Deconstructing Paranoia: An Analysis of the Discourses Associated with the Concept of Paranoid Delusion

David J. Harper PhD Thesis June 1999

Back to title page



In the first two parts of this dissertation I have explored how paranoia is constructed both in the wider context and in the specific context of interviews with professionals and psychiatric service-users. In this final part I want to 'stand back' from the study and constitute it as an object. In chapter 7 I will address some of the issues of reflexivity which have arisen throughout the course of the study and which have been noted in most of the chapters. In that chapter I will begin to elucidate some of the grounds for developing practical implications consistent with my analysis. In chapter 8 I develop this argument looking critically at the notion of 'applicability'. I define a number of interest groups and suggest implications that make sense in the context of current developments in each of the target areas. Once again, then, these chapters remain analytic although this is at a broader level with no quotes from the interviews.

Chapter 7

Reflecting on this account

Reflexivity: Term used by social constructionists to refer to the application of the theory back onto itself and its practices

Burr (1995, p.185)


This chapter is about reflexivity. Burr (1995) notes at least three different uses of this term. First, to refer to the way social constructionist theory reconstitutes the role of 'respondents', the status of their accounts and their relationship to the researcher. Second, to acknowledge that in giving an account, that account is both a description of an event and part of that event. Thirdly and most commonly it refers to the fact that social constructionism is itself a social construction and thus the knowledge-making practices of social constructionists need to be attended to (eg their use of rhetorical devices in constructing arguments like this one). Why bother to think about such issues? Squire (1995), in summarising Wynne's (1988) discourse analytic work examining the diagnosis of multiple sclerosis, highlights the following points: that, although extracts are used from, say, interviews, these are somehow seen as more 'real' than other kinds of data; that extracts are often quoted out of context; that often one extract may be taken to stand for others deemed 'similar'; that it may be assumed that speech documents 'the self'; that discourses like 'traditional' medicine are reified and assumed to exist unproblematically; that medical diagnoses may be seen as 'right' or 'wrong'; and that authors make implicit claims to truth in their accounts through the use of certain rhetorical constructions. These are all important issues and need to be addressed in accounts like this one.

In this chapter I will be using the notion of reflexivity in all three of the ways noted by Burr (1995). First, I will briefly review the use of reflexivity in discourse analytic, social constructionist and deconstructionist work. Second, I will attempt to give an account of some of the contextual influences and interests which have shaped this account. Third, I will explore how reflexivity issues arose in the research. Fourth, I will look, with a reflexive eye, at how this argument has been rhetorically constructed. Finally, I will examine the potential charge against my account of 'ontological gerrymandering' -- this is an argument put to researchers who, whilst claiming to use a constructionist approach, provide a realist interpretation of others' accounts (or who move to realist implications) when their account too could be examined using a constructionist approach.


The call to reflexivity has been one of the challenges with which traditional positivist psychiatry and psychology have been faced through the emergence of social constructionist work. However, it has also been one of the most problematic. One criticism of the use of reflexivity is that it has not got far beyond the playful use of self-subverting utterances (Sherrard, 1991). Duncan Moss (1991) has commented that the repeated use of reflexivity is like breaking your legs to show that you need them to walk whilst Stainton Rogers (1992) reminds us that reflexivity is a box of tools and not a thing. Others have commented on the tension that exists between giving an account and reflecting on the construction of that account -- a tension that could be characterised as one of 'content' versus 'process'. Too much either of content or process will make the account confusing to the reader. Stainton Rogers (1991) notes a tension between telling the story she would like and her duty as an author 'to weave a coherent, plausible, user-friendly story for you that will stimulate and retain your interest' (p.10). Potter (1996b) has noted how novel literary forms which underline reflexivity are hard to pull off without rendering the text reader-unfriendly. Commenting on an ealier piece by Potter (1988), Woolgar (1988b) notes the twin demand of 'challenging empiricism and not putting people off' (p.53) in accounts of the analyst's struggles. Burman & Parker (1993) note that one of the dangers of reflexivity is that it can shift a focus to the account rather than what is being accounted for, leading to a 'wallowing' in the researcher's interpretive assumptions and processes which then detracts from the importance of the topic and possible political interventions.

There have been a number of attempts to solve, or at least express to the reader, the problems and challenges of reflexivity. There have been new literary forms, like 'plays' (eg Ashmore et al, 1989) and 'reflexions' by editors on others' accounts (eg Woolgar, 1988a). Other examples include reflexive boxes (Edwards & Potter, 1992; Curt, 1994), 'interruptors' (Curt, 1994) and even café conversations and faxes (Riikonen & Madan Smith, 1997). Of course, these experiments lay themselves open to parody (see, for example, Harper & Smith, 1995) and sometimes even the authors themselves seem tired and cynical about these games (Gill, 1995). Moreover, even though reflexive, authors are still pursuing a particular line of argument (Harper, 1995).

Of course, maximum reflexivity is not the only goal of a piece of work and, if trying to achieve this (if that were possible) means other goals (eg accessibility or clarification of political and practical implications) are not attained then the balance needs to be redressed. In trying to engage with other goals, my account has perhaps been at the cost of reflexivity -- I have only briefly touched on issues of reflexivity in the rest of this dissertation. This is not to play down its importance, merely to reflect the difficulties of including reflexive considerations in the account. I am attending to reflexivity in this chapter to highlight how important it is in all the other chapters.

