Analysing data using DP
Here you are, with lots of words to analyse, and possibly not quite sure where to start. Facing a detailed transcript and being expected to analyse it can be a daunting and overwhelming place to be. When you first look at your transcript, you might at times wonder if there is anything to analyse in your data or, conversely, whether there is too much to analyse. In this chapter, we will break down the analytical process into manageable chunks and provide scaffolding to help you gain confidence and competence as a DP analyst. DP provides a way of analysing what might appear to be an overwhelming mass of words on a page: it provides a lens through which these words appear organised and coherent. So we are going to work through the process of using that lens and help you to make sense of the words (your data) in front of you. The aim of this chapter is therefore to demystify the process of analysis. It will first lead you through the stages of analysis, with a summary table of the discursive devices (see Chapter 7 for further discussion of these). It will help you to get started and, even more importantly perhaps, how to stop. The chapter will also discuss more advanced analytical issues, and ways in which you can ensure you are producing a ’good’ DP analysis. Finally, there are suggestions for holding a data session with other researchers, to help you see how you can develop your analysis through group discussion and interaction.
In many ways, each stage of your research is part of the process of analysis, since we make decisions about what features of interaction to record, the process of transcription, which sections to code, and how to select your extracts for close analysis. So this is really the part where the analysis becomes very detailed and concrete, and in many ways this is the part of the book that most clearly distinguishes DP from other forms of discourse analysis (see also Chapter 2). As with many analytical approaches, DP does not have a clearly defined, universal, step-by-step approach. There is no recipe to be followed. It has been likened to a skill such as riding a bike (Potter & Wetherell, 1987): a theoretical framework rather than an analytic method. That said, when starting out in analysis it can be useful to have a supporting structure to help guide us along.
Doing discursive psychology: stages of analysis
It will be helpful if you have your own data extract in front of you, with video and/or audio files where available, as you work through this chapter. This will allow you to practise each stage of analysis as you go along, pausing and taking a break when you need to, and spending time becoming familiar with each task. You may have been developing your own research project and worked through the previous chapters, in which case your research question will help to guide you as you work through the analysis. You might, however, have been given a data extract to analyse for an assignment or class activity, in which case you may not have been involved in collecting or transcribing the data, and you may not even have been given a research question. In that case, you might have no idea of where (and how) to get started. Either way, it doesn’t matter if at first your focus is quite broad, or if you find yourself getting distracted by different issues. You should just concentrate on getting started, slowly but surely, and don’t rush to get to the end. You should also familiarise yourself with the theoretical principles of DP (see Chapter 1) to ensure that you adopt an appropriate social constructionist stance on discourse. For those of you who don’t yet have any data to work through, see Box 4.1 (Chapter 4) to find some data to practise on, or use the ’steak and fish’ data extract referred to in Chapters 2 and 5.
The following stages of analysis will take you through the process of DP analysis, but they come with a caveat: doing DP is not a simple step-by-step linear process. It can involve moving backward and forward through different stages (see Figure 6.1 below for an overview of the DP analytical stages). The stages are provided as scaffolding, to get you started on analysis, until you build in confidence and competence. While you might need to follow these stages meticulously at first, you should find that as you become more skilled in analysing data, the process becomes more fluid and iterative. Around stage 3, you might also need to familiarise yourself with the DP devices before you apply these to your own data (see Chapter 7).
We will work through the data extract that we first saw in Chapter 2 (the ’steak and fish’ example) to illustrate how the analysis should be carried out at each stage. This has been transcribed to include paralinguistic details about pausing, speed of talk, overlapping speech and emphases on specific words. The extract was also chosen from a longer section of mealtime interaction as a starting point for analysis, given that it includes people talking about likes and dislikes of food and detailed descriptions of foods they are eating. This is characteristic of how DP works with data: we focus first on one extract, one single short piece of interaction, and examine this in detail. Only when we find something — a psychological concept, for instance — that is being invoked or orientated to, and have a sense of how this is involved in a particular social action, do we then move to find other examples in our data corpus. In this way, our research is inductive (also termed bottom-up or data-driven) and rigorous, staying close to the issues that are arising in the data and grounding our analyses in specific examples.
Extract 6.1: The data excerpt: ’steak and fish’
When you start to analyse data, it is important to analyse a small amount at first. Doing too much at once can be overwhelming and could lead to a limited analysis. About one side of paper is sufficient to get started, and there should be enough detail in this (particularly if we’ve coded an extract that overlaps with our research question or area of focus, see Chapter 5) to be able to get started. You can use up to around three or four sides of paper if you are working with others — see the section on ’holding a data session’ — because you will be able to cover more ground when working collaboratively. Note that the extract shown above is very short, but it should be sufficient to illustrate the stages of analysis for the purposes of this chapter. It can also give you a sense of how detailed psychological business can be accomplished in just a few seconds of talk. You will also note that Extract 6.1 includes words and features of intonation (i.e., paralinguistic details), but no extralinguistic detail (such as eye gaze, facial or bodily gestures). We can add these in later as and when we need to; what is important here is that when analysing the data we refer not only to the transcript, but also to the original recording.
Stage 1: Read the data
This stage might seem very obvious, for we need to have watched, listened to or read through our data to have been able to transcribe it. It might also seem so simple that you might not even think of it as an analytical stage. Do not be fooled, however, as the first stage of reading through the data can often help you to notice things that you had not seen when transcribing or coding the extract. Just as the process of transcription can slow down the pace at which you listen to the audio (or video) file, so a careful reading of the data should slow down the pace at which you consider each line of text, and the words within these. Remember that ’data’ refers here to both the original video and/or audio recordings of the discourse as well as the transcribed extract. When you are working with textual data, you will have only the text and no audio or video.
When you have video or audio data, then, you should first watch and/or listen to this a few times before you move to your transcript (so if you are working with textual data, you can skip to the next paragraph). Remember to focus, at this point, only on that part of the recording that matches the transcript that is front of you. This might be less than a minute, or just a couple of minutes’ worth of interaction. Play through the data at a slower speed if possible (see the section on software for transcription, in Chapter 5, for how to do this), to avoid you skimming through it too quickly. As you watch the video, pay attention to visual aspects — where the speakers are, who is looking at whom, whether there are any hand gestures or body movements, and so on — as well as verbal aspects. For example, consider how words are said and where the talk occurs in relation to gestures or body movements. This will help you to familiarise yourself with the interaction as it happened, and playing this over and over can sometimes make features of the interaction stand out more clearly. For instance, if we are used to hearing particular words or phrases we might not notice them as being particularly unusual. It is only when we focus on them as they are said as part of interaction that we can begin to question their taken-for-granted status.
