Discursive devices - Methods

Discursive Psychology: Theory, Method and Applications - Sally Wiggins 2017

Discursive devices

In Chapter 6, we worked through the process of DP analysis, from the first reading of the data to refining the analytical focus. We also first encountered the discursive devices: the core analytical tools of DP that enable us to examine the discursive production of psychological and social actions. In this chapter, we will examine the devices in detail, with detailed descriptions and worked examples, to help you to familiarise yourself with each one. The devices have been highlighted in bold type in the worked examples to enable you to more easily identify them. They have been split into three sections: basic, intermediate or advanced. This separation of the devices is somewhat arbitrary in that they might have been clustered together in different ways, but this ordering should match onto the level of competence you are likely to need when using them. You could think of this chapter as your DP training ground. As with any kind of exercise, little and often works best, so you don’t need to tackle it all at once. With each device, read through the worked example, then consider whether they are identifiable in your own data, and have a go at using them to begin to make interpretations about what is going on in the interaction.

The important thing to remember is that these devices cannot be used like tools in a mechanical sense; we do not simply identify the existence of a phrase or word in interaction and then immediately point to its function or outcome. Similarly, this is not like a chemical reaction or mathematical equation where x (assessment) + y (second assessment) = z (agreement). The devices are to be used, instead, as instruments that can sensitise us to the subtleties of interaction. We might then use the analogy of a Sherlock Holmes detective, who uses various techniques (ways of observing the world, a sensitivity to details) and devices (the magnifying glass, the fingerprint kit) to examine a situation and interpret what is going on. So, as DP analysts, we also need to use the devices with some care and skill, and this is why we need to practise. Each interaction is unique, and this is what makes DP analysis so interesting. Not only can we find something new each time, but it also provides a different perspective on the social world that you can practise and apply every day.

Basic devices

We will start with what have been broadly clustered as the basic devices: these are the ones that are typically easier to identify and which are fairly common in everyday interaction. In that sense, their relative frequency provides for greater opportunities to practise them. They are as follows: pronoun use and footing shifts; assessments and second assessments; silences, pauses and hesitations; hedging; extreme case formulations; minimisation; lists and contrasts.

Pronoun use and footing shifts

Pronoun shifts involve identifying when and where speakers use different pronouns to refer to themselves or other people. The most common ones are: I, you, he, she, it, we and they. They can also be used in terms of ownership: my/mine, his, hers, its, ours, theirs. Pronoun shifts are an example of the broader category of footing shift: this is a concept originating in Erving Goffman’s (1979) work and refers to the movement across participant roles that are produced in the talk. For example, one can speak as the author of our talk (i.e., saying our own words), the animator (i.e., the ’sounding box’: speaking as if relaying someone else’s words), or the principal (i.e., the agent or person responsible for the talk). The great thing about footing shifts is that they allow us to understand the layers of frames of reference in discursive practices. That is, we are rarely just speaking ’as ourselves’; our words also draw on issues about who is responsible for what we say, whose words we are using and who is saying it. So pronoun and footing shifts can be used to manage identities of the speakers, and accountability for what is being said. They can also be found when the factuality of an issue might be in question, to position the speaker as believable or simply reporting ’the facts’.

Worked example

Extract 7.1 illustrates the use of pronoun shifts in discourse, taken from research examining how people account for having chronic fatigue syndrome (CFS). The section is from an interview where Angela (someone with CFS) is talking with her husband, Joe, to the interviewer (Mary) about what happened when her illness began. See Box 5.4 in Chapter 5 for a full transcription key.

Extract 7.1: Taken from Horton-Salway (2001: 250, extract 1)


The different pronouns in this extract provide a way of orientating to a particular subject or object. For instance, the neutral third-person pronoun ’it’ refers to Angela’s illness but does so without labelling it too specifically (we can compare ’it’ with ’your illness’, for example). It also treats the illness as a ’thing’, an object or real state of affairs. By referring to it in this way, Mary invokes the reality of the illness without making any specific claims about what ’it’ might be. So neutral third-person pronouns can be used to make a vague, undefined reference to something while still treating that ’thing’ as real. The other pronouns are first-person (I, my) and second-person (you). Note that in the English language, the same word (you) is used for second-person pronouns, whereas in other languages (such as the French tu and vous) we can make a distinction between singular and plural uses through word choice alone. This makes the English ’you’ more ambiguous, though there are regional variations where speakers might say ’youse’ (or ’yous’) colloquially to indicate more than one person. Use of the first- and second-person pronouns can be seen in this extract to position the illness as located within an individual —something that Angela had, and to which Angela was subject — but also as something that Angela might be individually responsible for, or at least having primary access to the experience of the illness (’I had a sore throat’, ’I had the very worst headache’). We then see a footing shift on line 9, in which Joe invokes the first-person plural pronoun (we); this then re-orientates Angela’s account to include him as part of the ’accounting for the illness’ social action. The author of this line might then be both Joe and Angela, even though Joe is the animator. So while Joe is not claiming to also have the illness, the shift to ’we’, rather than singular ’I’ or ’you’, brings him back into the picture. It presents the illness as a mutually understood event; something that they were both involved in, and thus which Joe might also have access to, as part of that experience.

Assessments and second assessments

As with pronoun use, assessments are defined primarily linguistically, that is, they are identified as those instances in which a description is provided that makes a specific judgement or appraisal of something (and that something could be a person, object or event, for example). Assessments are sometimes referred to as evaluations in the DP literature. Second assessments are assessments that follow or are found soon after an initial assessment, and they typically follow what is called a preference structure. This is where there is a normative pattern for agreements (i.e., second assessments that match the evaluative direction of the initial assessment; see Pomerantz, 1984, for more details). This means that second assessments which agree with the initial assessment are typically produced soon after, with no hesitation or prefacing and may upgrade the initial assessment, whereas second assessments which disagree with initial assessments are typically preceded by a pause or hesitation, are often prefaced in some way and may partially agree or disagree with the previous assessment. Assessments can also be presented as if the reason for the assessment was located within the subject (termed ’subjective assessments’, the person doing the assessment), where personal pronouns are used (e.g. we loved the film), or as if located within the object or the thing being assessed (termed ’objective assessments’, e.g., the film was brilliant). Assessments therefore not only invoke our stances in an interaction — how we might be assumed to be a particular kind of person, or have a particular kind of opinion, for example — but also invoke our involvement in the thing that we are assessing. In other words, giving an assessment also makes a claim that we have experienced or have knowledge of the thing that we are assessing.

