understanding people come to share through social interaction, rather than a hidden process taking place in their minds. For example, by giving an answer, the respondent is showing they understand the prior turn to be a question. People design their turns at talk to achieve mutual understanding and ‘repair’ mishearings, confusions and other kinds of interactional trouble as a sequence unfolds. The smooth progress of a conversation can, at any point in or between turns, be halted as people correct and repair what they say to handle what they take each other to know (‘you know Jenny- uh- the woman in my office’), alter the design of a question to handle their entitlement to ask (e.g. ‘Can you- um would it be possible’) or get another person to repair what they are doing to achieve intersubjectivity (e.g. ‘huh?’). In other words, there are myriad practices to ensure that communication does not ‘break down’.

Finally, since the achievement of intersubjectivity is visible on the conversational surface – which is ‘rich’ (Edwards, 2006) and often ‘broken’ and under repair (Jefferson, 2018) – we can record, analyse and explicate it. As such, analysis in CA is grounded firmly in what the people in the interaction are doing, primarily through the use of the next turn proof procedure (Edwards, 2004), where the analysis of what one person’s talk is doing is based on the response in the next speaker’s turn – that is, how the people in the conversation make sense of it and construct a shared understanding.

In order to do this type of analysis, conversation analysts work with naturally occurring data (Speer, 2002). These are data that pass what Potter (2002) refers to as the ‘dead social scientist test’ and are encounters that are not set up by researchers, such as interviews, focus groups, experiments or simulations. For example, telephone calls between friends or a GP appointment do not have to be arranged or orchestrated by the researcher; instead the researcher will video or audio record these interactions ‘in the wild’, with participant consent (Golato, 2017). Indeed, many institutional encounters are already recorded as part of the ordinary work of the setting (e.g. police interviews; telesales) and can be made available – through strict protocols – to researchers. Access to recordings allows the researcher to view and listen to data repeatedly and transcribe them using a standard system (Jefferson, 2004). As a specialist rather than verbatim transcript, transcription is part of the process of analysis itself (Hepburn and Bolden, 2017). The Jefferson system uses notation to capture how and where talk happens, including pace, volume, pronunciation, pauses, breaths, overlap onset and sequential position. We include a transcription glossary at the end of the editorial. The system has been developed to include features of emotion (e.g. laughter and crying – see Hepburn, 2004) and the multimodal resources people use in interaction (e.g. eye gaze and body torque – see Mondada, 2018). Just as the standard system for music, notation enables those who understand it to know what notes to play and how, the dots, dashes and symbols of conversation analysis tell us how to hear and see the interaction to inform our understanding of what is being said, how it is being said and what action is being done (Stokoe and Albert, in press). The authors of papers in this special issue use or sometimes adapt these systems to help readers understand how interaction happens, as well as to enable readers to – very transparently – judge the validity of the analysis directly.

Conversation analysis in institutional settings

Although the first studies in CA were of ‘the search for help’ in a suicide prevention centre (Sacks, 1995), a great deal of this research has been concerned with the fundamental machinery, structures and actions of ordinary conversation, examining interactions without any particular preconception or interest in a predefined problem. However, CA is a powerful method to apply to social problems, and thus, an enormous field of research has developed across institutional settings. CA has generated many insights into how communication works in specific institutional contexts and also how the work of those institutional contexts themselves function. This research not only focuses on analysing the actions through which institutional tasks are accomplished but also how differently designed actions shape outcomes and how such analyses can comprise evidence-based solutions to problems (for overviews, see Antaki, 2011; Lester & O’Reilly, 2018).

One of the best known studies in recent years brings the qualitative and quantitative together in a CA-augmented randomised controlled trial examining interaction in healthcare (Heritage et al., 2007). While CA is largely a qualitative method, it is also used by some as a robust way of coding for statistical analysis and working with large and comparative datasets (e.g. Dingemanse and Enfield, 2015; Stivers, 2015), though there is also debate about quantification in CA (e.g. Robinson, 2007; Schegloff, 1993). In their study, Heritage et al. (2007) looked at how understandings about interaction from CA could inform an intervention to reduce patients’ unmet concerns in doctor appointments. To address the issue of patients not reporting additional concerns in appointments, medical school training recommends that doctors ask, ‘is there anything else we need to take care of today?’ However, even after being asked this direct question, many patients still did not report their further concerns. Based on the understandings of turn design and preference structures from CA, Heritage and colleagues designed two experimental conditions. In one group, GPs continued to ask patients if they had ‘anything else’ to address during their appointment whilst the second group asked if the patient had ‘something else’ to address. The hypothesis was that because ‘any’ is, in grammatical terms, negatively polarised and designed for a ‘no’ response and ‘some’ is positively polarised and designed for a ‘yes’ and elaborated response, people would be more likely to say yes to the latter. The ‘something’ condition resulted in significantly more patients reporting additional concerns that could be dealt with in the appointment (90% versus 50%). Changing just one word in this question significantly reduced patients’ unmet needs, changing the outcome for patients and GPs, and demonstrated the power of applying CA to deal with real-world problems (Heritage and Robinson, 2011; Robinson and Heritage, 2014).