There are a number of pitfalls in producing accounts attending to reflexivity. Thus Potter (1988) notes that the call to reflexivity may imply that maximum reflexivity becomes a postmodern code for progress towards truth. Another problem is that, by attending to issues of reflexivity, authors may act as if they have secured some kind of epistemological insurance. Gill (1995) notes that:

current discourse analytic adventures into reflexivity have often served to reinforce, rather than to challenge, the authority of the analyst, and have been used as a vehicle for disavowing explicit commitments, whilst pretending to give voice to other positions.

Gill (1995, p.182)

Thus whilst some claim that DA is a constant struggle to make sense of discursive practices while avoiding reading practices which tempt the analyst into a definitive version, there are often few signs that analysts or their commentators have struggled in producing their accounts, as Woolgar (1988b) notes, in his reflexion on Potter's (1988) paper. A further problem is that by agonizing about subjectivity and power, analysts can become paralysed, fearing that any interventions will be exploitative or paternalistic (Burman & Parker, 1993). It is interesting in this respect to note that in her review of discourse analytic reflexivity discussions, Squire (1995) found an interplay between what she terms 'extravagant' postmodern readings (which emphasise the provisionality of readings) and 'pragmatic' readings which lead less problematically to certain kinds of political intervention, especially amongst feminist workers. She also notes a pattern in DA accounts of reflexivity of authors being reflexive but fearing that they will go 'too far' which raises questions about what 'too far' means and what is it that analysts fear will happen if they do. I will attempt to negotiate a path between these influences in developing a reflexively aware account here.


What is the point of reflexivity? And what am I aiming to do through this chapter?

Discourse analysts should adopt a notion of reflexivity which stresses the need for the analysts to acknowledge their own commitments and to reflect critically upon them. By seeking to explain and justify the basis for their readings or analyses, discourse analysts become accountable for their interpretations and the social and political consequences of these interpretations.

Gill (1995, p.182)

Here, Gill stresses the need to look at the effects of reflexivity rather than abstract notions of endless reflection. For her, the key point is to attempt to make the account and its author accountable. But even here there are difficulties since some reflexive discussions can have a confessional, almost breast-beating tone. Parker (1994) asks who the subject that we are getting closer to in qualitative research and in discussions of reflexivity is? Who is the author that is being made accountable? He notes that discussions of reflexivity can assume different notions of subjectivity: 'uncomplicated subjectivity' (a broadly humanistic view); 'blank subjectivity' (where subjectivity is seen as a product of clashing discourses); and 'complex subjectivity' (which combines individual and cultural views of subjectivity, some using psychoanalysis). For him, it is important to 'ground' reflexivity in context:

The way that social research is contextualized now will also look a little more complex, for the 'context' is, in this account, not an objective background against which the researcher renders an account of the phenomenon in question. Rather the context is the network of forms of subjectivity that place contradictory demands on the research. In social psychological research there is an array of competing interests and agendas that frame the production of proposals; the expectations and demands of 'subjects' or co-researchers; and the career investments and projected autobiographies that exist in tension in the academic world.

Parker (1994, p.250)

He includes considerations of race, class and gender as further influences. Bola et al. (1998) argue that reflexivity is not just about listing the social locations one occupies in order to render them unproblematic, rather it is to think through 'the implications of that social positioning for the research itself' (p.106). In other words, what difference has one's structural position made? They note the danger of failing to understand the legitimate reasons for any distance between researcher and researched and the danger of apportioning responsibility for any difficulties on the researched.

Here, then, we begin to see how an account could be made more accountable, not just through a literary construction of an author who 'confesses' but also through an attempt to explicate the interests and other forces at work in the production of the account. In this chapter I will attempt to do this by first addressing some of the interests that I feel shape this work. Second, I will look at how issues like participants' expectations emerged in the actual practice of the research (eg interviews). Thirdly, I will examine some of the forces at work in the actual writing of the account.

7.3.1 Interests and agendas, accounts and accountability

One of the aims of this account so far has been to provide a critical commentary on the history and current professional and cultural use of the diagnostic category known as paranoia. Part of the analysis has involved describing the interests which might be at work in texts dealing with paranoia. I noted in chapter 1 how the only way to address reflexivity was to clarify what some of the interests constructing my account might be so readers might take them into account in reading the text. With cautions in mind, about some of the excesses of reflexivity, I will attempt to draw attention to some of the ideological influences which may be at work in this account.

One set of influences are biographical ones like my choice to do a Ph.D and 'projected autobiography', a small amount of funding for two years from the then Regional Health Authority, my previous research in the area (Harper, 1992), my training and work as a UK National Health Service clinical psychologist, my interest in social constructionist approaches to therapeutic work (see, for example, Harmsworth et al, 1996) and a variety of ethical and political views, for example my belief that many of the 'psychological problems' pathologised within the mental health system are entirely understandable and reasonable and need forms of intervention (eg social and political) other than, or at the very least in addition to, therapeutic aid. With Smail (1996c) I want to pronounce against biological (or at least simplistically biological) psychiatry:

I just know that the biological approach to psychological distress is bollocks ... I personally just can't be bothered to argue about it any more with a new generation. I don't care what 'new evidence' is supposedly advanced: I've seen it all before. Fatuous, self-important professors in white coats staring at computer images of people's brain waves, etc., expounding their half-baked ideas to mesmerized television pundits who swallow the story whole and breathlessly reproduce it for the viewing millions. It's all crap and I'm too old and too tired to be doing with it ... But that is not the attitude! ... These ideas are dangerous and destructive and they have to be dealt with.

