Chapter 8 ? Analysing Discourse Martin Barker The seemingly inexorable rise of the concept of ?discourse? has made it almost unavoidable for cultural studies researchers, particularly since its invitation to theorise culture as ?like a language? coincides with so many impulses within our field. But not without substantial costs. Looking at the cultural studies field from my angle as an audience researcher, some troubling features within discourse work come into view. For all the multiplicity of approaches, and the attendant variations in attached modes of ?discourse analysis?, there are some powerful unifying features in ?discourse talk?; and these features presume the very things that as an audience researcher I have to question. Very crudely, if the predominant theories of discourse are correct, my research field becomes ?impossible?. There are embedded assumptions about the ?powers? of discourses, about how discourses ?work?, which are powerfully disabling. There is a further problem, seemingly unrelated to the first, of the ?convenient sample?: that is, the choice of cases which suit a researcher?s pre-given position and purpose, and which cannot allow a test of these. How do researchers know what ?texts? or bodies of materials to choose, for analysis? To whom are they relevant other than to the analyst? This too has dangerous entailments for the possibility of audience research. It is time, in my view, to expose these assumptions and to unshackle discourse research from their influence. These issues have become particularly alive for me in the last four years, as I began with colleagues to plan for, conduct and assemble, and then analyse a vast body of materials within the international project on audience responses to the film of The Lord of the Rings. This project, which is being published in a range of forms and places, has required us to find or develop very detailed methods of discourse analysis in order to bring into view the differing orientations of a great range of kinds of audiences in varying cultural and country contexts. In this chapter I draw upon the insights I have gained, from being involved in these processes, without either directly addressing our detailed solutions or reporting any of the resultant findings. For any who are interested, I have footnoted some of the main places where these can be found. The Growth of Discourse Work Over the last 30 years discourse theory and analysis have grown from a minor specialist area to one of the most pervasive and multifarious academic fields. WorldCat is the nearest we have to a complete database of all publications in the English language. A simple search at five year intervals for book titles containing the word ?discourse? suggests a steady rise to its current prominence: Table 1: English language book titles including the word ?discourse?. Source: WorldCat 1965 1970 1975 1980 1985 1990 1995 2000 2004 70 144 157 327 360 449 651 762 616 This accelerating growth across the period 1980-2000 is striking, albeit it may be stalling now. But if we consider related journals, which emerge as a field consolidates and becomes organised, the picture becomes more complicated: Discourse (founded 1980) Discourse Processes (1981) Text: an Interdisciplinary Journal for the Study of Discourse (1981) Discourse and Society (1990) Discourse Studies (1999) Discourse Analysis On-Line (2003) Critical Discourse Studies (2004) Journal of Multicultural Discourse (2006) The first ?clump? of development marks the emergence of sociolinguistics, and of the cognitive sciences ? but my suspicion is that these are only weakly related to the kinds of book publication then appearing. In that period of accelerated book publication between 1975-85, a large number of the books derive from the crisis in academicist Marxism, and its replacement of ?ideology? by ?discourse?. The delay before the second round of journals emerges arises precisely from the fact that this was less a research tradition, than an expression of altered political concerns. The ?field? that emerges thus acquired some distinctive, peculiar qualities: ? A considerable amount of renaming went on ? in various parts of the field, from ?ideology?, or from ?text?, or from ?structure of feeling?, to ?discourse?; ? This field is very self-aware, philosophically and epistemologically ? it is not easy to work in it without entering fundamental debates about the relations between ?discourses? and the non-discursive; ? It is a field of contentions, with sharp and continuing clashes between, for instance, sociolinguists and conversation analysts on the one hand, and discourse theorists and critical discourse analysts on the other; ? It is a field very concerned to be culturally and politically relevant yet deeply worried about its warranty for taking political positions. In fact we might characterise it as a motley domain, made up of scholars who probably cannot agree on any fundamental definitions, yet all of whom are drawn to certain questions, which are seen as of particular relevance today. These questions concern the nature and role of language and other meaning-systems in the operation of social relations, and in particular the power of such systems to shape identities, social practices, relations between individuals, communities, and all kinds of authority. And the reason for the centrality of this topic of power surely arises from the ways in which discourse work emerged from the collapse of academic Marxism, the rise of alternative social movements theorised by near-simultaneous academic constituencies, alongside the ?cultural turn? in various fields of the social sciences. Just about every writer about discourse theory acknowledges the diversity within the field, although they may cut the cake differently.1 Whilst acknowledging the helpful distinctions various authors have made, I have found it most useful to ask a series of questions of different kinds of discourse work. This has led me to distinguish seven main tendencies2 and to tabulate their different answers to my questions as follows. Table 2: THEORIES OF DISCOURSE Saussurean Structuralism Lacanian Post- structuralism Foucauldian Theory Rhetorical Psychology Conversation Analysis Critical Discourse Theory Volosinovian Dialogism What are its key founding claims? Language is an arbitrary system of differences The formation of language is the formation of gendered self Institutions construct their subjects as objects of knowledge Language is always a function of contexts of talk Spoken interaction is a complex achievement Linguistic forms are constitutive of ideological positions People use language to form groups & conduct social struggles What forms of language are seen as primary? Forms of classification, descriptive systems Constructions that relay and relate 'Self' and its many relations to 'Others' Language embodying institutional processes and knowledge Ordinary talk embodying social arguments and beliefs Ordinary talk expressing patterns of exchange Embedded grammars expressing power/ ideological relations Shifts between linguistic struggle and conceptual sedimentation What is the main purpose/ function of speech in society? As samples of current synchronic structure To embody Self in its relation of Desire to Others To constitute subjects through institutions, shaping their responses To respond appropriately to social contexts To be and to embody complex social relations To fix or to challenge the fixity of linguistic codes To form and carry through social projects, and form communities How is linguistic and conceptual change explained? Secondarily, as a function of shifts in synchronic order With difficulty, mainly by 'consciousness- raising' techniques As a function of shifting institutional imperatives As social contexts change Change is the norm, fixity only relates to general speech patterns As a result of struggles between 'langues' and 'paroles' As an outcome of social struggles, emergent communities How is language related to thought? Language is thought, virtually Language is identity/ies, virtually