Worked example

Extract 7.2 is taken from a family mealtime, where Laura has cooked a chicken dinner for her daughter (Beth, 11 years), brother (Bill) and Bill’s wife, Doris. The family members are coming to the end of the meal.

Extract 7.2: Taken from Wiggins and Potter (2003: 522, extract 4)


There are two assessments in this extract — ’that was lovely’ (line 1) and ’it is lovely’ (line 3) — that display an orientation to the food as being the source of the assessment (with the use of the third-person pronouns ’it’ and ’that’). These are what might be called ’objective assessments’ in that they foreground the object (they are not, however, objective in the sense of being more accurate or truthful). As such, they work well here as compliments, because they locate the ’loveliness’ of the food as being in the food itself, not the consumer. Note, then, how there is also a subjective re-formulation of the assessments on line 4: Laura asks if ’you enjoy(ed) that’ (though we do not know here if the ’you’ is plural or singular), which shifts the focus back onto the person eating the food. The assessments work here then as compliments about the food and as demonstrating an appreciation (in combination with the ’thank you’ and the direct naming of Laura on line 1) to the cook. Laura’s re-formulation then avoids taking too much credit for the food and treats the assessment as an indirect request for more food. So there are a number of social actions going on in this very short clip. We can also identify a second assessment here (from Beth, line 3), which is produced soon after the initial assessment (by Doris, line 1). This conforms to the preference structure for agreement in assessments — there is no hesitation from Beth and her assessment is not prefaced or softened. Doing dispreferred actions as preferred can often be the source of humour in television sitcoms, for example where a character might violate the norms of interaction and this is treated as awkward or observational humour (see Stokoe, 2008).

Silences, pauses and hesitations

The noting of silences, pauses or hesitations — those gaps in interaction when there is no speech — is possibly one of the transcription features that most clearly flags up a transcript as having been prepared for conversation analysis or DP. Silences are typically measured in tenths of a second, with (.) indicating less than one-tenth of a second pause, (0.2) indicating two-tenths of a second pause, and so on. The term ’pause’ is most commonly used in DP research because it suggests that the lack of speech is a temporary state and that the conversation will resume at any point. ’Silence’ potentially suggests something more intentional or meaningful, so this might refer to a much longer pause (for instance, where the pause is more than just a few seconds long). Referring to something as a ’hesitation’ suggests that there is a deliberate withholding of talk by a speaker, so this term is used with caution. When transcribing, we also need to be careful as to where a pause is indicated on a line at talk; if it is within a speaker’s turn at talk, then it can appear as if the pause is theirs, and thus it is expected that they will continue speaking immediately after the pause (see also Chapter 5 for further discussion on this). Pauses are really important, then, in terms of clarifying the delivery of talk and noting any patterns or structure in the organisation of interaction. For instance, Jefferson (1989) suggested that in everyday conversation, a pause of one second or more was noticeable and could indicate interactional trouble of some kind. In institutional talk, where there is more likely to be a certain pattern to the way in which people respond to each other, then pauses may be longer or differently structured. There is no standard set of functions that pauses or silences can be used for, however, but they can alert us to something particularly interesting going on in the interaction.

Worked example

Extract 7.3 illustrates how pauses can be slightly longer than expected (cf. Jefferson, 1989) when used in institutional talk, but that they can also highlight other social actions and psychological business being managed. This extract is taken from an audio-recorded weight management group, within the National Health Service in Scotland, involving people who are attending for weight issues. The group leader (Melanie) is a dietitian and has been talking to the group about a ’balance of good health’ diet and how that might compare with their current food intakes.

Extract 7.3: Taken from Wiggins (2009: 378)


There are a number of pauses in this extract, including some short ones (such as (.), (0.4), or (0.6)) and longer ones ((2.4) and (1.0)). The short ones can occur within someone’s turn, such as the (0.4) on line 12, as well as between people’s turns, such as the (0.4) on line 9. In that sense, we cannot simply spot a short or long pause and be able to identify whether it signals a hesitation or trouble of some kind; we also need to examine where it is and how it is embedded with the surrounding talk. The (2.4) pause on line 4, for example, occurs immediately after Melanie’s question to the group. Given that the group consists of a number of people (around 10—12), this pause could be accounted for in terms of it being addressed to many people (not just one person) and so it may take time not only for people to answer, but also to make sure that they will not be interrupting someone else by answering first. So longer pauses may be more likely in institutional question—answer sequences than in everyday chat with friends. We can also note how there are some slightly longer pauses in Paul’s turn on lines 10—15. These serve the purpose of ’holding’ his turn, in combination with their placement after the stretched-out ’an::d,’, which also has the comma that marks continuing intonation. Pauses also indicate where there might be a change of speaker. For example, someone else might have started speaking at line 14, given that Paul’s immediately prior turn looks like it is ’complete’ and there is a one-second pause. Since no one else does, then Paul continues to give a little more detail about his account.


Hedging is what occurs when a turn in talk is marked in some way as provisional, tentative or conditional on some other events. This is something that can be seen in academic work as much as in everyday talk, for instance, when we ’suggest’ our results or state that ’we would argue that…’. It has also been considered in literature on politeness strategies. Hedged talk is thus talk that helps to manage a speaker’s accountability, in that it avoids making a specific or certain claim about something, and can be softened or retracted in the event of disagreement. Hedging often appears alongside other discursive practices, such as pauses or hesitations, or particular terms such as ’well-prefaced’ accounts (where a turn begins with ’well, …’ and typically marks a disagreement with the prior turn). It can also be used alongside delicate issues, such as in the worked example below.

Worked example

Extract 7.4 provides an illustration of the different forms of hedged talk that can be seen in social interaction. This extract is taken from a focus group with UK psychology undergraduates, who were asked to discuss the issue of asylum. SG is the moderator, P2 is one of the participants.