Stokoe and colleagues also demonstrated the power of words to shape outcomes through research on mediation services and police negotiations. For example, in examining the intake calls between members of the public and neighbour dispute mediation services, mediators who asked if callers – who often resisted the offer of mediation – were ‘willing’ to engage rather than if they thought it might ‘help’ them or if they were ‘interested’ were more effective in getting callers to become clients (Sikveland et al., 2019; Sikveland and Stokoe, 2016; Stokoe, 2013). The researchers showed that ‘willing’ worked in this context because it opens up the possibility not just for callers to agree to a service but to characterise themselves as someone who is ‘the willing type’, someone who wants to address neighbour dispute, the reasonable rather than the troublesome or difficult party (Sikveland and Stokoe, 2016). Even swapping words that are near-synonyms can be significant, as Sikveland and Stokoe (2020) found with the use of ‘talk’ and ‘speak’ in police crisis negotiations with people in suicidal crisis. Given that the success of any negotiation requires keeping persons in crisis talking – quite literally taking conversational turns rather than hanging up the phone or much worse – they found that when negotiators used the word ‘talk’, people in crisis resisted the negotiation. ‘Speak’, on the other hand, reduced resistance and enabled smoother progress towards a safe outcome.

In both cases, ‘willing’ and ‘talk/speak’ were identified as components of effective practices through the work of collecting recordings, transcription and analysis. Although they were already part of some but not all mediators’ and negotiators’ practice, the fact that they did not appear in any training guidance and were not used by all practitioners starts to show the value of applied CA research. CA has provided valuable insights into a range of other institutional settings by examining the talk ‘in the wild’. In healthcare settings, for instance, researchers have examined how doctors conduct end-of-life conversations with patients in palliative care (see Pino and Parry, 2019), how practitioners speak to women who have experienced birth trauma (Kitzinger and Kitzinger, 2007) or how GPs manage patient’s refusals to referral for weight loss support (Albury et al., 2022) (for a fuller review of CA in healthcare, see Barnes, 2019). This research has helped to not only explicate the institutional practices and actions in interaction within healthcare settings but also recommended and informed changes in communication practices to improve patient outcomes (Barnes, 2019). In counselling and psychotherapy research, for instance, conversation analysts have demonstrated how the conversational practice of formulation, that is, summarising the gist or upshot of what the other has said (Heritage and Watson, 1979), achieves institutional goals. Vehviläinen (2003) noted through formulations psychoanalysts could achieve a key goal of psychoanalysis, giving insight into the client’s unconscious processes, by proposing interpretations of client experiences grounded in the client’s own talk. Voutilainen et al. (2011) considered how formulations can support client change over time. Hutchby (2005) demonstrated how child counsellors show active listening, a core counselling skill, through their skilled use of formulation, which translated the child’s talk into a therapeutic object, making it ‘therapizable’ (Antaki, 2008: 35, emphasis our own). Like Heritage and colleagues, Hutchby highlighted the gap between the professional guidance and the reality of interactions in practice, noting that at best the guidance on how to ‘actively listen’ was vague, and at worst misrepresentative of what actually happens (e.g. advising the counsellor not to ask questions). Here, CA was able to uncover the detail of what active listening looks like in these interactions and what the consequences were, including unintended ones such as the child resisting the counsellor’s interpretation of their family issues. Work in this area has looked at a range of topics and practices, such as empathy, emotion and resistance (see, e.g., Kupetz, 2014; Muntigl and Horvath, 2014; Peräkylä et al., 2008; Voutilainen et al., 2018; Weiste and Peräkylä, 2014) and by making visible these processes can provide clear evidence to improve and inform practice (O’Reilly and Lester, 2019).

Conversation analysis for social work

Communication is the ‘backbone of social work’ (Hall et al., 2014: 2) and conversation analysis gives us a way of understanding how that communication functions in practice, the impact of the words we use and the way we use them, and how we achieve (or not) the aims and goals of our interventions with clients. However, as noted by Flinkfeldt and colleagues (this issue) in their scoping review of CA and social work, the application of CA in social work research is a relatively small and developing field. Members of the international research network, Discourse and Narrative Approaches to Social Work and Counselling (DANASWAC), have played a major role in taking forward research on social work interaction using CA and related approaches over many years. For example, in their edited collection Constructing Clienthood in Social Work and Human Services, Hall et al. (2003) examined how identities – such as client, parent and child – are constructed and function in practice. A further book focusing on discourse in social work used CA and other approaches, to analyse key topics in social work such as resistance, advice giving and accountability, through close examination of social work encounters (Hall et al., 2014). And more recently, in Interprofessional Collaboration and Service User Participation (Juhila et al., 2022), members of this research network turned their focus to the interactional conduct in interdisciplinary meetings, examining how ideals of service user empowerment are or are not realised on the ground.