Smail (1996c, p.16)

There is an ambivalence in the research literature about the articulation of personal commitment and interestedness. On the one hand, this can be seen as a way of allowing readers to see ideological influences at play. Thus Scheper-Hughes notes:

My particular sympathies are transparent; I do not try to disguise them behind the role of an invisible and omniscient third-person narrator.

(1992, p.25)

On the other hand, it could be seen as showing how the research is too interested and is, in fact, biased. On yet another hand, though, the articulation of interest could be seen as a particular construction of the author to achieve certain ends. For example, Wilden (1972) notes the dangers of the 'negative academic' who is, in fact, parasitic on traditional modes of research in order to be able to distance themselves from it, becoming what Bynum et al (1985) term iconoclastic 'Young Turks'.

There are also a number of broader historical factors which have created opportunities for work of this type. For example, the emergence of intellectual critiques of psychiatry noticeable in the 1960s and 1970s, although to a large extent assimilated by contemporary psychiatry (Nance, 1992; Samson, 1995) still provides an opportunity for critical discussion of psychiatric practice. However, there are a number of contextual changes which mean that present-day critiques are slightly different from those of the past. For example, the political changes in UK health and social service provision, the introduction of supposedly more accountable personal service management systems and the anti-professional rhetoric of policy-makers on the Right opened the British medical establishment up to a good deal of criticism during 18 years of Conservative government and some of these cultural influences will no doubt continue under a Labour administration. Samson (1995) has detailed a number of influences which, he argues, continue to threaten to fracture medical dominance.

In addition, the support in government rhetoric for citizen advocacy and the move by a range of voluntary groups to establish power bases not previously open to them (eg the provision of treatment and day and residential care) has meant that critiques of traditional models of mental illness and service delivery have been welcomed. This move has also been matched by other professional groups like clinical psychology who, having fought for their autonomy from the psychiatric establishment, are increasingly offering psychological models of intervention for people with diagnoses which previously only psychiatrists would have treated (Pilgrim & Treacher, 1992; Samson, 1995). Indeed some of the most recent critiques of psychiatry have come from clinical psychologists like Bentall et al. (1990) and Boyle (1990) while Samson (1995) sees this as an attempt by clinical psychology to occupy previously psychiatric terrain. In the meantime, cognitive-behavioural interventions have grown in popularity with professionals, users and purchasers of services. Such treatments owe a great deal to the 'scientific' validation of cognitive behavioural methods by clinical psychologists who either accept the validity of psychiatric diagnostic categories like schizophrenia (Birchwood & Tarrier, 1994) or who suggest focusing on individual symptoms like delusions (Haddock & Slade, 1996). Interestingly some of this work has been pioneered by psychiatrists (Kingdon & Turkington, 1994) and it is highly likely that these approaches will be assimilated into psychiatry since the Royal College of Psychiatry now explicitly requires trainee psychiatrists to learn about psychological interventions and have published books on this topic (eg Mace & Margison, 1997).

Another development has been the increasing interest in mental health of qualitative and social constructionist research which we noted in the introduction to Part I. In contrast, there is a relatively small space opened up in contemporary debate for social constructionist and deconstructionist work (see Parker et al., 1995).

7.3.2 Reflexivity and the practice of the research project Getting the research off the ground

One very practical influence on the research was actually getting potential participants which, I noted, in the introduction to Part II as a problem. I was faced with considerable organisational obstacles such as professionals forgetting about my research thus making the taping of initial interviews impossible. Also some potential participants declined possibly because they were guarded about further professional contact. Thompson-Pope & Turkat (1989) experienced similar difficulty, with only fourteen out of sixty clinicians being willing to help with their research and thus only being able to identify three potential research participants from a health district of 250,000. Moreover, professionals would tend to suggest only those service-users who they thought would agree to be interviewed. It is possible therefore that my sample of service-users might have a higher than representative number of people who were still in contact with health and welfare services. Given the limited number of participants there was no possibility of trying to ensure diversity in terms of age, race, class, gender and relationship with psychiatric services. I faced some similar difficulties in interviewing professionals as I described in the introduction to Part II. This led to considerable practical difficulties. For example, I was successful in gaining two years' funding for my release from work for half a day a week to conduct the interviews, transcription and analysis. However, by the end of the two years I had only just completed transcription of the interviews. The practice of interviewing

Interviews self-evidently reflect a power imbalance: one person asks questions, another has to answer. A number of authors have noted how the interview dynamic creates a number of roles for participants (Gilbert, 1980) -- they may feel positioned as confessant, patient, student and so on. At the same time, the interviewer too is positioned. In this research I held a dual position: as both researcher and as a clinical psychologist who worked in the locality(1). This made the interview situation even more complex. For example, service-users might not have felt I was independent of the professional system and may have felt constrained in what they could say to me. For me, there was an additional danger of conducting the interview with service-users as if it were a clinical interview and with professionals as if it were a case discussion or supervision. To avoid this, I tried to adopt a respectful journalistic stance although it was hard to avoid a therapeutic frame. This is not necessarily a problem though -- Coyle (1998) and Coyle & Wright (1996) have discussed the utility of a counselling interview when researching potentially distressing topics.