Language is related to regimes of truth which measure thought Language is determined by public contexts and knowable, thought is un- researchable Language is the 'front-end' of back-region unseeable processes Language shapes and 'determines' thought People orient to languages, actions provoke us to check them by critical thought How is 'power' in language conceptual- ised? As a direct function of structures, 'langue' precedes parole Language is a function of desire which powerfully constructs the world and identities As a function of regimes of truth, constituting forms of resistance also 'Contexts', both lived and ideological, are powerful Power is over the situation of talk, directing the flow of exchanges As embedded grammars which remain unexposed, hence unchallenged As persuasion in real situations, tactical control of concepts and exchanges What can discourse analysis reveal, and what can it prove? Reveal synchronic systems ( their power is already 'known') Reveal the operation of gendered constructions (their power is already 'known') Reveal institutions at work (their existence is already 'known') Reveal contextual/ ideological determinants, and strategies of response Reveal how people manage their social interactions, proving little beyond that Reveal social/ ideological grammars (their power is already 'known') Reveal new aspects of processes and terms of struggles, and how people orient to them Exemplars of each approach's ways of analysing John Hartley, Approaches to News, Judith Williamson, Decoding Advertisements Frank Burton & Pat Carlen, Official Discourse John Tagg, The Burden of Representation, Stephen Heath, The Sexual Fix Jonathan Potter & Margaret Wetherell, Discourse and Social Psychology Charles Antaki, Analysing Everyday Explanation Gunther Kress & Tony Trew, 'Ideological dimensions of discourse' Marc Steinberg, Fighting Words; Chik Collins, 'To concede or contest?' This way of thinking the field has at least two advantages. First, it addresses the relations between definitions of discourse, their ontological and epistemological assumptions, and the associated questions and methods. Second, it brings in the one approach that other accounts tend to leave out ? that deriving distinctively from the work of Valentin Volosinov. Work deriving from Volosinov still considers issues of power to be crucial, but because of its general theorisation of language, the nature of that power has to be considered an empirical question. By contrast, and with the one exception of Conversation Analysis, the other strands tend to treat ?power? as the central ?given? of discourse.3 It is this I wish to address. But rather than address this purely at the level of theory and definitions, I prefer to look in detail at actual examples of discourse analysis. For this purpose, I have chosen two books as test-cases. (1) Chris Barker & Dariusz Galasi?ski, Cultural Studies and Discourse Analysis (2001). Barker and Galasi?ski offer a useful test-case. Their book is, if you will, a manifesto, urging upon cultural studies scholars the benefits of complementing a presumed-to- be-agreed set of theoretical positions, with methods for close empirical analysis of cultural materials.4 The recommended method is critical discourse analysis. But these methods will not test any of the core theoretical claims ? these are ?known? on other grounds. They include a range of philosophical positions (about self and identity, about the nature and role of language in society, and about the wish for cultural studies to ?give voice? to disadvantaged and silenced groups), and which derive from a pantheon of recognised theorists (Saussure, Pierce, Wittgenstein, Barthes, Hall, Foucault, Lacan, Butler, etc). The role of methods is limited, it seems, to detailed illustration. The rest is ?interpretations?, which are essentially a matter of position. At the heart of the authors? approach is a stance that I would want to challenge, as either endlessly ambiguous or just plain wrong, a stance which is captured in their summative acceptance of the ?argument that language is constitutive of subjectivity, identify and our cultural maps of meaning? (p.47). There are many issues buried in here, but the one which concerns me here is the deterministic language. Throughout the book it is possible to find repeated instances of words assuming specific kinds of causal relations at work within culture: people are apparently ?constructed?, ?impelled?, ?constituted?, ?interpellated?, and so on. The book?s first three chapters lay out what the authors regard as this ?agreed territory?, followed by an account of critical discourse analysis, and then some extended applications to a CIA document; to some interviews with men about fatherhood, and with elderly Poles about their attitudes to Ukrainians. For the sake of focus, I look at their first example. Over fourteen pages, they scrutinise a Credo posted on the CIA?s own website, using various CDA tests for social grammars (in particular focusing on the recurrent use of ?we?, and the self-attributions this implies). This scrutiny unquestionably leads to a richer description of the document. But then comes a claim which goes beyond description. Here is what they say: Let us take a look at the thematic and information structure of the CIA Credo, beginning with the first sentence: ?We are the Central Intelligence Agency?. As we pointed out earlier, the theme of the clause is the recurrent ?we?. However, unless we can actually see a group of people, we cannot simply accept the ?we? as given. Who, we might ask, are the people saying this? There is no way to tell. So why not start by saying ?this is the Credo of the Central Intelligence Agency?. The answer lies precisely in the given status of the theme. The text proposes that we, the readers of the text, know who is talking. This is a strategy quite consistent with opening a Web site and working on an image of legitimacy, law-abidance, and transparency. What follows from this is an interesting exercise in locating the CIA as the given of the text and the rest as the new. In other words, the CIA assumes that its audience knows merely of its existence and nothing about what it does. This is a fascinating finding when one considers that the Credo, like other corporate texts, is displayed for public consumption on the Internet. Yet, we would speculate that this apparent glitch in the form of the Credo is probably well worth it. Thanks to it, the CIA not only establishes itself as a known, taken-for-granted part of American life, but presents itself as a unity defending American values. The average American can sleep peacefully knowing that the CIA is out there making sure that American interests are well-served. (2001: 83) There are several problems with this account, and its implicit moves. The most obvious is the slipped-in rhetoric of this ?average American? who appears to be someone who can be ?spoken? to by the CIA. An abstract figure, what qualities might s/he have? And what is it that makes her/him so open to being constituted into comfort by this document? It is not just that we do not know, but there is no interest in finding out. This aside, other implicit assumptions underpin this move. The most troublesome are these. It seems that the capacity to be affected by these discursive elements is a function of a motiveless encounter. The reader is not looking at this web document for any reason, or with any purpose ? they are just looking. It also seems that they have never before seen a document of this kind. If they had, generic knowledge that ?this looks like one of those Mission Statements? might kick in, making otiose the distinction which is vital to their move: that ?we? might only know of the CIA?s existence but know nothing of its nature. If we recognise this as a Mission Statement, we probably know something about the rhetorical functions of such Statements, and a number of things about the kinds of organisation that produce them. This example not only illustrates just how deeply embedded are these moves and implications. It