Extract 7.4: Taken from Goodman and Burke (2011: 115, extract 2)


As can be seen from the highlighted words and pauses here, there is no single way in which discursive practices can be understood as being hedged. They can be identified, however, through the way in which a turn in talk appears hesitant, includes a number of (longish) pauses within a speaker’s turn, and uses words such as ’I think’, ’I don’t know’, ’could be’ or ’sometimes’. In Extract 7.4, for example, P2 appears to be talking around the issue and not giving a direct answer. Talking about asylum, racial issues and minority groups is a delicate issue; there is the risk that the person might appear racist or prejudiced. The hedging therefore enables the speaker to raise these issues while also distancing themselves from any particular stance. It softens the impact of the discourse and marks the talk as delicate or sensitive in some way.

Extreme case formulations

Extreme case formulations, or ECFs, are words or phrases that are both semantically extreme and orientated to as extreme (Edwards, 2000; Pomerantz, 1986). They go beyond phrases such as ’very good’ or ’excellent’, which make a positive assessment of something but do not necessarily frame this as being an extreme case. So to work as an ECF, a word/phrase must be treated as, or hearable as, going to extremes. That is, it does more than just exaggerate or emphasise something. It is used to defend a claim or demonstrate investment in a particular account. They go beyond description and are used to manage a speaker’s identity in relation to what they are saying; as being a particular category of person, for example. As such, ECFs might seem at first to be easy to ’spot’ — in that we look for an exaggeration of something — but we need then to check how they are situated within the local interactional context to be certain whether they are being treated as an extreme case. In early work on ECFs, they were identified primarily in those situations when someone was making a complaint (Pomerantz, 1986). In later work, it was argued that despite being ’factually brittle’ (Edwards, 2000: 352), in that it would be easy to argue against an ECF being the case, they are typically unsoftened because they are designed to be heard as going to extremes. In other words, they are typically not produced as a ’fact’ but work instead to manage a speaker’s stake or investment in what they are saying. See Box 6.2 for more information on ECFs.

Worked example

In Extract 7.5, there are three examples of ECFs in a short space of time. This extract is taken from an interview with a man (’R9’) who had been a refugee in the UK for seven years and had recently moved to Glasgow. He had just been asked why he moved to Glasgow, and he then proceeds to give accounts of what happened to him following his move.

Extract 7.5: Taken from Kirkwood et al. (2013: 755, extract 4)


The ECFs featured in this extract are first used as a contrast pair — between ’all my teeth’ (line 3) and ’fully gone’ (line 6) — which works to emphasise the damage done by the attacks (line 1). Note also that the participant provides further evidence by pointing directly to his mouth so that the interviewer can see this for himself. The ECFs work not just as a description of dental records, but also as a way to manage the speaker’s claim as having being brutally attacked twice; it adds not only to the plausibility of his account, but also implies the force behind the attack, even if this was not explicitly stated. The third ECF — ’nothing to nobody’ (line 12) — then further adds to this management of the speaker as being a victim and blameless in the series of events being described. It would be very easy to challenge this ECF — he must have done something, even if it was just to say ’hello’ or to look at someone — but it works here to emphasise the speaker’s investment in the claim as being genuine. The interviewer’s responses (lines 2, 4, 9, 11 and 13) then orientate to the account as genuine and as hearably extreme.


In its simplest form, minimisation is present in discursive practices through the inclusion of words such as ’just’, ’only’, ’little bit’, and so on. It is the practices through which the volume or extent of something is treated as minimal or insignificant. It is important to note, however, that some of the words used for minimisation can also have other uses, so we need to do more than simply look out for these words in the data. The word ’just’, for example, can also be used to mean ’exactly’, as in ’it’s just right’ or ’recently’, as in ’we just got here’. Where minimisation does occur, this can be used to downplay the importance of an object, event or behaviour. It can therefore be used to manage one’s accountability for something: to minimise the extent to which it could be treated as serious behaviour (for example, in police interrogation or court testimony). As such, it tends to be found in discursive practices when people are describing something or have been asked to account for something. Minimisation can also be used in combination with contrasts (e.g., contrasting one smaller thing with something much larger); see ’lists and contrasts’ devices below.

Worked example

In Extract 7.6, we see the use of a minimisation device being used in an online discussion group for people who have recently had (or are about to have) bariatric surgery. In the first ’post’ of this extract, one of the forum members is asking other members what to eat in the post-operation period. Note that all the exclamation marks were in the original text taken from the online discussion forum.

Extract 7.6: Taken from Cranwell and Seymour-Smith (2012: 875, extracts 2 and 3)

1. Post 16, Jenny

2. Hey!!! I just had surgery on the 8th!!!!! I am SO Sore!!!!!!!!!

3. But I am excited!!!!!! And I am never hungry!!! Should I eat

4. just to eat, or don’t worry about it?!?!!

5. Post 17, Liz

6. Hey Jenny congrats!! YES you have to eat!!! I know it’s a fight

7. I am having aswell, but what the doctors told me was even if

8. I just take a couple of bits of my protein food, but NEVER miss

9. a meal completely.

There are two instances of ’just’ being used here as a minimisation device. In the first example, Jenny is asking a question about what to do (i.e., what to eat following surgery) and managing the dilemma of whether she might be eating ’just to eat’ (line 4). Given that bariatric surgery is aimed at helping people to effectively manage their weight, then eating for the sake of it (with no other purpose) is potentially a risky thing to say in this forum. The minimised ’just’ helps to soften this issue, and combined with the rest of this post, suggests that a small amount of food might be all that is needed. This is then confirmed in the response by Liz who explicitly tells Jenny that she has to eat (line 7); in doing so, she orientates to Jenny’s turn as being reluctant to eat when she isn’t hungry. Note then, the minimisation device (’I just take’, line 9) by Liz. Not only does this also downplay the amount of food that she eats herself — thus rhetorically working against the counter-claim that she might be eating too much or being greedy — but it also specifies the type of food (protein) that is ’taken’. So the minimisation device works alongside these other constructions (and you may also note the potential ECF on line 9: ’NEVER’) to both provide support for Jenny while also accounting for Liz’s own food intake.

Lists and contrasts

The use of listing in talk — where items are presented together in a sequential order as if reading from a list — are a particularly useful rhetorical resource in everyday and institutional discourse. Lists typically appear as three-part units (the ’three-part list’; Jefferson, 1990), and this normative pattern is fairly robust across different social settings — so much so, that in the absence of a specific third part, people will often use ’generalised list completers’. For example, a three-part list might be something like: ’he went on, and on, and on’ and an example with a generalised list completer (in bold) could be: ’there were wolves, raccoons and all sorts of animals’. One of the great things about three-part lists is not only that they serve to emphasise something and make it seem more factual or ’real’, but they also project a completion point. In other words, they indicate to the other speaker that the person’s turn is likely to come to an end: the third-part is so normative as to be expected. This has been evidenced in, for example, political speeches where the audience applause coincided precisely with the end of the three-part list (Heritage & Greatbatch, 1986). The use of contrasts is a related rhetorical device that also works to emphasise one thing over another and highlight the relevance of a particular feature. It occurs when one aspect of discourse is directly compared with another, to emphasise particular characteristics or the distinction between one or more objects.