As well as constraining what was said, my position also created other kinds of relationships with the interviewees. Thus, at times, it was clear that some professionals saw me as a fellow professional -- for example Dr Lloyd, Dr Howard and Dr Williams all asked me for my clinical opinion of the service user we were discussing. Service users too saw me in different ways, thus Ian read from a prepared statement when I interviewed him and ended it by asking how I could help him. Although I discussed the issue of consent with participants and they signed the appropriate documents it is hard to know if people, especially service-users, really did understand what they were consenting to and whether they really felt able to say 'no'. Boundary issues became complicated when I subsequently ended up receiving referrals for clinical work for three of the service users in the months following my interviews with them. I agreed to work with them since access to psychology colleagues might have been problematic (and I did not want to disadvantage them). Our previous research relationship did not appear to result in obstacles to our work.

Another symbol of the power imbalance and a major influence on the course of the interview conversations was the fact that I taped the interviews. I became used to some service-users being more forthcoming the first time I met them (without my tape-recorder) and more hesitant the second time (when I brought it). Reactions varied with one participant (Shaun) asking for it to be turned off and then discontinuing the interview after 10 minutes whilst Ian appeared to use the public status of the conversation to make a political statement. Sharon repeatedly looked over to the tape recorder during the interview. Interviewees reflected the constructed nature of the conversation when they asked, as Mike did, if the interview was going okay as I turned the tape over to record on the other side.

By virtue of the fact that it was I who asked the questions, I was a dominant influence over the structure of the conversation. My questions were influenced by my research interests. However, since the interviews were located both physically and discursively within psychiatry it was hard to avoid using psychiatric language to ask questions. The times that I did not sometimes led to misunderstandings with participants. Thus I was far from 'free' to ask just what I wanted. Often it was I who introduced certain medical concepts although I would, in other contexts, prefer other forms of language. This demonstrates how talk, and particularly its institutional context, imposes constraints on what can be said and produces certain kinds of knowledge. This point is demonstrated by Terre Blanche's (1995) study which found that many terms used in psychiatry were highly correlated with each other and thus one word could, in a sense lead automatically to another simply by the momentum of the text (eg 'schizophrenic', for example, was linked to 'chronic'). Thus I would not be surprised if the word 'treatment' was closely linked to 'continuing', 'stopping', 'beginning' and so on. Other researchers attempt to seek to use interviewees' own words and concepts. However, both users and professionals tended to use psychiatric terms which is not surprising since some of them had had substantial contacts with psychiatric services. At times, though, lay terms crept in -- for example Peter and Sharon's use of the word 'nerves'. My use of psychiatric terms like 'paranoia' (which was constrained because that was the topic of the research and I did not particularly want to broaden my questioning beyond that) presented the risk that I was participating in the reification of delusions by discussing them in this way. At other times, my professional culture and class position were revealed when, for example, towards the end of my interview with him Ian told me he did not understand what 'paranoid' meant when I (and he) had used it throughout the interview. This was not restricted to service-users and Edward Jackson (CPN) asked me a similar question at the end of his interview. Moreover, power was illustrated in the relative shortness of some users' responses to questions (eg John and Alan) due, to some extent perhaps, to being used to medical interviews and my class position although others' responses were longer.

In trying to assume a respectful journalistic but therapeutically informed stance in the interviews I was occasionally tempted to think I was being 'neutral' in interviews. I sought to be as curious as possible with the aim of providing contexts for linguistic variation (cf Potter & Wetherell, 1987) but did not seek to be confrontative. At times, I actively stated to users that I held no strong opinion about their beliefs. Despite this opinion it was instructive to be reminded that my interviewees did not share this view of the interview. Thus, despite my protestation of neutrality, Alan was clearly wary about whether I had an open mind about his views and I have examined some of the issues around our communication in chapter 5. Professionals too did not see me as someone without a history or context -- Dr Williams asked me questions about cognitive behaviour therapy and saw me as someone with expertise in this area. At other times, by asking questions which showed I was not taking things for granted I was seen as arguing from a particular position. For example, in response to one of my questions, Terry said 'I can see your point' (Terry Reid, CPN, line 41).