also suggests that they gain their persuasiveness because they are backed by those wider philosophical position-takes. Discourse analysis naturally generates ?images of the audience? which require no testing. Let?s see how this works in a more determinedly ?empirical? book. (2) Carla Willig (ed.), Applied Discourse Analysis (1999). Willig?s collection is an example of the best that critical discourse analysis offers to cultural studies. I don?t mean that cynically. The book contains much that is very valuable and instructive. Even so, each of its essays reveals blind-spots. In her introduction, Willig is acutely aware of the many epistemological and political problems of ?applying? discourse analysis to live problems. But two things are strikingly missing from her worries: any examination of the issue of the ?power? claims in discourse theory; and any consideration of how the truth-claims of discourse analyses might be tested. These absences re-emerge in the essays, which I want to examine5. I spend longer on the first, only because close attention to that saves time on the remainder. Steven D Brown explores the discursive organisation of self-help books ? the kind that tell you how to cope better with stress, and to make the most of yourself, especially at work. He asks: ?how do these texts exert their effects?? (24). He notes their near-didactic organisation, their constant ?prompts, suggestions, encouragement? (23) to examining yourself, seeing stress as something to be addressed within yourself (as against, for instance, challenging stressful working environments). This leads to him locating an idea of a ?serviceable self?, a managed production of oneself that will cope, be flexible and productive, and have a sense of self-worth from achieving this. But all the way through the essay, the analysis of the books is accompanied by an unanalysed figure. ?The reader? enters at the moment when s/he opens one of these books, after purchase: ?Let us leave aside for a moment what motivates people to buy or read a self-help book about stress. Consider instead what happens as they work through the text.? (23) Thereafter this figure gradually accretes attributions. Some are soft and casual: regimes are ?presented to the reader? (35, no implications of response); the books ?encourage readers to develop a particular relationship with themselves? (37, slight implications). Others slide further: ?The reader must accept the serious nature of the choices that he or she is making ?? (31); ?the use of devices such as heat serve to make stress visible in a way that is immediately explicable to the reader? (32); ?computational metaphors ? impress on the reader the importance of understanding mental operations during stress ?? (34 ? with the curious implication in the last two that these rhetorical resources are evidently effective, a claim I find curious to say the least). This is perhaps Brown?s strongest version: A grand gesture of extending wisdom and guidance is played out. The reader must further accept the serious nature of the choice he or she is making, and that this involves assuming an active role: ?This is not a book for hypochondriacs. It is for people who enjoy being healthy and are prepared to help themselves to remain healthy? (Eagle, 1982: 6). The work of staying healthy is purely a matter of personal responsibility. The texts offer help, but on the proviso that readers fully accept that the problem lies within themselves. (31) Thenceforth, this ?reader? becomes increasingly a textual construct, an ?addressed? and ?positioned? empty figure ?only becoming nearly three-dimensional in the closing paragraphs, where Brown asks: ?Clearly we need to understand just how this transformation then plays out when it becomes an accepted feature of labour relations?. (38) But notice how readers? motivations, and uses of the books, have now been subsumed within a disciplinary model: they meet these books within the context of labour relations. Indeed he makes the curious assumption that those most likely to read these books are the victims of such work environments, rather than (as I suspect) their managers, trainers, supervisors. The books? ?power? is thereby virtually guaranteed. Thus Brown?s safety clause (?Readers do what they will with them? (40)) becomes a rhetorical closure, instead of an invitation to possible testing research. And that to me is the real issue. Let me summarise the problems I am pointing to. Brown?s interpretation of these books only works if we share certain premises with him. (1) Premises concerning unity and coherence: he is forming them into a genre, assuming that the average ?reader? will receive them as working and meaning in the same way. That makes two further assumptions: (a) it assumes that readers will not see important distinctions among them ? perhaps in style, or in applicability of metaphors, examples, and regimes ? and thus generate their own genres, responding to some and perhaps rejecting others. (b) It assumes that there will thus be cumulative influence: ?[t]he more readers begin to revise their grasp of their personal circumstances in terms of stress? (24), the more they will be ?positioned?. Treating them as unified allows Brown to claim a ?discourse?, and thence to presume without evidence their ?power?. (2) Premises concerning persuasiveness: he has presumed on the effectiveness of their rhetorical organisation. For example he writes that ?it is the very vacuousness of the terms stress and energy which makes the mixing up of these discourses possible? (37). That permits him to move to arguing that this emptiness allows the books to be ?all things to all people?. The assumption here is that what he as discourse analyst can see as vacuous, not only is not visible to ?readers?, its invisibility is the very ground of their incorporation. Thus, discourse analysis perversely builds in a presumption of the effectivity of the patterns it ?discloses?. (3) Premises concerning investigative completeness and testability: all these are posed in ways which hide the possibility of testing. The implications are there, but are never sign-posted. They remain half- buried, with the protective stricture that this is ?only an interpretation?. This for me points to two further sub-questions: (a) the issue of researchers? responsibility for consequences and implications of their strong claims. It cannot be right that at the first point of critical enquiry an analyst is entitled to say ?This is only an interpretation, you are pushing my account too far?. That would lead to the most sterile form of relativism imaginable. (b) A further assumption, less evident in this case but vital later, is that his own method of analysing the books is trustworthy. Of course, the essay format makes it hard to demonstrate methods in depth. But that cannot remove the questions involved here: what larger investigation of their meaning-making processes underlies his presentation? What guided his list-making, and how complete was this? Has Brown examined the books rigorously, in the sense of attending to their overall organisation and direction, rather than lifting for quotation favoured but marginal elements? Has he examined differences as well as similarities within his ?genre?? All these have to be taken on trust in a way that they would not be in more ?conventional? modes of research. These issues are replicated, albeit with differences, in other essays. Timothy Auburn, Susan Lea and Susan Drake offer an account of the discursive practices at work in police interviews. They look in particular at moments of explanatory disjuncture, when an interrogator points to discrepancies in a suspect?s account, and demands they account for these. Using recorded interviews, they draw attention to a series of rhetorical devices used by the police ? such as urging suspects to be honest with themselves, or to see how another person would look at their account, or to think how an expert would evaluate their explanation. From this they develop an account of the police interview as the discursive