Once you have become more familiar with the video/audio data, you can then move to your transcribed data. Read through your extract at least two or three times to gain a sense of the sequential nature of the discourse; what comes first and what happens next. Take your time to become familiar with the transcription symbols and how these relate to the way in which the words were spoken. This is where it really pays off to have done at least some of the transcription yourself, as then you will already have a close familiarity with your data and the transcription symbols. If you have textual data, you may not have features of intonation, but there may be peculiarities in the way in which words are written (such as the use of capitals, ellipses, emoticons, and so on). This might also give you time to notice any patterns or repetitions (for instance, the slight mirroring of ’I don’t like’, line 3, and ’I like’, line 8, in Extract 6.1) or phrases that appear to stand out (such as ’disnae fire my rockets’ or ’incarcerated’ in Extract 6.1). You do not need to write anything on the transcript at this point, focus instead on becoming familiar with the content and organisation of the data: what is said, and in what order.
This first stage, ironically, is often the point at which novice DP researchers can panic, if they think that nothing appears to be of interest in the extract they’ve chosen, or if they think they have found something and then rush to identify a ’theme’. So the first lesson is: don’t panic. There is almost always something to analyse in your data, even if the extract you start with may not be the ideal choice. The second lesson is: don’t rush to find a theme. This stage is about becoming familiar with your data and getting used to using the DP lens. Unlike many qualitative approaches, DP (and other forms of discourse analysis) focuses more on issues or social practices rather than themes.
Stage 2: Describe the data
Once you have become more familiar with your extract, you can then begin to make initial notes on the transcript. The aim of this stage is to focus your attention on describing, in detail, what is going on in the interaction; to begin to use the DP lens to focus on how discourse is constructed/ive and how it is situated (the first two principles of DP; see Chapter 1). We cannot do this effectively just by listening to or watching the data; seeing the transcript typed out on the page is essential for examining the turn-by-turn detail of interaction. This is why the first step of reading of the transcript is important. It allows us to examine exactly what was said, how it was said, and when it was said: so your aim is to focus on the what, how and when of different parts of the interaction. This helps us to avoid making assumptions about the data. By describing it in detail, we focus on what is actually going on rather than what we assume is going on. A common mistake when first analysing discourse is to move too quickly to make interpretations or analyses of the data (see also the ’pitfalls’ sub-section of this chapter). Whether through habit (we might try to work out what our friends ’mean’ when we talk to them) or through an urge to do the analysis quickly, it can at first seem obvious about what is going on in the talk. This usually leads to weak analyses or analyses that are not consistent with a DP approach, so take it one step at a time, and you’ll save time in the long run. As you become more familiar with DP analysis, you will find that stages 1, 2 and 3 merge together, and that you will start to make notes as you are reading and will be quicker at identifying social actions. Getting started, however, it can be helpful to separate out these stages to ensure that you work through carefully and rigorously.
In this second stage, then, we move from reading the transcript to making our first notes about the data. This is part of the process of examining the interaction in terms of social actions (which will be stage 3), but before we can identify ’social actions’ we first need to be clear about what exactly is going on in the interaction. When making initial notes on the transcript, it can be helpful to start working through the transcript line by line, in sequential order, to help us to get started and ensure that we do not miss out any aspects of the data. Sooner or later, however, we will start to look backwards and forwards through the text in a more fluid manner. As we noted in Chapter 1, one of the core principles of DP is that it treats talk as situated sequentially and within a particular interactional and rhetorical context. In order to interpret one turn in talk, then, we also need to examine what comes before and what comes after the turn in talk that we’re focusing on. This is what is referred to as ’indexicality’, when the interpretation of a word or phrase is dependent on the context of its production; in this case, what occurs immediately before and immediately after that section of discourse. This is why DP needs to examine interactions within the context in which they were produced (whether this is a family meal, an online discussion forum or a telephone helpline).
For this stage, you need to focus on describing in detail what is going on in the interaction. As you work through each line of the transcript, make notes about the talk or text in terms of:
· What was said/written: focus on the type of word or phrase that was used, for example, in terms of word form (is it an adjective or noun?) and cultural context (is it a culturally-specific word or slang phrase?). How does it draw on specific categories of people or objects? This helps us to focus on the content of the discourse and on what is being constructed.
· How it was said/written: is there anything about its prosody or delivery (e.g., rising intonation, stretched-out talk, loud or whispered talk, ’smiley’ voice)? Does it overlap with other talk or actions? What else is going on in the interaction? Is there any laughter, crying or other displays of affect (note that we do not refer to these as emotions, as that would imply a known internal state)? This focuses us on the style or structure of the discourse, on how the talk is constructed.
· When it was said/written: look at where the word or phrase is positioned within the speaker’s turn at talk (i.e., at the beginning of their turn, or following another specific phrase), and within the surrounding interactional context (i.e., the preceding and the following turns). This focuses us on the organisation and situatedness of the discourse.
You might find it more comfortable to work through the data on paper, with written notes on the transcript itself and in the margins. This is why we use wide margins and double-spaced transcripts, to leave room for analytical notes. Or you might prefer to work in an electronic word document, using the comment function to add notes. Either way, don’t worry at this stage about whether you need to interpret something or make any ’meaningful’ notes. You do not need to write something for every line of talk, and you may need to go through the extract a few times to gain some practice. The most important thing here is that you are starting to ’do noticing’, that is, to slow down the pace at which you look at interaction, focus on the details and begin to describe the talk clearly. In doing so, you should focus on what is said and what happens in the discourse. Do not try to identify why someone said something, or what they might be thinking or feeling. If you find yourself making interpretations that infer cognitive or emotive states, then stop to question what it is about the discourse that triggered that interpretation. Sometimes interpretations like these can alert us to important aspects of the interaction, so we can use them to our advantage, not be concerned that we aren’t ’doing it right’. Remember too that most notes that you make on your data will not necessarily end up in your final written analysis, but they are an essential stage in the analytic process and can help you to develop ideas about the data.
Figure 6.2 is the ’steak and fish’ excerpt with my initial notes on it, to give an example of what this note making might look like.
We can then continue through the transcript in this way, line by line, and as we do so we can begin to gain a sense of the social actions that may be unfolding in the interaction. Our aim in the second stage, then, was to break down the interaction into small chunks, to start to see how discourse is put together, like an engineer taking apart a piece of machinery to understand the component parts. It is in the next stage of analysis that we will use the discursive devices to help us identify what actions are being accomplished.
Box 6.1: Activity: Familiarising yourself with the discursive devices
Before you begin stage 3 of the analysis, you might want to work through Chapter 7 to familiarise yourself with the discursive devices (sometimes referred to as the DP devices). The devices themselves are like a set of tools that are used to examine particular features of discourse, and like many tools, they require some practice to become comfortable with how and when they should be used, and to become sensitised to their flexibility and complexities. Just as with learning other skills, the more time you spend practising, the more competent you will become. So do not be put off if it seems strange or awkward at first. This is a natural process of working with something new. If you have not already done so, go to Chapter 7 for a full set of descriptions and worked examples for each of the devices, separated into basic, intermediate and advanced sections.
Stage 3: Identify social actions and psychological constructs
The third stage of analysis is the one that most distinguishes this process as DP analysis, by focusing explicitly on the interplay between social actions and psychological constructs. It is therefore probably the most important stage and the one that requires the most care and effort. We will zoom in on small sections of the data at a time, just two or three lines for example, and work closely with these. We will continue to use the first two principles of DP — how discourse is constructed/ive and how it is situated — and in this stage we will also add in the third principle: how discourse is action-orientated. This stage of analysis, then, is the process of explicating social actions: of showing what actions are accomplished through discursive practices, how they are accomplished, and how psychological business is managed in the process of doing these actions.
