Abstract: The contribution outlines a research programme which I have coined the "sociology of knowledge approach to discourse" (Wissenssoziologische Diskursanalyse). This approach to discourse integrates important insights of FOUCAULT's theory of discourse into the interpretative paradigm in the social sciences, especially the "German" approach of hermeneutic sociology of knowledge (Hermeneutische Wissenssoziologie). Accordingly, in this approach discourses are considered as "structured and structuring structures" which shape social practices of enunciation. Unlike some Foucauldian approaches, this form of discourse analysis recognises the importance of socially constituted actors in the social production and circulation of knowledge. Furthermore, it combines research questions related to the concept of "discourse" with the methodical toolbox of qualitative social research. Going beyond questions of language in use, "the sociology of knowledge approach to discourse" (Wissenssoziologische Diskursanalyse) addresses sociological interests, the analyses of social relations and politics of knowledge as well as the discursive construction of reality as an empirical ("material") process. For empirical research on discourse the approach proposes the use of analytical concepts from the sociology of knowledge tradition, such as interpretative schemes or frames (Deutungsmuster), "classifications", "phenomenal structure" (Phänomenstruktur), "narrative structure", "dispositif" etc., and the use of the methodological strategies of "grounded theory".

Key words: sociology of knowledge, discourse, politics of knowledge, symbolic interactionism, frame, classification, narrative structure, grounded theory, Foucault, Berger, Luckmann

Since the impressive work of Michel FOUCAULT in the 1960s and 1970s, discourse research in the social sciences has been oscillating between the comprehensive theoretical interpretation of social macro-discourses (e.g. the Foucauldian tradition, work inspired by LACLAU and MOUFFE, Cultural and Postcolonial Studies) and the analysis of concrete "language in use" in the field of discourse analysis (including linguistic pragmatics and ethnomethodologically rooted conversation analysis). Recent attempts to build bridges between these rather heterogeneous paradigms have aimed to reduce problems localised on both sides, either in an "all too abstract macro analysis in discourse theory" not really fitted to reach the level of empirical research, or in an "all to micro perspective" on discourse unable to go beyond local micro-data analysis. Although I agree with this diagnosis, I suggest a different strategy for discourse research in order to bring the latter "down to earth" in empirical sociology: Rather than focusing on the analysis of "language in use", it is preferable—and possible—to translate some Foucauldian insights on discourse into sociological theory building. With this move, it is possible to elaborate a sociology of knowledge approach to discourse based on the social constructionist tradition of Peter L. BERGER and Thomas LUCKMANN, and to adopt and adapt interpretative or qualitative traditions of data analysis. As an analysis of knowledge production and circulation, this approach is closer to the original Foucauldian programme of analysing discourses as "practices of power/knowledge" and meaning production, than the established focus on "language in use" research. But going beyond FOUCAULT, such an approach introduces a more sociological conception of actors and practices in discourse theory and research. The empirical practice of discourse research can thus reclaim modifications of qualitative data analysis in order to meet the necessities of discourse perspectives. In the following, I will first discuss the relation between discourse theory and sociology of knowledge. Then I present some basic assumptions of the sociology of knowledge regarding discourse. The third part of the article discusses some devices, methodological concepts and qualitative strategies for analysing "discourse data" (texts, visual data, ethnographic data) which draw on concepts such as the reconstruction of interpretative schemes or frames (Deutungsmuster), classifications, phenomenal structures, narrative structures, dispositifs, theoretical sampling or "coding". I argue that these concepts are well suited to provide a qualitative sociological perspective of discourse (see KELLER 2004, 2005). [1]

