Liz Stokoe and Saul Albert 'Just a Method in Search of a Problem?: The Power of Conversation Analysis

‘Just a Method in Search of a Problem?’

The Power of Conversation Analysis

Elizabeth Stokoe and Saul Albert

From Tadic and Waring (Eds) Critical Conversation Analysis (2024)

Introduction

This chapter explores the power of conversation analysis to reveal

ethical, moral and other deeply problematic assumptions about commu-

nication that have personal, institutional and societal consequences. Such

practices would otherwise remain largely unknown, incorrectly imagined,

unevidenced, disattended or obscured, since the worlds of communication

and conversation and the phenomena of the ‘psychological thesaurus’

(Edwards & Potter, 2005) – i.e. experience, cognition, attitude, memory,

etc. – are usually investigated using artificial or post-hoc methods such as

simulations, experiments, surveys and interviews (see Stokoe, 2020).

Despite a long tradition of revealing the systematic interactional work-

ings of power, prejudice, resistance and so on, most conversation analysts

will recognize the caricatures of their research methods that circulate

within critical social science and discourse analysis (Stokoe et al., 2012).

These include accusations of ‘pointless empiricism’; the ‘dangerous adop-

tion of relativism’; focusing on ‘nothing but the text’; failing to ‘deal with

subjectivity’; analyzing features of interaction that ‘are of no consequence’;

an ‘aversion to theory’; and thus of failing to address the ‘big and impor-

tant questions’ (e.g. Frosh, 1999; Parker, 2005, 2012; Sarangi, 2018).

According to Oliver et al. (2005: 1277), ‘the conversation analyst wants to

learn about talk’ while ‘the critical discourse analyst wants to learn what

this talk says about other aspects of the participant’s life’.

We do not recognize conversation analysis in these characterizations.

Since its inception, conversation analysis (CA) (along with ethnomethod-

ology, discursive psychology and membership categorization) has expli-

cated the interactional practices – sequential, categorial, multimodal,

embodied – that constrain and afford the workings of power, resistance,

justice and social change. In contrast to the way it is sometimes carica-

tured, CA’s engagement with and understanding of institutionality, racism

and other ‘-isms’, prejudice and power, provides a robust methodological

apparatus for their systematic analysis (e.g. Previtali et al., 2023; Rawls

et al., 2020; Yu & Sterponi, 2023; see the other chapters in this book).

The aim of this chapter is not to revisit debates about critical inquiry

but to argue for and illustrate the power of conversation analysis to expli-

cate, expose and challenge erroneous but widely held views about com-

munication that underpin the interactional procedures, asymmetries and

actions that are consequential for people’s everyday encounters with social

injustice. We will show, through illustrative cases, how, for instance, insti-

tutional encounters deviate from the written guidance and standardiza-

tion that are presumed to underpin and ensure their parity and integrity.

We will explore the deeply problematic practice, used by many organiza-

tions, of applying normative criteria about what counts as ‘good’ com-

munication in their training and assessment of employees – from written

assessment criteria to algorithmic coding in speech analytics and conver-

sational artificial intelligence (AI). Relatedly, the final section of the chap-

ter will show how communication training can be improved using

conversation analytic evidence, including for dealing with ‘-isms’ (Stokoe,

2015).

The fundamental precursor to each case we offer below, and the start-

ing point in our argument for the power of CA, is standard conversation

analytic research design: using audio and video recordings of actual

encounters illustrated with detailed technical transcripts.

The Power of Recordings and Transcripts

A commonly articulated reason for doing qualitative research is to

enfranchise research participants by eliciting what is important to them in

their own voices and words as ‘active participants’ in research activities

(e.g. Hutchinson et al., 2004; Johnston et al., 2021: 544). It is somewhat

ironic, therefore, that so much qualitative research collects and analyzes

participants’ voices by constructing unfamiliar, simulated research set-

tings, or by removing the participants’ words and voices from their natu-

ralistic interactional settings. Qualitative approaches often position

themselves in contrast to ‘alienating’ processes of quantification that risk

losing vital qualities of the phenomena in question (Jovanović, 2011).

However, qualitative analytic accounts about people’s lives and experi-

ences are often generated through semi-structured interviews that are set

up for recording and verbatim transcription without any intrinsic and

endogenous connection to participants’ everyday communicative prac-

tices. Like survey-generated data, interview-generated accounts are neces-

sarily post-hoc and consistently oriented to the researcher’s agendas,

questions or topics (Rapley, 2001) – the ‘shadow cast upon the social

world by the prior conceptual apparatus deployed by the person who con-

structed the [questions]’ (Harré, 1993). Interview interactions are shot

through with epistemic and presentational problems, from the inaccuracy

and biases of researchers’ memories in producing field notes and analytic

accounts, to the particularities of the researchers’ stakes in doing the

interview in the first place (Potter & Hepburn, 2005).

A related problem with most interview data is their forms of written

representation. Quoted data extracts that follow painstaking theoretical

and cultural/historical contextualization are typically stripped of their

interactional context. Researchers often emphasize the reflexive and co-

produced nature of interview data while, at the same time, failing to pres-

ent and analyze how the interviewers’ own questions may guide or

constrain the participants’ responses. For example, Charmaz and

Thornberg (2021) discuss what they cite as an exemplary case of qualita-

tive interview research in which adult victims of school bullying describe

their experiences to researchers that have set out to understand the par-

ticipants ‘as social actors and the social interactions they have been

involved in’ (Thornberg et al., 2013: 312). However, only short transcript

extracts are presented, without including the questions asked or the fea-

tures of ‘active listening’ that ‘interviewers were instructed and trained’ to

do (2013: 312). Thus, there is another kind of stripping away, which is of

interviewer and participant voices and embodied conduct that work in

aggregate in the sequential organization and production of accounts.

What should these short transcript extracts include? There is much

discussion of transcription ‘as theory’ (Ochs, 1979) and what to repre-

sent or not. Oliver et al. (2005: 1273–1274) refer to a transcript contin-

uum, from ‘naturalism, in which every utterance is transcribed in as

much detail as possible’, to ‘denaturalism, in which idiosyncratic ele-

ments of speech (e.g. stutters, pauses, nonverbal involuntary vocaliza-

tions) are removed’. On this continuum, Jefferson’s (2004) system for

conversation analysis is ‘natural’: it seeks to represent how turns of talk

are designed, where they are placed sequentially, timed gaps and pauses,

the onset and end of overlapping talk, and aspects such as pitch, empha-

sis, laughter and crying (Hepburn & Bolden, 2018). Some have criticized

CA’s use of Jeffersonian transcription for contradictory attachments to

incompatible ontologies (e.g. Ashmore & Reed, 2000); as a means of

empiricist standardization (Sharrock & Anderson 2017); as structuralist

or behaviorist (Atkinson, 1988); or as encoding ‘conventional sociologi-

cal preconceptions’ (Hester & Francis, 1990: 398). Others object to

‘excessive’ and ‘surplus’ levels of detail (Bogen, 1999). For yet more

others, the Jefferson system is not detailed enough, in terms of the pho-

netic details needed for auditory or acoustic analysis (Walker, 2012).

Relevant to ‘giving voice’ in qualitative research, Oliver et al. (2005)

further suggest that ‘natural’ transcripts may encode bias by, say, attempt-

ing to transcribe accents. However, analysts have principled ways of work-

ing with accent and generally only encode different ways of pronouncing

the same word where that is clearly interactionally relevant, such as when

‘participants’ pronunciations are the focus of the analysis’ or ‘when

aspects of the speaker’s pronunciation become an oriented to feature of

interaction’ (Hepburn & Bolden, 2018: 53–54). As Schegloff (1997) writes,

in conversation analysis, ‘it is the orientations, meanings, interpretations,

understandings, etc. of the participants... which are privileged in the con-

stitution of social-interactional reality’ and so ‘have a prima facie claim

to being privileged’ (1997: 166–167, emphasis in original). This principle

is realized in the Jefferson transcription system.

We want to step back from these debates to make different points

about giving voice to participants and the power of conversation analytic

transcription. We present two examples. The first comes from a British

televised interview in June 2020 between a journalist, Sophy Ridge, and

the UK government’s Health Secretary, Matt Hancock, in which Hancock

named two British Asian politicians (Priti Patel, Home Secretary, and

Rishi Sunak, Chancellor), in response to a question about the number of

Black people in the cabinet.