work of producing an administrative knowledge of the events; and suggest that discourse analysts should ?take sides? by teaching this kind of understanding to groups (gays, mental health advocates, trade unionists) who suffer regular harassment by the police. All this is interesting, but the revealing ?slip? comes when they sum up a police line of questioning as follows: This concern about the lack of intersubjective agreement on ?what really happened? is warranted by a particular selection of features of the event which give rise to inferences that there are discrepancies between the two available accounts. The production of a discrepancy in turn relies upon carefully crafted fact constructions so that the discrepancy becomes a plausible inference from the selection and meaning of the ?facts? as part of a wider narrative of the events. (54) Pause on this. Just what would a ?carelessly crafted? or ?implausible? version look like? Is there any way in which, for discourse analyses of this kind, rhetorical moves can fail? I do not think they can, because these accounts assume the productive coherence (as against the tactical, and contestable nature) of official discourses. (It is interesting that the extract which precedes and leads to the above quote ends with the suspect simply repeating ?No I didn?t do it?.) Val Gillies explores women smokers? argumentative strategies for not giving up in the face of their acceptance of health arguments about smoking. Using quotes from four women, she draws attention to the way they talk about ?addiction?. This is a discourse, she argues, whose ?most powerful effect ? is to provide a deterministic explanation that emphasises the smoker?s lack of control over her actions? (71). This couples with her talk of discourses such as ?addiction? as ?containing? and ?positioning? individuals. Their use of these languages shows they are victims because of their discursive domination. However with one woman at least, ?Mary?, she notes that ?addiction? is only one among a number of other strategies. Mary also says she is ?not as bad as some others?, is ?able to say no? if offered a cigarette, and ?isn?t bothered if people ask her not to smoke?. All this might be seen to suggest that Mary is not ?positioned? by these languages, but is calling ? almost at random, and certainly without adherence ? on a range of discursive resources, and it does not matter to her that they might be seen as incoherent and contradictory. But Gillies does not go this way; instead, she takes from Mary a passing reference to a ?gradual brainwashing thing? to reassert that Mary has absorbed and been constructed as victim by a discourse of ?lack of control over herself? (74). This allows Gillies to go to examples of Health Education literature in which, she argues, there is a ?prevalence of a discourse of addiction? (81), and even beyond that to much wider ?concepts of self- control and restraint? (82) within which the discourse of addiction then finds a home. Mary?s references to addiction thus become symptoms of discourses located by other means. But we do not know, and apparently do not need to know, if Mary has ever encountered ? let alone absorbed ? any of these wider discourses. This brings into focus what it means to identify a ?discourse?. What standards of evidence are required to ?name? something as a coherent, effective discourse? And what standards then apply to knowledge about people?s encounters with those, sufficient to count as having been ?positioned? by them? That, of course, is among the tasks that audience research has set itself. Willig?s own essay presents some outcomes of a larger project on the discourses of safe sex. Drawing upon interviews with heterosexual men and women, she identifies a series of discursive frames which they use to explain how they make or perhaps would make decisions about ?safety?. These include: marital safety (?I wouldn?t be with him/her if I didn?t trust them?); trust (?it would be very hurtful to suggest I don?t trust her?); and problems of interrupting a romantic encounter at a critical moment. She also identifies the devices that people use to distinguish ?innocent victims? of STDs from other people. Willig opens by counterposing a discourse analytic approach to conventional social cognition approaches, concluding with these comments: ?Social cognition models have received limited empirical support. They can account for up to 50 per cent of the variance in declared intentions to adopt health behaviours but only control up to 20 percent of variance in actual behaviour?. (112) This is an apt and perfectly valid criticism of social cognition approaches, and it references the long tradition of research into the gaps between people?s declared attitudes and their behaviour. What is striking is the absence of any wish to mount equivalent tests of a discourse approach. Why? After all, in theory, discourse theory has a distinction quite closely matching that between attitudes and actions. In the book?s introduction, Willig distinguishes two regions of discourse work: the investigation of discursive practices and of discursive resources.6 Discursive practices are the ?local? communicative regimes which individuals and groups use, in ordinary communication. Discursive resources, on the other hand, are more widely distributed. Because of this, they are more obviously researchers? constructs, but still make strong claims on reality in that we try to understand local discursive practices through them. There is nothing wrong with this double articulation, providing it remains double, and thus open to tests. But in Willig?s research the distinction collapses. Having discovered that her respondents use these explanatory frames, she concludes that they are ?predominantly disempowered? by them: ?The marital discourse positioned spouses as safe by definition, which meant that talk about sexual safety constituted a challenge to the nature of the relationship itself. Those who position themselves within this discourse are required to take sexual risks with their partner in order to negotiate a trusting relationship?. (118) First there is that slippage between apparent choice (?position themselves?) and apparent domination (?positioned?). But then there is the fact is that this argument elides the very distinction which Willig insisted on earlier in relation to social cognition approaches: between intention and action; between talk and behaviour. In the absence of other kinds of testing investigation, we simply don?t know how far, or for whom, this kind of talk might be disempowering. One could well imagine a process of management if someone was nervous, in which another discourse ? say, about contraception ? could allow trust to be made compatible with safety. In other words, Willig is assuming a complete congruence between her interviewees? ?local? talk (discursive practice) and a wider discourse. I wonder what ?proportion of variance? she sees herself as explaining ? The final essay raises a different dimension of my problems with discourse analysis. David Harper presents some selected aspects of his larger doctoral study on the discursive processes involved in establishing and then evaluating drug therapies of mental illness. It is important to note in this case that we have here only a very small part of a larger study. Harper bases his investigation on interviews with a mix of psychiatric professionals and users of their services. He first lists a range of explanations offered by professionals to account for failures of drug regimes, and identifies within the talk that proposes these a body of ?symptom-talk? which thereby engenders a structured distinction between ?surface? phenomena and ?underlying? pathology. Everything thereafter turns on one medium-sized quotation from a consultant psychiatrist, ?Dr Lloyd?. Lloyd appears to offer several distinct explanations why a drug regime failed to alter the belief systems of a psychotic patient. Harper teases out these explanations, showing how they in turn reference ?sociological, behavioural, cognitive, personality