At present, various notions of discourse are in circulation in the humanities. They can be grouped into six categories (see KELLER 2004): (1) In Germany, Jürgen HABERMAS contributed extensively to the dissemination of the term "discourse". But in the Habermasian tradition, discourse is hardly an object of inquiry, to be empirically analysed. Instead, it is regarded as an organised and ordered deliberative process to which a normative ethics of discourse is applied. This use, which is current today primarily in the political sciences, has created—and still creates—some confusion in German debates on discourse research. The traditional political science approach to discourse is mainly interested in the relationship between arguments (ideas) and interests: in short, discourse matters if the better argument wins. However, this argumentative approach to discourse up to date rarely analyses the politics of knowledge. (2) Discourse analysis is a master frame for the micro-orientated analysis of language in use, which is based on pragmatic linguistics and conversation analysis. (3) Corpus linguistics builds up enormous corpuses of text data around selected themes (such as political issues) in order to look for statistical correlations. (4) Critical Discourse Analysis (Norman FAIRCLOUGH) and its German counterpart Kritische Diskursanalyse (Siegfried JÄGER) are both based in linguistics, but with slightly different discourse-theoretical elaborations; they direct discourse research to the ideological functions of language in use. (5) Discourse theories—like those of Michel FOUCAULT or Ernesto LACLAU and Chantal MOUFFE—are designed to analyse the social macro-levels of power/knowledge relationships or the articulation of collective identities. (6) Culturalist discourse research could be the label for a field of research derived from three different traditions: Symbolic Interactionism (i.e. the analysis of the construction of social problems in public discourses), the investigation of language use and symbolic power inspired by BOURDIEU, or the analysis of "circuits of representation/culture" in Cultural Studies. While approaches 2 and 3 are interested in questions of micro/macro processes of language use, and 4 is directed towards ideology, approaches 5 and 6 are closely related to questions of knowledge production, circulation and transformation, or in more general terms: they are related to questions of symbolic structuring of meaning and the generation of symbolic orders including their material groundings and effects. The main difference between the two strands seems to be that the latter approach gives greater importance to social actors. [2]

Recent years have seen an increasing interest in discourse research in the social sciences as well (see KELLER 1997, 2004). Yet, current research still faces one major problem: How to enter the practice of discourse research? Once the theoretical grounds are prepared, building on FOUCAULT or the LACLAU/MOUFFE tradition, how to do, step by step, the concrete empirical research? Methodological devices offered by traditional discourse analysis—the analysis of "talk and text in action" (Teun van DIJK)—do not serve well to address the interests of social sciences (sociological) discourse research at more comprehensive or meso/macro levels. This constellation has given rise to attempts to bring together the best of both worlds of discourse research: theoretical groundings offered by discourse theory, and empirical concepts and strategies from the toolbox of discourse analysis (WETHERELL 1998; JØRGENSEN & PHILIPPS 2002). Nonetheless, as I suggest, this attempt to ground discourse research is not as new as it claims to be—it has been present for more than fifteen years now in approaches such as the already mentioned "Critical Discourse Analysis" or "Kritische Diskursanalyse" (FAIRCLOUGH 1995; JÄGER 1999). Considering empirical research presented by both critical approaches I currently see two main problems: [3]

The first problem is closely tied to the interest in ideological functions of language which all to often results in a rather reductionist "proof" of the presence of ideological notions and functions in a concrete set of spoken or written language (discourse). There is no place for any surprising results or insights to be derived from such empirical research, because the discourse theorist always knows how ideology works in advance. [4]

The second problem is closer to the solution proposed for problems of discourse research mentioned above. The methodological devices which are offered by Critical Discourse Analysis and Kritische Diskursanalyse stem from linguistics and may be well suited for questions of linguistic research (including linguistic pragmatics and conversation analysis). But they are hardly suitable to grasp the larger dimensions of knowledge and knowledge/power which FOUCAULT was interested in.¹⁾ An approach to discourse informed by the sociology of knowledge promises to grasp these latter dimensions. [5]

