Recommend a Qualitative Design

Everleigh

DiscourseAnalysisApproaches.pdf

Home >Psychology homework help >Recommend a Qualitative Design

C H A P T E R 4

Discourse

Analysis

Approaches Amanda Holt

Introduction

This chapter is about approaches to discourse analysis. Discourse analysis is a methodological approach that aims to highlight the ways in which ‘knowledge’ is socially constructed. It aims to expose the implicit values and hidden assumptions that underpin taken-for-granted knowledge – knowledge that legitimises existing institutional practices that may be considered unjust. This chapter begins by briefl y describing the historical and intellectual context of this methodology, and outlines the assumptions about the world and the assumptions about psychological knowledge upon which it is based.

The chapter then discusses two distinct theoretical approaches to discourse analysis that are used in psychology: ‘Foucauldian discourse analysis’ and ‘discursive psychology’. The chapter explores the commonalities between these two approaches and considers the ways in which they might be combined. The latter part of this chapter includes detailed descriptions of how to use discourse analysis. Examples throughout the chapter illustrate the differences in the fi ndings reached by each approach. Practical issues such as recruitment and sampling, data production, analytical coding and presenting research are discussed. The chapter concludes with a summary of some common research uses and applications.

History (or histories?)

It has been suggested that there are at least 57 varieties of discourse analysis (Burman & Parker, 1993). These are encompassed within a number of broad the- oretical traditions that foreground language, such as social semiotics, ethnometho- dology and conversation analysis (Gill, 2000). The method itself transcends subject disciplines: you will fi nd it being frequently used in fi lm, media and cultural studies, languages and literature, politics and law, the visual arts, as well as more traditional social sciences such as sociology and criminology. Therefore, it is impossible to provide a coherent narrative that can outline ‘the history of the method’.

qualitative research methods - final.pdf 76 14/06/2011 14:07

Frost, N. (2011). Ebook : Qualitative research methods in psychology: combining core approaches. ProQuest Ebook Central <a onclick=window.open('http://ebookcentral.proquest.com','_blank') href='http://ebookcentral.proquest.com' target='_blank' style='cursor: pointer;'>http://ebookcentral.proquest.com</a> Created from ashford-ebooks on 2021-08-19 17:31:16.

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Ontology of the approach 67

However, within psychology, the 1970s is often pinpointed as the time when many psychological researchers were drawn to the discourse analytic methods that were already established within other subject disciplines. Furthermore, many historical overviews of social psychology (e.g. Burr, 2003; Richards, 2009) suggest this ‘turn to language’ was the outcome of an epistemological ‘crisis’, which resulted in disillusioned social psychologists actively searching for alternative methodologies. This disillusion was mainly fuelled by a dominance of experimental methods within social psychology – methods that, in the endless quest to isolate ‘variables’, tended to ‘bracket off’ the very social processes that social psychologists were interested in. However, in the spirit of this chapter, which is attempting to encourage students to question taken-for-granted assumptions, it is important to remember that this version of history is one of many possible versions. Over time, it has managed to achieve legitimacy as the history of discourse analysis in psychology. But this is not The Truth. It is one truth of many ‘truths’. This is an important point, to which this chapter will return.

Ontology of the approach

Discourse analysis is underpinned by a constructionist ontology. As such, it is at odds with, and attempts to challenge, the realism that underpins more mainstream research methods (such as quantitative experimental methods). Such mainstream realist methods tend to be based on the assumption that pre-existing ‘structures’ determine social life. Such structures might be assumed to exist ‘inside our heads’ as more psychological paradigms, such as cognitivism, might suggest. Or they might be assumed to exist ‘out there’ in the world, as more sociological paradigms, such as Marxism, might suggest. In contrast, constructionist methodologies (such as discourse analysis) make no assumptions about the social world, and instead aim to expose and highlight the constructedness of these assumptions. They also aim to question the implications of taking for granted such assumptions. (Hence the term ‘post-structuralism’, a term that is often associated with an intellectual approach that rejects, or challenges, the assumption of ‘pre-existing structures’!)

As well as making assumptions about the social world, mainstream realist methods also make assumptions about being a person. That is, they take as their (unquestioned) starting point an essential, rational and boundaried subject (or person). While in our everyday interactions it is perfectly reasonable to draw on common-sense assumptions about the world and the people in it, this does become problematic if we do the same thing in our research and fail to ever question the nature of what it is we are researching. One of the good things about constructionist methodologies is that they won’t let us take for granted alleged ‘truths’ about the world. And, once we start questioning these ‘truths’, and start seeing them as only one of many possible ‘truths’, then we are liberated to consider alternative ways of thinking about (or constructing) the world and the people in it. Constructionists would argue that it is here where research can produce real change – rather than the

qualitative research methods - final.pdf 77 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

68 Chapter 4 Discourse Analysis Approaches

superfi cial change that more realist research might produce (which, in the process, leaves underlying fundamental ‘truths’ unquestioned and, therefore, still in place).

All of this means that one of the defi ning features of discourse analysis is the way in which the questioning of taken-for-granted ‘truths’ is intrinsic to the research question itself. That is, the research question should always aim to be looking to uncover particular assumptions that we make about the world and to consider what the effects of it might be (see the research example box).

Given the centrality of constructionism to discourse analysis, the potential for its combination with other analytic methods will inevitably be shaped by the extent to which those methods are also underpinned by constructionist ontologies. For example, traditional grounded theory, such as that developed by Glaser and Strauss (see Chapter 2), assumes a realist ontology in that the knowledge produced is assumed to be grounded in the data that pre-exists the researcher looking at it. However, Charmaz (2006) has developed a grounded theory more in keeping with constructionist principles, with which discourse analysis methods would certainly be compatible. Similarly, the more constructionist approaches to psychoanalysis are also compatible with discourse analysis (for an overview, see Branney, 2007). Nevertheless, in some cases it may be very useful to combine discourse analysis with quantitative methods, if done appropriately. For example, a brief quantitative survey that looks at ‘stop and search’ patterns among different ethnic groups may be a useful precursor to interviewing police offi cers and using discourse analysis to examine the ways in which they account for their seemingly discriminative practices.

R e

s e

a rc

h E

x a

m p

le Developing a research question that is underpinned by a constructionist ontology Jane Ussher’s (2003) research focused on women’s experiences of pre-menstrual dysphoric disorder (PMDD), which was classifi ed as a psychiatric disorder in the DSM-IV in 1994. However, her research question did not aim to explore whether menstruation causes PMDD, as realist ontological paradigms (using experimental methods) might dictate. Instead, Ussher aimed to expose this notion as one particular and dominant ‘truth’. Therefore, she developed a research question that asked: ‘What are the effects of this particular hegemonic truth for the women who are the subjects of it?’ (i.e. for women whose experiences are medicalised and consequently constructed as pathological). Ussher also aimed to explore what possibilities for change may be enabled or disenabled by this particular ‘truth’ for the women in question. What this example shows is the way in which Ussher’s decision of whether to investigate this issue through a constructionist lens or through a realist lens is a political one, since each will have ‘effects’ by either maintaining or disrupting the current ‘truth’.

qualitative research methods - final.pdf 78 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Epistemology: why do discourse analysis? 69

In relation to Ussher’s (2003) work, then, she wants to look at the ‘effects of’ a ‘medical discourse’ that is dominant enough to shape the way that women understand their experiences. One could also speculate on the assumptions upon which this discourse is founded: the idea of Cartesian dualism (i.e. the mind/body split) and, within this, the idea that there is a causal relationship between biological changes and psychological effects. These are only ideas, yet they have been talked

However, within the fi eld of discourse analysis itself, there have been a number of discussions regarding the extent to which different approaches (e.g. Foucauldian discourse analysis and discursive psychology) can be combined. This is something to which this chapter now turns.