(1) Ridge/Hancock 7.6.2020 (See https://youtu.be/lEp0ENBmQLg?t=265)

01 Ridge: .h How many: (0.2) Black people are: (.)

02 in the current #↓cabinet.#

03 (2.6)

04 Hancock: U::m: (0.5) thē:: ↑W’ll.

05 (0.2)

06 Hancock: ↓#Uh- (in-)# (0.3) W- g- there’s a- whole

07 ↑series of people from a: (0.3)

08 Black an’ maha- minority ethnic

09 background? .hh u::h the Chancellor of

10 the Exchequer? (.) the Home Secretary?

11 .hh u:m as:- (.) £to na(H)me but tw(Hh)o£

12 [so:: th- it’s uh y’know it’s uh actually- ]

13 Ridge: [I’m just talking about Black- Ss:- >Sorry<]-

14 (.) I’m talking about Black people

15 specifically ...

This interview was lauded as a ‘gotcha’ moment in the British media since,

at the time, a factual answer to Ridge’s question would have been ‘zero

Black people’. The aftermath of the murder of George Floyd had prompted

a public debate about the disadvantages and disparities experienced by

Black people in British society. Media headlines such as ‘Matt Hancock

Names Two Asians When Asked To Identify Black Cabinet Members’

(Singh, 2020), highlighted Hancock’s inability to name any Black cabinet

members, and academic analysis (e.g. Bankole, 2020) enumerated and dis-

cussed the impact of the lack of Black people in powerful governmental

roles. However, the verbatim quotes from Ridge’s interview used to illus-

trate these articles omit informative interactional levels of detail.1 For

example, Hancock’s response is filled with disfluencies, delays and other

marks of structural dispreference such as the long silence before answer-

ing (line 03), hesitation (U::m), abandoned components and restarts (thē::),

and the ‘well’ preface (line 04). Hancock’s response between lines 04–12

is also littered with repair. After a turn-initial u::m and a pause on line 04,

he starts to reply (thē::), but abandons this beginning and restarts with

↑W’ll. After a 0.2 second gap (line 05), he again abandons development of

the prior TCU and starts a new one, with many more repair initiation

components (↓#Uh- (in-)# (0.3) W- g-, line 06). Apart from anything that

can be inferred from these details, such delays and disfluencies work to

extend the time available before a politician is obliged to provide a

response. These disfluencies also tend to cluster around particularly trou-

blesome and therefore informative moments in the interaction.

Firstly, Hancock’s response, when it finally comes – that there is a

whole ↑s’ries of people – orients to the preference requirement of the ques-

tion to supply a number (How m’ny:, line 01), but expands the constraints

of the question by expanding the categorial component ‘Black’ contained

in the initiating action from a more specific category of Black people (line

01) to a broader category of people from a: (0.3) Black an’ maha- minority

ethnic background, lines 07–09). Furthermore, this maha- minority repair

on line 08 is an example of what conversation analysts call ‘same-turn’ or

‘self-initiated self-repair’ (Schegloff et al., 1977), a common type of repair

found in naturally occurring interaction that, in this case, pinpoints the

sequential location of Hancock’s trouble in responding to Ridge’s ques-

tion. Whether the self-repair to Black an’ maha- minority ethnic (usually

contracted to BME) at line 08 is a self-correction from ‘majority’ to

‘minority’, or another incidental misspeaking, repair marks this term as a

source of specific interactional trouble. Both the terms ‘BME’ and ‘BAME’

are deprecated in official guidance (UK Government, 2021; see Aspinall,

2021) because they aggregate groups racialized by different identity cate-

gories such as skin color or geographical location, while excluding other

persecuted minorities racialized as White such as Gypsy, Roma or

Travellers of Irish Heritage (Bunglawala 2019).2 It is ironic, then, that

Hancock chooses, then stumbles over this term in the context of being

made accountable, as a White Cabinet minister, for the lack of Black

members, before tentatively naming two of his British Asian Cabinet col-

leagues (e::r the Chancellor of the Exchequer? (.) the Home Secretary?,

lines 09–10) – notably using their official titles instead of their identifiably

Indian names (Rishi Sunak and Priti Patel).

Finally, Hancock’s laughter inflection in line 11, produced following

his own response to a ‘serious’ question, resists the premise or trajectory

of Ridge’s questioning by treating it as laughable (Glenn, 2003: 142–144).

Since media reports, quotes and print representations do not usually tran-

scribe delays, disfluencies and repairs, even when politicians are held

accountable for their failure to answer a question, they are not always held

accountable for the disingenuity and deliberate evasiveness evident in the

interactional detail of their answers.

The second example is a leaked audio recording of an off-air conversa-

tion in January 2018 between two British Broadcasting Corporation

(BBC) journalists, John Humphrys and Jon Sopel, about another journal-

ist colleague, Carrie Gracie. Gracie had resigned from the BBC (a ‘bomb-

shell’ decision, according to BuzzFeed) the previous week, in the context

of revelations about a gender pay gap at the organization. On 12 January

2018, The Guardian newspaper headline stated that ‘John Humphrys

jokes about BBC gender pay gap in leaked off-air recording’. Extract 2a

is part of an orthographic transcript of the leaked recording published on

the news website Buzzfeed (Di Stefano, 2018), followed by Extract 2b,

which is the same segment transcribed for conversation analysis.

(2a) Buzzfeed transcript: John Humphrys and Jon Sopel, Jan 2018

Humphrys: ‘Ah... Can you hear me Sopel?’

Sopel: ‘Humphrys, I can hear you’.

Humphrys: ‘Good, slight change of subject — first question will be

how much of your salary you are prepared to hand

over to Carrie Gracie to keep her, and then a few

comments about your other colleagues, you know,

like our Middle East editor and the other men who are

earning too much...’

Sopel: ‘I mean, obviously if we are talking about the scope

for the greatest redistribution I’ll have to come back

and say, ‘well yes Mr Humphrys, but...’’.

Humphrys: ‘And I could save you the trouble, because I could

volunteer that I’ve handed over already more than you

fucking earn, but I’m still left with more than anybody

else and that seems to me to be entirely just –

something like that would do it?’

Sopel: ‘Don’t’.

(2b) Jefferson transcript: John Humphrys and Jon Sopel, Jan 2018

01 Humph: Ah.h

02 (0.2)

03 Humph: >↑C’n ↑you ’ear me.<=Sopel.

04 (1.3)

05 Sopel: ↑.hHumphrys: I c’n £↓hear£ you.

06 Humph: Good.=slight change of um:: (.) slight

07 change of subject, .hhh um: u- a- first

08 question will be: um how much hhh

09 (0.4) of your salary

10 [#you are prepared t’hand over to]=

11 Sopel: [ £#.h.h .h .h .h .h .h#£ ]

12 Humph: =Carrie Gracie to keep her, .hhh um::

13 an’ th(h)en uh- [- y- a few- a £few£ ]=

14 Sopel: [ . hh heh hhe ]=

15 Humph: =[comments about (0.2) your- your]=

16 Sopel: =[ hheh h hh .heh ]

17 Hump: =other [ c o l l]eagues: you know: l- um=

18 Sopel: [.HHH]

19 Humph: =like our: o- [our-] [our Middle East

20 Sopel: [.HH ] [Yeh,

21 Humph: =[editor: and .hh the other men who are=

22 Sopel: [(yeh.)

23 Humph: =earning too much.=d’you know ↑just- .hh=

24 Sopel: =I mean ↑obviously- (0.2) ob- obviously

25 if we’re talking about- the scope for

26 the greatest redistribution,

27 Humph: Mmm.

28 Sopel: .hh I’ll have t’come back an’ say well-

29 (0.2) ↑ye:s mister Humphrys.=[but

30 Humph: [An’ ↑I am:

31 an’ I- I- I’d (.) could save you the

32 [trouble because I could volunteer]=

33 Sopel: [ £#.h.h .h .h .h .h .h#£ ]=

34 Humph: =[.hh <I’ve um:> handed over alrea:dy=

35 Sopel: =[£.HHh£ heh

36 Humph: =.h more than you fuckin’ ↓earn

37 [.hh um:: I’m sti:ll left with more ]=

38 Sopel: [ £#.h .h .h .h .h .h ]=

39 Humph: =[than anybody e:lse.=an’ n- that]=

40 Sopel: =[ .h .h .h hh #£ .shih ]=

41 Humph: =that- seems to be- t’be entirely just.

42 .hhh uhh so(h)mmat like [that,

43 Sopel: [£uh heh heh£ heh

44 [ heh #eh eh .h eh eh ] £.HHHHH£

45 Humph: [Would do it, d’y’think,]

46 Sopel: £Don’t.[hh£

47 Humph?: [uhhh. Dear g(h)od.

The conversation analytic transcription not only provides more informa-

tion about what was said, how and precisely when, but also reveals the

extensive affiliative laughter particles produced by Sopel throughout

Humphrys’s talk. This was not included in Buzzfeed’s transcript. Sopel’s

affiliative laughter is audible at lines 11, 14, 16, 18, 20, 33, 35, 38, 40, 43

and 44, including immediately as Humphrys produces more than you

fuckin’ ↓earn at line 36.