So what is a social action?
Social actions are the kinds of things that we do in talk and interaction all the time: make requests, ask questions, flirt, discuss the weather, complain, praise someone, make promises, and so on. We are never ’just talking’; there is always something else going on. Sometimes these social actions are referred to as the ’functions’ of talk, though the term function suggests something a bit more rigid and formulaic (as if by using word X then function Y will be the resulting outcome). By using the term social action instead, the emphasis is on the combination of discourse and the interactional context (including who is speaking and how things are spoken) and the subtle ways in which actions are enacted across a number of turns in talk. In other words, it is not just that a single word can have a particular function, but that sequences of words and phrases, used when people are interacting with each other, together combine to perform a social action. In other words, the social action is the outcome of interaction.
One of the skills you will learn in doing DP, then, is to be able to recognise the kinds of actions that frame our talk and social interactions. What you will see is that social actions are produced through the subtle and intricate ways in which we talk and interact with other people. For example, just saying ’I do’ or ’I will’ does not make a marriage; these words must be said in the right context, with the appropriate people, and following a particular script. As such, it is not possible to identify the actions in advance of the analysis. We cannot simply point to the transcript and say, ’look, here’s a complaint’. The words on their own are not actions: they become actions through being used in a particular context. While we might have an intuitive sense that someone might be trying to persuade us, or flatter us, or blame other people, we need a rigorous analytical approach to help ’unpack’ the actions in talk and to examine how psychological matters are being invoked as part of these actions. If we break this process down a little, to examine talk in terms of its component parts, then we can start to identify the actions that are being performed and how these are put together. Examining the discourse in terms of ’social actions’ can take a bit of practice; it requires a different approach to talk than we are probably used to and being an expert at talking doesn’t make us immediate experts in analysing talk. So we need to analyse the interaction using participants’ orientations — to examine how they make sense of each other in talk — not our categories as analysts. For this reason, then, identifying social actions is our goal, not our starting point.
In order to help us to analyse discourse in this way, DP makes use of what have been referred to as ’discursive devices’. These are features of discourse, ways of talking and writing, that are recognisable and recurrent across different interactional contexts, and which help to perform social actions. These devices are the tools of DP, as it were, to enable us to examine the discourse as discourse. They shine a spotlight on particular parts of discourse and are particularly useful in terms of examining social actions and psychological concepts. For example, a footing shift can mark the change in speaking as an individual to speaking on behalf of oneself and an/other person/people. Sometimes this can be achieved through using a different pronoun. I have done this a few times in this book, for instance, when I move between talking about what ’we’ will do in a chapter, to ’your’ data, and also to personal accounts of what ’I’ have found to be useful. The point with discursive devices, then, is that they can help us to use the DP lens, to make the shift from describing our data in everyday language, to analysing it using the DP framework. Since the initial development of DP in the late 1980s and early 1990s, researchers have been slowly adding to the list of devices (see Table 6.1), and so there is every chance that new researchers might identify new devices that can be used in subsequent research. You never know, it could be you.
For now, though, let us look at the most common devices used in DP research. These have been broadly separated into three groupings in Table 6.1: basic, intermediate and advanced. The reason for separating them is that many new researchers to DP can be overwhelmed by the list of devices at first, and can have trouble knowing where to start. We will start, therefore, with the basic devices. These are the ones that are a fairly regular occurrence in everyday talk and interaction, and therefore most likely to be the ones to identify first in your data. The intermediate ones are less common but should be straightforward to identify when they are present. The advanced devices are a mixture of features of discourse but most of them require more experience of DP to be able to analyse data using them effectively. Think of these as your power tools, to be used once you’ve become more comfortable with the way of doing DP and when you’re ready for something that requires more skill, but which can also give you extra analytical insight.
Now, the trick with using the devices is to examine them while also paying attention to the constructed and situatedness of the discourse. Consider how and where they feature in the data, as much as you note which devices are being used. So we are not just ’spotting’ the devices (see ’pitfalls’ sub-section), but instead are using them as a means to interpret and analyse the data and provide an explanation of what is going on. As with stage 2, you can work on your data transcript on paper or electronically. Rather than working through line by line as we did in stages 1 and 2, however, focus on a few lines at a time. This will enable you to examine how different devices enable specific social actions to be accomplished, and to work across the interaction as a whole. You won’t need to use all of the devices and, as noted above, it may help to focus on the basic devices first. What you can do, then, is to identify any of the devices that appear to be prevalent in the section of data that you are focusing on. Your annotated transcript may then look something like Figure 6.3.
The first thing we might notice in the extract in Figure 6.3 is the three assessments (lines 3, 5 and 8). If we consider how these assessments are formed (in the form of ’I like X’ or ’I don’t like X’), they are subjective assessments; that is, they construct the assessment in terms of a personal reaction rather than attributable to the food (e.g., ’I don’t like steak’ rather than ’the steak is horrible’), because of the use of a personal pronoun (’I’). All three assessments also refer to a category of food rather than a particular item: ’steak’ or ’fish’ rather than ’this steak’ or ’that fish’. It is worth noting that immediately prior to this extract, Lesley had been commenting on how much fat was on each of their pieces of meat, and this may be what Bob is referring to on line 1 (i.e., ’quite a lot of’… fat). The one-second pause on line 2 also suggests trouble, because it is a reasonably long pause (anything over one second long is quite a long time in everyday conversation). As Bob appears to stop mid-sentence, in that he doesn’t quite finish what he was saying (line 1), the pause works to ’hold’ his turn. So Bob’s negative, subject-side assessment, category-focused assessment (’I don’t like steak’) is then situated following this trouble-marked pause. One of the issues that we might investigate further is whether the social action being accomplished by this assessment is a complaint or simply a report of an internal state. These two actions can actually be combined, and so this is an early interpretation about what might be going on here. We do not have to come to any conclusions at this stage; what we are doing here is identifying possible social actions and the construction of psychological states. At this point, a possible social action could be a complaint and the psychological state seems to be food preferences.
If we look further at the extract, we can also note that Lesley provides an assessment on line 5 (in the form of ’don’t do either’), so this works as a second assessment in that it follows soon after an initial assessment by another speaker. It agrees with the first assessment so it is produced without an extensive delay and is unmarked by any hedging or accounting for the assessment (which we might expect if it disagreed; see Pomerantz, 1984). While Lesley utters this, Bob then adds elaboration to his assessment (lines 6—7), and because of the overlapping speech early in Lesley’s turn, we could argue that Bob is not responding to Lesley’s turn but simply continuing his own. Staying with the focus on Lesley’s turns, she then introduces another assessment (’I like fish’, line 8) followed immediately by a claim that ’we never have enough fish’. This is an extreme case formulation (ECF, see Box 6.2) because it suggests that no amount of fish would ever be enough. Produced together, these are an interesting turn in talk. They make relevant a different food preference from the one previously discussed (i.e., fish rather than steak) and, as such, shift the focus from what the family are eating at the present time to what they might eat (in either the past or the future). By prefacing the ECF with an assessment, however, Lesley invokes and makes relevant her claimed food preferences for the purposes of talking about the family’s food habits. In doing so, a psychological state (in this case, what someone ’likes’ or doesn’t like to eat) is used to perform a social action: a potential complaint that the family’s eating pattern does not match her own food preferences.