Epistemology : why do discourse analysis?

While realist approaches to social research view language as a means to access ‘The Truth’, constructionist approaches view language as constitutive of truth. This means that it is through language that meanings are negotiated and ‘realities’ are produced. In effect, nothing pre-exists language. Thus, our knowledge about the world is produced through the organisation of language and particular behaviours (or ‘practices’) into particular discursive formations that comprise ‘discourses’. Discourses serve to seemingly provide a coherent and credible ‘truth’. To take an example, ‘cognition’ – and the internal structures on which it is assumed to be based – does not pre-exist language: it is talked into being by the use of words such as ‘operation’, ‘perception’ and ‘higher/lower order processing’. It is also practised into being through the performance of controlled laboratory experiments, through a lecturer’s PowerPoint slide, which shows a fl ow chart with ‘memory’ at the top and the categories of ‘episodic’ and ‘procedural’ underneath, through the writing and publication of journal articles, and so on. That is, the way it is talked about and the way things are done are organised into a specifi c formation which constitutes a ‘cognitive discourse’. Similarly, ‘social class’ – and the external structures on which it is presumed to be based – does not pre-exist language.

Question

In what ways is a ‘discourse of social class’ talked into being and practised into being? It might help to think about the different ‘agents’ (e.g. social scientists – including university students! – doctors, politicians) and ‘institutions’ – (e.g. research institutes, social services, the police, mass media) that produce and reproduce particular discourses of social class. You might take this exercise a step further and think about which groups are privileged by these discourses and which are disadvantaged (and in what ways).

qualitative research methods - final.pdf 79 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

70 Chapter 4 Discourse Analysis Approaches

into being to such an extent that they are taken for granted. They are also practised into being by a woman visiting her doctor, or through the institutional practice of classifying PMDD within the DSM-IV. The seemingly coherent and credible ‘truth’ that emerges from this medical discourse is that menstruation causes psychiatric disorders. Such a ‘truth’ will have all sorts of implications for the women who are subject to them, and the medical institutions that gain or lose from them. And the fact that different groups of people gain or lose from the dominance of different discourses means that when we are thinking about discourses, we also have to think about power.

So, given the centrality of language to power relations in constructionist ap- proaches, we need a research methodology that similarly foregrounds language and power in social analysis. Discourse analysis does this, and two distinct approaches have emerged within psychology – one that arguably foregrounds power, known as Foucauldian discourse analysis, and one that arguably foregrounds language, known as discursive psychology. It is to these two approaches that we now turn.

Foucauldian discourse analysis

Foucauldian discourse analysis (FDA) is an approach associated with the work of Ian Parker (1992) and Burman and Parker (1993). It is derived from post-structuralism and, in particular, from the work of Michel Foucault and his conceptions of power (see Foucault, 1975/91; 1976/90; 1991). FDA aims to examine how ‘objects’ (things) and ‘subjects’ (people) are constructed in discourse and to explore what the effects of this might be for people who are subjected to them (hence the term ‘subjects’). One of the key ideas in FDA is the notion of subject positions. This term refers to the possible social locations that either afford or delimit particular ways of being a subject. For example, if you recently experienced a burglary, a particular subject position will have been made available to you: that of ‘victim’. Once this subject position is taken up, a number of ways of being will be opened up to you (such as access to victim support resources or eliciting sympathy from friends), but other ways of being will be closed down (such as sleeping soundly for a few nights!). This is one reason why there has been a movement to rename women who experience sexual violence as ‘survivors’, rather than ‘victims’ – because the ‘discourse of victimhood’ upon which the subject position of ‘victim’ is based enables some very limiting and disempowering ways of being.

A further example of this can be found in Ussher’s (2003) work, discussed above: the dominant medical discourse that ‘menstruation causes psychiatric disorders’ offers up very few subject positions to the women who experience bodily changes beyond that of ‘ill, unstable or mad’ (2003: 142). Such a subject position may be reinforced by (and may, in turn, reinforce) particular institutional practices, such as doctors prescribing medication or psychological treatment to such women. Thus, this subject position offers both rights (such as access to particular forms of ‘treatment’) and duties (such as permitting the medical regulation of one’s body). As

qualitative research methods - final.pdf 80 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Epistemology: why do discourse analysis? 71

this example illustrates, the subject positions offered may afford multiple and contradictory experiences and practices, and these may need to be negotiated carefully by the subject in question.

In order to enable analysis of this process, an FDA approach to data analysis asks the data a number of questions:

■ How do subjects utilise particular discourses that are available to them to construct their experiences?

■ What ‘subject positions’ are made available to subjects within the discourses that are drawn upon?

■ How do subjects negotiate these subject positions? ■ What ways of being do these subject positions enable or delimit? The role of

power is particularly explored in this fi nal question: who wins from this discursive process … and who loses?

Discursive psychology

A second approach within discourse analysis in psychology is commonly referred to as discursive psychology (DP), and is associated with the work of Jonathan Potter and Margaret Wetherell (1987; Wetherell & Potter, 1992). While FDA enables a more macro-level analysis, DP takes analysis to a more micro-level by focusing almost entirely on the immediate interactional setting that produces the data. In this sense, DP is particularly concerned with the ‘action orientation’ of discourse in its recognition that language is a social practice that has a performative function. In many ways, it focuses in on the third of the four questions asked by FDA outlined above (i.e. how do subjects negotiate subject positions, albeit specifi cally in the context of the communicative setting?).

Potter and Wetherell’s infl uential Discourse and Social Psychology (1987) takes particular interest in the role of accounts, which they identify as explanations of behaviours that might be ‘unusual, bizarre or in some way reprehensible’ (1987: 74).2 Researchers working in this area (not all would necessarily identify as being ‘discursive psychologists’) have suggested a number of verbal and non-verbal rhetorical devices that may enable excuses and justifi cations in a speaker’s explanation of their actions. These devices are fashioned from the speaker’s culturally available linguistic resources. A summary of some of these devices can be found in Table 4.1.

DP also identifi es wider cultural explanatory frameworks that are taken for granted as ‘truths’. These are known as ‘interpretive repertoires’ and are similar in

2 That is not to suggest, of course, that accounts cannot be rather more ordinary than this, but the point is that accounts nevertheless involve the negotiation of a subject position that is morally justifi able. The manoeuvres that are made to get to such positions are of particular analytical interest to discursive psychologists (they also tell us much about the moral context of the interaction).

qualitative research methods - final.pdf 81 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

72 Chapter 4 Discourse Analysis Approaches

fl avour to the notion of ‘discourses’ used in FDA. However, Gill (2009) argues that ‘interpretive repertoires’ are perhaps more fl uid concepts in that they tend to be more specifi c to the context of their use and are in a continual state of fl ux (in contrast, the FDA concepts of a ‘medical discourse’ or a ‘consumer discourse’ are more singularly encompassing and, as such, suggest greater rigidity).

The strength of a DP approach is its recognition that accounts will vary according to their situational context. Therefore a focus on the immediate data-producing setting is an important aspect of DP discourse analysis to illustrate how subject positions are not only grounded in wider material and institutional power relations (as an FDA approach would suggest) but are also ‘local, highly situated and occasioned’ (Wetherell, 1998: 401). For example, in an interview setting where a researcher asks a participant about her/his experiences, the participant will draw on a specifi c set of linguistic resources that are tailored to the researcher, the environment, the dialogue, the purpose of the interview, and so on.