A search of the LexisNexis database for reports of the leaked conversa-

tion, while formulated in some articles as ‘the Humphrys-Sopel exchange’

(Daily Mail, 12.1.2018), focused much more heavily on Humphrys rather

than Sopel. It was Humphrys’s name that appeared in the headlines and his

conduct that was the focus of the public outcry (e.g. ‘BBC stars at war as

several female presenters threaten to walk out after £650,000-a-year John

Humphrys is recorded JOKING about gender pay gap’ (Daily Mail,

12.1.2018)). Our point is not, of course, to lessen the offensiveness of

Humphrys’s conduct, but to show that Sopel might also have been held to

account. Media coverage taking into account the co-production of this epi-

sode rather than focusing on one individual might also have explained more

about the institutional culture and systemic inequalities in gender pay at the

BBC that Gracie cited in her resignation.

Together, these two examples begin to reveal the power of conversa-

tion analytic methods to expose far more of what happens in social inter-

action, and to specify and explicate matters of practical, political and

ethical accountability. While, of course, neither of our technical tran-

scripts ‘contain everything’ (the naïve realism arguments mentioned

above), the first points to the specific sources of trouble and highlights the

form, structure and content of a politician’s practices of dissembling in a

political interview about race. The second straightforwardly demonstrates

the collusion of a second party in an offensive ‘hot mic’ moment for which

only one party was held accountable. Similarly, conversation analysts have

systematically shown the serious problems of relying on standard ortho-

graphic records of spoken interaction in high-stakes contexts such as

legally regulated evidence- and testimony-gathering interactions. In a

recent overview, Richardson et al. (2022) make the point that, while there

are many transcription developments in conversation analytic and related

disciplines, these ‘have not (yet) translated over to improvements in tran-

scription practices in institutional contexts’ and ‘there is little recognition

in legal contexts of the long realized basic principle of the non-equivalence

of spoken and written text’ (2022: 1–2).

The Power to Reveal Variation in Standards

Conversation analysts have regularly exposed the gap between written

instructions, diagnostic instruments, guidance, regulations, the letter of

the law – and what the actual interactions that are apparently accountable

to these texts look like in encounters (e.g. Houtkoop-Steenstra, 2000;

Lavin & Maynard, 2001). There is a tacit assumption that in settings where

equity is particularly important such as police interviews or healthcare

interactions – where encounters are routinely recorded as part of institu-

tional or organizational practice for the public record or for (explicitly

inspectable) ‘training and quality purposes’ – that standardized interac-

tions will reflect the paperwork trails they generate. However, myriad con-

versation analytic studies have shown how actual encounters deviate from

guidance or instruction, and from one encounter to the next, even in exper-

imental or survey research which ostensibly generates standardized ‘hard’

data compared to the ‘soft’ accounts of qualitative research, a presumption

that Stokoe, Antaki et al. (2021) refer to as ‘the softness of hard data’.

Researchers have identified, for example, ‘radical departures’ from ‘stan-

dardized experimental procedures’ in research laboratories (Gibson, 2011;

Hollander, 2015), the impact of experimenters’ talk on ‘the trajectory of

the experiment’ (Wooffitt, 2007), and even how the spoken (re)formulation

of standardized (written) questions for gaining research consent shapes

participants’ responses (Speer & Stokoe, 2014). Without recordings and

transcripts, however, we would not know about how the details of interac-

tion – in-breaths, pauses, gaps and delays, self-repair – tacitly convey inter-

actional trouble, preference, affiliation, alignment and so on.

These gaps between guidance and practice can have dramatic and

morally worrisome implications for research and procedures with a

direct bearing on social justice. For example, in interviews between

police officers and suspects – a high-stakes encounter for those involved –

one might imagine that police officers follow the letter of the law when it

comes to asking questions, reminding suspects of their rights and so on.

However, police interviewers routinely deviate from the mandated word-

ing in the Police and Criminal Evidence Act (1984) in important ways,

such as how they cite suspects’ right to remain silent (Stokoe et al., 2016),

and how police reformulations of suspect testimony comprise evidence

for one of the ‘hardest’ data categories around: whether they broke the

law (Stokoe & Edwards, 2008). These variations in practice are, of

course, subject to bias, and may further entrench existing inequalities in

law enforcement.

For example, Walker et al.’s (2021) analysis of 585 case files of reported

rape suggested that there are intersecting inequalities in the trajectory of

cases reported to police for survivors minoritized by their sexuality,

gender, mental health status, race or ethnic background. However, studies

of police case files alone may not show how such inequalities are generated

in practice. Using conversation analysis, Richardson et al. (2018) analyzed

evidence-gathering interviews between police officers and ‘vulnerable’

adult or child victims of sexual assault and rape. In particular, they asked

to what extent police officers followed formal guidance: Achieving Best

Evidence in Criminal Proceedings: Guidance for Vulnerable and

Intimidated Witnesses (UK Ministry of Justice, 2011), which was pub-

lished as a special measure to enable the best conditions for these groups

of witnesses to provide evidence. One instruction in these guidelines is for

police officers to establish that alleged victims understand what it means

to report their experiences honestly. As the document puts it, police offi-

cers are instructed to ascertain whether victims understand the difference

between ‘truth and lies’. This is what the guidelines say:

Certain vulnerable adult witnesses

Para 3.22 In cases where discussion of truth and lies is appropriate, it

is important to demonstrate that the witness understands the difference

between the two. The witness could be asked to give examples of truth

and lies. If this is not possible, the interviewer can ask some questions

about this difference. If such questions are asked, they should follow the

guidance set out elsewhere on styles of questioning, and focus on the

intention to deceive rather than mere mistakes. After such questions, it

is appropriate to conclude with a statement like: ‘Please tell me all you

can remember about what happened. Don’t make anything up or leave

anything out. It is very important to tell the truth’.(UK Ministry of Justice, 2011: 73)

The guidelines also state that questions about truth telling should not

happen ‘later in the interview because this might run the risk of the [wit-

ness] concluding that the interviewer had not believed what they had said

up to that point’ (UK Ministry of Justice, 2011: 72). Yet conversation

analysis revealed that police officers deviated from the guidance inappro-

priately in several ways – not only by eliciting confirmations rather than

demonstrations of understanding, but also by eliciting multiple demon-

strations and confirmations of understanding, and by reintroducing ‘truth

and lies’ conversations at potentially sensitive points in the interview. The

latter two cases do exactly what the guidelines warn against – implying

dishonesty on the part of the victim, creating opportunities for variations

in practice to generate aggregate forms of discrimination and inequality.

As in policing, there is also evidence of structural inequalities in

healthcare outcomes that could benefit from more explanatory interac-

tional accounts. The UK government’s Marmot Review (2020) shows

widening inequalities between different regions of the UK, where intersec-

tions between ethnicity and socioeconomic background produce particu-

larly poor health outcomes, although the lack of adequate data and

analysis presents a continuing barrier to more informative explanation

(Marmot, 2020). In healthcare provision, it is probably also assumed that

diagnostic instruments and surveys of the kind delivered by clinicians will

be administered as instructed, in a standardized way. However, Antaki

and Rapley (1999) analyzed recordings of how a ‘Quality of Life’ ques-

tionnaire was administered by a psychologist to service-users. The ques-

tions on the sheet had three response options. The instructions for ‘reading

the items’ were for the psychologist to ‘pay close attention to the exact

wording’ (emphasis added). But, in the empirical reality of encounters, the

psychologists often reformulated those three options into one ‘yes/no’

question with a positive tilt, which led to service-users responding in par-

ticular ways that had consequences for the type and level of support they

subsequently received. Once again, these conversation analytic approaches

to the detail of talk and social interaction have the power to reveal varia-

tions in standards between guidance, procedures and practice, and to

highlight opportunities for bias that might explain otherwise opaque

inequalities in health outcomes.