Worked example

Extract 7.7 provides an illustration of both listing and contrasts. This data is taken from a research paper that analysed audio recordings of antenatal classes in the UK, where expectant parents went for advice and support during pregnancy and in the lead-up to the birth of their child. In this clip, the class leader (CL) is discussing expectations of what is achievable for parents in the first week after the baby is born.

Extract 7.7: Taken from Locke and Horton-Salway (2010: 1219, extract 3)

1. CL: If you are up and dressed by lunchtime in the

2. first week after the baby then you are doing

3. really well, and that is what you should be aiming for.

4. Not cooking, not shopping, not washing, not racing out,

5. taking care of yourselves. If you think back, my eldest

6. child is 17, and we had to stop work at 28 weeks.

7. You had to go and be signed fit to carry on working

8. beyond 28 weeks. So consequentially most people didn’t

9. go beyond 32 weeks because it was too much of a hassle

10. had to go every week to be signed fit. Which is what

11. we had to do.

There are two examples of contrasts here, though it is difficult to highlight these in bold as the contrast works across the discourse rather than being tied to specific words. The first contrast is between what should and should not be the activities for new parents (lines 1—5, specifically lines 4—5). The lowering of expectations — being ’up and dressed by lunchtime in the first week’ (lines 1—2) is further emphasised through the contrast with seemingly everyday activities (cooking, shopping, washing, line 4). The second contrast is between ’what we had to do’ (lines 10—11) and parental expectations in the present time with regard to how long you can work during pregnancy. Again, the contrast serves to strengthen the current situation through exemplifying what it could have been like. The use of a list (line 4), although it has more than three-parts, additionally supports the rhetorical features of the contrast. Discursive devices can thus be seen to work collaboratively together in interaction, and should be analysed as such.

Intermediate devices

Having established some familiarity with the ’basic’ discursive devices, this section provides descriptions, worked examples and practice tasks for the intermediate devices. These do not typically occur so frequently in talk, or may require a little more skill to apply them, but they are all the more powerful in how they can manage psychological business in interaction. The intermediate DP devices are: affect displays; consensus and corroboration; detail versus vagueness; disclaimers; metaphor; narrative structure; reported speech; script formulations.

Affect displays

Affect displays involve the apparent display of emotion, such as laughing, crying or sighing. Unlike emotion categories, which use words to describe emotions or emotional states, affect displays are those that invoke the emotion or embodied practice itself. Some of this work focuses on conversation analytic concerns such as explicating the sequential production and organisation of laughter (e.g., Jefferson, 1984) and crying (Hepburn, 2004), though it has also been used to examine how issues such as ’gustatory pleasure’ are produced and made relevant in talk (Wiggins, 2002). In DP, the concern is not to make claims about whether or not an emotion has been ’experienced’, but rather to examine how displays of affect work to manage psychological business at a particular moment in social interaction. This device can be analysed, then, for where and when it is located in talk and interaction, and how it is orientated to or made relevant. Past research examining affect displays (see references above) have noted that they are typically highly organised in interaction; orientated to by both the speaker and others in interaction as socially relevant.

Worked example

Extract 7.8 is taken from a study that examined calls to a child protection helpline in the UK (CPO = child protection officer). In this particular fragment the caller becomes audibly upset, as shown by the ~ in the transcript to indicate ’tremulous’ or wobbly voice.

Extract 7.8: Taken from Hepburn and Potter (2007: 93—94, extract 2)


One of the issues noted in Hepburn and Potter (2007) is that calls to this service can often be highly charged and emotional, given as they are dealing with potential cases of child abuse. What is important for the CPOs is therefore that they gather sufficient information from the caller in order to be able to follow up the case and intervene where necessary. If callers become upset, however, there is a risk that they will terminate the call and thus the helpline organisation cannot take any further action. Being sensitive to potential ’affect displays’ is therefore an important skill for CPOs, shown here in lines 11—14 where the silences and quietly spoken ’take your time’ (line 13) give space for the caller to continue, while also gently affiliating with the caller, and acknowledge and validate the upsetting nature of the call. Being able to examine affect displays, where and how these are produced, and how they are attended to by others in the interaction can thus provide unique insights into how psychological business is handled. In this case, the psychological business is ’being upset’ and the delicate family issue of a mother claiming to be scared of her own son.

Consensus and corroboration

The related concepts of consensus (reporting something as if many or all people are in agreement) and corroboration (reporting something as if supported by an independent source) are particularly useful in building up the factuality of accounts. As with other devices — such as reported speech — they work by invoking other people in support of whatever it is you are claiming. They are often used in accounts that manage attributional and accountability issues, of the untangling of complex social relationships or the unpacking of an event. So this is a device that might feature in institutional accounts (such as in newspaper reports or political rhetoric) or when the factuality of an account is under question. Consensus in itself can provide corroboration; if many people are in agreement, then this supports the factuality of the claim. With consensus, however, there is a risk that it could be challenged as collusion: that the ’many others’ or ’all’ who support your account may not be working independently. In the case of witnesses or court cases, this could have serious implications. This is where corroboration comes in, as it only needs one apparently independent account to provide support for and validate the account. This is, however, at greater risk from being challenged: it is much easier to undermine one independent witness than many.

Worked example

In Extract 7.9, we see an example of consensus being used alongside a contrast device. It is taken from an interview with people where they were shown news media clips and asked how they made sense of competing views, particularly in relation to political debate.