Table 4.1 Rhetorical devices suggested by researchers interested in discursive analysis

Rhetorical device

Example and function Reference

Active voicing He said ‘ don’t do that …’ Increases facticity of account, establishes objectivity and rhetorical distance from account

Hutchby & Woof fi t t (1998)

Contrasting discourse

I was like this … now I’m like this … Emphasises transition by listing competing descriptions

Smith (1978)

Disclaimers I’m not being sexist but … Disclaimers anticipate (and reject) potential negative claims

Hewit t & Stokes (1975)

Extreme case formulation

It was ‘phenomenal’ Strengthens claims by taking claims /evaluations to their extremes

Pomerantz (1986)

Use of passive language

I found myself … Precludes possibilities of agency and choice

Abell et al. (2000 )

Use of ‘realise’ An interactional resource that suggests authenticity Edwards (1997)

Temporal markers

Then I … when you … fi rst I … Marks temporal relationships between events (of ten to signal developmental progression)

Shiffrin (1987)

Vague descriptions

Vivid details can easily be undermined: vague descriptions produce just enough material to sustain action without opening to at tack

Pot ter (1996)

qualitative research methods - final.pdf 82 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Epistemology: why do discourse analysis? 73

In order to enable analysis of this process, a DP approach examines how dis- cursive resources (including the use of rhetorical devices and interpretive repertoires) are used to ‘do things’ in a particular context and to examine their particular effects. In particular, the analytical process involves identifying such discursive resources in a text and looking at how they vary, and how they are consistent across different texts and within the same text (Potter & Wetherell, 1987).

Commonalities between Foucauldian discourse analysis and discursive psychology

There are clearly commonalities between these two approaches to discourse analysis. Both FDA and DP foreground the role of language in the construction of social reality by ‘de-centring the subject’. They also both emphasise the importance of refl exivity in the research process – that is, as a researcher you need to be ref- lexively aware of how your own cultural, social, political, linguistic and episte- mological location shapes your production of research knowledge. Refl exivity is consistent with an epistemology which recognises that knowledge claims are ‘ideological, political and permeated with values’ (Schwandt, 2000: 198) and the

Identifying ‘ interpretive repertoires’ in the sex and relationships articles in a UK women’s magazine

In her DP analysis of Glamour magazine sex and relationship advice articles, Gill (2009) identifi es three interpretive repertoires that she claims serve to privilege men and heterosexuality. These are the ‘intimate entrepreneurship’ repertoire, which draws on the idea that a successful relationship (constructed as a ‘goal’) is founded on strategy, planning and tactics (similar emotion-less ‘skills’ to those deemed necessary in the western workplace). The ‘men-ology’ repertoire draws on the importance of studying and learning about ‘what men want’ – in effect, making relationship-building an educational project that requires expertise. The third repertoire is ‘transforming the self’, which draws on notions of self-change, particularly in the fi eld of sexual and bodily ‘confi dence’ and ‘attitudes’.

Gill argues that these three interpretive repertoires ‘intermingle and co-exist’ (2009: 361) in the same text and work together to perpetuate not only unequal gender relations (in their promoting the servicing of men) and hetero-normativity (it is always men who are the relationship goal) but also a neoliberal ideology in the way that they impel women to survey and regulate their selves. However, as Gill points out, by each drawing on notions of women’s agency, choice and empowerment, the three interpretive repertoires work together to effectively disguise the rather questionable overall message of Glamour magazine.

R e

s e

a rc

h E

x a

m p

qualitative research methods - final.pdf 83 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

74 Chapter 4 Discourse Analysis Approaches

refl exive process involves continual re-examination of initial analyses and an explo- ration of alternative interpretations (Alvesson, 2002). It also means acknowledging that the researcher’s interpretation is a privileged one that silences possible others, and a consideration of the implications of this. However, while many researchers (e.g. Finlay, 2003) suggest that the use of refl exivity in social research turns a potential ‘problem’ (of subjectivity) into an ‘opportunity’, the use of refl exivity is not un- problematic. Its use has often been suggested as a way of hiding the lack of democracy that characterises most kinds of social research – after all, it is the researcher’s reading that is ultimately privileged (Burr, 2003). It is also important to recognise that refl exive insights are also subject to the same discursive processes as participants’ insights, and so any refl exive analysis must be recognised for the range of ‘functions’ that it may perform (Gough, 2003a). That is, it may tell us less about the ways a researcher’s professional and personal identity shapes research and more about the discursive moves made by a researcher to claim a moral subject position

R e

s e

a rc

h E

x a

m p

le Using discursive psychology (DP) to examine transcripts of children’s calls to a telephone helpline Hepburn (2005) analysed two transcripts from calls to the NSPCC helpline; the fi rst concerned a 12-year-old girl who reported concerns over the possible sexual abuse of her best friend. In this transcript, Hepburn identifi ed the use of ‘active voicing’ and the provision of vivid details as particular rhetorical devices used by the girl to persuade the counsellor of the facticity of her account:

7 She said ‘oh: I dunno what you’re 8 ta:lking about. (.).hh my little 9 [sist]er must ‘ave wrote that’ an 10 Coun: [°Mm°] 11 I says .hh (0.3) ‘like your- but I know 12 your handwritin:. what-who is the devil.’ (2005: 259)

Hepburn then identifi ed the way in which the caller justifi es why she has not told her friend’s mother about her concerns (because ‘he’ll hit ’er’ (2005: 260)) – an option that Hepburn suggests would be the ‘normal’ course of action for somebody located in this particular subject position (as a 12-year-old girl). Hepburn argued that the discursive analysis highlights how children are compelled to convince they are telling the truth and to justify why they are taking up valuable time and resources. In turn, this tells us much about the problematic institutional requirements of calling such helplines, which appear to demand more of children and young people, in terms of their credibility and motives, than they do of adult callers.

qualitative research methods - final.pdf 84 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Epistemology: why do discourse analysis? 75

for themselves (for example, they could do this by ‘refl ecting on’ their interest in social justice or their concern for the well-being of the participants).

FDA and DP also share a similar stance towards issues of reliability and validity. Clearly, the positivist meaning of these terms is not appropriate for research methods that are underpinned by a constructionist methodology, but there does need to be some way of evaluating discourse analytic work. Useful conceptual suggestions include ‘quality’ (Willig, 2008), ‘transparency’ (Coyle, 2007) and ‘coherence’ (Harper, 2006), concepts that have been suggested in more general debates about evaluation in qualitative research (see Elliott et al., 1999; Elliott, Fischer & Rennie, 2000; Madill, Jordan & Shirley, 2000; Reicher, 2000). Stenner (1993) suggests that discourse analysis produces a reading, rather than an interpretation, as there is no supposition of an ‘outside truth’ against which the analysis can be assessed. Thus, judgement of its quality must be in terms of its usefulness, rather than any kind of accuracy. This brings attention to the fact that the aim of discourse analysis is not to produce generalisable fi ndings that can establish universal laws of behaviour (as more mainstream psychological scientifi c methods do). Instead, it aims to highlight particular processes that are anchored to their context – and perhaps to provide a guide for political practice in the light of such fi ndings.