This section has exposed three problems that can lead to injustices

and inequalities, and three ways in which CA has the power to expose and

explain these issues in actionable ways. The first problem is the tacit

assumption that, even in mundane settings, people can ‘just read out’ the

words written in a document (including scripts, survey questions, experi-

mental instructions) without succumbing to the interactional imperative.

In fact, apparently ‘just reading out’ is an interactional accomplishment

itself (Stokoe & Edwards, 2007). The second problem is that we do not

know how, in the heat of a sequence of talk, people turn guidance into

spoken talk. For example, what does ‘The witness could be asked to give

examples of truth and lies’ look like in actual police interviews? Finally,

even though so much of institutional life is now audio- or video-recorded,

and thus inspectable, egregious breaches of standards, guidance and even

the letter of the law often remain hidden in verbatim transcripts. CA has

the power to reveal the inadequacy of standardization and written guid-

ance for securing the assumed validity and equity of institutional

encounters.

The Power to Assess Communication

The assessment of communication styles and practices is also a

common site for bias and discrimination that conversation analytic

research has the power to expose. Communication guidance, training

manuals and role-play assessment methods can easily reproduce harmful

stereotypes because they often draw on assumptions rather than an evi-

dence base about what constitutes good communication practice. The

outcomes of communication assessments can be highly consequential for

those being evaluated – will they be employed, promoted or fired, depend-

ing on how they perform against a list of assessment criteria? This begs

questions about what should appear on a list of ‘communication skills’.

Conversation analysis has the power to challenge and improve the inter-

actional integrity and real-world usefulness of assessment practice.

One of the most common environments for assessing communication

is in simulated encounters or role-play, in which simulated interlocutors

(including actors) aim to replicate communicative scenarios and those

who are being assessed attempt to meet the assessment criteria for effec-

tive interviewing, medical history-taking and so on. Yet, conversation

analytic research shows that what happens in role-play does not corre-

spond to the institutional encounter it is designed to mimic (Stokoe,

2013a). This means that assessing someone’s communication skill based

on what they do in role-play is potentially problematic.

For instance, Stokoe (2013a) examined how police interviews with

actual suspects compared to interviews with actors playing the parts of

suspects in training and assessment sessions. The opening of such inter-

views is legally prescribed by the Police and Criminal Evidence Act (PACE,

1984). So, in real and simulated interviews, the interviewers’ first task is

to turn the written law into spoken turns at talk with the suspect,

including legal instructions to ‘tell the suspect about the recording pro-

cess’, ‘give their name and rank and that of any other interviewer present’

and ‘remind the suspect of their entitlement to free legal advice’. These

written directives were formulated differently depending on whether the

interview was an assessment with an actor-suspect, or part of the criminal

justice process with a real suspect.

One difference, which may seem inconsequential, was that, in real

interviews, police were likely to turn the instruction to ‘give their name

and rank and that of any other interviewer present’ into a statement like,

‘I’m PC 3456 Smith based at Anytown police station’, whereas in assess-

ment role-plays, they more typically said, ‘Um my name is PC 3456 Smith,

please feel free to call me Bob’. The latter component of the role-play

seems designed to enact other guidance to ‘build rapport’ and lessen the

formality of institutional identifications – while at the same time meet the

constraints of formal identifications for the benefit of the tape (Stokoe,

2009).

Furthermore, when more than one interviewer was present, the two

officers were more likely to introduce themselves across a compound turn

construction unit or TCU (Lerner, 1991), with the first interviewer adding

‘Also present is my colleague...’ after ‘I’m PC ...’. The second interviewer

then collaboratively completed the overall action. Stokoe also found that,

in real interviews, officers do not typically supply their first name nor issue

an invitation to the suspect regarding how to refer to them. Compound

TCUs were largely absent in assessment role-plays. That is, each officer

took their own full turns to introduce themselves, presumably to ensure

that the ‘assessable’ communication practices were clearly visible and

attributable just to them by the observing assessor. In real interviews, iron-

ically, officers work as a team and display this teamwork interactionally.

Why do officers state ‘My name is X’ when they know they are being

evaluated but use ‘I am’ elsewhere – and does it matter? Atkins (2018)

found the same alternative formats in real versus simulated doctor–patient

conversation. In real encounters, general practitioners introduced them-

selves saying, ‘I’m Dr X’. However, when being assessed on their consult-

ing skills, they were more likely to say, ‘My name is Dr X’, to pass their

‘objective structured clinical examination’ (OSCE). Furthermore, if a can-

didate used the ‘I’m Dr X’ format in the simulation, they were likely to be

marked negatively for doing so. As Atkins et al. (2016) point out, ‘the

linguistic problems and differences that arise from interacting in artificial

settings are of considerable importance in assessment, where we must be

sure that the exam construct adequately embodies the skills expected for

real-life practice’.

It appears, then, that different occupations are trained to make introduc-

tions in the same way by saying, ‘My name is X’. Doctors and police officers

use this format when being assessed. But they use ‘I’m X’ in real encounters –

with no evidence that this is less effective for the comprehensibility of the

action or ‘poor communication’. It is therefore difficult to know what under-

pins the reification into guidance and evaluation of one practice over another.

Training, guidance and scripts for, and assessment of, communicative

encounters are often apparently built from people’s attempts to make explicit

their tacit knowledge and build that into guidance and recommendations.

This leads to failures in identifying the right ‘assessables’.

The point is not that stating, ‘My name is’ rather than a categorial ‘I’m

X’, or some other way of referring to oneself is intrinsically flawed, or not

found endogenously in other non-simulated data. Rather, we argue that pre-

scribing ways of talking falls foul not just of the principle of recipient design,

but also the evidence of how people actually talk in any given situation.

This brings us to another example of communication assessment

based on simulation. We examined what happens when ‘mystery shop-

pers’ telephone veterinary practices with the goal of evaluating the provi-

sion of good service – as achieved through conversations with receptionists

(Stokoe et al., 2020). The point of mystery shopping as an assessment tool

is to evaluate the process of service rather than post-hoc scoring on a

scale. Unlike the police and general practitioner (GP) contexts, mystery

shopping relies on those being assessed not knowing that they are talking

to, in our case, a real pet owner or a simulated client.

Our research on mystery shopping pulls together several themes. First,

when used in healthcare, mystery shopping studies claim to reveal bad prac-

tice, such as that pharmacists deviate from regulatory guidance when selling

medicines over the counter. But what guidance are mystery shoppers using

to ask for medicines in the first place? The methodology claims to ensure its

validity and parity of assessment by training shoppers with scripts, but how

the written script is (re)formulated in encounters is unknown. Second, the

research showed that mystery shopper calls differed systematically from real

calls. Real calls were longer; the reason for the call was different (real callers

asked for appointments and treatment; mystery shoppers asked about the

cost of services); and the outcome of the call was different (real callers made

appointments, requiring the receptionist to navigate the appointment system

and meet their needs; mystery shoppers did not).

While a mystery shopper is hard to identify on a one-off occasion,

larger datasets revealed stark differences. The implications for equity and

workplace assessments are also stark: when mystery shoppers report back

to organizations on the communicative competencies of their staff, their

reports cannot reflect what staff typically do on the phone and the contin-

gencies they routinely manage. Scripts and reports are based on normative

notions of ‘good communication’ or ‘effective service’. Yet staff will be

promoted, given bonuses, be mandated for more training and so on based

on a deeply flawed assessment methodology (e.g. Miller & Turner, 2005).