These are early, tentative interpretations about the data. What they do, however, is to illustrate how we can use the discursive devices (here, we used assessments, pauses, and an extreme case formulation) to help develop our interpretation and understanding of the data.
We would then continue to work through our data in this way, bit by bit, and make notes about what kinds of social actions are being performed and which psychological states are being invoked, managed or negotiated within the interaction. You should aim to work through a reasonable amount of your data in this manner. There are no fixed guidelines about how much of your data set you should use at this stage, but you need to cover enough to generate a number of possible analytical issues. This could be, for example, around a quarter or a half of your data set. The aim of this third stage is to identify possible analytical issues to focus on — selecting one and finding examples of that in the whole data set in the next stage — so repeat this process of examining a small section of data until you have found a number of possible areas. For example, in the mealtime clip above, I identified a possible complaint about food, and also the use of food assessments, framed as individual food preferences. So these might be two areas (both of which deal with psychological issues: responsibility for food choice in the family and preference for particular foods) that could be developed into the focus of our analysis for this data set.
Box 6.2: The joy of ECF
The extreme case formulation (ECF) device is often incorrectly and over-used by new researchers to DP, primarily because it can appear to be quite obvious and common across numerous data sources. So this is a quick how-to guide on ECFs to help you use them with academic precision rather than wanton abandonment.
To summarise: ECFs are semantically extreme words (nouns, adjectives, adverbs) or phrases. They can be used for a number of social actions: to defend a claim, to support ’object-side’ (’out-there-ness’) accounts (i.e., attending to factuality), or when accounting for a behaviour. They can be used to normalise one person’s behaviour and pathologise another’s. They were first noted by Sacks (1992: lecture 3, fall 1964), with key analyses also provided by Pomerantz (1986) and then by Edwards (2000; see also Whitehead, 2015). An ECF is similar to hyperbole in linguistics, but distinguishable in terms of the way in which they work in an interactional context (Norrick, 2004). That is, ECFs only become ECFs through indexicality (see Chapter 1) in that they are hearably extreme; that is, they are typically orientated to as non-literal descriptions, as going to extremes. They are, however, what Edwards (2000: 352) terms ’factually brittle’. It would be easy to challenge and undermine them with a single exception to the extreme case they imply and yet they still maintain their rhetorical power even when they are challenged and are subsequently softened to a less extreme version.
An ECF is not, however, an assessment such as ’very good’, ’excellent’ or ’awesome’. These are positive assessments but on their own do not mark the target item as special in any way. Like any DP device, the ECF should not be treated as a ’thing’ that can be easily defined and then spotted in your data. They are ultimately a participants’ resource, something that is used to perform particular psychological business, such as defending a claim or showing investment in an account of events. When checking whether or not something is an ECF, then, you need to: (1) consider whether it is an end-of-the-line, semantically extreme version of events, and (2) examine how it features in the interaction: what social action is it accomplishing? How does it manage a speaker’s investment in their account? How is it responded to in the subsequent interactional turns?
I noted earlier that the third stage of analysis should not be rushed. You will need to go through various sections of your coded data corpus working through the analysis in this way, examining social actions and psychological business. Sometimes this can be helped with time away from your transcript doing another research task, such as reading or transcription, so that you move from one research process to another in an iterative manner and allow each part of the research to inform the other parts. Reading other published reports of DP empirical research can also help to shed light on your data, even if the topic area or data set used is very different from your own. I also find that reflecting on the data (and often, just one small phrase or section of the interaction, and ’playing’ that over in my thoughts) while I am engaged in other things can often bring new insights. Sometimes this happens when I’m travelling or engaged in a completely different activity, such as walking the dog. It can also happen when I’m talking to other people or listening to other people’s conversations on the train (yes, eavesdropping). The point is that analysing data isn’t just about time spent looking at your transcript. Since DP is all about understanding the discursive practices that make our individual lives relevant to social practices, then immersing ourselves in social settings can help us to stay grounded in the practical relevance of our data. You can find comfort, therefore, in knowing that you can still analyse when you are socialising and doing all the things you usually would do; just don’t forget to return to your data!
Stage 4: Focus on a specific analytical issue
By the time you reach the fourth stage of analysis, you will have read and worked through your data extract many times. Even in just a short section of interaction (say, 30 seconds for audio or video data, or a few lines of interaction in an online forum), you may be surprised at how many social actions are being performed, and how psychological business is relevant for topics that on first appearance might seem quite commonsensical or trivial. Once you have worked through some possible interpretations of social actions and psychological concepts in your data, you then need to develop a structure from your initial notes. This is the stage of moving from notes on your transcript to a focus on a specific analytical issue, the one which might then turn into the focus of your report or research paper. Remember that this is an iterative process and that we might follow up one issue only to find that it develops into another issue and we follow through a different part of our data. It is around this stage (and the next one) that I find things can start to get explosive: we may think we’re focusing on one issue (such as ’food assessments in mealtime talk’) that at first seemed quite specific and narrow, only to find it explodes into multiple issues (such as ’subjective versus objective assessments’, ’food preference talk’, ’assessments as part of food offers’, and so on). Each of these issues might also be refined further: how children rather than parents use food preference talk, for example. What we had thought, at first, to be a simple topic suddenly becomes much more detailed and intricate. This is one of the reasons why our research question can change when we’re using DP, and in the first stages of our research, you might find that your research question is a guide rather than a rigid focus. It is also one of the reasons why DP research can be so interesting and exciting to take part in; you never know where you might end up.
The best way to manage this stage of analysis is to write down all the emerging analytical issues in a separate word document or piece of paper. You might do this as you are working through stage 3, by noting down any features or aspects of the talk that appear to be particularly interesting or relevant for your research question. This might be quite an overwhelming list, and you might, like me, be sure that you will return to all of these issues one day (even if the lists are many years old and in your ’archived data’ folder). The point here is to see the potential in your data, to gain an understanding of the range of analytical issues you might focus on, and how these might overlap or relate to each other.
Once you have your list, then you can return to the research literature and your own research question. What was it that inspired this project in the first place? What were you looking for or aiming to analyse? What does the literature suggest is an important (or missing) area of research? Consider each possible ’issue’ in your list in light of these considerations. Which might be the most fruitful for analysis? Which ones, ultimately, are you most interested in or intrigued by? There are no hard-and-fast rules for which analytical issue to focus on, but remember that you will need to report these decisions and choices in the writing-up stage. So consider each one carefully, and make an informed choice on the basis of existing literature and your own understanding of the data so far. You should then identify one or two issues to focus on first.