Differences between Foucauldian discourse analysis and discursive psychology

In terms of distinctions between FDA and DP, perhaps most apparent is their differing emphases: while DP focuses on the immediate situational interaction and how subjects adapt their talk according to its needs, FDA is more focused on the relationship between discursive formations and wider social and institutional practices. Thus, it has been argued that their differences mainly concern the different kinds of research question that they each address, rather than any profound differences in the practice of data analysis (Burr, 2003). As you may already have noticed, in practice there is little difference between ‘discourses’ in FDA and ‘interpretive repertoires’ in DP. Similarly, both concern themselves with subject positions and the ways in which moral contexts shape these. However, while some theorists (e.g. Parker, 1992) have suggested that these two techniques should not be

Question

You want to investigate the reasons why Members of Parliament have chosen their particular line of work, and have used ‘life story interviews’ to produce rich qualitative data that, among other things, taps into their account of their occupational history. This data could be analysed in a number of ways, but you have decided that discourse analysis is most appropriate. Would you choose Foucauldian discourse analysis, discursive psychology or a combination of the two? What factors drive your reasoning?

qualitative research methods - final.pdf 85 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

76 Chapter 4 Discourse Analysis Approaches

C A S E S T U D Y

Combining Foucauldian discourse analysis and discursive psychology

Depending on the topics of a particular research study, it can be useful to integrate both approaches. For example, my research aimed to explore parents’ experiences of their child’s involvement in offending (see Holt, 2009), their receipt of a Parenting Order through the courts as a result of this (see Holt, 2010a), and their subsequent attendance of ‘parenting classes’, which was a condition of the order (see Holt, 2010b).

A DP approach was essential to enable an understanding of how the specifi c research interview setting (which includes the questions that are asked, the mode of interaction and the researcher–narrator relationship) shaped the data produced. Consequently, each participant’s account needed to be understood within each situated and local (or ‘micro’) interview context, which necessitated the need for a DP analysis at this immediate level.

However, the institutional context of this research project made an analysis of wider power relations essential: Parenting Orders are disproportionately issued to mothers, lone parents and those with fewest economic resources. Furthermore, receiving a Parenting Order has real material affects: the parent is compelled to attend court, pay court costs and engage in parenting support sessions.

However, and perhaps most importantly, it is only through acknowledgement of this wider ‘macro’ context that the assumptions inherent within the DP approach become comprehensible. For example, Riley, Sims- Schouten and Willig argue that the DP approach assumes that ‘people want to appear in a good light, want to blame others, or that they want to disclaim responsibility’ (2007: 142). But it is only when the subject’s position is contextualised by reference to wider relations of power (i.e. having been made responsible through the courts and experienced blame through being labelled a ‘bad parent’) that such discursive actions within the immediate interview context become meaningful.

Questions for refl ection

1. What might be the problems with combining the FDA approach with the DP approach? How important are concerns over theoretical consistency when undertaking social research?

2. How might the usefulness of this particular research study be evaluated? How can this study facilitate social, political or personal change? Is it ethically acceptable for the researcher to decide whether and what kind of change is necessary?

3. What kind of topics/research questions might be inappropriate for combining FDA and DP? When might using an FDA or a DP approach singly work better?

qualitative research methods - final.pdf 86 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Methods: how to do discourse analysis 77

integrated, given their distinct theoretical foundations, others (e.g. Wetherell, 1998) suggest a more pragmatic approach that integrates both the need to attend to the discursive practices employed in the immediate setting and the wider social and institutional practices being drawn upon.

Methods : how to do discourse analysis

Now that we have outlined the theoretical context of discourse analysis, this section outlines the more practical aspects of employing discourse analysis. In particular, it discusses pragmatic issues including recruitment, sampling and data collection, interpretation and analysis, and the presentation of discourse analytic work. As the practice elements of FDA and DP are similar, they will be discussed together. Where there are differences, these will be highlighted.

Pragmatics

Recruitment and sampling

One of the strengths of discourse analysis is its applicability to a wide range of texts. Newspaper articles, policy documents, online texts (e.g. blogs, message boards, emails) and visual data (e.g. photographs, videos) can all be analysed using discourse analysis. However, given that discourse analytic approaches do not attempt to generalise from a sample to a wider population, it is not considered important to obtain data from a large sample. Indeed, given the level of detail that is required in the analysis, it is prudent (and perhaps more ethical) to avoid collecting data from a very large number of participants that is unlikely to be used. A rough guideline might be something like 20 interview transcripts or ‘cases’ elicited for analysis.

One issue that can arise when using discourse analysis is the ethical question of obtaining informed consent from participants. As Elliot (2005) acknowledges, outlining the nature of the research to participants becomes problematic once the research epistemology shifts from the ‘common sense discourse’ of naturalism to a more marginalised and complex constructionism. This raises a number of dilemmas: for example, should the researcher go into detail about the kind of analysis their data will be subject to? Given its shift away from the more ‘mainstream approaches’ that the participant is likely to expect, does the researcher have a duty to explain – at the risk of sounding opaque – that they are less interested in the participant’s experiences in any realist sense, than in their constructions of

Question

What might be the ethical implications of Hepburn’s published work (see the research example box on page 74), in which children’s and young people’s calls to a helpline were ‘translated’ into an exercise in identifying discursive resources in the context of conversational demands?

qualitative research methods - final.pdf 87 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

78 Chapter 4 Discourse Analysis Approaches

experiences, which will not be assumed to possess any real existence of their own? Does the research topic affect the importance of a researcher’s duty to ‘epistemologically explain themselves’?

Eliciting and gathering data

If interviews are going to be used, they do not necessarily need to be produced through face-to-face encounters; indeed, as I have argued elsewhere (see Holt, 2010c), in many ways discourse analysis lends itself very well to data produced from telephone interviews. This is because telephone interviews require the participant to fully articulate their responses since alternative means of communication (e.g. body language) are not available (and, in any case, cannot easily be produced within a typed transcript).

Regardless of mode, it is generally recommended that interview data is produced through unstructured interviews, since more structured interviews are likely to curtail both the detail that is required for a sophisticated analysis, and the incon- sistencies and contradictions that occur in talk that discourse analysis aims to highlight. Thus, an interview guide or topic list, which provides a few open-ended questions and/or discussion themes, is most appropriate. An open approach during interviewing is also necessary to enable participants to conceptualise and construct experiences in ways most meaningful to the participant (rather than to the researcher). Harper et al. (2008) also suggest that adopting a ‘devil’s advocate’ position in a research interview might facilitate the use of counter (and counter-counter) claims. That is, a more adversarial framework in interviews is likely to enable the excuses and justifi cations that are central to discourse analysis. However, such an approach could be problematic and should perhaps be considered in the light of the competing need to facilitate a positive social experience for the participant!

It is also useful to make notes immediately following the interview, using a research diary. This might include impressions of the interview, points of anxiety and any atypical incidents that occurred. This practice provides you with the opportunity to document any immediate analytical ideas that occur, as well as enabling you to refl ect on your own role in the interview process. You might also take this process a step further and ask participants for their responses to their experiences of the research process. Such a request might constitute one way in which refl exivity can be made a more democratic process by enabling the participants as well as the researcher to take part in refl exive processes (Gough, 2003a).