We also found that real callers included components that mystery

shoppers did not, such as accounts for their requests, or the name, breed

and weight of their pet. Only real callers referred to their pet by name, and

mystery shoppers were more likely to initiate repair on, delay or otherwise

‘stumble’ in responding to queries about the pet breed and weight. These

small details have implications for the development of conversational AI

systems such as Apple’s Siri, Amazon’s Alexa and Google’s Duplex

(Leviathan & Matias, 2018) that attempt to emulate naturalistic spoken

interaction. There is an emerging body of conversation analytic research

that shows how constrained the ‘conversations’ are in such interactions

(e.g. Albert et al., 2019). Our findings suggest that, worse still, even

humans who are simulating other humans do not produce social actions

in the same way when the stakes are different. So, the developers of arti-

ficial agents face an even tougher challenge.

Our final example in this section stays with the practice of assessing ser-

vice provision on the telephone. The use of natural language processing

(NLP) and AI to build tools for ‘speech analytics’ and ‘sentiment analysis’ has

grown exponentially, with many software-as-a-service (SaaS) products on the

market to assess service provision at scale. To analyze, say, contact center

calls, audio-recordings are uploaded, run through speech-to-text software,

and text outputs are assigned codes and scores of various kinds (e.g. positive,

neutral, negative; confusion, happiness, anger). This labeling can be applied

at different levels of granularity, from a turn-by-turn code to a score for an

overall encounter. However, they do not typically capture the micro-analytic

details of interaction or concepts such as ‘adjacency’ – and certainly not con-

versational AI ‘moonshots’ like compound TCUs (Stokoe et al., 2021).

In pilot research (Parslow et al., 2021), we were invited to work with

a start-up to test out their speech analytics platform. We uploaded audio-

recorded telephone calls from institutional settings including general prac-

tice. These calls had been transcribed using Jeffersonian conventions and

were already analyzed as part of earlier projects (e.g. Sikveland et al.,

2016). We then made some basic observations about the accuracy of the

platform’s speech-to-text transcription, speaker ‘diarization’ (partitioning

and attributing the audio to separate interlocutors), and the kinds of codes

attributed to parts of the interaction that comprised an overall service

provision score.

As an example of the difference between our transcript and the soft-

ware system output, consider Extract 3. Extract 3a is the Jefferson tran-

script of a call between a caller (C) and their general practice reception

(R); Extract 3b is the SaaS output of the same recording. Both are anony-

mized. The caller and receptionist are discussing the caller’s son.

(3a) Jefferson/GP call

01 R: Ri:ght.

02 (1.0)

03 R: Uh::m hh do you feel it’s ur:gent.=I’ve got

04 to ask you this because I can’t make the

05 decision.

06 (1.3)

07 R: [You see.

08 C: [If you’ve not got any appointments I can

09 take him to the walk-in centre.

10 (0.8)

11 R: Are you sure:¿

12 (0.4)

13 C: Y:eah,

14 (0.9)

15 R: I c- >I mean< there is that,

16 (0.4)

17 R: Uh:: or I can put him on a cancellation list,

(3b) SaaS / GP call

The contrast between the two transcripts is stark. Not only is Extract 3b

full of errors, including which speaker is producing which turn (e.g. it is

the receptionist not the caller who asks, are you sure), but one mis-

transcription (the caller’s solution to take her son to the walk-in centre

becomes contribute to the walking dead and is, perhaps unsurprisingly,

tagged as ‘negative’).

We found many basic errors in transcription, with knock-on conse-

quences for the system’s inability to correctly attribute turns, interpret

gaps and pauses, handle repair or overlapping talk, and – perhaps most

importantly – identify action and sequence. For example, the ‘reason for

the call’ (e.g. ‘I’ve got a bad eye’; ‘my window frame is broken’) was fre-

quently tagged as ‘negative’, despite this action in a service encounter

having little relevance to service provision. Furthermore, items like ‘thank

you’ were typically tagged as ‘satisfaction’ or ‘positive’, regardless of who

produced the turn and its sequential position. Extract 4 is an example

from the GP-patient call:

(4) SaaS / GP call

Here, both the caller’s and call-taker’s ‘thank yous’ – despite being

composed differently (thank you; thanks very so much) and formulated

twice by the caller (much obliged) – receive the same tags. The system

makes no distinctions between, say, a pre-closing ‘thank you’ and appre-

ciation, even as the caller does both things. Since the call-taker’s ok, thank

you is also tagged as ‘satisfaction’ and ‘positive’, we can see how scores

may be inflated or depressed depending on diarization errors. And, of

course, conversation analytic research shows that ‘thank you’ is an unreli-

able indicator of customer satisfaction and, depending on its sequential

location and who says it first in pre-closings, can actually convey dissatis-

faction (Sikveland & Stokoe, 2017).

Like other examples in this chapter, we see how a basic lack of atten-

tion to what some regard as ‘excessive’ and ‘surplus’ details, as well as a

deeply flawed set of assumptions about real conversation and communica-

tion, embed themselves into a wide variety of research, assessment and

organizational practices. Communications training and assessments are

often vital for addressing obstacles to career progression for disadvan-

taged groups in critical professions such as policing (see e.g. Mayor of

London, 2020) where there are persistent unequal outcomes in recruit-

ment, retention and promotion, with significant implications for social

justice.

These research findings discussed above, about the differences

between simulated versus actual professional settings, have been dissemi-

nated in public lectures and science communication environments, leading

practitioners to reach out to us to discuss what happens in communication

assessment in their organizational contexts. For example, one medic

wrote to discuss the implications for inequality in the aforementioned

simulation-based OSCE examinations. The medic had observed – and,

indeed, research has shown (e.g. Elewa-Ikpakwu & Ayoola, 2020; Wass

et al., 2003) – that minoritized medical students perform less well on such

(simulated) communication assessments. What might this mean if the

assessment situation does not authentically reproduce actual practice and

assessment criteria are based on ‘examiners’ assumptions about what is

good communication’ (Wass et al., 2003: 800)? Conversation analysis has

the power to shine a light on these common flaws in communication

assessment practices and the unfair consequences for those assessed.

The Power of Evidence-Based Communication Training

The final example of the power of conversation analysis is connected

to the previous section on communication assessment and focuses on the

related issue of communication training. In 2008, Stokoe (2011) began to

develop Conversation Analytic Role-play Method, or CARM, which is a

communication training method based on conversation analytic research.

With Derek Edwards, Stokoe had conducted analyses of initial inquiry (or

‘intake’) calls into community mediation (and related) services focusing

on how identity matters – cashed out in categorial terms, including gender,

race, ethnicity, age and class – impacted the trajectory of the dispute

between parties (e.g. Stokoe, 2009). These data were collected as part of

a project on community and neighbor conflict and dispute.

Towards the end of the project, Stokoe began to focus particularly on

whether callers became clients of community mediation organizations by

the end of their encounter with a mediator. Such intake calls are treated

as separate from and outside of an actual mediation and had previously

received no attention from either researchers or the mediation industry,

both in the UK and USA. But because community dispute resolution ser-

vices secure funding partly based on recognized need – on the size of their

client base – it became important for mediators to learn what might

improve their conversion rates, from caller to client, as Stokoe’s research

demonstrated (e.g. Sikveland & Stokoe, 2016; Stokoe, 2013b).

Starting with findings about mediation, Stokoe designed a communi-

cation training method underpinned by selected aspects of conversation

analytic research, often co-produced with organizations who funded the

research-to-training cycle. This approach was in stark contrast to tradi-

tional forms of training based on simulation or role-play, discussed ear-

lier, since it took research findings about effective practice as its basis.

CARM involves identifying examples of more and less effective practices;

for example, for describing and offering mediation to skeptical callers.

Audio and video extracts from recordings are anonymized and played

synchronously with transcripts to enable participants to ‘live through’ real

encounters without knowing what would happen next. Workshop partici-

pants discuss what they might do next to handle the situation. For exam-

ple, if party A makes a particular sort of comment, how might party B

respond most appropriately? Finally, party B’s actual response is revealed

and discussed, and the workshop moves on.

The workshop with the largest reach, delivered to organizations across

the UK and USA, started out as a response to a request from community

mediators. Stokoe was awarded funding to develop CARM and offer two

workshops to mediation organizations, including a ‘bespoke’ session

designed to meet whatever need they had. One asked for a session on hate

speech and how to identify and respond to it. To meet this need – and

move away from role-play or experience/scenario-based training – Stokoe

developed a workshop based on different strands of her research in which

age, gender, race and ethnicity categories became live issues in the inquiry

calls, as well as accusations and denials of multiple forms of prejudice (e.g.