Stage 5: Collect other instances of this analytical focus in your data corpus
You should now have an issue or focus for your analysis. In our worked example above, this might be ’how and where objective assessments feature in family mealtime talk’. Already, then, we have focused down our research question to something even more specific, and we should be prepared for this potentially changing again. The next step is to read and search through the rest of your data corpus to collect together more instances of this issue. Not all of your data will be fully transcribed, so work with either the transcripts or the original recordings. Just ensure that you are thorough and take a systematic approach so that you are as inclusive as possible. In the meal example above, I would identify all sections where a subjective (e.g., ’I like fish’) or an objective (e.g., ’the steak looks fatty’) assessment is used. As when coding the data, when you select out these sections always include a few more lines of interaction before and after the target section, erring on the side of caution and including more interactional context until you are sure what is and what is not relevant.
This is the equivalent of a second stage of coding and you may have to go back through your original corpus (rather than your first coded corpus) again to check that there aren’t any instances that you might have missed the first time. Also be flexible with your search as some instances might be ambiguous (some descriptions, for example, are treated as assessments in interaction) and others might use different words or phrases than you might expect, so just using the ’search’ function of word documents could miss valuable examples. You are aiming to compile another, new, word document that contains all of the transcripts for the sections of data that include your specific analytical focus. This may mean that you need to do a little more transcription at this stage — to represent those sections of the recording that you have identified in this stage, but which have not yet been transcribed — and as such that can help you to gain a fresh look at the data through blending the transcription/coding/analysis stages.
Stage 6: Focus and refine the analysis (aka ’how to stop analysing’)
At this final stage of analysis, you should have a word document — with associated video/audio files where relevant — which includes all of the possible segments of data that incorporate your specific analytical focus. Your task, then, is to repeat stage 3: to work through each data segment and analyse it in turn. You may find that some segments seem to follow a pattern, while others might seem very different from the rest (these might turn out to be deviant cases; see the section on validating analyses). If you are struggling to analyse or make sense of some segments, it is fine to leave these and come back to them later. You should aim to get a sense of how the whole of this new corpus ’works’: are there any patterns across the data segments? Is there coherence in how your analytical issue fits with the data? What new insights can you gain from analysing lots of examples of the same sorts of issues?
This is now the time to start writing out your analysis, even before you have ’polished’ it up and come to any conclusions. It may seem counter-intuitive to be writing before you have finished your analysis, but the writing process can help to refine certain aspects of the analysis, and can flag up issues that perhaps don’t really ’work’. In the same way that a story can be refined and clarified in the telling, so the writing up of the analytical notes can show more clearly what adds depth to the analyses and what is merely noting devices in the data.
Once you have gone through your list of data segments, making analytical notes and checking your analysis against the original recordings, then you can start to select some for writing up. There are no fixed rules about which ones to select: at this stage you should choose the data segments that featured the most analytical insights and which you found most interesting and engaging. Later on, there will need to be a more careful selection process to ensure that the data extracts we use in our written reports are representative of the coded corpus. For now, though, your aim is to start writing and not to spend too long at this stage pondering which segment to focus on. Copy and paste one of your data segments — the ones you have just been working on — into a new word document. You need to create a clear space to begin writing, to allow you to write as much as you want or need to, without cluttering up your documents with your analytical notes. Underneath the extract, begin writing out your analytical notes: what did you notice about the data? What social actions and psychological states are being managed? How are these achieved through the use of discursive devices? This stage of writing out your analysis can take some practice, and you will probably have to draft and re-draft many times. This is something that even very experienced researchers do. It is part of the process of refining our analyses and working through different interpretations and conclusions about the data. So do not worry if it looks very sketchy at first. What you are doing here is beginning the process of ’fixing’ your analysis and working towards producing your final report, which is where we will continue in Chapter 8. Box 6.3 provides a student perspective on this process.
Box 6.3: Transitioning from note-taking to dissertation writing
’At first, the analysis phase can seem like a really daunting process, especially when your data set is so lengthy. However, I soon found that I had much more confidence in formulating my analytical notes, purely because I was able to work from such an extensive data set. With in-depth data, I was able to make highly detailed notes, focusing on key patterns and consistencies among the data, as well as highlighting differentiations within the data corpus: something which was fundamental in my own study of social identity construction. Although the revision of notes can be somewhat monotonous and time-consuming, I found that perseverance allowed the transition into the actual dissertation writing to be far more manageable, as my notes were so enriched.’
Robert McQuade, undergraduate student (2010—2014), University of Strathclyde
Advanced DP analyses
The six stages of analysis outlined above are the scaffolding that will hopefully enable you to get started with DP and continue to develop your own competence and confidence in using this approach. At some point, however, you may be ready to advance a little further, and to tackle more complex issues that allow you to develop your DP skills in terms of both theory and analysis. The nature of discursive work is such that the field is developing all the time — empirical research is applying analyses in new areas, developing new discursive devices and pushing the boundaries of our understanding of psychological issues (see also Chapters 9 and 10) — so immersing yourself in the literature is crucial at this stage. This could be literature that not only relates to your topic area, but also to DP or other discursive or interactional research. This can be daunting at first in that there can seem so much to read. That is why this chapter has shown you how to apply the principles and analytical devices of DP without the need to do mountains of reading first. Many of you will, of course, have read and will want to read research in this area. Reading widely in DP and other research in your topic area will enable you to really develop your analytical skills and produce a much better report at the end.
While reading (lots of related research) is the crucial first step to developing advanced DP analysis, you will also need to examine areas that the previous stages do not address in detail. There are potentially many such areas, of course, and critics of DP might well have an extensive list of their own. Areas might include, for example, the boundaries between the individual (as psychological, physical, emotional, for example) and the social or regional vocal accents and how these might be attended to interactionally. Below are three areas relating to advanced analytical issues that might help inspire you to develop your own analyses further.
Subject-object relations and embodiment
As we saw in Chapter 1, DP emerged out of a concern with the cognitivism of much of (social) psychological research, and as such the early studies in DP often focused on core aspects of cognition research: attitudes, memory, attributions, emotions and so on. In more recent years, however, researchers in a number of different research disciplines have been considering the relationship between people’s physical bodies and the world around them.
This kind of work is sometimes referred to as embodiment or embodied action (the state of being in a physical body and the related processes and practices that this involves) as well as affect (emotional reactions or experiences). This is an intriguing area as it requires us to re-consider the inside/outside dualism of psychology, and with how certain aspects of the physical body (such as emotions, pain, hunger) are treated as purely visceral states and separable from discourse or social worlds. It also relates very closely to the issue of subject-object relations: the ways in which discursive practices create a separation between people (subjects) and things in the world (objects). See Chapter 9 for suggested DP reading on ’emotions’ and Chapter 10 for future developments of DP in this area.