Texts such as policy documents, political speeches or extracts from e-media are arguably easier and less time-consuming to obtain than interview data. However, as with any secondary data, queries concerning its quality and credibility need to be addressed. Furthermore, it is obviously more diffi cult to perform a DP analysis on such texts when the interactional context of their production is more ambiguous; indeed, Potter and Wetherell (1995) stress that DP should not be used on any texts that do not constitute ‘naturally occurring talk’, and this includes ‘research interviews’ (since the researcher is setting the interactional agenda). However, many researchers have usefully applied DP to analyse data from research interviews, and I would add

qualitative research methods - final.pdf 88 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Methods: how to do discourse analysis 79

Analysis and interpretation

Many researchers have suggested a series of ‘steps’ (e.g. Parker, 1992; Willig, 2008) and ‘worked examples’ (e.g. Gill, 2000; Wood & Kroger, 2000) to aid the practice of discourse analysis. However, the notion of a procedural prescription is both theoretically problematic and diffi cult to articulate, as it is a rather intangible and idiosyncratic process. Perhaps what is more useful is Walton’s (2007) observation that discourse analysis is less to do with following prescribed steps and more about conducting analysis in the spirit of post-structuralist inquiry – that is, to avoid making ‘truth-claims’, to utilise appropriate analytic concepts and to report fi ndings in a way that is consistent with its theoretical and epistemological underpinning. Nevertheless,

that a very blurry line distinguishes ‘naturally occurring’ and ‘unnaturally occurring’ talk, and that the discursive moves that people make in a research interview are as worthy of exploration as in any other interactional setting.

Applying discourse analysis to online data: the production of ‘pro-ana’ identities

Giles’ work illustrates how online communities are an increasingly important site for the negotiation and management of identities, and thus provide fertile ground for discourse analysts. In his research, Giles (2006) aimed to explore the ways in which young people who identify as ‘pro-ana’ (as in ‘pro-anorexic’) construct their identities in online ‘pro-ana’ communities. Such communities promote and attempt to normalise young people’s engagement in anorexic practices (such as self-starvation and self-induced vomiting), and could therefore be seen as a ‘resistance’ to the dominance of medical discourses in ‘offl ine’ settings that serve to pathologise anorexic behaviours. Thus, data from such communities provided Giles with a rich and unique text to enable the analysis of particular identity- making practices that are silenced in more everyday mainstream settings.

Giles examined 20 ‘pro-ana’ websites, and accessed archives from their discussion forums and message boards. He exported a sample of threads that involved confl icts/debates on to a Word document, which in total produced a data set of over 150,000 words. Analysis focused on the ways in which ‘ana identities’ were constructed, particularly in relation to other ‘outsider’ identity categories such as ‘mia’ (bulimia) and ‘dieter’. Analysis also centred on the ways in which such ‘outsiders’ were managed within these communities. In conclusion, Giles refl ected on the ways in which such websites provided individuals with a unique way of collectively (re)producing counter-discourses that served to undermine mainstream healthcare practices that construct anorexia within a dominant medical discourse.

R e

s e

a rc

h E

x a

m p

qualitative research methods - final.pdf 89 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

80 Chapter 4 Discourse Analysis Approaches

what follows is an outline of some of the ‘stages’ (as much as they can be considered as such) and some of the key decisions that are involved when I have conducted discourse analysis. While clearly infl uenced by the academics’ guidelines suggested above, what follows perhaps illustrates the intangibility of doing discourse analysis, in the sense that mere words may not be enough to convey an analytic process that, I would argue, mostly takes place in one’s own head.

Transcription

If interviews are being used, they must be audibly recorded and a very lengthy stage of transcription is likely to follow. While often left unremarked on by social researchers, transcription is an important stage of the analytic process since the intensity of engagement with the text means that many analytical insights are produced at this stage (which should be noted, perhaps continuing the research diary started during the interview phase). Thus, while laborious, it is important for the researcher to perform the transcription themselves. Furthermore, transcription is also the fi rst stage where data starts being shaped by the researcher’s own theoretical decisions about what is important and what can be omitted (Dunne, Pryer & Yates, 2005). A further ‘shape-shifting’ decision concerns how ‘fi ne-grained’ the analysis should be and, consequently, how fi ne-grained the transcription should be to enable this. For example, many discourse analysts adopt Jefferson’s (2004) notation technique, but decisions need to be made on what aspects of the talk should be included: you may decide to include pauses, laughter and the emphasis of words, but decide to omit intakes of breath and the timed length of pauses. This, then, is the fi rst phase where participants’ words slowly become yours.

Developing themes

After transcription, it helps to read each transcript/document a number of times to develop a sense of its context and the story ‘as a whole’ while continuing to make notes on emerging themes. It might also be useful to compare these notes with your earlier notes in your research diary to see whether they still make sense, or whether their importance has changed over time (one key distinction between constructionist methodologies and positivist methodologies is that, with the former, the analytic focus will frequently shift until the end of the analytical process). This stage is possibly the most diffi cult of the entire research process, as the lack of clarity and certainty can produce much anxiety in the researcher. (The best advice I ever received concerning this aspect is simply to ‘keep your nerve’!)

If approaching the data using an FDA approach, the process involves identifying words and phrases that, together, constitute wider ‘discourses’. These can then be highlighted and coded together under an appropriate name (e.g. ‘medical discourse’, ‘parent-blame discourse’, ‘lifestyle choice discourse’). A qualitative data software package (e.g. NVivo) might be useful to help manage this process by saving each highlighted and coded phrase into its own fi le and storing these fi les in an organisationally coherent way (e.g. by using hierarchical ‘trees’ to differentiate ‘higher-level’ discourses and ‘lower-level’ ideas, which individually constitute parts

qualitative research methods - final.pdf 90 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Methods: how to do discourse analysis 81

of a discourse). However, such software is not essential and it does not perform the analysis for you. Indeed, many researchers fi nd doing discourse analysis using software a rather alienating experience and prefer using a series of Word documents using the copy + paste function. Other researchers have told me that they ‘go old skool’ with paper and highlighter pens.

As the analysis progresses, codes may change in conception (which may involve a continual and tiresome renaming of them!). Codes may also become more or less encompassing, resulting in their deletion or blending into other codes. It is also important to identify patterns based on wider structuring axes such as gender, age and ethnicity. For example, you might want to explore whether particular groups of participants (e.g. older men) are using particular discourses, while others (e.g. younger women) are not (and question if there are any exceptions to such patterns). You might also identify and code the materialities that are being drawn on (e.g. community, work and employment, family), and consider how these might be constraining and/or enabling the participant. It is also important to be alert to silences: what is not being drawn on or spoken of? It is also important to be aware of your own reactions during this process: what are you expecting that has not been spoken of or drawn on? Why should this be? Thus, it is important for researchers to interrogate themselves as much as the data. At this stage, it is useful to consider the codes as performing a temporary role in helping to organise the data, rather than as demarcating any kind of permanent conceptual division between categories. They are likely to continually shift and regroup as analysis continues and as you develop a deeper understanding of similarities and differences, and continuities and discontinuities – both within and between individual participants/cases.

Using a DP approach, the analysis is likely to be at an even more detailed level, as you identify specifi c discursive resources (such as rhetorical devices and interpretive repertoires) and consider their function within the interview setting. What was said immediately prior to it and what was the response ‘doing’? As with FDA, you will need to consider the implications of the speaker’s use of particular discursive resources, for both the speaker themselves and their audience. Based on your preparatory reading of discourse analysis, which is essential to this process, it might also be useful to make a list of some of the established rhetorical devices that you might want to identify (see my own attempt to do this in Table 4.1). An important point to remember is that analysis should always ‘stay at the level of the text’ and avoid making assumptions about what the speaker is ‘trying’ to achieve at a psychological level. This is because, in keeping with the ontological assumptions of the methodology, intentions and motivations should be considered as ‘constructions’ (rather than as ‘pre-existing entities’) and therefore truth claims should not be made about them.