Another important contextual shaping factor was the physical location of the interviews. For the professionals I mainly interviewed them in their workplace -- either their own office or attached offices or meeting rooms. For the users I interviewed two people in rooms attached to their ward, one person in a day hospital, three people in offices in a hospital where they visited their CPN and three people at home. It was hard to delineate any gross effects of this -- thus one of the users I interviewed at home spoke the least whilst some on the wards spoke the most. However, the physical setting might well have contributed to the shaping of the conversation, thus making it hard to divorce the interview from notions of psychiatric and other professional interviews. Power was evidenced in other ways -- thus, for example, all of the service-user interviewees were unemployed and/or on sickness benefits whilst they were being interviewed by a middle class professional. It is likely that this added to the kinds of effects noted already. Reflexivity and transcription

Others have written more about some of the issues involved in transcription (see, for example, Banister et al., 1994; Hollway, 1989). They note that transcription is a constructive process which involves the researcher's assumptions, memory and the particular stage at which the research is at. There are potentially a wide range of issues to bring up here (including, for example, the practical problems of poor noise quality) but I wish to focus on only two here as exemplars which point to the constructed nature of transcription -- a construction where certain aspects are highlighted whilst others are concealed. The first relates to my transcription of laughter. On a number of occasions there was some laughter in the interviews. Usually on these occasions I smiled (and sometimes laughed quietly) -- possibly because I was too aware of doing a 'proper' interview -- whilst often the interviewee laughed out loud. Since my transcription notation was concerned only with what was audible on a tape recorder the resulting transcripts either included '<laughter>' or simply indicated that the other person laughed whilst I didn't which, on analysis made it seem as if I was acting a little sternly in the interviews and not laughing at things that were funny whilst, in reality, I had probably smiled but of course this was un-transcribable.

The second issue relates to my use of names in the transcripts. As I have noted previously, all the names used were pseudonyms. However, at an early point I was faced with a dilemma about whether to use interviewees' first names in the transcripts since full names seemed clumsy and long-winded. This raised an interesting issue since I knew the first names of the CPNs and users, but only two of the psychiatrists and none of the GPs. In most settings where I have worked as a professional, doctors will rarely give their first name when they are introduced (indeed, staff introducing them will rarely give the doctor's first name) whereas most other staff and service users will -- doctors may well give first names to professionals they know but rarely to users of services. To call the doctors by their first names in transcripts might have served to conceal the power that was manifested in their title of doctor and their relationships with users. Users, for example, generally called some workers (eg CPNs) by their first name but I did not hear one user call their GP or psychiatrist by their first name. Of course, by using my own first name in the transcripts I may well be working to conceal my own status and power -- although some users called me by my first name in the interviews (eg Ian and Don). Reflexivity and analysis

Despite the presence of guiding concerns and questions in the research, like the questioning of taken-for-granted notions and the revealing of 'hidden' assumptions, I often felt at sea in developing both a theoretical and epistemological position and an analysis of the various materials I examined (interview transcripts, examples from popular culture, professional texts etc). Potter & Wetherell (1987) in their ten steps to analyse discourse note there is an 'intermission' after transcription of interviews:

At this point the researcher usually sits back in contentment and surveys a whole set of cardboard boxes with bits of paper in them. Some contain documents of various kinds, others are full of transcript and cassette tapes. There are literally hundreds of thousands of words waiting to be transformed into exciting research findings, and with this thought contentment can easily be transformed into total immobility and panic. Where should one start?

Potter & Wetherell (1987, p.166)

They follow this disarming passage with the comment that 'we find it reassuring to begin with some coding' -- this rhetorical move serves to cover over some of the rich difficulties they note. Hollway (1989) by contrast, comments:

Making sense of the transcript in terms of the research questions is the most harrowing part of all. The more unstructured it is, the greater the anxiety that it is going to be impossible to analyse rigorously.

Hollway (1989, p.21)

My experience seemed to accord more with this latter perspective. As I noted earlier, problems in getting interviewees meant that funding for time off from work ran out just as I finished transcribing. I spent a good portion of the year following the transcription of interviews in a state of analytic paralysis and this was, no doubt, influenced to some extent by this change. On two occasions, practising the use of Narrative Therapy styles of questioning with colleagues, I used my difficulties in doing my PhD as my 'problem'!

Moreover, there seemed too much material to analyse and Hollway (1989) has noted some of the difficulties in making an analytic way through this amount of data. Would a decision to focus on one aspect rather than another simply be arbitrary? There seemed relatively little in the literature about the process of choosing themes. A relatively routine criticism within DA circles is that themes and discourses do not just 'emerge' from transcripts, rather they are constructed by the analyst. Like Hollway (1989) and as I have described earlier in my discussion of my analytic method, I found both that my codes and categories changed and that the connections between them changed too. In the end I made a selection of three analytic themes for the interview analysis chapters. At the time and, on reflection, these themes felt the right choice in terms of the aims and preoccupations of this study and I could (and have elsewhere in the thesis) justified this but this does not get away from the fact that the analyst has to make choices. There is a dual danger here in representing more open-ended approaches to analysis: on the one hand it can be made to seem arbitrary when it was not but also it can be made to seem tidier and more matter-of-fact than it was (Hollway, 1989). There is a need to represent that this analysis was, in Woolgar's (1988a) words a struggle. Moreover, the analytic choices I made had certain consequences. Thus I did not focus on other aspects which I could have done (eg the various rhetorical strategies around what users and professionals saw as the causes of paranoia). Looking back, I can see that some interview transcripts were used more than others because they were better examples of some of the themes in which I was interested. Here too, there were further decisions since I chose to focus on broader narratives rather than individual cases -- Georgaca (1996) in contrast, used a more temporally contextualised and individually-focused analytic approach.