Extract 7.9: Taken from Dickerson (2000: 388—389, extract 3a)


The consensus terms used here — ’anyone can say that’ (line 6), ’Joe Bloggs on the street can say that’ (lines 6—7) - work to diminish the status of the politician’s (Margaret Beckett’s) remark. Not only does it suggest that it was an ’idiotic’ (line 4) thing to say, but also that it didn’t require any special skill or knowledge, in that ’anyone’ can — and be entitled to — say that. The term ’Joe Bloggs’ is a British colloquial term to refer to a hypothetical ’average’ person. This further supports the claim that it was an idiotic remark, in that any ’average’ person could say it. Consensus is provided here from ’anyone’ and ’Joe Bloggs’ — in themselves also drawing on vagueness and lack of specificity — even if in practice it would be easy to find someone to challenge that claim. Note, however, that the extract here also includes an extreme case formulation: ’anyone can say that’. As such, what is also going on here is that this turn in talk is hearably going to extremes, emphasising the point in a manner that is not designed to be taken literally. As noted above, it would be easy to find one person to refute the claim (i.e., one person who wouldn’t say that), but that is not the point. The point is that this turn makes a particular case about the politician in a way that undermines the status of her speech.

Detail versus vagueness

The opposing categories of vivid detail or systematic vagueness can be used to manage a number of psychological concerns in social interaction. They are often involved in what Potter (1996) has described as constructing the ’out-there-ness’ of an object or event; producing it as a fact or reality, independent of our accounts. Using detail in an account can be used to add credibility to your category as a ’reliable witness’ or someone whose report is more factual. This can be useful on occasions when our category entitlement as trustworthy is under threat. Alternatively, being ’systematically vague’ — i.e., vague or lacking in clarity on particular issues — can be a way of inoculating against claims that you might have a stake in what you are saying (see also ’stake inoculation’ device). In other words, there are times when being too detailed and specific might appear as if you are too heavily invested in the account. Trying too hard, as it were. Sometimes details can be challenged or seen to be inaccurate, and so vagueness can be more robust on such occasions; it is less easy to contradict someone if their account is not specific.

Worked example

Extract 7.10 is taken from a study examining student blogs on an educational online noticeboard, where the blogs were designed to enable students to reflect on what they know about a topic before learning more in class. Issues of student identity and of knowledge claims are therefore relevant here, and these blogs were made publicly available to fellow students and tutors.

Extract 7.10: Taken from Lester and Paulus (2011: 676, extract 1)

1. I don’t really know much about supplements. I’ve always

2. had the impression that vitamins are good for you. I know

3. that my mom takes a daily multivitamin and another type of

4. mineral supplement for health reasons, which was recommended

5. by her doctor. So, these types of things don’t scare me, and

6. are probably something I’ll consider taking in the future.

7. However, like a few people have said, things that are

8. supposed to give you energy do make me nervous. As far as

9. protein shakes go and other types of protein supplements that

10. many athletes consume, I don’t know much about them. I’ve

11. never had any personal experience with them, but alot of my

12. guy friends have taken things of that nature in the past

Alongside the use of possible disclaimers (e.g.,’I don’t really know much’, line 1), the presence of vague references to the fact that ’vitamins are good for you’ (line 2) and that ’things that are supposed to give you energy’ (lines 7—8) serve to provide support for this account while also situating the account as based on second-hand knowledge. Note, for example, how the reference to ’my mom’ (line 3), ’a few people’ (line 7) and ’a lot of my guy friends’ (lines 11—12) could be considered corroboration for her account. So the vagueness refers to what it is that vitamins actually do, and as such the student here is downplaying her own status as expert, but also providing an account that is fairly robust. For instance, it is harder to undermine the claim that ’vitamins are good for you’ than if the student had claimed, for example, that ’vitamins give you x-ray vision’.


Disclaimers are typically short phrases at the start of a turn in talk that try to mitigate the speaker’s stance on a particular issue. They do so by explicitly denying a potentially negative interpretation of what they are about to say, even if the rest of their turn contradicts this. Disclaimers are often in the form of, ’I’m not racist/sexist/homophobic, but…’ or ’I have nothing against (insert category of person), but…’. As such, they are often used in situations where someone’s identity or category membership is under question. They can be used in conjunction with a number of other devices, such as contrasts, category entitlements and assessments.

Worked example

In Extract 7.11, a couple are being interviewed in their home about services in their local area. This is taken from a study about how denials of prejudice are often achieved collaboratively with other speakers.

Extract 7.11: Taken from Condor et al. (2006: 452, extract 3)


There are a number of disclaimers in this extract, some of which (on lines 5 and 6) are working to defend against the potential for Jack’s turn in lines 1—3 to be heard as xenophobic. So in producing the disclaimer, ’he’s not xenophobic’ (line 5), Hilda works to resist that particular categorisation of Jack by challenging it explicitly. There then follows a series of disclaimers, which deny that what was said was racist (’it’s not racist’, line 6) as well as denying that both Jack and Hilda are racist (’we’ve never been racist’, line 6). Similarly, the phrase ’we’ve got nothing against’ (line 8, completed on line 9) also works to undermine any potentially negative interpretation of what is being said. And yet, what follows on lines 10—11 still marks boundaries between a seemingly homogeneous ’they’ who are coming ’here’. This extract also makes use of footing and pronoun shifts to manage these claims.


The use of metaphors in discourse refers to the way in which descriptions equate one thing with another to make a comparison or produce a particular rhetorical effect. For example, the phrase ’your words are poison’ is not a literal statement but constructs someone’s speech in a way that suggests that it causes severe harm or damage. Metaphors can therefore be used to produce categories of the world, and of people themselves (hence it is linked with the category entitlement device), in that it constructs particular versions of the world that often have quite visual or figurative references. As such, metaphors can be useful to highlight some features while blurring or concealing others; they can overly simplify the distinction between one category and another. They can also, therefore, overlap with extreme case formulations, which are other devices which foreground a non-literal interpretation and could also be easily undermined or challenged. Edwards (1991) provides a detailed discussion of how the production of categories, and metaphors, can be considered discursive rather than perceptual or cognitive.

Worked example

The example in Extract 7.12 is taken from a study that examined the ways in which speakers managed their subjective investment in the complaints that they were making, that is, how they produced their complaints as properly ’complainable matters’ rather than an example of how they might be characterised as moaning, whinging or generally a complaining type of person. The section of talk is taken from a phone call between friends, one of whom (Lesley) has recently had her home burgled and the police have been to her home to search for fingerprints. It is the burglar who is being referred to as the ’visitor’ here.