Since the early days of the sociological classics in knowledge analysis—such as Karl MARX, Emile DURKHEIM, Max WEBER, Max SCHELER, Karl MANNHEIM, Ludwig FLECK—the sociology of knowledge has seen a rather heterogeneous development. Its latest impressive manifestations appeared in social studies of science and technology. In the following, I refer to the sociology of knowledge tradition mainly the seminal book on the "Social construction of reality" by Peter L. BERGER and Thomas LUCKMANN (1980), originally published in 1966 at the same time as FOUCAULT's "Order of things" (1974). BERGER and LUCKMANN proposed a synthesis of different strands of sociology of knowledge approaches ranging from MARX and DURKHEIM to the phenomenological approach of Alfred SCHÜTZ. Inspired by arguments of pragmatism and symbolic interactionism (i.e. MEAD's theory of socialisation), they developed the theoretical groundings of a comprehensive sociological analysis of the social production and circulation of knowledge. This perspective ranges from processes of generating, objectifying and institutionalising knowledge as "objective reality" to the mechanisms of the individual's more or less creative adoption of knowledge patterns taken from the collective "stock of knowledge" (SCHÜTZ & LUCKMANN 1979).²⁾ The concept of knowledge refers to everything which is supposed to "exist" (including ideas, theories, everyday assumptions, language, incorporated routines and practices). The "social construction of knowledge" is conceived as an ongoing activity, performance and process; it is not the intentional outcome of any individual effort, but rather an effect of everyday action and interaction. The collective stocks of knowledge appear as institutions (like language itself), theories and other socio-cognitive devices, organisations, archives, texts and all kinds of materialities (e.g. practices, artefacts). Together, they constitute a historical Apriori for embedded individual actors. These actors' minds constitute the world not as transcendental subjects, but by using the knowledge devices at hand or, if routine (inter)action and interpretation is disturbed, by "creating" new ones in extended processes of social interaction. [6]

The BERGER/LUCKMANN tradition in Germany at present uses the label of "Hermeneutische Wissenssoziologie" (hermeneutical sociology of knowledge) (HITZLER, REICHERTZ & SCHRÖER 1999) to mark its difference to other social science approaches to knowledge. Since it has always—and lately more and more explicitly—accorded great attention to the connection between language and knowledge, it has been presented recently by some of its proponents as the "communicative paradigm" in knowledge research (LUCKMANN 2002; KNOBLAUCH 1995). In taking up the foundational work of BERGER/LUCKMANN, including their tenet that everyday knowledge should be the central point of reference for sociological knowledge analysis, the Hermeneutische Wissenssoziologie has unfortunately concentrated mostly on micro levels of knowledge analysis. It directed its interests to ethnographies of "small life worlds of modern man" (Benita LUCKMANN) or actors' interpretations of their everyday activities. Norbert SCHRÖER (1997) ultimately identified Hermeneutische Wissenssoziologie with this latter interest in actors' local knowledge. Against such reductionist adoptions of the BERGER/LUCKMANN tradition I propose an extension to include all social levels of institutional and organisational circulation of knowledge. This was originally proposed in their seminal work through their use of concepts such as objectification, institutionalisation, and legitimisation. The notion of discourse is well suited to analyse social processes, practices and politics of knowledge in modern societies as discourses. It helps to provide a more subtle theoretical understanding of the otherwise rather static idea of "stocks of knowledge". Before explicating further details of the sociology of knowledge approach to discourse (Wissenssoziologische Diskursanalyse), let me sum up its major promises:

We can't enter the world and see "discourses" in the way we see, for example, a piece of cake, a building, or even a concrete set of social interaction. "Discourse" is not an ontological entity. In the empirical world, we can't collect anything but disparate elements or utterances, occurring at different instances in time and social as well as geographical space. Discourse so far is nothing but a theoretical device for ordering and analysing data, a necessary hypothetical assumption to start research. The last decennia have seen a rich development of reflections on "doing discourse research" following more general interest rather than concrete language-in-use approaches (see KELLER, HIRSELAND, SCHNEIDER & VIEHÖVER 2001, 2003; KELLER 2004). Impressive theoretical propositions have been made by LACLAU and MOUFFE on the role of practices of articulation for the constitution of collective identities, or in the Cultural Studies focus on the ways in which actors produce and actively consume circulating representations. But they remained rather silent about their methods. Therefore I still consider FOUCAULT's work to be the central source of inspiration for elaborating discourse research. The books, articles and conversations signed by "FOUCAULT" present a delightful set of proposals and toolboxes open to various interpretations. [8]