Throughout the analysis, you need to ask the data many questions. Such questions might include:

■ How does the interview setting, and the researcher’s and the participant’s responses to each other, produce that particular account?

qualitative research methods - final.pdf 91 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

82 Chapter 4 Discourse Analysis Approaches

■ What discursive resources are used to enable the participants to move forward (or not) in their account?

■ How do the accounts of people of different genders, or ages, or cultures, differ from each other?

■ How are earlier episodes that are talked about made sense of in relation to later episodes?

■ How do the discourses (or interpretive repertoires) drawn on when explaining one experience relate to those drawn on to explain other experiences?

■ What subject positions are available within these particular discourses, and what does that mean for practice (ways of doing) and subjectivity (ways of being)?

■ How does a participant’s orientation to particular contextual factors circumscribe their specifi c experiences, particularly in terms of the extent to which agency can be exercised?

Possible responses to these questions should be explored for some considerable time, and may be generated not only through the researcher but also through lengthy discussions with colleagues, supervisors and other academic audiences in presentations – all of which should enable the researcher to refl ect on the quality of their work and its potential usefulness.

Analysis of experiences of second-time motherhood

First reading

I read through the paper transcript and began to scribble notes (in black ink, in the margins) on thematic ideas. I focused on questioning what kind of relations of power and effects of power seemed to be underpinning the narrative. Key themes identifi ed at this stage concerned the gendered binary between self (as mother) and husband (as father); tensions between the subject positions of ‘mother’ and ‘employee’; and relation- ships between gender, work, subjectivity and social class.

Second and subsequent readings

Next day, I reread the transcript and (in green ink) made further notes – looking for new themes that I had not ‘read off’ on fi rst reading. Second reading enabled a richer understanding of the narrator’s experiences, and ideas seemed to link together more effectively. I consciously ensured that a more theoretical framework informed my thinking this time, and I started making links between freedom, choice and responsibility: ‘freedoms’ seemed very limited and very gendered, and I started questioning the narrative in Foucauldian terms – how was the narrator regulated through motherhood, and how did resistance play out in the text?

R efl

e ct

io n

o n

P ra

ct ic

qualitative research methods - final.pdf 92 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Methods: how to do discourse analysis 83

While predominantly using FDA by analysing how power was played out in the narrator’s life (as depicted in her account), I also attended to a micro-level analysis using DP by also thinking about how power was played out in the immediate interactional setting of the interview. For example, I noticed tensions where the narrator made claims about the world and then immediately negated them using statements such as ‘well, that’s my perception’ and I wondered what the implications of this rhetorical device might be?

Noticing that my ideas ran thinner as a reading through the transcript progressed, third and fourth readings started at different points in the transcript, yielding a little more, albeit diminishing, analytical insight.

NVivo analysis

My readings produced a number of different kinds of categories: some could be conceptualised as structuring axes (e.g. ‘gender’), others as effects of power (e.g. ‘regulation’), and others as agents (e.g. construc- tions of ‘self’, ‘husband’). I imported the transcript into NVivo and, reading through the transcript, I highlighted and coded (and continually recoded) these categories, creating (and re-creating!) ‘tree hierarchies’ in the process. At this stage, I wanted to create a convincing story that operated around core conceptual ideas (e.g. regulation, resistance) and that would be consistent with a constructionist ontology and that centralises (a Foucauldian conception of) power.

Write-up

Analysis continued ‘on the page’: themes were weaved together to produce an analytic narrative, which was punctuated by transcript extracts to provide supporting ‘evidence’ (perhaps a throwback from positivist research practices?). Often, it is only during write-up that one can test whether the analytic narrative is intelligible, and a key element of this stage was the continual monitoring of this from an ‘objective’ perspective (while simultaneously accepting that there is no such thing!).

A fi nal anxiety concerned the writing of the ‘Introduction’ and ‘Conclusion’. This ‘top and tailing’ seemed a further throwback from scientifi c methodology, which performs the role of maintaining the notion of linearity in the research process: I wondered if my analysis might have more integrity if I left it a little rough around the edges? For me, doing discourse analysis is ‘top and tailed’ with an intense anxiety – both when beginning the process and ‘committing’ to slicing up the text in a particular way, and then again on ending the process, which for me was characterised by an anxious re-rereading of the work and a reluctance to hit the ‘send’ button.

qualitative research methods - final.pdf 93 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

84 Chapter 4 Discourse Analysis Approaches

Presenting the research

As with all stages of the research process, the presentation of discourse analytic research needs to be consistent with the theoretical and epistemological positioning of the approach. It is no good conducting discourse analytic research in a constructionist spirit if, at the fi nal ‘presentation stage’, an awareness of and attention to language eludes the researcher (resulting in a research presentation that merely reinforces – rather than challenges – the status quo). Given that a key underlying assumption of discourse analytic research is that research does not merely provide an objective mirror on the social world, but that it actively constructs the social world, then every word needs to be scrutinised. Researchers have an authority and credibility that works to legitimise their construction of the social world, and with such power comes responsibility.

While by no means exhaustive, the following guidance suggests some of the things that researchers should be aware of when presenting their research.

Be continually aware of and alert to your use of a ‘realist’ language

Clearly, if the analysis is performed in the spirit of constructionism, then this needs to be refl ected in the write-up. Truth-claims must be avoided – thus, participants do not ‘think’, but ‘construct the world as …’. However, writing in this way is actually quite diffi cult to achieve for two reasons. First, because realism shapes everyday communication, it is diffi cult to avoid making slips in the write-up (for example, my own entrenchment in a ‘positivist research discourse’ often makes itself known by my frequent unchecked use of terms such as ‘causes’ and ‘signifi cant numbers of …’ whenever I discuss research). This may be even more of an issue when presenting research orally at conferences when you are not able to scrutinise your every word quite as carefully as you might at a keyboard. Second, it is diffi cult because, to communicate to a wide audience (particularly those with the power to make changes, such as policymakers and practitioners) you cannot avoid drawing on language, terms and concepts that are compatible with mainstream psychological discourses. If you didn’t, your research would quickly become incomprehensible (remember that discourse analysis – and therefore the language it is steeped in – is a very marginal methodology). This poses a dilemma, for if one wishes their research to enable social change, one must also implicate themselves in the maintenance of dominant ways of thinking and doing. At least for a while.

Make yourself visible

Kitzinger (1987) writes powerfully about the rhetorical devices that are used in the writing up of realist social science research. She describes how the ‘discourse of scientifi c progress’ pervades the text, enabling past research to be dismissed through phrases that describe its ‘antiquated viewpoints’, ‘outdated references’ and ‘archaic practice’. She also describes how the scientifi c method is conceptualised in a way that makes it appear a neutral, objective and apolitical domain of technical expertise. Part of this discourse involves removing the researcher from the research process,

qualitative research methods - final.pdf 94 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Methods: how to do discourse analysis 85

which is most obviously achieved by using passive sentence constructions and writing up research in the third person. Any reference to the researcher, or to the ways in which the researcher’s own identity might have shaped the research process, is avoided. While this is clearly in keeping with the spirit of realism, such practices are inappropriate when writing up discourse analytic research. Instead you should clearly document the role of the researcher (through the use of ‘I’) and enable a refl exivity that clearly positions aspects of the researcher’s identity that she/he considers salient to the research process. For example, in their research into rape survivors’ accounts of their experiences, Wood and Rennie (1994) positioned themselves by writing about their own experience (or not) of forced sex and how sharing this with the participants may have shaped participant responses. Similarly, in their study of ‘gay men/heterosexual women’ friendship dyads, the authors identifi ed ‘as three gay men’ (Shepperd, Coyle & Hegarty, 2010: 210) to bring attention to the way their own personal sexual politics shaped their analytical emphasis (i.e. by attending more to heteronormativity than to sexism).