and biological psychiatry?, all the time surrounded by ?fluid? cautions and qualifications. (134) As an explanation this fails, he argues, because nothing could refute it. But it does work to sustain Lloyd?s expertise: ?The extract warrants the continued use of medication here despite there being no change in psychotic symptoms?. (136) The remainder of the essay moves between drawing out implications for courses of action that could be adopted by various interest groups; and a self-reflexive angst over claiming his account as ?true?, and thus empiricist. My concerns are partly theoretical, partly tactical, but driven by one question: why should we trust his ?reading? of Lloyd?s account? Theoretically, his cautions against ?truth? largely let him off the hook, and the essay format colludes with this. We have no grounds for determining what will count as a completed analysis. One of the features of expertise is that not everything that is believed or known can be said explicitly at one point. So, what else indeed did Lloyd say in the interview? The use of that single quoted paragraph may have denied him reasonable space to make sense. What could have made his account less ?fluid?? This points up a problem for us as analysts: what can we responsibly do, in setting up and carrying out interviews and in analysing and presenting them, that will make our accounts of them fair, and will enable readers to assess them? I?ve tried to delineate a number of issues with discourse analysis that emerge through the practices of these writers. They can be summarised as follows: (1) The problem of the unity and coherence of the ?research object?, leading on to (a) the problem of readers? genres, and (b) implicit claims of cumulative influence. (2) Presumptions about persuasiveness and associated concepts of power. (3) Issues of investigative completeness and testability, leading on to (a) researchers? responsibility for their claims? implications, and (b) visibly trustworthy methods of analysis. Challenging the ?power? of discourse. How might we go about remedying these problems? There is first, I think, a question of attitudes. The simple excitement that many discourse theorists have felt at the emergence of their field, thence its tendency to intellectual imperialism, need now to be tempered. Discourse theory does not explain the world, it helps us to understand parts of it. And it is the relations between those parts and the rest, that is at stake here. It will mean, therefore, being a bit more modest and sensible than has always been the case. Take this opening sentence: Language organised into discourses (what some contributors here call interpretative repertoires) has an immense power to shape the way that people, including psychologists, experience and behave in the world. (Burman & Parker, 1993: 1) Really? Immense power over which people? When and where, precisely? Under what circumstances? With what determinable and checkable outcomes? Researchers should back off from this kind of talk, and take some responsibility for spelling out how such claims might be tested. To challenge this kind of talk is not to attempt to rubbish all the work that has gone on under such inflated circumstances. It is to ask that discourse researchers ? as indeed any other kind of qualitative researcher ? consider why anyone should trust them. Why should any reader trust their selection of materials for analysis, their mode of analysis, and the ways in which they draw conclusions from those? The notion of ?trustworthiness? is, for me, the qualitative researcher?s equivalent of quantitative researchers? ?triad? of validity, reliability and generalisability. But to avoid ?trustworthiness? becoming simply a rhetorical claim, we require a set of distinct of principles, properly following which will enhance the strength that discourse work can claim. Each of the following is intended to be an answer to the problems I have enunciated above: 1. The defensible corpus: in all kinds of quantitative research, the issue of the reliable sample is a first consideration. In qualitative research the principles cannot be anything like the same, but that does not excuse us from having good grounds for our selections. I propose to call the bounded group of items a qualitative researcher studies her/his ?corpus?, in order to mark this off from the quantitative researcher?s ?sample?. I would argue that a corpus should be subject to various tests which amount to measures of the analysis? trustworthiness. In selecting a corpus for analysis, I propose that to the extent that there are defensible grounds for its selection, this alone adds to the stature of the analysis. So what are such defensible grounds? Suppose a discourse researcher chooses to study a TV interview with a famous person (there are a number of examples of such analyses). On convenience grounds, this is attractive. The materials are nicely bounded, were produced and distributed independently of the analyst, and were (presumably) seen by a large number of people. But choice for convenience alone must increase the provisionality of any claims. Take an interestingly complicated example: Abigail Locke and Derek Edwards (2003) analysed the Grand Jury cross-examination of Bill Clinton during the ?Lewinsky Affair?. They are particularly interested in the ways in which Clinton defended his own position by attributing emotional insecurity to Lewinsky. This apportioned some of the blame, and thereby exonerated himself. It might seem that the focal, indeed televised, nature of the cross-examination guarantees the value of this corpus. And indeed at one level it surely does. But at another, it remains problematic ? because they impose their own framework of relevance on it. Most of the essay is a close analysis of particular attributions, but here is how they close their account: ?Lewinsky?s disposition towards irrationality and heightened emotions ? provided the basis for various alternative accounts he was able to offer, of key and controversial events and readings of events. Rather than exploiting a young and vulnerable White House intern, and persuading her to lie under oath, he was helping and counselling an emotionally vulnerable friend with whom he had responsibly ended some regrettably ?inappropriate contact?. .. Clinton?s accounts of interactions with Lewinsky worked to soften or rebut any notions of perjury and exploitation ...? (2003: 253-4) This notion of the ?work? Clinton?s talk achieved, made visible by their alternative account of what might have been said instead, raises their description to the level of significance. But at a price. Missing, is any sense of the questions asked, and of this being a Grand Jury investigation. Here was a piece of theatre, where both attackers and defenders shared an interest ? in not harming the status of the Presidency per se. Clinton?s line of defence, I would argue, is made possible by the inquisition?s institutional context. Whether they or I am right or wrong, depends upon a wider contextual knowledge ? and that is just my point. If there is independent evidence of the cultural importance of a corpus ? which needs to identify to whom and under what circumstances it was important ? to that extent the analyst has two advantages: s/he will already have a sense, from knowledge of the nature of the people concerned, of what aspects may be most relevant to attend to; and s/he will have the strongest grounds for the relevance of the outcomes of the analysis. 