FOUCAULT insisted on the relevance of general or "higher level" research questions to illuminate the "history of the present". He analysed the genealogy of modern configurations of the subject, the power/knowledge relationship, or processes of normalisation of bodies, sexualities and so on (bio-power). His major work on discourse theory, the "Archaeology of knowledge" (FOUCAULT 1988 [1969]), is very successful in constructing a theoretical idea of "discourse". But it neither addresses questions of relations between discourse and other social phenomena, nor does it talk about methodical devices for empirical research. Rather, FOUCAULT reflects on the guiding assumptions implicit in his previous works, without pretending that he ever followed the road map of discourse theory presented in "Archaeology". Neither did he do so later. While the "Archaeology" argued for historical snapshots of power/knowledge regimes, his later concept of "Genealogy" accentuated the diachronic intertwining of discourses, practices and dispositifs (cognitive/material infrastructures) in historical power struggles or struggles for truth (see FOUCAULT 1974b [1972], 2002 [1973/1974], 1992 [1978]). Here too FOUCAULT was rather arcane about his actual practice of "doing research". He never actually did the kind of discourse analysis for which he prepared the theoretical grounds in his "Archaeology of knowledge". Nevertheless he made a few points concerning his strategy of questioning his data (see FOUCAULT 1991):

Given these basic assumptions, it seems that such a perspective on discourse could fit well with strategies of qualitative research in the social sciences. The approach of Wissenssoziologische Diskursanalyse does not pretend to offer "a true Foucauldian application" of discourse research. It rather presents a research programme that adopts some of FOUCAULT's general proposals for understanding discourse as a social phenomenon. It anchors them in the larger framework of BERGER/LUCKMANN's sociology of knowledge and thereby transforms this latter framework as well. This programme basically proposes:

The purpose of the research programme of Wissenssoziologische Diskursanalyse is to analyse ongoing and heterogeneous processes of the social construction—production, circulation, transformation—of knowledge. This comprises the analysis of symbolic order on institutional and organisational levels and arenas as well as the effects of such an ordering in different social fields of practice.³⁾ This perspective also covers the implication of social actors in the performance and "reception" of discourse. It defines discourse as identifiable ensembles of cognitive and normative devices. These devices are produced, actualised, performed and transformed in social practices (not necessary but often of language use) at different social, historical and geographical places. They unfold in time as well as they are embedded in historical contexts. Discourses in this sense constitute social realities of phenomena. At least they compete in the everlasting struggle over symbolic order. Insofar, discourses occur as "structured and structuring structures" (Pierre BOURDIEU): They emerge out of historically situated practices and "problematisations". They gain a certain—and never ultimately fixed—"internal stability and structure". They propose a symbolic-material structure of the world. Such discourses (as structures) do not exist in an imaginary (idealistic) "heaven" above society. Instead they are realised by social action, i.e. by social actors' practices and activities. Actors need motivation to enter a discursive field, but we should neither imagine them as complete masters of a singular discourse nor as transcendental subjects beyond their concrete historical contexts. Social actors are embedded in the historical a priori of established symbolic orders and institutionalised power/knowledge-regimes. In order to enter a given discursive field they have to draw on existing subject or "speaker" positions whose criteria of performance are beyond their control. [11]

The sociology of knowledge approach to discourse draws on the Symbolic Interactionist tradition of analysing public discourses and the social construction of collective action/problems which is closely related to the work of BERGER/LUCKMANN. Based on the tradition of pragmatist philosophy, symbolic interactionism has analysed public discourses as symbolic struggles over the (collective) "definition of the situation" (William I. THOMAS) and used concepts like "universe of discourse" (PEIRCE, MEAD), "social worlds", "arena" of discourse (Anselm STRAUSS), or "community of discourse" (Robert WUTHNOW), without, however, formulating a theory of discourse (see KELLER 2005, pp.64ff; KNOBLAUCH 1995). Having outlined the theoretical groundings elsewhere (see KELLER 2005, pp.189ff) I just want make a few brief points concerning theoretical issues: [12]