Be open to challenge

As discussed earlier, discourse analysis produces a reading rather than an inter- pretation, and so your analysis should not be presented as the defi nitive interpre- tation. Furthermore, openness to alternative readings should not only be welcomed, but encouraged by providing the original data in the appendix (or at least making large amounts accessible). If possible, provide a transparent account of the analytic process to enable the audience to evaluate the persuasiveness of the analytic account proffered (Langdridge, 2004), particularly in the light of your own acknowledgement of the personal and political values that informed the research (see above).

Provide data extracts in context

Following on from the previous point, unlike more mainstream realist research, which often provides brief sentences as ‘supporting evidence’ of a particular truth- claim, discourse analytic work often presents much larger chunks of data to provide a context for what is said (or written, or indeed placed if using visual data). When using DP in particular, often a participant’s response can be made meaningful only when it is looked at in terms of what was said previously. Similarly, the interviewer’s questions should be included within the extracts, as an analysis of the researcher’s talk will enable a contextualised analysis of the data as a whole.

As this chapter has made clear, there are a number of ways in which discourse analysis can be used. First, as the research examples discussed in this chapter demonstrate, it can be used to address many different research questions, spanning topics as diverse as clinical psychology (e.g. Ussher), feminist psychology (e.g. Gill), communication psychology (e.g. Hepburn), criminological psychology (e.g. Holt) and health psychology (e.g. Giles). Furthermore, it can be applied to a range of primary and secondary textual data, which need not necessarily meet the criteria of

qualitative research methods - final.pdf 95 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

86 Chapter 4 Discourse Analysis Approaches

R e

s e

a rc

h E

x a

m p

le The Social Construction of a Serial KillerBartels, R. & Parsons, C. (2009) The social construction of a serial killer. Feminism and Psychology, 19(2), 267–280.

Brief context and research aim

While there has been much psychological theory and research examining serial killing, this has tended to adopt a realist ontology by locating the ‘causes’ of serial killing within the agent himself (it is nearly always a ‘he’). The most dominant explanations are those that mobilise the notion of sexual fantasy: a pathological category of person is constructed who is unable to control his instinctual urges, and who instead continually ‘acts out’ his fantasies. Such ‘acting out’ results in serial homicide.

However, rather than merely perpetuate this individualising discourse by investigating it within a realist paradigm, Bartels and Parsons adopted a discourse analytic approach. By analysing the courtroom testimony of Dennis Rader,3 the research aimed to identify the ways in which Rader constructed himself and his crimes within the courtroom setting.

Method

Data were collected from a publicly available 78-page transcript of a court hearing that took place on 27 June 2005. The content of the transcript concerned Rader’s confessions to and accounts of ten murders. Although Bartels and Parsons claim to have adopted a DP approach, and refer to Potter and Wetherell’s (1987) germinal work as their touchstone, their identifi cation of ‘discourses’ certainly draws on ideas consistent with FDA.

In outlining their method, Bartels and Parsons describe their rereading of transcripts before focusing their attention on ‘the ways in which events, categories and emotions were discursively constructed and what functions they appeared to serve’ (2009: 270). Their work was presented under three headings, labelled as ‘discourses’, and included extracts from the transcripts to support their analysis. Reference was also made to particular rhetorical devices and their functions within the courtroom setting. The analysis is followed by a discussion that summarises their fi ndings, and suggests possible implications of the research and possible areas for further research.

3Dennis Rader was a notorious US serial killer who went by the name of BTK (Blind, Torture, Kill) and who was indicted for ten counts of murder in 2005.

qualitative research methods - final.pdf 96 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

Methods: how to do discourse analysis 87

Findings

Bartels and Parsons identifi ed three discourses in Rader’s talk. These all worked together to uphold the construction of serial killer as a sympathetic character who is taken over by sexual urges that he cannot control.

The fi rst discourse identifi ed is ‘perpetrator as sympathetic’; this discourse was constituted by a number of specifi c rhetorical devices. For example, in describing his crimes, Rader refers to how he tried to make Mr Otero [one of his victims] comfortable by placing ‘… “I think a parka or a coat underneath him”…’ (2009: 270). Bartels and Parsons suggest that not only does this discursive move enable Rader to construct himself as sensitive to another’s discomfort, the use of the term ‘I think’ functions ‘to refl exively display him as an honest person [that] serves to validate his account’ (2009: 271).

The second discourse identifi ed is ‘perpetrator as a serial killer’ and the authors identify Rader’s references to expert literature (e.g. ‘… “I don’t know if…[…]… you read much about serial killers, they go through what they call the different phases”…’ (2009: 273), to enable him to position himself in the category of ‘serial killer’. While by no means a fl attering subject position, Bartels and Parsons suggest that it nevertheless affords Rader the possibility of mitigating responsibility for his crimes, by ‘[drawing on] the inference that he can’t help doing what he did because it’s what serial killers do’ (2009: 274).

The fi nal discourse identifi ed is ‘perpetrator as driven by sexual fantasy’, which Rader draws on to construct his actions as sexual, thus obscuring the violence of them. For example, Rader talks about when ‘… “I was in the sexual fantasy”…’ (2009: 275) and the use of the term ‘in’ suggests a different state of consciousness over which he had no control.

Discussion

Bartels and Parsons refl ect on other research fi ndings such as those in Coates and Wade’s (2004) work, which identifi ed similar discourses within judges’ talk in sexual offences cases. These similarly served to minimise the perpetrators’ responsibility by constructing them as ‘overwhelmed by psychological forces’ (Bartels & Parsons, 2009: 277).

They also refer to past research that identifi ed such discourses within the talk of sex offenders, and they refl ect on the implications of this for practice. In particular,

R e

s e

a rc

h E

x a

m p

qualitative research methods - final.pdf 97 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

88 Chapter 4 Discourse Analysis Approaches

representativeness that one fi nds in scientifi c methodologies. One might take discourse analysis to a macro-level or a micro-level, and increasingly research is becoming more pluralistic as FDA and DP are integrated – perhaps with one or other approach having greater analytical emphasis. In terms of outcomes, there are a number of ways in which doing discourse analysis can produce a change for good – perhaps by exposing the implicit values and hidden assumptions that underpin particular kinds of knowledge that legitimise existing unjust institutional practices. Alternatively, it may have something to say more directly about the way communication is ‘done’ by particular professionals, and how it needs to be altered to work better for service users. Table 4.2 summarises and clarifi es this diversity of application with reference to the fi ve short research examples that were referred to in this chapter.

R e

s e

a rc

h E

x a

m p

they question how professionals from the legal and psychological fi elds ‘may attend to Rader and other serial killers in ways that are in accordance with this construction, for example, in terms of how they are treated, sentenced and researched’ (2009: 277).