The processes of analysis and of 'writing up' the research were the same in this piece of work. My experience of writing the chapters (and, for some chapters, the journal articles that formed their basis) reflected my experience of doing the analysis. Thus, at times I forgot what DA and deconstruction were. Whenever I started a chapter I felt a continual movement between feeling I had nothing to say or feeling there was too much and that I would not be able to structure a coherent and linear argument; a tension between restlessness and paralysis. Henwood & Pidgeon (1992) have talked of the cycle between openness and closure in qualitative analysis. There was a search for structure for the chapters to organise my thoughts and data and then a revision when it needed developing.

I am writing about this experience not to 'confess' my shortcomings as an analyst -- I seek only to be a 'good enough' discourse analyst following Scheper-Hughes' 'good enough' ethnography (1992, p.28). Rather, I am trying to reveal some of what Figueroa & López (1991) note is lacking in much discourse analytic work: attending to the methodological process by which the analysis was arrived at.

Further challenges in the analysis were posed by the fact that I knew some of my professional interviewees already through my work (three of the four psychiatrists, two of the CPNs and some of the GPs) although not as friends. I was, as I have already noted, both linguistically and physically part of the local psychiatric culture. Despite this, it was still possible to be critical of the assumptions implicit in psychiatric accounts -- although user and humanistic psychological accounts were probably more difficult to unpack. My role as both practitioner and discourse analyst led to a curious dualism and I began to feel a little like Sampson's (1993) tight-rope walker. Barrett speaks of this experience of living in two worlds -- in his case as both a psychiatrist and an anthropologist:

It was an undoubted asset but also an impediment; my familiarity with the world of the psychiatric hospital and its language often made it difficult for me to perceive the taken-for-granted assumptions on which that world was built. It was only with the assistance of my anthropologist colleagues, who continually insisted that I maintain a sense of curiosity about what I normally regarded as self-evident, that I was to make use of my cultural competence in an analytic way.

Barrett (1996, p.xvii)

For me too, the discussions with my supervisor, the reading of others' work, writing at home (rather than at work), writing for publication and discussions with others helped with this process.

Burman & Parker (1993) note the ethical issues involved in interpreting others' words. Stainton Rogers (1991) also notes that 'in order to weave my story, I must inevitably do violence to the ideas and understandings as they were originally expressed' (p.10). Questions in, and following the analysis for me were: 'have I been symmetrical and equally respectful and non-blaming of all participants?'; 'has there been any creeping intentionalism?'. Many commentators have noted the care they have to take. Most often the only comment they make on this is, in relation to those they have interviewed or from whom they have gathered information, to 'hope they will forgive me for the use to which I have put it' (Rose, 1989, p.xiv) or to trust that they 'have done them no further violence in the rough and impressionistic strokes I have left on this canvas' (Scheper-Hughes, 1992, p.xiii).

Of course, this danger is highlighted when DA is used, as it has been here as a critique and this can induce defensiveness in participants when the research is fed back to them (Marks, 1993) -- this issue will be addressed in more detail in the following section. Indeed in a reviewer's comment on a draft of an earlier article for the British Journal of Medical Psychology (Harper, 1992), I was informed that my account confounded a useful conceptual debate with:

less convincing arguments in favour of the author's views about a conspiracy theory involving the serving of professional interests ... [the] arguments presented are extremely biased in favour of conspiracy theory occasionally to the point of absurdity.

Anonymous (1992)

This passage is ironic (especially given my discussion of this in chapter 5) in that I, as a researcher into paranoia, was implicitly accused of being paranoid. It is doubly ironic since I can rebut this challenge by making the same claim of the writer since I did not claim there was a conspiracy in my paper. So, I could argue, 'this person sees accusations of conspiracies everwhere'! This paradoxical situation introduces us to another reflexive aspect of this research: that it looks at the deeper meanings -- underneath the surface, as it were -- of talk, utilising a hermeneutics of suspicion (Burman & Parker, 1993; Ricoeur, 1970) in a manner that some might term paranoid (Bywater, 1990; Fisher, 1992). There are further ironies here. For example, what of my 'firm beliefs' as a researcher which I examined in relation to others' work in the previous chapter? My response to this would be that I am aiming here to detail my interests so readers can take them into account. Also, I am not claiming to write, as some traditional researchers do, about paranoia as something others do, rather that it is a form of discourse (as are the other discourses which construct it) that all can and do use. Moreover, my discussion of judgements of plausibility, for example, included an analysis of my judgements as well as those of others.

Here I have examined some of the ways reflexivity influences analysis. In particular I have highlighted how analysis involves choices and how those choices have consequences. Of course one choice is about what kind of analysis one undertakes and this has implications for whether or not research will be fed back to research participants. Reflexivity and the dilemmas of feeding-back

Burr (1995) notes that one of the consequences of constructionist work has been that since 'respondents' are of 'equal status' with researchers they need to be built in to the research. For example, many workers see participants' validation of a researcher's account as a criterion for quality control (Potter & Wetherell, 1987). However, Sherrard (1991) notes that such validation rarely happens. Moreover, the process of feeding back to participants may not not change the analysis since the analyst's interpretation may be given most weight and there may only be an illusion of democratisation (Marks, 1993) or the process may simply lead to a proliferation of readings with no final settled version being possible, only a continual state of motion (Potter, 1988).