Stokoe, 2003; Stokoe & Edwards, 2007).

To illustrate the CARM approach, Extract 5 comes from a mediation

session (not an initial inquiry) in which three parties in a dispute are co-

present (Macy, Gary, Henry), with two mediators (M1, M2). This extract

is over halfway through a 90-minute session, and the ostensible cause of

the dispute is a shared garden space which has been, according to Gary

and Henry, ‘colonized’ by Macy’s plants. Here, Macy (MA) is explaining

what she sees as the root of the problem.

(5a) KM-3

01 MA: I remember (0.3) uh having a word with Gary

02 about something he accused me of being uh-

03 (0.5) like a Barbra Streisand.

04 Which meant I must be aggressive (0.5) Jewish,

05 (0.3) which I am neither,

In the CARM workshop, the audio recording and transcript stop here.

Participants discuss questions about what kind of accusation this is (and

their evidence for it) and what – if they were the mediator – they may say

in response. This case is interesting not least because of the way Macy

turns ‘Barbra Streisand’ into a category (note the indefinite article a, at

line 02) and unpacks what she takes to be the category-bound predicates

that turn the categorization into a possible instance of anti-Semitism.

Most workshop participants – the community mediators – say that

their response would be to explore the categorization further, including

asking Gary for his account. But this is what happens next.

(5b) KM-3

06 (0.4)

07 MA: Um::

08 (0.4)

09 M2: I- I think- [I’m not- I’m not sure that=

10 M1: [Is this helpful?

11 M2: =this is helpful

The mediators, in overlap and in alignment, do not expand upon Macy’s

accusation but instead negatively evaluate it as something that is not help-

ful to address in mediation.

The participants in the CARM workshop use such cases as the basis

for further discussion, including about the nature of mediation itself. Like

findings from a survey of mediators reported in Rendon (2007: np), some

mediators ‘believed that prejudice and discrimination issues are irrelevant

distractions from mediation’s goal of case resolution, and therefore should

be ignored or put aside’, while others argued that ‘the mediator should

meet the issues “head on” by bringing them to the table’. What became

possible in CARM sessions, however, was that mediators – sometimes

from the same organization – were able to discuss actual cases, and see

actual mediator responses, and discover that not everyone shares an

8understanding of what counts as an ‘-ism’ nor even what the purpose of

mediation is.

The ‘-isms’ workshop was initially delivered to mediation centers

between 2011 and 2013. It has since evolved, including into a co-produced

document by Loughborough University’s Discourse and Rhetoric Group

and its ‘Maia’ women’s network, led by Stokoe. The one-page guidance,

‘How to say when it’s not okay: Tackling prejudicial -isms in workplace

conversation’, is designed as a standalone how-to, as well as providing a

quick-reference to accompany the regularly refreshed CARM workshop.

The guidance describes eight response types to an ‘-ism’ based on research

(e.g. Pino, 2017; Robles, 2015; Stokoe, 2015; Stokoe & Edwards, 2007;

Whitehead, 2018). It also foregrounds the fact that what people say they

would do to ‘be an ally’ is not the same as what typically happens in inter-

action (see also Kitzinger & Frith, 1999).

Discussion and Conclusion

Since the earliest days of EMCA, a common criticism has been its

ostensible lack of what Coser (1975) referred to as ‘substance’ in his key-

note address to the American Sociological Association. This rather dis-

missive attitude and – for us – unfounded claim has persisted across the

decades. Indeed, the title of this chapter is a direct quote from a reviewer

who characterized conversation analysis as ‘a method in search of a prob-

lem’. However, its power to shine empirical light on the pervasive interac-

tional production of social problems of injustice and inequity makes this

a description to embrace.

The bringing together of ‘critical’ and ‘conversation analysis’ is not

new (e.g. Blain, 2020) and, for some in EMCA, the notion of adding ‘criti-

cal’ to ‘conversation analysis’ will be conceptually flawed, tautological or

even oxymoronic. Indeed, although the most-cited debate about conversa-

tion analysis versus critical analysis was undoubtedly between Schegloff

(1997), Wetherell (1998) and Billig (1999), there have been other long-

running debates between ethnomethodologists and conversation analysts

themselves about, say, the project of ‘feminist’ conversation analysis (e.g.

Kitzinger, 2000, 2008; Speer, 1999; Stokoe, 2000, 2008; Tennent &

Weatherall, 2021; Whelan, 2012; Wowk, 2007).

The aim of this chapter has been to provide evidence that and how

conversation analytic methods, research designs and findings provide the

strongest possible evidence for understanding and exposing power, ethics

and injustice in practice. CA transcription has the power to describe and

document, in detail, the co-production of systemic forms of bias, and to

empower public debates to hold people and institutions to account. It has

the power to show the variation in ostensible ‘standards’ and highlight

how certain deviations from ‘standard practice’ in communication may

lead to unjust and iniquitous outcomes. It has the power to highlight the

ways in which communication assessment and simulation training often

rely on biased and simplistic assumptions, and can provide evidence as the

basis for more accurate, fair, social justice-oriented communication train-

ing. Despite Billig’s (2019) criticisms of conversation analysis, we agree

with his call for ‘more examples’ and ‘less theory’. We underline the

importance of examining everyday life ‘as it happens’ (Boden, 1990). The

devil, as we have seen, is in the detail. Finally, we agree with Edwards’s

(p.c.) proposal that, rather than worrying about our accountability to the

rhetoric of ‘big questions’, we should argue that those who make those

critiques should be accountable to our data which shows the workings of

real-life inequalities, the myriad practices of exclusion, and the otherwise

hidden reality of the damage turns at talk can do.

Notes

(1) See for instance: https://twitter.com/ridgeonsunday/status/1269537028040990722;

https://www.huffingtonpost.co.uk/entry/black-cabinet-matt-hancock-protests_uk_

5edc9ffdc5b6aedebbc6a27a

(2) These terms have also been criticized for their use in Diversity, Equity and Inclusion

(DEI) initiatives as a euphemism ‘used by institutions to signal their recognition of

“diversity” whilst simultaneously cloaking and reproducing the racism of the institu-

tion’ (Skeggs, 2021: 125).

References

Albert, S., Housley, W. and Stokoe, E. (2019) In case of emergency, order pizza: An urgent

case of action formation and recognition. CUI 2019: Proceedings of the 1st

International Conference on Conversational User Interfaces, Dublin, Ireland, 22–23

August 2019CUI ‘19. https://doi.org/10.1145/3342775.3342800

Antaki, C. and Rapley, M. (1996) Questions and answers to psychological assessment sched-

ules: Hidden troubles in ‘quality of life’ interviews. Journal of Intellectual Disability

Research 40 (5), 421–437.

Ashmore, M. and Reed, D. (2000) Innocence and nostalgia in conversation

analysis: The dynamic relations of tape and transcript. Forum: Qualitative

Social Research 1 (3). http://www.qualitative-research.net/index.php/fqs/article/

view/1020/2199.

Aspinall, P.J. (2021) BAME (black, Asian and minority ethnic): The ‘new normal’ in col-

lective terminology. Journal of Epidemiol Community Health 75.

Atkins, S. (2018) Assessing health professionals’ communication through role-play: An

interactional analysis of simulated versus actual general practice consulta-

tions. Discourse Studies 21 (2), 109–134. https://doi.org/10.1177/1461445618802659

218 Part 3: A Final Argument

Waring & Tadic.indb 218 27-02-2024 15:38:18Atkins, S., Roberts, C., Hawthorne, K. et al. (2016) Simulated consultations: A sociolinguistic

perspective. BMC Medical Education 16. https://doi.org/10.1186/s12909-016-0535-2

Atkinson, P. (1988) Ethnomethodology: A critical review. Annual Review of Sociology

14, 441–465.

Bankole, M. (2020) Black British citizens want more than complacency from this govern-

ment. The Conversation. Available online: http://theconversation.com/

black-british-citizens-want-more-than-complacency-from-this-government-140297

Billig, M. (1999) Whose terms? Whose ordinariness? Rhetoric and ideology in

Conversation Analysis. Discourse & Society 10 (4), 543–558.

Billig, M. (2019) More Examples, Less Theory: Historical Studies of Writing Psychology.

Cambridge University Press.

Blain, H. (2020) Discourses in action: operations of race, sexuality and gender in chinese

talk-in-interaction. Unpublished PhD thesis, School of Languages and Linguistics

University of Melbourne.