(1) In the SCHÜTZ-BERGER-LUCKMANN tradition, signs (and knowledge) are seen as typified and typifying concepts. They emerge from social processes of inter-action and language use. They are temporarily stabilised through communication by social actors' interaction and institutional devices. Actors' use of communicative genres and signs presupposes a given "universe of discourse" (with related sub-universes); these processes permanently perform and transform this universe. Discourses are structured processes of sign/knowledge production and reproduction in society. [13]

(2) Wissenssoziologische Diskursanalyse proposes to consider the relationship between discourse (as structure) and the singular language use or other discursive event/practice as "duality of structure" (see GIDDENS 1992).⁴⁾ This means that every action/social practice draws on structuring devices in order to be performed, and in turn performs, reproduces or transforms those elements. No structure without action, no action without structure. GIDDENS proposed to distinguish between signification rules for the discursive constitution of meanings, normative rules for "correct" practice, resources of social action and material resources. In adopting this idea, I suggest that discourse as structure

Discourse therefore instructs social actors' symbolic practices (whether they are discursive or not). Discourse is not prescription or determinist rule; it proposes positions for actors engaging in knowledge production and circulation as well as opportunity structures for using symbols to say things or to create the dispositional prerequisites for power effects. [15]

(3) It is important to keep in mind that it is not discourse itself which performs actions or social practices, but rather social actors involved in different social fields and symbolic struggles. If we consider discourses as more or less institutionalised structures of knowledge production and circulation, it should be clear that there are pre-constituted subject positions for "articulation" (LACLAU/MOFFE). Surely these can vary between the rather fixed positions in scientific disciplines to the rather open participation arenas of public discourses. Actors may engage in very different discourses and for short periods of time. In addition to the production of discourse, we have to pay attention to the subjectivities and identifications proposed in discourses which construct symbolic structures of the world, e.g. in proposing collective identities (we—the others etc.). However, we should never equate this elements of discourse with the actions and interpretations of those who were addressed. [16]

(4) We can further distinguish between discursive and non-discursive practices of discourse (re) production, practices proposed by a discourse (as part of a dispositif) and extra-discursive practices in social fields. [17]

(5) Finally, I propose to consider and to analyse both specialised discourses (such as scientific disciplines) and general public discourses as discourses, which means to ask for their materialities. Discourse research can be done on different levels of abstraction and is able to consider differences between sub-discourses as well as similarities. A scientific discipline is no monolithic entity; as we approach it, we may discover a battlefield of competing discourses within which are very different despite having, on a more abstract level, something in common. Research interests are guiding devices in order to focus homogeneities of discourse or heterogeneities of sub-discourses. [18]

Discourses unfold in time and social as well as geographical space. The sociological or social scientific analysis of discourse starts from general sociological research interests. It then addresses questions ranging from micro-levels of discursive practices to more general issues about the discursive structuring of symbolic orders to wide-ranging reflections on the relationship between discourse, extra-discursive events and social change (see KELLER 2000, 2003, 2005, pp.273ff). Discourse research is interested in:

Since real world empirical research in the social sciences is subject to restrictions of manpower, time and money, it would not be feasible to address all those questions at once. Therefore a concrete analysis of discourse has to select some research interests to concentrate on. [20]

Having presented so far some theoretical and methodological reflections, the question of how to do discourse studies in practice still remains. I assume that discourse research always has to be considered as a process of data construction and interpretation. The notion of an "analytics of interpretation" proposed by Hubert DREYFUS and Paul RABINOW seems appropriate to define FOUCAULT's programme (see DREYFUS & RABINOW 1987). However, what does it mean for a discursive turn in the sociology of knowledge? First of all it points to the analytical "business as usual" part in discourse studies. Social sciences' discourse research starts with a theoretically informed research question and a heuristic circumscription of the social phenomenon under examination. In the following step explorative interviews might be conducted to gain further information on the object, appropriate units of analysis (data-format: e.g. documents, flyers, monographs, visual images, newspaper articles) have to be defined and subsequently to be collected. Later on, the leading questions as well as the data sample might be modified, transformed or even replaced by others. These steps, isolated here for reasons of clarification, are actually more or less intertwined. [21]