Gestures and multimodality
Given its close alliance with conversation analysis, DP has also increasingly drawn on the use of technology to capture the visual aspects of interaction. It is easier, for instance, to record gestures and facial expression as well as the interaction between people and physical objects in the environment. This has opened up challenges to transcription (see Chapter 5) and of how we might analyse the extralinguistic features of interaction. The DP devices, for instance, rarely refer to aspects of interaction that are not primarily discursive. So there is an opportunity here for DP to develop in areas that address the multimodality of psychology in everyday life; of how our social actions are not just about what we say, but how we say these things in a specific environmental context. Even online interaction involves physical objects and the management of different identities in offline and online settings. How then might your analysis deal with other aspects of the social setting that are included in your data?
Change and temporality
Research tends to be treated as a ’snapshot’ of time: a static moment. Yet DP has the capacity to offer a much more fluid and longitudinal approach. We could consider, for example, the issue of individual change versus consistency. Is a person treated as being the same over a lengthy period of time, in terms of their identity or behaviours, for example? How is consistency versus change treated as being more or less accountable in relation to certain topics or issues? How do discursive practices change over a period of time? Given that DP is based on the assumption that discursive practices are produced within a particular interactional and social context, then we might argue that discourse is context-dependent rather than person-dependent. This has sometimes led critics to suggest that it leaves no space for the notion of personality or consistent traits; that people might be observed using the same phrases or ways of talking across different interactional settings. By considering discursive practices in terms of processes rather than static objects, we can offer sophisticated analyses of everyday social life and the management of psychological business.
Validating the analysis (aka ’how do I know when I’m doing it right?’)
By this stage, your analysis should be developing nicely and you probably have a large folder on your computer with video/audio files, transcripts, coded transcripts and analytical notes. So far, so good. You may, however, have been wondering about something along the lines of ’how do I know when I’m doing DP correctly?’ and ’how do I know that I’m not just making this up?’ These relate to issues such as the quality of the analysis (how good it is) as well as the broader issues around providing evidence and justification for your analyses. These might collectively be understood in terms of how we might validate or warrant our analysis, to ensure that it can stand up against scrutiny.
From a positivist approach, these issues would be referred to as reliability and validity. These terms (and the way they are used) rest on the assumption, however, that there is an objective world against which we can measure our own analytical claims. One of the things you will have noted about DP is that it is a relativist approach, based on the premise that there is no absolute ’truth’ or single version of reality. As such, when we refer to validity we are not referring to whether or not our analysis is ’correct’ or whether it is the only way in which the data might have been analysed. Instead, validity should be understood in terms of how grounded our analysis is in the data and the phenomena we are researching. In other words, our analysis must provide a coherent, plausible and insightful interpretation of that data. Does our analysis stay focused on the object we are investigating? Does our analysis provide an authentic representation and interpretation of the data?
A similar rationale applies to the concept of reliability. In positivist terms, reliability might refer to the extent to which others could replicate our research and find the same results or come to the same conclusions. It relies on the notion that there is a stable world (reality). With discursive research, however, no data are ever the same. Even if we choose an identical situation, with the same participants and words spoken, there will always be subtle variations in talk that can have important implications for the function of discursive practices. So when we refer to reliability in DP terms, we are concerned here with the extent to which others might come to the same conclusions, on examining our data and analyses. In other words, is our analysis convincing to others, and does it provide a way of understanding the data that goes beyond our own personal interpretations?
Like other relativist, critical and/or social constructionist approaches to research, DP uses a different set of criteria for warranting analytical claims, which is consistent with its theoretical and epistemological underpinnings. There are five key procedures and techniques we can use to ensure that we are staying true to our data: (1) transparency in methods, (2) participants’ orientations, (3) coherence and deviant cases, (4) identifying by-products and the consequences of discourse, (5) analytical insight and contributing to the literature. These procedures and checks enable us to ensure that our DP analyses are valid and reliable and we will discuss each of these in turn here. See also Box 6.4 for a checklist to help you validate your own analysis.
Transparency in methods
One of the most important things we can do as researchers is to be clear about our procedures, both at the stage of conducting the research and in the reporting stages. This means that we should be careful in the decisions made at each stage of data collection and analysis. Who might our participants be? Where will we collect the data (and where will the cameras be located)? How much data will we collect? Each decision has consequences for the quality of analysis and our research conclusions, so it is important that we take our time to carefully consider all options, and ensure that these are aligned with our research questions and theoretical position. Reporting these methods clearly — even if at the reporting stage you will not need to note each minute detail of decision-making (see Chapter 8) — is important for what is termed ’transparency’. This means that our procedures are clearly set out so that other people can understand exactly what we did and why. At the reporting stage, this also means being transparent about our data: this is why we include transcripts in journal articles, books and presentations. If we are making analytical claims about the data, then it is important to be able to make visible those claims using examples from the ’raw’ data. This is why transcripts are included in published reports and, where possible, still images from videos or sound files on websites. As technology develops, so the opportunities to have closer links between the research data and the reader will hopefully increase. Including transcripts in published work also provides opportunities for ’reader validation’, where readers can check the validity of the interpretative claims being made, as sophisticated users of discourse themselves.
Participants’ orientations and next turn proof procedure
One of the things that distinguishes DP from other discursive approaches is a concern with how discursive practices have particular actions in specific interactional contexts. This means that we are not trying to identify the abstract ’meanings’ of words or provide an expert diagnosis of what someone is saying. Instead, the aim is to examine how participants (i.e., the people whose discursive practices we are analysing) themselves orientate to and make relevant different interpretations of discourse. This is what is known as an ’emic’ approach: taking an insider’s, rather than an outsider’s, perspective. While we are not aiming to get ’inside the participant’s head’, we should focus on analysing discourse in terms of how people within the interaction make sense of it. For example, if we are not sure about our interpretation of something (in Extract 6.1 above, is Bob’s ’I don’t like steak’ utterance a complaint, for example, or a taste assessment), then we should look to see how it is treated by the other participants, either in the next immediate turn or in the unfolding interaction soon after. This is what is referred to as the ’next turn proof procedure’, where the ’proof’ of the interpretation is based on what happens in the next turn. The point is that we should not look to validate our analytic interpretations in terms of whether we think it is correct or ’sounds right’; instead, we should look to see how our analysis is supported by the data and in the way in which participants make visible their interpretation of each other’s talk.
Coherence and deviant cases
Your analysis should be coherent: it should be clear and understandable, with each part of the analysis fitting together in an organised way. An analysis often has many parts with each part focusing on a different aspect of the discourse (such as the structure or content of the interaction). Separately, they may not add much insight, but taken together they should provide a coherent narrative, an interpretation of the data that is consistent, clear and structured. We might think of the analysis like a jigsaw puzzle, with all of the separate pieces fitting together in just the right way to provide a bigger picture. It is not so simple as that in practice, of course, but the analogy highlights the need for us to think of our analysis as comprised of many parts, with each part being important, but not sufficient on its own (see ’device spotting’ in the ’pitfalls’ section). You should therefore be thorough in your analyses. While you may not transcribe all of your data in Jefferson detail, you should ensure that your analysis fits with all relevant sections of your data (i.e., the coded corpus) and not just a small selection.