2. The defensible method: qualitative (therefore including discourse) methods always suffer from the difficulty that they are harder for other analysts to check. Many things contribute to this. Pressures of time, the virtual disappearance of the monograph, increasing disciplinary specialisation: all contribute to a tendency to produce smaller, more enclosed pieces of research. Journals impose tight word limits, and that restricts how far authors can make plain their methods of using their materials. (Actually, there are solutions ? web journals need not restrain length, and can include subjunctive pages, and it is not inconceivable for an author to point in any publication to a personal webspace which could display more fully the elaborated methods of analysis.) This pulls us in two directions. In one direction, with a very small corpus, it is possible to show more detail of the materials, and to show the methods of analysis in action; but it carries the higher risk of ?privileged choice? ? that is, favouring cases which suit a conclusion reached on other grounds. But in the other direction a larger corpus is harder to display, and therefore the methods used to examine it tend to greater opacity. My argument is this: the conveniently small corpus is at great risk of never being more than illustrative. In the act of becoming more than this, it inevitably grows. We have to face and find solutions to the problems of managing (both analytically and presentationally) large bodies of discursive materials. Consider a possible study. From time to time, in any culture, certain expressions rise up (often, interestingly, out of fictional contexts) to encapsulate attitudes and relationships.7 Some examples: ?Gizza job? (out of Boys from the Black Stuff); ?Loadsamoney!? (from Harry Enfield?s popular-Thatcherite); ?You might well say that, I couldn?t possibly comment? (from House of Cards), ?We wantses it, my preciousssss? (Gollum in The Lord of the Rings). These have the great virtue that we know ? we will have chosen them for consideration ? because they have such high salience. And a great deal can be said, discursively, about these, in all their minimalism. Their very pithiness itself is in fact an important consideration. But an analysis of any such epithets that did not tackle the variety of contexts of use would not pass beyond the illustrative. We learn much more if we can also consider who uses them in what ways, and with what evaluative accents ? indeed, with what bodily expressions they were associated (by turns, mock head-butt; loud money-brandishing; suave pseudo-diffidence; ingratiating sneakiness).8 Such a broader analysis could disclose how apt such an expression is to incorporation into wider discursive constructions and debates, and that is precisely the point. But then, from having begun with some of the smallest and most enclosed corpuses conceivable, in order to achieve significance we have to expand them greatly. We are forced back to tackling the problems of the ?inconvenient? corpus! How do we manage the analysis of a large set of materials, and then present them satisfactorily? Of course particular fields within discourse research have strict and shared procedures for presenting samples of talk, particularly transcription rules. These mainly govern the stages prior to analysis. 3. Taking responsibility for implied claims: it is arguable (I would argue it) that all analyses of texts and discourses will inevitably make some substantive claims about things beyond themselves. Most typically, these are claims about reception.9 Who are the ?people? who will receive the discursively organised ?messages? which the analyst has disclosed, and what is the possible impact of these? ?People? here needs to be in quote-marks, since ? again, perhaps inevitably ? our analyses use ?figures of the audience?. We do not name actual people (individuals or groups) but more likely kinds of people. The moment we move beyond the loosest and least satisfactory use of terms like ?the audience?, or ?the spectator?, we begin to impute characteristics to them. The difficulty is that these imputations are simultaneously pseudo-empirical, and theoretically-charged. An example to explain what I mean. Michael Stubbs (1996) has recently proffered a new way of doing discourse analysis, which exploits the power of modern computers to permit the comparison of grammatical forms with vast datasets and thus disclose patterns and regularities, and offers interesting examples of what the method can achieve. In one chapter, he explores and compares two final messages from Robert Baden-Powell, one to the Scouts, the other to the Guides. Drawing on a corpus analysis of the two messages, he has valuable things to say about the ways in which the grammar of the two letters embodies, among other things, sexist ideas about the separate roles of men and women (and the ways in which these can be embedded in, for instance, talk about ?happiness?, which might at first sight appear gender-neutral). But then he has this to say about how the Guides might have responded to the inherent sexism of BP?s message to them: They express, quite explicitly, the view that women and men have very different places in the world, and many aspects of these views would now appear deeply objectionable, or perhaps just ridiculous, to many people. Their tone strikes us, over fifty years on, as patronizing and na?ve. And there is no reason to suppose that Girl Guides down the years have passively absorbed BP?s message. They may have actively contested it, given it subversive readings, laughed at or just ignored it. There is no direct way to investigate this, although one indication is that the Guides text has long been out of print. (1996: 84). Notice in here two linked tendencies. (1) The salient feature is gender. That may sound unproblematic, until we consider that it is also only gender. This is not middle- class girls and boys in the UK in early 20th century conditions, in the sphere of leisure relations; it is just ?girls? and ?boys?. And the address of the messages is thus presumed to be ?about? gender as such. That might not matter in itself, since Stubbs might argue that his gender-analysis could be extended and supplemented by attention to class (for instance, what vocabularies are assumed? What modes of ?official speech? are used, and so on?), except that the theoretical stance of this argument has already taken us further. (2) What we see here strongly recalls Stuart Hall?s encoding/decoding model. It begins by measuring the corpus against our concerns. So, first we determine that the messages are ?sexist?. That already carries an implication. If a reader were to have a passive encounter with these messages, they might well be ?inscribed? into a damaging self-definition. Of course, if they ?actively? resisted or negotiated or took the piss, that safely distances them. It is curious how this model of activity/passivity, despite frequent critiques, persistently inheres in models of this kind. I would argue the case differently. In the range of options open to middle-class women at that time, Baden-Powell?s communication could well have counted as a radical one. Here they were, being addressed in their own right. They had a role, and one demanding conscious attention and work ? all contributing to a feeling that this was a positive rather than a demeaning message. And from other historical researches we know that women could actively collaborate in promulgating (what we might see as) ?sexist? definitions of themselves (see for instance Women in the Third Reich). Stubbs? analysis is thus compromised by his model. These, then, are my proposals for reforming the use of discourse analysis within cultural studies. It should come as no surprise to realise that in essence I am arguing that discourse work needs always to be conducted within an explicit recognition that talk of all kinds arises within the circuit of culture. The recognition of that circuit, embracing history, production practices, textual form, reception and recirculation, is one of the great achievements of cultural studies.10 None of this means that only projects which achieve all the above are worth doing. Rather, I am arguing for being honest about limits and boundaries. Discourse work, like any other worthwhile research, is strong to the extent that it recognises its inevitable inclusion of implicative claims, which it cannot itself test. It must therefore acknowledge the provisional nature of its findings but try to articulate what might take matters further. A Very Brief Example Although this example from