Boden, D. (1990) The world as it happens: Ethnomethodology and conversation analysis.

In G. Ritzer (ed.) Frontiers of Social Theory: The New Synthesis (pp. 185–213).

Columbia University Press.

Bunglawala, Z. (2019) Please, don’t call me BAME or BME! – Civil Service. https://

civilservice.blog.gov.uk/2019/07/08/please-dont-call-me-bame-or-bme/

Charmaz, K. and Thornberg, R. (2021) The pursuit of quality in grounded theory.

Qualitative Research in Psychology 18 (3), 305–327. https://doi.org/10.1080/14780

887.2020.1780357

Coser, L. (1975) ‘Two methods in search of a substance’. Presidential Address, annual

meetings of the ASA, 1974. American Sociological Review 40 (6), 691–700. https://

doi.org/10.2307/2094174

Di Stefano, M. (2018) Here’s the audio recording of John Humphrys and Jon Sopel’s ‘jokey

exchange’ about equal pay. BuzzFeed.News, 13 January. https://www.buzzfeed.com/

markdistefano/humphrys-sopel-audio-recording

Edwards, D. and Potter, J. (2005) Discursive psychology, mental states and descriptions.

In H. Molder and J. Potter (eds) Conversation and Cognition (pp. 241–259).

Cambridge University Press.

Elewa-Ikpakwu, C. and Ayoola, G. (2021) Response to: Double jeopardy: Black and

female in medicine. The Clinical Teacher 18 (3), 311–311.

Frosh, S. (1999) What is outside discourse? Psychoanalytic Studies 1 (4), 381–390.

Gibson, S. (2011) Milgram’s obedience experiments: A rhetorical analysis. British Journal

of Social Psychology 52 (2), 290–309.

Glenn, P.J. (2003) Laughter in Interaction. Cambridge University Press.

Harré, R. (1993) Social Being (2nd edn). Wiley-Blackwell.

Hepburn, A. and Bolden, G.B. (2018) Transcribing for Social Research. Sage.

Hester, S. and Francis, D. (1990) Ethnomethodology, conversation analysis, and ‘institu-

tional talk’. Text 20 (3), 391–413.

Hollander, M.M. (2015) The repertoire of resistance: Non-compliance with directives

in Milgram’s ‘obedience’ experiments. British Journal of Social Psychology

54, 425–444.

Houtkoop-Steenstra, H. (2000) Interaction and the Standardized Survey Interview: The

Living Questionnaire. Cambridge University Press

Jefferson, G. (2004) Glossary of transcript symbols with an introduction. In G.H. Lerner

(ed.) Conversation Analysis: Studies from the First Generation (pp. 13–31). John

Benjamins.

‘Just a Method in Search of a Problem?’ The Power of Conversation Analysis 219

Waring & Tadic.indb 219 27-02-2024 15:38:18Johnston, O., Wildy, H. and Shand, J. (2021) Projecting student voice by constructing

grounded theory. Australian Educational Researcher 48, 543–564.

Kitzinger, C. (2000) Doing feminist conversation analysis. Feminism & Psychology 10,

163–193

Kitzinger, C. (2008) Developing feminist conversation analysis: A response to Wowk.

Human Studies 31 (2), 179–208.

Kitzinger, C. and Frith, H. (1999) Just say no? The use of conversation analysis in develop-

ing a feminist perspective on sexual refusal. Discourse & Society 10 (3), 293–

316. https://doi.org/10.1177/0957926599010003002

Lavin, D. and Maynard, D.W. (2001) Standardization vs. rapport: Respondent laughter

and interviewer reaction during telephone surveys. American Sociological Review

66 (3), 453–479. https://www.jstor.org/stable/3088888

Lerner, G.H. (1991) On the syntax of sentences-in-progress. Language in Society 20 (3),

441–458. https://doi.org/10.1017/S0047404500016572

Leviathan, Y. and Matias, Y. (2018) Google duplex: An AI system for accomplishing real-

world tasks over the phone. Google AI Blog. https://ai.googleblog.com/2018/05/

duplex-ai-system-for-natural-conversation.html

Mayor of London (2020) Action plan—transparency, accountability and trust in policing.

Greater London Authority. https://www.london.gov.uk /publications/

action-plan-transparency-accountability-and-trust-policing

Miller, G. and Turner, S.H.R. (2005) Applying the mystery shopping technique: The case

of Lunn Poly. Tourism Research Methods 119.

Ochs, E. (1979) Transcription as theory. In E. Ochs and B. Schieffelin (eds) Developmental

Pragmatics (pp. 43–72). Academic Press.

Oliver, D.G., Serovich, J.M. and Mason, T.L. (2005) Constraints and opportunities with

interview transcription: Towards reflection in qualitative research. Social Forces

84 (2), 1273–1289. https://doi.org/10.1353/sof.2006.0023

Parker, I. (2005) Qualitative Psychology: Introducing Radical Research. Open University

Press.

Parker, I. (2012) Discursive social psychology now. British Journal of Social Psychology

51 (3), 471–477.

Parslow, S., Stokoe, E. and Albert, S. (2021) Using conversation analysis to challenge the

notion of emotional tone in sentiment analysis. International Pragmatics

Conference.

Pino, M. (2017) I-challenges: Influencing others’ perspectives by mentioning personal experi-

ences in therapeutic-community group meetings. Social Psychology Quarterly 80 (3),

217–242

Potter, J. and Hepburn, A. (2005) Qualitative interviews in psychology: Problems and

possibilities. Qualitative Research in Psychology 2, 38–55.

Previtali, F., Nikander, P. and Ruusuvuori, J. (2023) Ageism in job interviews: Discreet

ways of building co-membership through age categorisation. Discourse Studies 25

(1), 25–50.

Rawls, A.W., Whitehead, K.A. and Duck, W. (eds) (2020) Black Lives Matter:

Ethnomethodological and Conversation Analytic Studies of Race and Systemic

Racism in Everyday Interaction. Routledge.

Rapley, T.J. (2001) The art (fulness) of open-ended interviewing: Some considerations on

analysing interviews. Qualitative Inquiry 1 (3), 303–323.

Rendon, J. (2007) Facing prejudice in mediation: What should the mediator do? Available

online: https://www.mediate.com/articles/rendon3.cfm

220 Part 3: A Final Argument

Waring & Tadic.indb 220 27-02-2024 15:38:18Richardson, E., Stokoe, E. and Antaki, C. (2018) Establishing intellectually impaired

victims’ understanding about ‘truth’ and ‘lies’: Police interview guidance and practice

in cases of sexual assault. Applied Linguistics. https://doi.org/10.1093/applin/amy023

Richardson, E., Haworth, K. and Deamer, F. (2022) For the record: Questioning transcrip-

tion processes in legal contexts. Applied Linguistics. https://doi.org/10.1093/applin/

amac005

Robles, J.S. (2015) Extreme case (re) formulation as a practice for making hearably racist

talk repairable. Journal of Language and Social Psychology 34 (4), 390–409.

Sarangi, S. (2018) Modes of en’gaze’ment and analytic accountability in discourse and

interaction studies. Language and Social Interaction Working Group, 8th

Meeting.

Schegloff, E.A. (1997) Whose text? Whose context? Discourse & Society 8 (2), 165–187.

https://doi.org/10.1177/0957926597008002002

Schegloff, E.A., Jefferson, G. and Sacks, H. (1977) The preference for self-correction in the

organization of repair in conversation. Language 53 (2), 361–382. https://doi.

org/10.2307/413107

Sharrock, W.W. and Anderson, R.J. (2017) Has ethnomethodology run its course? Available

online: https://www.sharrockandanderson.co.uk/wp-content/uploads/2019/10/Run-its-

Course-VII.pdf

Sikveland, R.O. and Stokoe, E. (2016) Dealing with resistance in initial intake and inquiry

calls to mediation: The power of ‘willing’. Conflict Resolution Quarterly 33 (3),

235–253. https://doi.org/10.1002/crq.21157

Sikveland, R.O. and Stokoe, E. (2017) Enquiry calls to GP surgeries in the United

Kingdom: Expressions of incomplete service and dissatisfaction in closing sequences.