Extract 7.12: Taken from Edwards (2005a: 17, extract 9)


The ’complainable’ here is on line 5 with the comment about ’this fingerprint stuff’ from the police. While this in itself is analysable — in that we might expect the burglary to be the source of the complaint, not the police investigation of this — it is the metaphor on line 12 that is our focus here. The term ’galloping all over the cushion’ responds to an earlier issue noted by Lesley that the burglar had left muddy footprints all over a cushion in her house. The metaphor ’galloping’ produces this in a rather comical manner; it attends to the seriousness of the issue while also dealing with the irony of being concerned about a cushion (rather than, say, the break-in and any valuables stolen). The metaphor works here, then, as a way of packaging up the event in a particular way, to manage issues such as what counts as a complaint and who should be complaining (and when).

Narrative structure

Narrative, in its simplest interpretation, is like the telling of a story. But it is more than that; it is the production of an account with a coherent, sequential order. This may involve setting the scene (where and when did something happen), the timescales (how long did it take), the order of events (what happened first), and so on. The narrative structure of an account is often also combined with other features, such as the inclusion of vivid details or vagueness in specific places; each of these helps to support the credibility of the narrative. Narratives can be used in cases where people are accounting for something that they may or may not have had a personal liability in, such as accounts of illness (e.g., Tucker, 2004) or criminal activity (Auburn & Lea, 2003). Note that this is not the same as taking a narrative approach to analysis, which involves a different set of epistemological assumptions about discourse.

Worked example

Extract 7.13 is taken from a study into the group interactions in a sex offender treatment programme, and the way in which offenders account for their criminal activities. Part of the programme requires offenders to provide an account — their story — of events.

Extract 7.13: Taken from Auburn and Lea (2003: 287, extract 2)

1. PO1: Well talk us through that (unclear)

2. (2.5)

3. Off: Uu::hh, well we’d gone out out and we went down to a

4. club which was about 20 mile away,

5. uu::m (1) um by th- by the time my wife, my son, and

6. Sally had drunk a lot

7. I didn’t drink as much ’cos I was driving, I had a 20

8. mile drive back home (2) [sigh]

9. so when we got home then we had a cup of coffee and

10. I put on a, some records while we were drinking

There are a number of features to this narrative example, such as description of what had happened first (gone to a club), the location of this (20 miles away), and a sense of time in relation to the unfolding events (’by the time’, line 5; ’when we got home’, line 9). Each of these on their own would not necessarily make this a strong narrative, so it is the culmination of various features of the discourse that performs the action here. In doing so, and producing it as a ’normal night out’, the speaker is also managing his own accountability and involvement in the criminal behaviour.

Reported speech (aka active voicing)

Reported speech (sometimes known as ’active voicing’) refers to those features of speech that attribute the source of the talk to another speaker. This typically appears to be reporting what was said on a previous occasion. So the talk is presented as if it is a direct reproduction: not just the words of another speaker, but often presented as if it is in the same style or delivery. They can often be prefaced with phrases such as, ’and he was like…’ or ’she said…’. It can therefore be used to produce a particularly vivid account, to attend to the speaker’s own identity and accountability for what is said, and for footing shifts. It can help to increase the factuality of one’s account, to make it seem more realistic and also to minimise one’s own stake in what is being said. In other words, it can construct the speaker as if they are merely ’reporting back’ and not accountable for the content of the talk.

Worked example

Extract 7.14 is taken from audio-recorded Dutch criminal trials in the closing arguments between the prosecutor and defence lawyers. Reported speech can be particularly useful in such institutional settings where the aim is to ’get the facts’ and stay close to the original version of events. In this extract, you will see the English translation on the first set of each pair of lines, with the original Dutch words underneath.

Extract 7.14: Taken from Sneijder (2014: 477, extract 3)

1. on the other hand I understand J a little when he says like

2. aan de andere kant eh snap ik J ook wel een beetje dat ie

3. zegt van

4. yeah, but I e:h didn’t behave violently at all did I

5. ja maar ik heb toch e:h toch helemaal niet gewelddadig

6. gedragen,

7. eh (.) with the word violently,

8. eh (.) bij het woord gewelddadig,

9. you rather think of indeed (.) slapping, kicking,

10. denk je ook wel meer aan (.) inderdaad slaan, schoppen,

11. threatening

12. dreigen,

The reported speech here is enclosed in what Sneijder (2014) refers to as the pre-quotation and post-quotation talk; so the surrounding talk of the reported speech is as important as the reported speech itself. Framing the talk in this way then enables the speaker to partially align with the source of the reported speech (’J’, line 1) as well as using it as evidence to support their case. As such, it is not only that reported speech is used to emphasise or validate a particular account, but it is also the sequential placement of the speech that adds to its rhetorical effect. This can mean, for example, the speaker prefacing or building up to the reported speech, and using it as a ’punchline’ or, in extract 7.14, a basis on which to then ’unpack’ the words spoken.

Script formulations

Script formulations are descriptions that present a behaviour or event as if it regularly or frequently occurs: as if following a script (Edwards, 1994, 1995). As such, they can be used to present the behaviour as normal or expected, as not unusual in any way. Script formulations can also be combined with ’dispositional formulations’ (or a person script), where a person’s character or disposition is constructed as if routine and predictable. Note that this can also be seen in the emotion categories device. Scripting is theorised in discursive terms, rather than perceptual ones, so it is argued that something is produced in the unfolding sequence of interaction as being scripted in a particular way. So they are produced to rhetorically defend against alternative descriptions (or formulations). Script formulations often draw on modals, such as ’would’, reporting something as plural (e.g., arguments, parties), adverbs such as ’always’ or ’usually’, or verbs which highlight the regularity or iterative nature of something (e.g., ’gets’, ’loses’). See Edwards (1995) for a summary of scripting devices. When examining discourse for script formulations, however, the focus should be on how the script works alongside the rest of the interaction; we are not simply spotting words but examining talk for its action-orientation.

Worked example

Extract 7.15 is taken from a speed-dating conversation between a man (M) and a woman (F) who are meeting each other for the first time and have been asking questions such as ’what do you look for in a man/woman?’ Scripting is thus a characteristic feature of such conversations precisely because the participants are having to formulate what it is they do/are/like on a regular basis in order to try to persuade the other person that they might be a suitable future romantic partner.