work I have been personally involved in is very brief, you will be able to see, I hope, how it follows the principles I have laid out. I would want to stress that this is one way, but only one among many, to observe these. The Lord of the Rings international audience project was precisely designed to test some of the kinds of claims that circulate in both academic and public spheres, about how audiences might relate to, and be affected by, a film such as this: an enormously successful fantasy trilogy, based upon a very English story, while trumpeting its New Zealand production and settings, and yet made with Hollywood studio money. From the outset, then, it had a serious ambition to contribute to our knowledge of cross- cultural film reception. Among our several means of gathering materials, we used a questionnaire whose results were fed into a searchable database. The questionnaire combined quantitative (multiple-choice response and self-allocation scales, plus demographic information) with qualitative (free-text opportunities to explain the quantitative responses, along with questions about particular aspects of the film). In all, just under 25,000 completed responses were received, in 14 languages. In terms of scale of corpus, this was going to be complex to handle. In terms of questions, we simply could not assume in advance what the film would mean to different people. I use one key investigation I undertook, to illustrate how we tried to secure trustworthiness. A sequence of quantitative searches led us to identify a separation. Within the world set, we found that while the most common descriptor for the film was ?epic?, the one chosen by those most committed to the film was ?spiritual journey?. These were among twelve options we had offered audiences (with the further possibility to nominate their own) to characterise their overall sense of the story. A further set of quantitative explorations, using twelve countries with the highest overall rates of responses, found a complex patterning. An inverse relationship emerged between the strength of separation between common and committed responses, and the proportion of repeat-readers of the books in the country. Indeed, in five countries with low proportions of repeat-readers, the ?spiritual journey?/?epic? vanished. We therefore wanted to know (a) what these terms meant generally within the world set, and (b) how the separation of common vs committed choices worked within each country. The database allowed us to take random samples, and to gather together their answers to our first free-text question: ?What did you think of the film??. This had come immediately after we had asked people to tell us how much, on a five-point scale, they had enjoyed the film. Those grouped answers, gathered in this fashion, now constituted our corpuses: 100 each from the world set of those who had nominated ?epic?, or ?spiritual journey? (but excluding each other so as to minimise overlap) among their up-to-three terms to describe the story; then, 50 for twelve countries each from each of the most common choices and most committed choices (again, excluding each other). Each corpus included examples ranging from one-word expressions of pleasure (?Wow!?, ?Fantastic?, and so on) to quite elaborated explanations of why and how people had enjoyed the film. If you would like to see the procedures in detail, they are available online.11 In brief, the analysis involved: ? a close reading of the two world sets, from which a coding scheme was developed that could encompass everything said; ? producing formal definitions of the, in the end, ten codings; ? the systematic application of these, ensuring they covered everything; ? a first-level analysis of the codings to disclose similarities and differences both in frequencies, and in kinds of mention; ? from these, an examination of how, within each corpus, elements were linked and moves made between kinds of talk, with the aim of disclosing discursive connectors. From these, a portrait of the typical elements was constructed and then tested against the most explicit and elaborated examples. This stepped analysis was then repeated for the two sets from each of the twelve countries, in order to produce a portrait of their culture-specific patternings of choices. If you would like to see the outcomes in detail, they are published in the main book from the project.12 What I believe we were able to achieve through this means, was a trustworthy account of two things: 1. the different core meanings of The Lord of the Rings for those whose encounter with it was based deeply in the books, and their history; 2. the ways in which these core meanings were altered by the local circumstances of the book/film?s history and reception in some very different country-contexts. Although doing this was undoubtedly tough, I believe that this carefully ramified set of stages and procedures increases the trust that can be placed in our findings. It also had the effect of revealing to us things that completely surprised us. All this allowed us to go back to ask what could make sense of the peculiarities in responses in each country, and so not overlook what discourse analysis can easily suppose or take for granted. Summary: Key Points ? The chapter traces the rise and spread of discourse theory and analysis, a multi-faceted development and one of the fastest growing areas of cultural theory and methodology. ? It examines the ways it contains within many of its formulations complex claims about the responses of those on the receiving end of discursive forms and communications. The problems inherent in this are traced to the ways in which many formulations of ?discourse? presume on particular models of ?cultural power?, which are in themselves antipathetic to the very notion of audience research. ? The essay explores in detail a number of cases where such formulations are at work, within examples of discourse analyses of specific kinds of cultural materials, and explains why they may be problematic. ? It offers as a way forward a series of methodological tests which can be applied to cases of discourse analysis, which could reduce the subjectivity and strengthen the trustworthiness of discourse analytic claims, and make them more open to empirical testing by reception research. ? The essay very briefly introduces materials derived from the international Lord of the Rings audience research project, within which discourse analytic methods were used to examine responses to the film adaptation of Tolkien?s books. Further Reading Discourse analysis is widely recognized as a wide-ranging set of approaches, deriving from competing paradigms and models. A very good survey of the main approaches, at both theoretical and applied levels, is the pair of volumes edited by Margaret Wetherell, Stephanie Taylor and Simeon J. Yates: Discourse Theory and Practice, and Discourse and Data: A Guide for Analysis (both Milton Keynes: Open University Press, 2001). A range of journals carry important exemplars of the various kinds of work undertaken under the banner of discourse analysis, notably Text, Discourse and Society and Critical Discourse Studies. These journals also contain important debates between practitioners within the various major ?schools? of language and discourse work. A range of audience researches have at various points claimed to use discourse analytic methods, not always very systematically. Although written before the expression ?discourse analysis? became popularised, Ien Ang?s Watching Dallas: Soap Opera and the Melodramatic Imagination (London: Methuen 1985) remains an important example of the critical examination of language to reveal social and cultural understandings. Martin Barker and Kate Brooks? Knowing Audiences: Judge Dredd, its Friends, Fans and Foes (Luton: University of Luton Press 1997) contains an examination of various approaches, and outlines a set of procedures of an approach compatible with cultural studies? general audience research practices. Bibliography: Attwood, F. (2002), ?Reading porn: The paradigm