Up to this point, the concept of discourse works as a sensitising hypothesis for data collection, in order to find appropriate data sources (newspaper texts, books, speeches, media events, web presentations etc.). But only data analysis can show whether the original hypothesis for data collection was appropriate or ill suited. Answers to the questions of whether concrete phenomena of language do account for a particular discourse, and by what elements or "rules" and strategies the discourse is constituted, cannot be found a priori, but only in the process of analysis.⁵⁾ [22]

The point I want to make is the following: Many scholars working in the Foucauldian or post-structuralist tradition consider themselves to be "beyond hermeneutics" (DREYFUS & RABINOW 1987), in the sense of "beyond interpretation". I would contend, however, that discourse research, as far as it is concerned with social practices and symbolic ordering, cannot abstain or withdraw from interpretation. Against hermeneutics, FOUCAULT insisted on the description of the "positivity" of discursive events, rejecting any kind of interpretation that aims to discover the one and true meaning. Yet, a closer look reveals that his critique of hermeneutics was directed on the one hand against Marxist reductionism. He doubted the idea of the one and only truth of the text, which derived from the historical laws of class formation. On the other hand his reservations against hermeneutics are also valid for any idealistic assumptions drawing on a remote religious and/or philosophical past (the one and only truth lies in the authors—or gods—intention).⁶⁾ [23]

The meaning of "hermeneutical" in the German Hermeneutische Wissenssoziologie is, in contrast, much more modest. Drawing mainly on work of Hans-Georg SOEFFNER (see SOEFFNER 1989; REICHERTZ 2004; SOEFFNER & HITZLER 1994), it simply pleas for methodological reflection on the researcher's use of interpretative skills. It argues for a socially accountable data analysis instead of reasoning on THE truth contained in textual data. Accordingly, what is required then is a convincing argumentation for each step of the analysis. As a presentation edited by HITZLER and HONER (1997) stresses, social scientific "hermeneutics" is anything but free floating interpretation trying to discover the ultimate truth underneath the surface data. This further means that, depending on appropriate research questions as well as concepts of analysis, "data construction" and data analysis may take different directions. [24]

In the case of Wissenssoziologische Diskursanalyse I propose to distinguish two main dimensions of analysis. The first is concerned with the analysis of materialities, the latter focusing on the process of symbolic ordering (or meaning production). The former has to look for "material" dimensions as the key actors performing a discourse (actor positions) for instance, their "standing" and role in an arena of discourse, the relations (e.g. discourse coalitions) between them etc., the practices (and strategies) of discourse production and reproduction and the elements of dispositifs (e.g. institutionalised data production, socio-material infrastructures, networks of articulation and distribution). This might also encompass an ethnography of institutional settings and particular events of discourse production and performance. This part of discourse research is much like "ordinary" case study work in the social sciences. It builds up on complementary strategies of observation and data collection (e.g. by "expert" interviews). [25]

Paralleling the "material" level, discourse research is engaged in linking the social (institutional) dimensions of knowledge production and circulation with the symbolic order that is thereby achieved.⁷⁾ In order to analyse the symbolic or "meaning/knowledge" dimension of discourse (the way in which phenomena are configured), I suggest referring to the rich tradition of the sociology of knowledge and to the interpretative paradigm, rather than resting on linguistic concepts or methodical devices. The tradition of "qualitative" research offers methodical devices for the analysis of knowledge which are useful (not only) for doing discourse analysis. It further helps to illuminate the relationship between discourse and extra-discursive fields of social practice (e.g. everyday knowledge and action). Given the prevalence of textual data, what should one look for in actual analysis, if not linguistic devices? I propose to make a distinction between four analytical units or concepts: (1) Deutungsmuster (interpretative schemes, frames), (2) classifications, (3) phenomenal structure and (4) narrative structure (plots). Taken together these elements form the "interpretative repertoire" (POTTER & WETHERELL 1995) by which a discourse tends to achieve its symbolic structuring of the world. Before entering further questions of concrete data analysis, I will briefly outline some main assumptions associated with these concepts: [26]