As you conduct your coding and analysis of the data, therefore, you are likely to come across sections of the data that do not fit the pattern you have identified or which do not seem to be coherent with the rest of the analysis. These are known as deviant cases (similar to the notion of ’outliers’ in statistical terms): they are instances in the data that do not fit the pattern or analytical interpretation that you have just made. On first glance, then, you might think they are a problem and we need to ignore them (or worse, pretend that you did not see it in the first place). Do not be fooled, however, as the deviant case is actually a really useful analytical tool and can often be the saviour of your research. What the deviant case does is to mark some exception (a deviation) to the analytical pattern or observations you have previously made, but in doing so it can often confirm the validity of your analyses. So here we see the beauty of the deviant case: it can give your analyses a whole new strength. An example of a deviant case might be in terms of structure of the interaction (such as if someone does not return a greeting when you say hello) or about the content of the interaction (such as if someone uses a very different rhetorical device to support their account of their behaviour).
Identifying by-products and the consequences of discourse
In identifying a pattern or rhetorical strategy in discourse, we then examine the function(s) that these serve. For example, using a subjective category assessment (such as ’I don’t like steak’) can serve the function of turning down the offer of food without making any comment about the food itself (the refusal here is apparently based on our taste preferences, not a problem with the food, for example; though it may, of course, open up a different kind of complaint about not attending to personal preferences). While discursive practices have particular functions in interaction, however, they can also produce by-products or introduce new problems or issues that need to be dealt with. This might also be likened to the idea of an emergent property, which is a property of a system as a whole, but not present in the component, separate parts. In the example above, a by-product of using a subjective category assessment is that it suggests a fixed food preference that is unlikely to change. This then creates a dilemma for the person should they find themselves eating steak on another occasion (whether intentionally or not); how do they then account for their consumption? So identifying one function of discourse can often reveal other by-products. Identifying these by-products and the consequences of different discursive practices can be another way in which we can ensure our analysis is valid and coherent.
Analytical insight and contributing to the literature
DP analyses should provide insight into the data and phenomenon that you are investigating; it should add something new and different, provide an alternative way of looking at the phenomenon. You should not, therefore, be able to come to the same analytical conclusions if you had not first undertaken DP. This is the distinction between description and interpretation; it needs to provide another layer of understanding that could not have been reached by other means. When you are starting out in DP analysis this might seem a tough challenge, to add something new and insightful. Remember, however, that your data are unique (even if others have conducted research in a similar area), and that there is always something that you can add to the literature. Your analysis should also, therefore, be grounded in other research, whether related by topic or other discursive work. It should not be treated as isolated or separated from other work. It should refer back to earlier research and demonstrate how you are building on that research (whether by providing additional insights or alternative interpretations). In that way, you can ensure not only that your work adds something new and provides a fresh insight into a phenomenon, but also that it adds to a growing body of research, of which yours should now be part.
Box 6.4: Validation checklist
· Do the methods (of data collection and analysis) match the research question?
· Does the research connect with or adequately refer to existing research in this area?
· Is the data collection procedure clearly described and documented?
· Have ethical issues been appropriately considered?
· Is the analytical procedure clearly and fully described?
· Is the selection of data/extracts clearly justified?
· Is the analysis systematic and organised?
· Is there appropriate discussion of issues in the data with reference to extracts?
· Are any deviant cases (where relevant) discussed in relation to the analyses?
· Is the analysis coherent and insightful?
· Do the conclusions align with the analyses?
Analytic pitfalls to avoid
One of the daunting aspects of doing any analysis is the fear of not doing it ’right’ or making a mistake. This is the equivalent of falling off your bike or crashing your car, if we use the analogy of learning DP as similar to a craft skill (we’re using the word ’craft’ in its loosest sense here; dropping a stitch — the knitting equivalent of falling off your bike — doesn’t quite have the same gravitas). After the first lesson in how to get started (stage 1) and your second lesson in how to stop (stage 6), your third lesson is what to do if you do fall off and how to get back on again. Knowing what can go wrong won’t necessarily prevent you from making these mistakes, but it should make you more aware of being able to recognise these as mistakes and so put you in a position to be able to fix them for yourself. Besides which, it won’t be as catastrophic as crashing a car, so the odd mistake as you practise will not do you any harm.
With DP (and other forms of DA), there are a number of pitfalls that, with a little practice, we can avoid. Antaki et al. (2003) noted the following six pitfalls, and these neatly encapsulate many common problems that can be encountered when starting out in DP analyses:
1. ’under-analysis through summary’
2. ’under-analysis through taking sides’
3. ’under-analysis through over-quotation or through isolated quotation’
4. ’the circular identification of discourses and mental constructs’
5. ’false survey’
6. ’analysis that consists in simply spotting features’
We will use Extract 6.2, taken from an interview with a first-year university student (Louise) about friendship (see www.heacademy.ac.uk/resource/tqrmul-dataset-teaching-resources-louise%E2%80%99s-interview for access to the video data), to illustrate each of these pitfalls.
Extract 6.2: Friendship interview
Pitfall (1) is where you might find yourself describing or summarising the data — perhaps even pointing out some features of the interaction — but without adding anything new or providing any interpretation. For example, we might say of Extract 6.2 that ’Louise notes that she tries to make an effort with her flatmate and that she and her friends are quite bubbly and outgoing’. We might also point to the careful way in which Louise appears to be talking about how she has spoken to her flatmate, with a few short pauses and rising intonation. In doing so, we aren’t providing any analysis because there is no interpretation being made, no discussion of psychological states or social actions being performed in the talk (the description of Louise’s efforts do not count as a social action because these are things that happened prior to the interview and the discourse we are analysing).
Pitfall (2) is what can happen if you find yourself taking a moral or political position on what is being said by participants. There may be a temptation to make a judgement or assessment of the discourse; to either align or distance yourself from it. With Extract 6.2, we might say that Louise is right to try to include her flatmate or that she sounds like she is being a good friend. In data that are more emotionally charged — such as when dealing with traumatic or dangerous social encounters, such as reports of abuse or recorded interactions where people are being physically threatened — the stakes can feel higher and we may want to take a moral or political stance in order to defend someone or argue against a political regime. It is appropriate to make choices in the research that you conduct, to deconstruct racist rather than anti-racist discourse, for example. You can apply your analysis to support a particular issue in social policy or public debate, and in that sense to use your analyses for a political goal. But these decisions should come before and/or after the analysis; the analysis itself should be balanced and free of any attempts to ’spin’ the analysis in a particular direction.
Pitfall (3) is where you provide more quotes than interpretation (it is often recognisable by large sections of direct quotes and very little written analysis) or else provide a quote and let it ’speak for itself’. This is tempting to do when you are very new to DP and perhaps do not have the confidence to make your own interpretation of the data (or you are worried about misinterpreting or reading ’too much’ into the data). In Extract 6.2, we might cut-and-paste parts of the extract and then present this as ’here Louise accounts for how she tries to include the other flatmate in group activities’.