Extract 7.15: Taken from Korobov (2011: 474, extract 9)


There are a number of scripted features here, most of which refer to what ’other’ women are like to create a contrast between the speaker’s own stance and those of other people. This is another example of how devices can work together in a stretch of talk, and how identifying them in isolation does not constitute a full analysis. In this extract, the scripting refers to what women say (line 9), what they do (lines 10 and 16), what they think (line 13), and then specific characterisations of them as ’always complaining’, ’moaning’, ’nagging’ or ’talking about others’. The analysis could also examine how these script formulations work to produce particular gender categorisations: how women are produced in particular ways, and how the speaker’s own (gendered) identity is then partly produced through contrasting with this. Note that it is F who first proposes the category of ’complaining women’ (line 1) which she then uses to create distance between her own stance on the issue (line 18).

Advanced devices

The final section of discursive devices are not necessarily harder to identify, but they either occur less frequently in talk or they require a little more skill in using them in analysis. They are: agent-subject distinction; emotion categories; category entitlements; modal verbs; and stake inoculation.

Agent-subject distinction

In this device, we are considering how the agency of the speaker (or the agency of those being spoken about) is managed within the talk. Are they characterised as being agentic, as having choice, the capacity to act and make decisions or determine the course of behaviour, for example? Or are they characterised as subjects, as passive and to whom things are done? This can be as simple as the distinction between saying something like, ’I had gone to the shop’ (active/agent) rather than, ’I had to go shopping’ (passive/subject). In the latter formulation, the speaker is constructed as having little (or no) choice in whether they had to go shopping. The use of the agent-subject distinction, therefore, can be particularly useful where people are managing their accountability for a behaviour or event. It is typically analysed in terms of how the level of agency is being managed in talk; that is, it is rarely as straightforward as a dichotomy between being active or not. Situations in which someone’s agency might be particularly important to the consequences of the interaction are police or courtroom interrogation (Stokoe & Edwards, 2008) or neighbour mediation (Stokoe & Wallwork, 2003), for example.

Worked example

In Extract 7.16, S (suspect) is being interrogated by a police officer (P) in the UK about possible criminal damage to his neighbour’s house. It could be argued that the whole interrogation can be analysed in terms of agency and accountability, and to what extent the suspect is liable for the damage (and thus can be prosecuted).

Extract 7.16: Taken from Stokoe and Edwards (2008: 104—105, extract 3c)


One of the features of the questioning from P in this extract is to present questions in a way that invite a ’yes/no’ response from S. The way in which P formulates the questions is thus important in terms of having an admission from S that makes relevant particular actions. In this case, it is the hitting of the door (line 8) that is one of the criminal behaviours; and ’you hit’ formulates this not only as agentic (in that S is the cause of the action), but it also suggests a category of action (hitting, rather than, say, knocking or tapping) that implies a degree of force. The question here, however, is combined with another agentic behaviour (intention, line 9) and it is possibly to this latter suggestion that S gives his strong ’no’ response. Note that there is also an interesting use of an emotion category here (line 5), which we will examine in the next device.

Emotion categories

Emotion categories are those instances in which speakers verbally make relevant a particular category of emotions or emotional states. Note that this is different from affect displays, which suggest that the emotion is being displayed or experienced; by contrast, emotion categories can be delivered in a way that reports on past or current emotion states as if reporting on any other individual characteristic. For example, someone might refer to themselves as being ’so happy I could cry’, or someone else as ’she’s always so angry’. In DP, emotion categories are analysed as a discursive accomplishment — with how they are used, invoked and consequential in interaction — rather than whether they suggest an individual emotional ’experience’. They are often involved in rhetorical work to present a contrast with other categories, such as emotional as opposed to rational, logical or cognitively-based, or to set up a distinction between someone being genuinely emotional and ’faking it’, or as having the emotion because of some external event or driven by an emotional disposition. Emotion categories can therefore be used in a range of different social actions, such as managing one’s identity or accountability, managing one’s stake in a state of affairs, or supporting the factuality or credibility of a claim.

Worked example

Extract 7.17 is taken from a counselling session with a married couple (Mary and Jeff), where Mary is telling her account of when she decided to tell Jeff about an extra-marital affair that she had recently ended.

Extract 7.17: Taken from Edwards (1999: 276, extract 3)


The emotion categories used here — ’angry stage’ and ’very upset’ — are both used in reference to Jeff, in relation to his reaction to Mary’s news of her affair. Not only do they treat Jeff’s reaction as primarily emotional (rather than, say, rational), but the repeated use of the word ’obviously’ orientates to this as being a normal and expected reaction. The anger is not just anger, it is an ’angry stage’, and thus constructed as if being a temporary state. The sequential placement of these terms is as important as how they are described: immediately after the news of the affair and just before the account of ’arguing’ (line 6) and ’drifting away’ (line 7). What is being dealt with here, then, is Mary and Jeff’s accountability for the problems in their marriage. The use of emotion categories both invokes a justifiable reaction to the affair, but it also sets up a potentially ongoing issue of how those emotional reactions might then impede any repair of the relationship.

Category entitlements

Working up categories and category entitlements are one of the key ways in which identities are managed in interaction. These can be categories such as being an ordinary person, an expert in a particular skill, belonging to a particular group or geographical area, or someone who can be trusted. So categories do not always have to be broad social groupings, such as those defined by age, gender, ethnicity, nationality or some other demographic. They can be small, everyday and seemingly trivial things. Work in this area overlaps with what have been termed ’normalising’ devices, such as Jefferson’s (1984) examination of ’At first I thought…’ sequences, and building on Sacks’ (1984) work on ’doing being ordinary’ (see also the stake inoculation device). What is important, however, is that these categories are consequential for that moment in the interaction; they are bound up with the social action taking place. Category entitlements are then what follows from these categories; the kind of knowledge, experiences, skills or responsibilities that the category is entitled to own. For example, the category of ’expert’ is entitled to claim knowledge about a certain topic, but an ’ordinary person’ is expected not to possess supernatural powers. Category entitlements can also be used when people are managing the factuality of claims; how they treat each other as having a stake in what is being said, and thus potentially undermining its credibility (Potter, 1996). For instance, by categorising a person in a particular way, one can support or undermine their potential claims to different category entitlements.

Worked example

Even the category of ’being ordinary’ has enormous potential, and this is particularly clearly seen in reports of paranormal experiences, where people are working up not only the credibility of their account, but also the implications that this has for them as credible persons. Extract 7.18 is taken from a collection by Robin Wooffitt (1991) on people’s accounts of having had some kind of paranormal experience, such as seeing a ghost, claiming clairvoyency, or experiencing poltergeist activity. The speaker’s husband (David) was an RAF pilot who had recently died; she is talking here about a moment during the funeral.