(1) Deutungsmuster ("interpretative scheme", "frame"):⁸⁾ In German qualitative social research Ulrich OEVERMANN introduced the concept of Deutungsmuster in the early 1970s, and accentuated the link between interpretation (meaning attribution, cognitive ordering) and action. Different uses of this concept have since been established (see LÜDERS & MEUSER 1997). They either look for everyday knowledge, interpretation and action as guided by such interpretative schemes.⁹⁾ On a more collective level they are considered as socially typified historically embedded interpretation devices for occurring events, urgencies of action etc. "Risk" is a good example for one such modern frame which structures the perception of and action towards certain socio-technical complexes (e.g. nuclear energy, waste incineration, genetically modified plants). The concept of Deutungsmuster refers to typified clusters of disparate elements of meaning production, the core configuration of signs, symbols, sentences and utterances, which create a coherent ensemble of meaning. Qualitative research in Germany has used this concept for analysing biographical narratives or everyday routines in professional fields. Few efforts have been made, however, to give a more general account of the social genealogy of such frames.¹⁰⁾ Introducing Deutungsmuster into discourse research means asking for discourses as instances of the production and circulation of frames. How many and what kinds of "master" interpretative schemes one finds in a given discourse is open to empirical research. This is the case for questions of their interrelation as well as for questions of consistence and coherence. Empirically, Deutungsmusteranalyse is done by sequential analysis (Sequenzanalysen). This means it is a "sentence by sentence" interpretation, which uses the methodological tenet of avoiding assumption or prejudice (i.e.: assuming to know the meaning of the data immediately). Instead, this approach advocates an attitude of "artificial stupidity" (Ronald HITZLER) which creates an account of many conceivable/possible interpretations, and proceeds to reduce them step by step in order to establish the one which creates "the most powerful account". [27]

(2) Classification: Since DURKHEIM and MAUSS' pioneering work, scholars in the social sciences or social anthropology have largely reflected on the origins, meaning/importance and effects of classifications. The interest in classificatory devices and classifications, complementary to the question of frames, is due to their constitutive role for symbolic order in discourse. Whereas FOUCAULT's "The order of things" refers to (scientific) classification on a rather abstract level of analysis (FOUCAULT 1974), his archaeological work on modern concepts of madness and medicine (FOUCAULT 1972, 1973) deals in greater detail with the very questions of "practical" classification or classification as social practice (be they binary concepts such as the normal and the pathological, reason and madness, the knowing subject and the subject-object of knowledge, or the legitimate speaker and the excluded ). LACLAU and MOUFFE (1991) relate the role of classifications to the articulation of collective identities, to the functioning of distinctions between us and them, for example, as well as related attributes. Although such constitutions of subject positions by classification are a very important feature in discourse research, the research should not be reduced to this topic alone. Following its research questions, it should rather consider all kinds of classification which are performed by a given discourse. As far as I know there is, besides some general work on classification (BOWKER & LEIGH-STAR 2000) in the Anselm STRAUSS tradition and in social studies of science, little or no reflection on how to analyse classification in qualitative research. Surely it requires a kind of analytical attitude which "deconstructs" discursive (textual) units in order to rearrange them in the form of tables. [28]

(3) Phenomenal structure: The concept of "phenomenal structure" bears upon the idea of "Aspektstruktur" which Karl MANNHEIM introduced into the sociology of knowledge (see MANNHEIM 1969, p.234). It does not refer to some kind of "ontological entity" that is supposed to be behind representations or to some essential qualities of a phenomenon. Rather it assumes that the structure of a phenomenon is constituted by discourse. This phenomenal structure includes cognitive devices like the concepts used to name an object, the relations between those concepts, the introduction of causal schemes and normative settings, the dimensions, urgencies and legitimations for action, as well as the kind of practices considered to be suitable to a particular phenomenon. In order to gain the formal dimensions and the concrete elements of such a structure, it might be helpful to follow some methodical devices proposed by grounded theory (STRAUSS 1998). Grounded theory elaborates theoretical understanding of actions and interactions in institutional fields like hospitals (e.g. interactions between staff, patient and technical infrastructure). It proceeds by using different steps of "coding", writing "memos", etc. in order to generate typified concepts and relations between them which account for a given arena of social practice. Discourse research can use this procedure for elaborating condensed descriptions of phenomenal structures on the basis of its empirical data. [29]