Pitfall (4) is perhaps one of the toughest challenges for new DP researchers and there are two parts to this. The first is that, in rushing to identify a ’discourse’ or pattern in the data, you might find yourself listing the features of a pattern or discourse and then claiming that you have identified a pattern and the features that define it. In Extract 6.2, for example, we might say that people justify their behaviour around friendship in terms of individual characteristics (being ’quiet’ or ’bubbly’, for example), and that being quiet is important in friendship because Louise made reference to this issue. In doing so, we are identifying a possible pattern and saying that it exists because we have found it; hence it is a circular argument. Similarly, we might make inferences about the cognitive or emotional state of someone through reference to their use of cognitivist (e.g., liking something) or emotional (unhappy) talk. Not only is this a circular argument in that it provides no evidence to support these claims (it only states the conclusion as if it were the evidence for the conclusion), but it also makes the assumption that talk is a direct link to cognition, which goes against the theory underpinning DP (see Chapter 1 for a detailed discussion about this).
Pitfall (5) is what can happen if you extrapolate too far from your findings; if you notice something in your data and then interpret this as an indication of what happens in other settings and to a broader context. With Extract 6.2, for example, we might say that it is important to make an effort with new friends otherwise they will feel excluded. While we might want to, and think it important to, show how our research has a wider relevance (and indeed it does), making claims that go beyond your data set is not the way to do this. What we need to do instead is to show how our analysis provides an insight into a small part of a bigger picture; one study cannot find all the answers, but it may well provide some answers and a handful of more important questions yet to be explored.
Pitfall (6) is similar in some ways to pitfall (1), but here you might find yourself spotting the discursive devices and showing how and where these occur. This can be very satisfying when you are starting out in DP: ’Look — there’s an extreme case formulation’, or ’I found a three-part-list!’ All well and good for practising your skills at focusing on DP devices and for training yourself to look at interaction with the DP lens, but this is not in itself analysis. In Extract 6.2, for example, you might have spotted an example of direct reported speech on lines 12—13: ’how are you finding it here’. Just spotting this, however, does not tell us anything about how it functions in this extract; to add more, we might analyse it in terms of how it constructs Louise as a caring flatmate, and as if this were just one example among many that illustrate this as being the case. You need to take the next step, therefore, to examine how these devices are used in the interaction to accomplish particular social actions and make relevant specific psychological concepts.
So beware of the pitfalls, and don’t despair if you fall into a few along the way. This is very common and is not a reflection on your skills or competence with DP. Even proficient researchers sometimes get caught out. It makes us human. The main thing is that if (or when) you do fall down, dust yourself off and keep going. Your analysis will be all the better for being able to recognise when you are going wrong.
Holding a data session
One of the most effective ways to become competent and confident as a DP analyst is to take part in a data session with other researchers. Ideally, do so on a regular basis if possible, whether this is weekly, monthly or even just a couple of times a year. This will provide not only the intellectual support of working with others in your field, but also exposure to different ways of interpreting data and enabling you to refine your own analytical skills. Ensure that some of those taking part in the data session are familiar with, or experienced in, discursive research (and DP specifically); at least one other person needs to be competent so that you can stay focused on discursive analyses. It can also be helpful, though, to have a range of perspectives and levels of competence in a data session. New researchers often come with fresh questions and different ideas from those who perhaps are more familiar with the literature. Trying out your analytical ideas in a data session is a great way to develop your skills and competence. You can test out your initial observations among a group of people who are similarly there to discuss the data, without fear of being ’wrong’ or looking stupid. So plan a data session at the early stages of coding and analysing your data, as they work best as a means to explore different possible aspects of your data. If you wait until you have almost finished your analysis, you are likely to be less open to new ideas. Data sessions can also be valuable not just for the person who brought the data, but for the other group members as well. It enables you to hone your analytical skills and you never know when you might gain insights to your own data through examining someone else’s data. Box 6.5 includes suggested guidelines for holding a DP data session.
Box 6.5: Suggested guidelines for holding a data session
· A data session might involve between two and twenty people, but between four and ten is optimal, to enable everyone to be able to take part in a full discussion of the data.
· Agree on the language to be used in the data session in advance if it is an international group or involving visitors from other countries.
· The data session should focus on just one piece of data (usually provided by one person or group of people, if working collaboratively), to stay focused on a particular issue.
· Prepare sufficient transcription to cover around 2—4 sides of paper (double-spaced, wide-margins). Any more than this and it can be difficult to discuss it all and to collectively stay focused on certain sections of the data. Bring paper copies of the transcript for everyone in the data session; circulate electronic copies in advance if you can. Without the transcript, it can be very difficult for participants to make any comments on the data or to examine the detail of the discourse. If confidentiality is an issue, you can request that all transcripts be returned at the end. Regardless, everyone should be respectful of the data and not use them for other purposes.
· If the data are taken from a video or audio recording, the clip should be played and shown on a large screen if possible (i.e., through a data projector), or at least on a laptop or tablet that can be seen and heard by everyone.
· Data sessions might be structured in many ways, but a typical session could run like this (plan for about a 1.5 to 2 hour session):
1. Play the video or audio clip (where possible) and read through the data. Play the data three or four times to enable people to have time to read and get familiar with it. This might take around 10—15 minutes.
2. Allow time for people to make notes individually, in silence, to give time for ideas to develop. This might take around 15—20 minutes.
3. Discuss the data by, for example, making analytic observations, comments about issues that might be explored further, or noting similarities with other data sets/research settings. You might prefer to allow people to speak by going round each person in turn, or raising hands to allocate a turn-taking order. Or you may find that an unstructured discussion works well enough (this can be the case if you have only three or four people present, for instance). Decide at the start of the data session how you will structure this discussion, and on any etiquette that you wish to follow as a group.
· Encourage everyone to comment on the data and offer analytical suggestions, and use these as either stand-alone comments or as starting points for discussion on a particular issue. The data session is not a time to critique or assess other people’s analyses; no one should be judged or have their contributions treated as less valuable than others’ contributions. Similarly, no one should feel obliged to speak if they are not confident to do so or do not have anything to add to the discussion. Data sessions at their best are collegiate, supportive and intellectually stimulating settings. Refreshments help — sharing food can help to lighten the atmosphere and provides a good distraction if the discussion is slow to get started.
· Analysing DP is a skill to be learnt rather than a recipe to be followed, but there are six interrelated stages that can be used to help scaffold the analytic process.
· Good analysis takes time: don’t rush it!
· The stages of transcribing, coding, analysing and writing up your data are iterative and interwoven; be prepared to go back through your data as your analysis develops.
· Gain confidence with DP analysis before you move on to the advanced issues.
· Validating the analysis means checking for quality, rigour and coherence.
· There are common pitfalls when analysing data and these can help you to distinguish ’good’ from ’poor’ analyses.
· Working in a data session with other researchers can be a really valuable way of developing your analytical skills.
Edwards, D. (2005). Moaning, whinging and laughing: The subjective side of complaints. Discourse Studies, 7(1), 5—29.
Wood, L. & Kroger, R. (2000). Doing discourse analysis: Methods for studying action in talk and text. London: Sage.