Smith (1994) attempted to address some of these difficulties by adopting a stance of co-operative inquiry in which he shares his analysis of an interview extract with the interviewee, or co-researcher. Hollway (1989) holds out the possibility of democratic analysis by sharing interpretations with informants but in her research she self-selected her participants on the basis of their ability to engage in a form of self-analysis. An interesting point Smith (1994) makes is that whether to do this and if so, how, depends on the aim of the research project. If one is attempting to produce an account that fits with what interviewees think (as ethnomethodologists might seek to do) or which might even be useful to them as a description, then feeding back is essential. Burr (1995) over-generalises when she talks about respondent validation as an essential hallmark of social constructionist work. One might, instead aspire to selective validation (eg the validation of those lower down the status/power hierarchy (eg users of psychiatric services).

This raises a number of issues. For example, what if interviewees have different views on the analysis? Those in more powerful positions might be threatened by an account which revealed the employment of that power whilst those in more powerless positions might be happy with this. Kitzinger & Wilkinson (1997) have raised a number of related difficulties of validating interviewees' experiences. Further, Coyle (1995) notes, given the poststructuralist assumption that meaning is not fixed, that language use may have consequences the speaker did not intend -- thus they may invoke discourses and ideologies of which they were not aware.

It is worth noting that accounts are written with particular audiences and aims in mind. For example, here I am writing with the aim of producing an account that meets my own aims for this research and with conventional rules about academic thesis writing in mind. I would write a very different account for a popular magazine, for a professional journal and for a user movement newsletter. Moreover, they are not just different representations of 'the same' analysis but different analyses. So there is no one analysis that it is possible to feed back. In an earlier piece of DA work a resulting paper (Harper, 1994b) was very different from the thesis presentation (Harper, 1991). There is also the issue of what aspects of my analysis to feed back. What would I include and what would I leave out? Would I expect all my interviewees to read my thesis? In an attempt to feedback the results to one of the interviews for an earlier piece of work (Harper, 1991) I was faced with the dilemma of describing in detail the account in my thesis and risk being inaccessible or making it accessible at the risk of making the account seem banal. In the end I erred towards the latter with the result that the interviewee found the analysis banal. The analysis presented here is inevitably an academic one. To make it more accessible is a process of continual revision, distillling what the most important aspects are for different target audiences. At the moment, my target audience is primarily an academic one -- and, as Willig (in press) notes, this is characteristic of much critical discourse analytic work -- but I would certainly wish to be producing analyses for other target audiences (like service users) via other outlets (like user newsletters). There is also the danger that, as with reflexivity, saying your participants agree with your account gives you some insurance against challenges rather than letting the account stand for itself.

There is a dual danger in the feeding-back (or seeking of respondent validation): the researcher thinking the research was so important to the participants that they need to see the researcher's account in order not to be disempowered which I would see as a little narcissistic; and on the other hand, thinking the research was not important and did not have an effect on the participants. I think practical considerations play a part here. I would see some post-research discussion as important if the research had involved taking up a good deal of the participant's time, in order to explain my analyses and seek their views. However, I think that when one has spent less than an hour with most interviewees (sometimes only ten minutes) one would risk testing participants' patience and goodwill by seeking to feed back and take up even more of their time. The research is important to me but not necessarily to them. Moreover, as Smith (1994) notes, there are practical constraints on the ability to be reflexive. As in his case, I too had others to interview, a thesis to complete and, in addition, a day job. Of course, this whole dynamic might be reversed in an action research project formulated by users where discussing findings with participants would be a central part of the project since they would be its commissioners.


We cannot get iffy over other people's power-games with language, and then pretend we are not players in the game too.

Curt (1994, p.19)

As I have noted throughout this chapter, the analysis and arguments in this thesis are not abstract, they are the product of material practices of reading, interviewing, transcribing, writing, speaking and so on. Sampson (1996) has argued that embodiment pervades all human practices, including discursive practices. Thus whilst words construct the body, those words are themselves embodied. For him, we should not focus on talk about the body but rather on embodied discourse, including how discourse about discourse is embodied.

This raises the issue of how my discourse about others' discourse is embodied and focuses attention not just on the content of my argument and its rhetorical processes but also on material influences. A major material influence on my writing was the actual amount of time I spent on writing and how frequently I was able to write. Other calls on my time (eg my employment as a clinical psychologist and my personal life) meant that my writing existed in a particular organization of time and space and this influenced my writing in a number of ways. On the one hand this helped me to bridge the world of work in psychiatry and the world of analysis of it. However, it also meant at times that I was prevented from becoming immersed in the analysis of detail, a bit like trying get a good sleep but being constantly wakened.

Another influence on my writing was the need to develop a coherent, consistent and linear argument, to tell a particular story. In doing this, one marshalls evidence and makes a case. So the reader needs to read this thesis with an eye on the rhetorical strategies I use to produce linearity (eg of plausibility and persuasiveness). Stainton Rogers (1991) notes that the implication of telling a story is that the one that is read 'is a distorted and partial version that I have deliberately constructed for you in a particular way, for particular reasons' (p.10). Scheper-Hughes (1992) comments that what emerges from writing is 'highly subjective, partial, and fragmentary' (1992, p.xii). Stainton Rogers suggests that she is not telling it 'like it is' but rather saying 'look at it this way' (p.10). In this process, many things are left out (though this may seem hard to believe, given its length!) and, like Light (1980) I have had to focus on only a narrow range. But, as Scheper-Hughes notes 'we struggle to do the best we can with the limited resources we have at hand' (1992, p.28).