shift in pornography research?, Sexualities, 5 (1), 91-119. Barker, Chris & Dariusz Galasi?ski, Cultural Studies and Discourse Analysis: A Dialogue on Language and Identity, London: Sage 2001. Barker, Martin, ?Film audience research: making a virtue out of necessity, IRIS (French/American Film Journal), No.26, 1998, pp.131-48. Barker, Martin (with Thomas Austin), From Antz to Titanic: Reinventing Film Analysis, London: Pluto Press 2000. Barker, Martin & Ernest Mathijs (eds.), Watching the Lord of the Rings: A Study of Tolkien?s World Audiences, NY: Peter Lang 2007. Billig, Michael and Emanuel A Schlegoff, ?Critical Discourse Analysis and Conversation Analysis: an exchange?, Discourse and Society, 10(4), 1999: 543- 82. Burman, Erica & Ian Parker, Discourse Analytic Research: Repertoires and Readings of Texts in Action, London: Routledge 1993. Burton, Frank & Pat Carlen, Official Discourse: Discourse Analysis, Government Publications, Ideology and the State, London: Routledge 1979. Collins, Chik, ?To concede or contest??, in Colin Barker & Paul Kennedy (eds.), To Make Another World: Studies in Protest and Collective Action, Avebury 1996. Davies, M?ire Messenger, ?Micky and Mr Gumpy: the global and the universal in children?s media?, European Journal of Cultural Studies, 7(4), 2004, pp. 425-40. du Gay, Paul et al., Doing Cultural Studies: The Story of the Sony Walkman, Milton Keynes: Open University Press 1997. Edwards, Derek & Potter, Jonathan, Discursive Psychology, London: Sage 1992. Fairclough, Norman, Discourse and Social Change, Cambridge: Polity Press 1992. Fairclough, Norman, Media Discourse, London: Arnold 1995. Gillies, V. & Willig, C. (1997) ??You Get the Nicotine and That In Your Blood?: Constructions of Addiction and Control in Women?s Accounts of Cigarette Smoking?, Journal of Community and Applied Social Psychology, 7: 285-301 Hartley, John, Understanding News, London: Methuen 1982. Heath, Stephen, The Sexual Fix, Basingstoke: Palgrave 1982. Howarth, David, ?Discourse Theory and Political Analysis?, in Elinor Scarbrough & Tanenbaum, Eric, Research Strategies in the Social Sciences: A Guide to New Approaches, Oxford: Oxford University Press 1998: 268-93. Locke, Abigail & Derek Edwards, ?Bill and Monica: memory, emotion and normativity in Clinton?s Grand Jury testimony?, British Journal of Social Psychology, 42, 2003, pp. 239-56. Korobov, Neill & Michael Bamberg, ?Positioning a ?mature? self in interactive practices: How adolescent males negotiate ?physical attraction? in group talk?, British Journal of Developmental Psychology (2004), 22, 471?492. Korobov, Neill & Michael Bamberg, ??Strip Poker! They Don?t Show Nothing!??: Positioning identities in adolescent male talk about a television game show?, in A. De Fina, M. Bamberg, and D. Schiffrin (Eds.), Talk and identity in narratives and discourse, Amsterdam: John Benjamins Publishing 2006. Kress, Gunther & Tony Trew, ?Ideological dimensions of discourse, or: ?How the Sunday Times got its message across??, Sociological Review, Vol.26, 1978, pp.755-76. Partridge, Eric, A Dictionary of Catchphrases: British and American from the Sixteenth Century to the Present Day, London: Routledge 1986. Potter, Jonathan & Margaret Wetherell, Discourse and Social Psychology: beyond attitudes and behaviour, London: Sage 1987. Stef Slembrouck, ?What is the meaning of discourse analysis?, http://bank.rug.ac.be/da/da.htm (accessed 1 August 2005). Steinberg, Marc, Fighting Words: working-class formation, collective action, and discourse in early nineteenth-century England, NY: Cornell University Press 1999. Stubbs, Michael, Text and Corpus Analysis: Computer Assisted Studies of Language and Culture, Oxford: Blackwell 1996. Tagg, John, The Burden of Representation: essays on photographies and histories, Basingstoke: Macmillan 1988. Threadgold, Terry, ?Cultural studies, critical theory and critical discourse analysis: histories, remembering and futures?, Linguistik online, 14, 2/03. (Accessed 6 October 2005.) Wetherell, Margaret, Stephanie Taylor & Simeon J Yates, Discourse Theory and Practice: A Reader, Milton Keynes: Open University Press 2001 Williamson, Judith, Decoding Advertisements: ideology and meaning in advertising, London: Boyars 1978. Willig, Carla (ed.), Applied Discourse Analysis: Social and Psychological Interventions, Buckingham: Open University Press 1999. 1 See, as examples, Norman Fairclough (1992); Stef Slembrouck (2005); and Wetherell et al (2005). 2 Recently I have encountered two interesting variants. ?Positioning analysis?, associated in particular with the work of Michael Bamberg, claims to find a mid-way between CA and CDA. I do not yet feel confident enough to comment on this. See, for instance, Korobov & Bamberg (2004). 3 It is necessary to say something about the relations of this critique to the long-running, and perhaps inevitably unfinished debate between two traditions of work on talk: conversation analysis, and critical discourse analysis. In a series of often tetchy exchanges, scholars in the two camps have rehearsed arguments against each others? approaches. It might appear at a quick glance that my argument sides me with the CA camp. That would not be right at all. I am very largely persuaded by the critique of CA offered by, for instance, Michael Billig. In an exchange in Discourse and Society (Billig and Schlegoff (1999)), Billig argues that while CA presents itself as strictly empirical, concerned only with looking and seeing what are the organising characteristics of ?ordinary conversations?, in fact it is heavily based upon a ?foundational rhetoric? which among other things presumes a working distinction between ?ordinary conversations? and ?official or institutional talk? ? the former being presumed to be equal and participative, and thus not inflected by power-relations until specifically proved otherwise by formal qualities (such as imbalances in turn-taking). Emanuel Schlegoff?s cross response to this critique badly misses the point, I think because Schlegoff simply cannot accommodate the idea that research such as his necessarily involves theoretical commitments. But the problem is that my acceptance of Billig?s critique hits a limit when he himself stops just there, with that acknowledgement of theory-tasks (?CDA aims to make explicit such tasks, in order to enable a theoretically based choice between available rhetorics and attendings/disattendings? (575)). And it is clear that it is the question of ?power? that is the heartland of the Billig/Schlegoff disagreement. What Billig does not go on to do, is to consider how those commitments, and the findings which they thus prompt, might be tested. Without that, in the end, the CA/CDA choice is purely one of political preference. 4 I cannot in the space I have give as full an account of this book as it deserves. A much longer critical review of it is contained in Terry Threadgold?s (2003) essay. 5 In fact I have chosen not to explore one essay (by Joan Pujol). This is because the issues it raises are rather different, and would concern the sheer untraversable distance between her weighty theoretical framework deriving from Derrida and Ricoeur, and some hardly digested fragments of empirical material. 6 This distinction is derived from Edwards and Potter (1992). In another essay, Willig has explored this distinction further. See her (1997) essay with Gillies. 7 See Eric Partridge (1986) for a fascinating collection of such catchphrases. 8 I have recently been using the marvellous online database LexisNexis to explore the evolution of references to ?Gollum? before, during and after the appearance of the films of The Lord of the Rings. Methods of both analysis and presentation have posed real challenges. It remains to be seen, by others, how successful I have been. 9 See for instance my IRIS essay, and From Antz to Titanic. 10 For a clear statement of the nature and importance of this circuit, see Paul du Gay et al., (1997). 11 Go to www.users.mib.aber.ac.uk (Cross-Cultural Pleasures). Not there yet! 12 Martin Barker & Ernest Mathis (2007).