However, although Bartels and Parsons do not mention this, there are also wider political implications for their fi ndings: serial killing is unequivocally a gendered crime, with the majority of perpetrators being male and the majority of victims being female (Wilson, 2009). However, the perpetuation of a psychologising and individualising discourse – perpetuated by judges, practitioners and the perpetrators themselves – obscures this structural context. Applying an FDA approach to the Rader data may have highlighted this issue. However, this may well have come at the cost of a clear and focused DP analysis, which highlights the ways in which talk is situationally contingent and performs functions beyond refl ecting back ‘reality’.

qualitative research methods - final.pdf 98 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

T a

b le

4 .2

S um

m ar

y ta

bl e:

h ow

d is

co ur

se a

na ly

si s

ca n

be u

se d

ac ro

ss a

r an

ge o

f a pp

lic at

io ns

U s

s h

e r

(2 0

0 3

) G

il l

(2 0

0 9

) H

e p

b u

rn (

2 0

0 5

) H

o lt

( 2

0 0

9 ; 2

0 1

0 a

; 2

0 1

0 b

) G

il e

s (

2 0

0 6

)

R e

s e

a rc

q u

e s

ti o

n H

ow a

re s

ci en

ti fi c

ns tr

uc tio

ns o

f P M

D D

ay ed

o ut

in w

om en

’s

na rr

at iv

es ?

H ow

a re

s ex

ua l

re la

tio ns

hi ps

ns tr

uc te

d in

w om

en ’s

m ag

az in

H ow

d o

ch ild

re n

m an

ag e

in st

it ut

io na

l d em

an ds

he n

m ak

in g

ca lls

to a

lp lin

H ow

d o

pa re

nt s

co ns

tr uc

t ex

pe ri

en ce

s of

re nt

in g

a ch

ild

in vo

lv ed

in o

ff en

di ng

H ow

a re

‘p ro

-a na

’ w eb

si te

s us

ed to

e na

bl e

id en

ti ty

w or

k by

it s

us er

T y

p e

o f

d a

ta F

a c

e -t

o -f

a c

e i

n te

rv ie

w s

w it

h 10

8 w

om en

b ei

tr ea

te d

fo r P

M D

20 m

a g

a zi

n e

a rt

ic le

fr om

5 r

an do

m ly

le ct

ed is

su es

o f

G la

m ou

r ( U

K ) m

ag az

in e

Tr an

sc ri

pt s

fr om

t w

o te

le p

h o

n e

c a

ll s

to a

n N

S P

C C

he lp

lin e

T e

le p

h o

n e

i n

te rv

ie w

w it

h 17

p ar

en ts

u si

na rr

at iv

e in

te rv

ie w

A rc

hi ve

d th

re ad

s fr

m e

s s

a g

e b

o a

rd s

a nd

is c

u s

s io

n b

o a

rd s fr

om 2

0 w

e b

s it

e s

A n

a ly

ti c

p p

ro a

c h

F D

A –

w it

h a

fo cu

s on

e ro

le o

f s ci

en tifi

di sc

ou rs

D P

– w

it h

pa rt

ic ul

em ph

as is

o n

in te

rp re

tiv e

re pe

rt oi

re s

D P

– a

ls o

us in

g el

em en

fr om

‘c on

ve rs

at io

n an

al ys

is ’ s

uc h

as s

cr ip

t fo

rm ul

at io

F D

A –

b ut

a ls

o us

D P

to a

tt en

d to

et or

ic al

d ev

ic es

D P

– u

si ng

‘m em

be rs

hi p

ca te

go ry

a na

ly si

s’

te ch

ni qu

e –

b ut

a ls

o dr

aw s

on F

D A

c on

ce pt

o f d

is co

ur se

K e

y fi

n d

in g

s Fi

ve h

eg em

on ic

‘t ru

th s’

en ti

fi e d

in w

om en

’s

ta lk

th at

w or

k to

gu la

te fe

m in

in it

W hi

le d

ra w

in g

di sc

ou rs

es o

f f em

al e

em po

w er

m en

t, ar

tic le

s w

or k

in G

la m

ou r

m ag

az in

e to

p ri

vi le

m en

C hi

ld re

n st

ru gg

le to

ak e

th em

se lv

cr ed

ib le

in h

el pl

in e

in te

ra ct

io ns

Th e

po si

tio ni

ng o

f pa

re nt

s as

‘b ad

re nt

s’ p

ro du

ce s

pa re

nt al

r es

is ta

nc e

to th

ei r s

oc ia

l a nd

ga l r

eg ul

at io

O nl

in e

co m

m un

it ie

s ar

e fi e

rc el

y po

lic ed

b y

th os

e w

ho id

en ti

fy a

s ‘p

ro -a

na ’

P o

s s

ib il

it ie

fo r

p e

rs o

n a

s o

c ia

l a

n d

/ o

r p

o li

ti c

a l

c h

a n

g e

In di

re ct

c ha

ng e

th ro

ug h

hi gh

lig ht

in g

cu rr

en t

do m

in an

t d is

co ur

se s

an d

po ss

ib le

a lte

rn at

iv e

di sc

ou rs

es th

at fr

am e

m en

ta l h

ea lt

h pr

ac tic

In di

re ct

c ha

ng e

th ro

ug h

ex po

si ng

th e

w ay

s in

w hi

ch la

ng ua

w or

ks to

p ro

m ot

e ge

nd er

in ju

st ic

e in

th e

po pu

la r m

ed ia

A lt

ho ug

h an

e xp

lo ra

to ry

ud y,

it h

as fu

tu re

u til

it y

in o

ff er

in g

pr ac

tic e

gu id

an ce

fo r h

el pl

in e

pr of

es si

on al

s w

he n

re sp

on di

ng to

te ra

ct io

na l d

if fi c

ul tie

s w

it h

ch ild

re n

En fo

rc in

g pa

re nt

s to

‘t

ak e

re sp

on si

bi lit

y’

th ro

ug h

th e

cr im

in al

st ic

e sy

st em

w ill

t w or

k in

e ng

ag in

g pa

re nt

s of

‘y ou

of fe

nd er

s’ –

m or

e su

pp or

ti ve

p ol

ic ie

s ar

e ne

ed ed

W hi

le G

ile s

cl ai

m s

pa rt

ic ul

ar ‘s

oc ia

l g oa

l’ (2

00 6:

4 75

), th

er e

ar e

cl ea

rl y

im pl

ic at

io ns

fo r t

he r

ol e

‘p ro

-a na

’ w eb

si te

s in

de rm

in in

g he

al th

of es

si on

al s

an d

un de

rs ta

nd in

g th

e ef

fe ct

s of

ch s

it es

o n

th ei

r u se

qualitative research methods - final.pdf 99 14/06/2011 14:07

C o p yr

ig h t ©

2 0 1 1 . M

cG ra

w -H

ill E

d u ca

tio n . A

ll ri g h ts

r e se

rv e d .

90 Chapter 4 Discourse Analysis Approaches

Chapter summary

■ Discourse analysis is a methodology that incorporates a number of approaches and techniques. Two specifi c approaches commonly used in psychology are Foucauldian discourse analysis and discursive psychology.

■ Discourse analysis is underpinned by a constructionist ontology that questions taken-for-granted assumptions about the world and the people in it. It emphasises the role of language in constructing the world, and the role of power in recognising that these constructions have different political effects for different groups of people.

■ Foucauldian discourse analysis enables analysis of these constructive processes at a macro-level; a key aspect of analysis is the identifi cation of wider social and institutional discourses and practices that shape people’s experiences of the world.

■ Discursive psychology enables analysis of constructive processes at a micro- level; a key aspect of analysis is the identifi cation of the function of discursive resources used by speakers in interaction.

■ Discourse analysis practice needs to remain faithful to its constructionist approach to knowledge. This means attending to issues such as refl exivity, researcher positioning and accountability, and scrutiny of the language used to present the research.

■ Discourse analysis can be used to analyse a diverse range of texts. It can be used to achieve theoretical goals as well as more pragmatic goals in terms of making recommendations for policy and practice. It can also be used to highlight injustices and to offer alternative ways of thinking (that is, counter- discourses) about the world and the people in it.