Discourse Studies 19 (4), 441–459. https://doi.org/10.1177/1461445617706999

Sikveland, R.O., Stokoe, E. and Symonds, J. (2016) Patient burden during appointment-

making telephone calls to GP practices. Patient Education and Counselling 99 (8),

1310–1318. https://doi.org/10.1016/j.pec.2016.03.025

Singh, A. (2020) Matt Hancock names two asians when asked to identify black cabinet

members. HuffPost UK, 7 June. Available online: https://www.huffingtonpost.co.uk/

entry/black-cabinet-matt-hancock-protests_uk_5edc9ffdc5b6aedebbc6a27a

Skeggs, B. (2021) Necroeconomics: How necro legacies help us understand the value of

death and the protection of life during the COVID-19 pandemic. Historical Social

Research/Historische Sozialforschung 46 (4), 123–142.

Speer, S.A. (1999) Feminism and conversation analysis: An oxymoron? Feminism &

Psychology 9 (4), 471–478. https://doi.org/10.1177/0959353599009004013

Speer, S.A. and Stokoe, E. (2014) Ethics in action: Consent-gaining interactions and impli-

cations for research practice. British Journal of Social Psychology 53 (1), 54–73.

https://doi.org/10.1111/bjso.12009

Stokoe, E.H. (2000) Towards a conversation analytic approach to gender and discourse.

Feminism & Psychology 10 (4), 552–563. https://doi.org/10.1177/0959353500010004018.

Stokoe, E.H. (2003) Mothers, single women and sluts: Gender, morality and membership

categorization in neighbour disputes. Feminism & Psychology 13 (3), 317–344.

https://doi.org/10.1177/0959353503013003006Stokoe 2011

Stokoe, E. (2008) Categories and sequences: Formulating gender in talk-in-interaction. In

K. Harrington, L. Litosseliti, H. Saunston and J. Sunderland (eds) Gender and

Language Research Methodologies (pp. 139–157). Palgrave.

Stokoe, E. (2009) Doing actions with identity categories: Complaints and denials in neigh-

bour disputes. Text and Talk 29 (1), 75–97. https://doi.org/10.1515/TEXT.2009.004

‘Just a Method in Search of a Problem?’ The Power of Conversation Analysis 221

Waring & Tadic.indb 221 27-02-2024 15:38:18Stokoe, E. (2011) Simulated interaction and communication skills training: The

‘Conversation Analytic Role-play Method’. In C. Antaki (ed.) Applied Conversation

Analysis: Changing Institutional Practices (pp. 119–139). Palgrave Macmillan.

Stokoe, E. (2013a) The (in)authenticity of simulated talk: Comparing role-played and actual

conversation and the implications for communication training. Research on Language

and Social Interaction 46 (2), 1–21. https://doi.org/10.1080/08351813.2013.780341

Stokoe, E. (2013b) Overcoming barriers to mediation in intake calls to services: Research-

based strategies for mediators. Negotiation Journal 29 (3), 289–314. https://doi.

org/10.1111/nejo.12026

Stokoe, E. (2015) Identifying and responding to possible ‘-isms’ in institutional encoun-

ters: Alignment, impartiality and the implications for communication training.

Journal of Language and Social Psychology 34 (4), 427–445. https://doi.

org/10.1177/0261927X15586572

Stokoe, E. (2020) Psychological matters in institutional interaction: Insights and interven-

tions from discursive psychology and conversation analysis. Qualitative Psychology

7 (3), 331–347. https://doi.org/10.1037/qup0000162

Stokoe, E. and Edwards, D. (2007) `Black this, black that': Racial insults and reported

speech in neighbour complaints and police interrogations. Discourse & Society 18

(3), 337–372. https://doi.org/10.1177/0957926507075477

Stokoe, E. and Edwards, D. (2008) `Did you have permission to smash your neighbour’s

door?’ Silly questions and their answers in police-suspect interrogations. Discourse

Studies 10 (1), 89-111. https://doi.org/10.1177/1461445607085592

Stokoe, E., Hepburn, A. and Antaki, C. (2012) Beware the ‘Loughborough School’?

Interaction and the politics of intervention. British Journal of Social Psychology 51

(3), 48–496. https://doi.org/10.1111/j.2044-8309.2011.02088.x

Stokoe, E., Edwards, D. and Edwards, H. (2016) ‘No comment’ responses to questions in

police investigative interviews. In S. Ehrlich, D. Eades and J. Ainsworth (eds)

Discursive Constructions of Consent in the Legal Process (pp. 289–317). Oxford

University Press.

Stokoe, E., Sikveland, R.O., Hamann, M.G.T., Albert, S. and Housley, W. (2020) Can

humans simulate talking like other humans? Comparing mystery shoppers to real

customers in service inquiries. Discourse Studies 22 (1). https://doi.

org/10.1177/1461445619887537

Stokoe, E., Albert, S., Parslow, S. and Pearl, C. (2021) Conversation analysis and conversa-

tion design: Where the moonshots are. Medium, 3 June. https://elizabeth-stokoe.

medium.com/conversation-design-and-conversation-analysis-c2a2836cb042#:~:text=

Conversation%20analysis%20and%20conversation%20design%3A%20Where%20

the%20moonshots%20are,-Professor%20Elizabeth%20Stokoe&text=Conversation

%20has%20been%20described,different%20facets%20of%20life%20accomplished.

Stokoe, E., Antaki, C., Bracher, M., Chrisostomou, L., Henderson, E., Jones, G. and

Stewart, S. (2021) The softness of hard data. Medium. Available online: https://eliz-

abeth-stokoe.medium.com/the-softness-of-hard-data-475743d8a2f2

Tennent, E. and Weatherall, A. (2021) Feminist conversation analysis. In J. Angouri and

J. Baxter (eds) The Routledge Handbook of Language, Gender, and Sexuality (pp.

258–271). Routledge.

Thornberg, R., Halldin, K., Bolmsjö, N., Petersson, A. (2013) Victimising of school bul -

lying: A grounded theory. Research Papers in Education 28 (3), 309–29. https://doi.

org/10.1080/02671522.2011.641999

222 Part 3: A Final Argument

Waring & Tadic.indb 222 27-02-2024 15:38:18UK Government (2021) Writing about ethnicity. Available online: https://www.ethnicity-

facts-figures.service.gov.uk/style-guide/writing-about-ethnicity#bame-and-bme

Walker, G. (2012) Phonetics and prosody in conversation. In J. Sidnell and T. Stivers (eds.)

The Handbook of Conversation Analysis (pp. 455–474). John Wiley and Sons.

Walker, S.-J.L., Hester, M., McPhee, D., Patsios, D., Williams, A., Bates, L. and Rumney,

P. (2021) Rape, inequality and the criminal justice response in England: The impor-

tance of age and gender. Criminology & Criminal Justice 21 (3), 297–315. https://doi.

org/10.1177/1748895819863095

Wass, V., Roberts, C., Hoogenboom, R., Jones, R. and Van der Vleuten, C. (2003) Effect

of ethnicity on performance in a final objective structured clinical examination: qual-

itative and quantitative study. British Medical Journal 326 (7393), 800–803.

Wetherell, M. (1998) Positioning and interpretative repertoires: Conversation analysis and

post-structuralism in dialogue. Discourse & Society 9 (3), 431–456.

Whelan, P. (2012) Oxymoronic and sociologically monstrous? Feminist conversation

analysis. Qualitative Research in Psychology 9 (4), 279–291. https://doi.org/10.1080/14

780887.2011.634360

Whitehead, K.A. (2018) Managing the moral accountability of stereotyping. Journal of

Language and Social Psychology 37 (3), 288–309.

Wooffitt, R. (2007) Communication and laboratory performance in parapsychology

experiments: Demand characteristics and the social organization of interaction.

British Journal of Social Psychology 46 (3), 47 7– 498. https://doi.

org/10.1348/014466606X152667.

Wowk, M.T. (2007) Kitzinger’s feminist conversation analysis: Critical observations.

Human Studies 30 (2), 131–155.

Yu, B. and Sterponi, L. (2023) Toward neurodiversity: How conversation analysis can

contribute to a new approach to social communication assessment. Language, Speech,

and Hearing Services in Schools 54 (1), 27–41.

‘Just a Method in Search of a Problem?’ The Power of Conversation Analysis 223

Waring & Tadic.indb 223 27-02-2024 15:38:18