Existing research in this field has already given fascinating insights into the realities of communication in social work practice. For example, Juhila et al. (2014), examining ‘resistance’ in social work, challenged the idea that there is a category of ‘resistant’ clients and rather analysed how resistance is performed in specific interactions. They illustrated that clients work up and justify ‘resistance’ to social worker suggestions in specific ways, that social workers can also be seen to ‘resist’ clients’ preferred courses of action and moreover that clients and social workers can also align themselves to collectively ‘resist’, for example, unjust institutional procedures. Symonds (2017) outlined how practitioners were able to engage parents referred to a parenting programme through their talk by giving them space and encouraging them to share their problems, demonstrating empathy and formulating their position in a way that was relevant for the service. By focusing on the specifics of how conversations unfold, such detailed work can demystify the reality of social work practice and generate understandings of value for research, theory, policy and practice. This special issue continues this work, showcasing the value of CA to extend our knowledge of social work practice ‘in the wild’.

Introducing the special issue

Here, we bring together a collection of 14 exciting papers which showcase and develop this field further, highlighting the potential and value of conversation analysis as a research approach to understanding and informing social work practice. We open with Flinkfeldt and colleagues’ scoping review of conversation analysis in social work research which maps the development, contribution and future direction of the field. They specifically recognise the potential for research in this area to inform practice, calling for conversation analysts to be clear about the practice recommendations emerging from their research.

The next three papers broadly focus on how the professional relationship with social work clients is built to achieve institutional outcomes, such as client engagement or agreeing a focus for intervention. These papers also focus on what terms and topics like ‘small talk’ and ‘respect’ might actually look like in real talk, rather than the more idealised and generic definitions found across communication skills literature (see Sikveland et al., 2022). First, Iversen and colleagues, in their study of Swedish peer support helplines for older people, show how so-called ‘small talk’ is not separate or an additional extra to the core institutional business of social work but itself forms and facilitates the institutional tasks by promoting connectedness and providing a foundation to discuss being active as an older adult. Second, Mullins and Kirkwood examine how the skills of ‘warmth’ and ‘respect’ are done in social work practice with people who have committed sexual offences, offering a theoretical framework to understand these as interactional practices. They argue that understanding ‘warmth’ and ‘respect’ as concrete practices rather than abstract skills goes further towards more precisely describing and pinning down the work of social work. Next, Canty explores how social workers and clients can establish the purpose and focus of their interactions together, outlining the necessarily collaborative interactional steps through which a ‘social workable matter’ is achieved as an outcome of a meeting or interview.

In the next three papers, Williams and Symonds, Nilsson and Olaison, and Bostock and Koprowska examine the core social work practice of conducting and managing assessments, across older adult and child protection contexts. Returning to the role and manifestation of informal talk, Williams and Symonds question the implications of ‘conversational’ or ‘friendly’ assessment conversations when access to resources may be at stake and clients may inadvertently disadvantage themselves by not knowing what constitutes ‘assessment relevant’ information. Nilsson and Olaison report on the ways social workers work to persuade clients with dementia to engage with certain services and how clients can and do resist this. Bostock and Koprowska explore how social workers make use of written reports in interactions with families, highlighting the way disputes may occur and be managed. The analysis demonstrates how social workers may create instances of what they term ‘epistemic injustice’ through privileging the formal output of meetings – the written report – over family members’ own testimony.

The wide scope and variety of both CA and social work practice – and the modalities through which social work occurs – is demonstrated in the next three papers. First, Thell explores how counsellors pursue and construct a shared understanding of the client’s problem in an anonymous online text-based counselling service, highlighting the challenges this virtual context places on communication. Majilesi and colleagues and Monteiro demonstrate how to examine multimodal interaction in social work. Majilesi and colleagues note the importance of examining multimodal interaction in the context of physical care giving, highlighting how caregivers can respect clients’ personhood through their actions and not reduce them to an object of care. Next, Monteiro describes how social workers and clients end their sessions, using linguistic, bodily and material resources, with particular attention to how social workers manage the balance of care and control. These papers showcase the way that written and spoken interaction and embodied conduct collectively constitute the work of social work and care.

The delicate processes of decision-making are tackled by Osvaldsson Cromdal and Cromdal, and Webb, Wilkins and Martin. These two papers discuss and highlight the processes by which decisions are arrived at (or not) and importantly note how what is at stake for the practitioners, in terms of accountability and professional relationships, come into play in these key moments for all involved. Our final two papers then focus on a core skill in practice: reflection. Symonds and colleagues consider how social work supervisors encourage practitioners to reflect on cases within supervision discussions by asking about other people’s understandings. They note several competing interactional and moral constraints which impact how social workers can ‘reflect’ in these discussions. Finally, Dall and Caswell demonstrate how CA can be used as a potent resource to support practitioners in developing their reflective skills by making the routine visible and providing a space away from evaluative assessment to examine their own practice.

The span of this collection of articles demonstrates the variety, scope and capability of conversation analysis as an approach to examining social work practice. Analysing what actually happens ‘when social worker and client meet’ has the potential to improve outcomes for our clients and our practitioners by improving the evidence base and our understanding of these often hidden, deeply personal and crucial encounters.