Extract 7.18: Taken from Wooffitt (1991: 270, extract 3)

1. an’ I went in there (.) er:m w-with my mother-in-law

2. and uhm:(0.4) friends that were with me

3. (1.3)

4. .hhh (.) and I was looking at the coffin

5. and there was David standing there

6. (0.3) he was in Blues

The presentation of this account is in what Wooffitt describes as an X/Y format, where X is the mundane and ordinary behaviour or event, and Y is the paranormal or extraordinary thing. These two formulations are usually presented in close sequential order (the Y immediately follows the X) and in doing so this highlights the contrast between the two. In the example, the speaker is doing the ordinary thing of looking at the coffin at her husband’s funeral, when the extraordinary thing (seeing his ghost) is reported immediately after. So it is the combination of both the content of the talk (what is said) and its sequential order (when it is said) that does the work here of producing the account as if unexpected. The speaker’s category of ’an ordinary person’ is thus invoked here through a category relevant behaviour — doing something very ordinary — so categories do not always have to be directly named or explicitly stated to be invoked in this way. The upshot of the X/Y formulations is that it enables people to defend against the claim that they might have been looking for something strange or unusual by showing that the weirdness came to them.

Modal verbs

Modal verbs are those that implicate the speaker’s degree of ability, obligation, intention or permission to be able to perform that activity. These are verbs such as: might, could, would, should, can, must and will. They invoke a range of different psychological issues in themselves, such as being responsible or accountable for something or being able to do something, and in that sense it might be easy to identify them in your data. They do, however, require a more sophisticated analysis: we cannot simply point to them and show that someone is managing their obligation to perform a task, for example. They are thus in the ’advanced’ list because they need to be interpreted alongside other features of talk; to examine the possible implications of these terms, and then to interpret how they are being managed by the participants themselves. This is one of the core requirements of DP: that we examine participants’ orientations in talk, rather than analysis based on our own assumptions or purely on the basis of linguistics.

Worked example

Extract 7.19 is taken from research into an online veganism discussion forum, where participants were often seeking advice or suggestions about what to eat when following a vegan diet. One of the issues that is raised elsewhere on the forum is that there may be a risk of vitamin (or other) deficiencies if one does not follow an appropriate diet regimen (e.g., when not eating enough of a particular category or type of food).

Extract 7.19: Taken from Sneijder and te Molder (2004: 607, extract 3)

1. Date: 3 June

2. From: Victor

3. Hi, I do hope you found good info about a healthy diet as well.

4. Veganism and a healthy diet are certainly not mutually exclusive.

5. [... 20 lines omitted ...]

6. And if you have a glass of fruit syrup (without added sugar)

7. with every meal, or another source of vitamin C, then

8. there shouldn’t be any problems, certainly not with iron or zinc

9. particular

The modal verb ’shouldn’t’ (line 8) is used here to manage the accountability that people might face when following a vegan diet: is the vitamin deficiency likely because they have not followed the diet correctly, or because a vegan diet, by definition, misses out on food groups that would provide those vitamins? This is a delicate issue in a vegan discussion forum, as the forum members themselves identify as following a vegan diet; anything that potentially troubles the ’vegan’ category is thus problematic. What can be seen in this extract, then, is the forum member (Victor) constructing his response in an ’if…then…’ format: if (you do this) then (there shouldn’t be any problems). The implication, then, is that the vegan diet itself is not likely to cause vitamin deficiencies: it is the responsibility of the individual to eat correctly and following guidelines to ensure that they stay healthy.

Stake inoculation

Stake inoculation refers to a range of practices whereby discourse is constructed to defend against claims that the speaker might have a stake in, or be overly invested in, what they are saying. This is similar to saying that the speaker might be ’biased’, ’subjective’ or ’twisting the truth to serve their own purposes’. It usually occurs, however, in a subtle rather than a direct way, and so contrasts with disclaimers where the issue is tackled directly. In many ways, stake inoculation is one of the more fundamental issues of a DP analysis, though it can cover a range of different devices and, as such, there is no single way in which it can be identified. This is why it is in the ’advanced’ category rather than the basic one. DP is concerned with how people treat each other as being factual, believable, accountable and so on, rather than with whether accounts are ’facts’ in themselves. So using the concept of stake inoculation, we can examine discursive practices to see how they are put together to inoculate, rhetorically defend against or resist counter-claims that they may be less-than-factual (Potter, 1996).

Worked example

Extract 7.20 is taken from a study that involved the interviewing of paranormal investigators about their experiences of paranormal events. As noted below, however, being able to state that something was, or was not, paranormal also involves the issue of credibility as a witness (see also category entitlements).

Extract 7.20: Taken from Childs and Murray (2010: 26, extract 1)

1. Ad: ↓ye::ah so I mean par- ee:<paranormal< li=""></paranormal<>

2. (0.2) we ca:n’t say (.) b’cause we can’t

3. prove it wa:s, but we can’t prove at the

4. mo:ment, we can’t prove it wa:snt,

5. (.) neither. (0.2)

The phrase, ’we can’t prove it wasn’t’ (line 4) is an example of stake inoculation whereby the speaker defends against the claim that he might be biased simply because he is a ’paranormal investigator’, and thus potentially looking out for such activity. The use of the word ’proof’ is also particularly helpful here, given that it hints at scientific endeavours in which evidence is required to prove a theory or assumption. So while the speaker here doesn’t directly say, ’you can believe me’, the use of this stake inoculation device works to help achieve the social action of being a believable or credible witness.

Key points

· The discursive devices are analytical tools that can help us to ’unpack’ interaction.

· Not all of the devices will be needed for your analysis; be guided by what you find in your data.

· The DP devices often work together in discourse; more than one device is typically used to examine any section of talk.

· The important thing is to use the devices as an analytical tool to examine the social actions and psychological issues that are being managed in the unfolding sequence of interaction.

Recommended reading

Potter, J. (1996). Representing reality: Discourse, rhetoric and social construction. London: Sage.

Wood, L. & Kroger, R. (2000). Doing discourse analysis: Methods for studying action in talk and text. London: Sage. Particularly Chapters 7 and 8.