One particularly difficult issue to address around the topic of reflexivity is of linearity. One is continually tempted to step back from the writing and to reflect on the written account or to go off at apparent tangents and to make a connection elsewhere. What is lost for the sake of developing a linear logical argument? We have noted that there is never a definitive 'finished' version. Rather, analyses are, to some extent, provisional since they are the products of a continual iterative cycling between reading (of one kind or another) and writing. This process is punctuated by practical requirements like the writing of a thesis. So, this thesis is a snapshot of something that is actually dynamic. However, I would not wish to claim that this process is infinite and it is likely that there are some limits to interpretation -- thus Curt (1994) note the emergence of a limited muber of stories in any one concourse.

Given this provisionality, there is a danger of linear accounts promoting a textual singularity (Stainton Rogers, 1991) implying that there is a pure explanatory system. Moreover there are a number of difficulties when we write as singular authors since this leads both to monologue and to the temptation of being-an-I rather than being-an-us (Pujol & Capdevila, 1995) -- and of course 'I' have already fallen into this trap in writing this piece from 'my' standpoint (Harper, 1995b). This is a danger since analytic ideas often come from dialogues of one kind or another (eg reading others' books, E-mailing, talking and corresponding with others etc). When in dialogue with others -- as I was at the 'Workshopping Social Knowledge' conference in Barcelona in 1994 -- it becomes harder to develop a theoretical monologue (a kind of unitary rational text) where we advance a series of points. When in dialogue this process is undermined since others argue, take different positions, force us to use their language and concepts or colonise ours, push us to extremes and so on. It is much harder for language to remain theoretically 'pure' when others are involved in the discussion. On paper we might individually, or in groups, advance particular theoretically consistent arguments but conversations range over vast territories of debate, are uncertain, ambiguous, refuse to come to a conclusion, and, more importantly, are unauthored.

As well as a wish to be reflexively aware, another of my aims is to develop practical implications and to be accessible to different audiences. For me it is the political and ethical ends we have that are more important than theoretical 'purity' though, of course, I am constructing an opposition here. This is easier to say than to practice, since theories, values and politics are continually in conflict internally and externally. A more dialectical and pragmatic approach might be to acknowledge that different positions in discourse are not right or wrong but simply enable certain possibilities of thought and action together with certain dangers.


As the clamour of competing voices and conflicting versions of reality increases, so the plausibility and political utility of claiming special truth-statuses for one or other of these voices decreases.

Curt (1994, p.20)

Rhetoric is both the means of enquiry as well as its object. Because of this, and not despite it, there is the possibility of critique. The analyst, at a remove, can join in the arguments, which are being examined, and can argue about the arguments ... In the last analysis, or rather in the first analysis, critique depends upon the argument produced by the analyst.

Billig (1991c, p.23)

Ontological gerrymandering is a charge levelled against those discourse analysts critical of the political relativism they fear DA can lead to. Gill (1995) notes that relativists criticise realist DA researchers for being only selectively relativist: ie following a relativist and constructionist style of analysis but then implicitly claiming their account is true and refusing to see some things (eg death and furniture) as equally constructed -- what are known as 'bottom line arguments' (Edwards et al., 1995). Thus, in the mental health arena, Hak (1992) has criticised critical sociological evaluations of psychiatry because 'in the end this evaluation turns out to be nothing but the comparison of the psychiatric account with another account, the sociologist's' (p.141).

The danger is that practical and political interventions then become problematic: if all accounts are equally valid how do we choose which to act on? Gill, quoting Hine, talks of the danger of 'epistemological correctness' (1995,p.173) and notes that some relativist discourse analysts seem to want to return to the kind of disinterested enquiry reminiscent of the positivism DA was supposed to be an alternative too. Certainly, the debate about relativism versus realism is a peculiarly academic one involving few practitioners -- contributors to Parker's (1998) edited collection on this topic are predominantly academics -- perhaps because practitioners are not in a position to choose whether or not to intervene, they are already intervening; for them the choice is how to intervene. Moreover, by setting the debate up as one of realism versus relativism we create an unhelpful opposition -- a more helpful approach would be to look at the positions people take in particular contexts rather than talking in generalities. There are, for example strong social constructionist versions of therapy which, although drawing on relativistic positions would still count as an intervention (Anderson & Goolishian, 1992). For me there is no contradiction between the use of social constructionism and having a political position. Reflexivity is a means and not an end. The charge of gerrymandering is only accurate if I claim some accounts are more real than others. I do not want to claim this -- rather I simply want to claim that some accounts are better than others (Kitzinger & Wilkinson, 1997). As Gill (1995) notes:

There is no evidence that appeals to reason or truth are uniquely effective as strategies for bringing about change ... What we -- as feminists -- want is not truth but justice.

(1995, p.178, emphasis in original)

I am not too concerned about producing an account which is more true than others, but I would want to claim that my account and some of its implications are better than others (and I will develop this further in the next chapter). So, the question becomes 'are my analysis and implications persuasive, taking what you know of my assumptions and politics into consideration?'.

1. Aitken & Burman (in press) discuss some of the complexities involved in negotiating boundaries around the positions of researcher and clinician.

Back to title page