(4) Narrative structure (narration, plot, story line): Different elements of symbolic order—such as Deutungsmuster (interpretative schemes), classifications and phenomenal structure—are tied together by narrative elements such as a story line or a plot which explains who is doing what and why. These may be stories of progress, of true and false knowledge as in scientific debates, they may be stories of heroes, "criminals", causalities, moralities, responsibilities, undesirable consequences, danger and promise of paradise as in much public discourse. Story lines organise the genealogy of discourses (through histories on reasons for change) as well as its present symbolic order. There is a rich literature on the theory of narration, as well as on concepts and procedures for their analysis, e.g. an analysis of the structure of or relations between "actants" and their implication in the story established by a discourse (see RICOEUR 1988; GREIMAS 1970; VIEHÖVER 2001, 2003). Social actors make use of story lines in order to form discourse coalitions through different fields of practice. By not considering the narrative element of "ways of world making" (Nelson GOODMAN), one risks producing an account of disparate elements, which looses sight of the network of relations that is specific to a discourse. [30]

In proposing these elements for sociology of knowledge research on discourse I do not pretend to provide ultimate concepts or devices for discourse analysis in the social sciences. Other research questions, other traditions of qualitative research may enrich or replace concepts proposed above. It is important to keep in mind that strategies of qualitative data analysis cannot be used in one to one translation. They need to be adapted to the specific issues of discourse studies, because the level of analysis in discourse research is both identical with, and beyond the singular [unique] (typified) set of data. A given document for analysis is a "fragment of discourse" (Siegfried JÄGER), it can only give partial answers to general research questions. Therefore, discourse research requires larger corpuses of data and has to develop its own "arts of combining", of building up general arguments in step-by-step procedures. Moreover it has to link the analysis of materialities with symbolic order by following the logic of "abduction" (Charles S. PEIRCE).¹¹⁾ [31]

Some elements of the "grounded theory" toolbox might be very helpful in organising data collection and analysis (STRAUSS 1998), for example to reduce data samples by "controlled" strategies. "Theoretical sampling" informs the collection of data as well as its selection for analysis. The concepts of "minimal" and "maximal contrast" are very instructive for the exploration of fragments of discourse. They suggest to start an analysis with some data or document and then to look for the next piece of data (such as a book, a news text, a policy document etc.) either by criteria of "similarity at first glance" or "complete difference at first glance", the former being useful to develop precise reconstruction of core elements, the latter being helpful to explore the range of heterogeneities in a discourse or discursive field. Since today more and more discourse data is available as digitalised text, it becomes easier to work with computer added qualitative data analysis. One should keep in mind that programmes at hand are useful tools to organise research and data analysis; but they do not replace researchers' tasks and interpretative strategies. [32]

The first reflections on the programme of Wissenssoziologische Diskursanalyse were presented in the late nineties (see KELLER 1997, 1998, 2001), and since then they have been informing empirical research in different fields of the social sciences and stimulating a wider debate on the relationship between "sociology of knowledge" and "discourse theory and research".¹²⁾ Whereas the present empirical research might, as I hope, contribute to further reflection on the practice of discourse research, the more theoretical contributions may lead to an open dialogue between different strands of knowledge analysis, especially between poststructuralism, cultural studies and the interpretative paradigm as represented by BERGER and LUCKMANN. Rather than presenting a conclusion I would insist on opening up the debate on social scientific analysis of power/knowledge regimes and processes in (the transformation of) modern, global societies. To me it still holds true what BERGER and LUCKMANN (1980, p.99) stated quite a while ago: "Sociology of knowledge faces a large open field of empirical problems." [33]

1) KELLER (2005) contains an extended discussion of discourse theories from FOUCAULT to Cultural Studies, including LACLAU and MOUFFE. <back>

2) See KELLER (2005) for a discussion of the sociology of knowledge tradition in relationship to discourse research; see KELLER, HIRSELAND, SCHNEIDER and VIEHÖVER (2005 forthcoming) for current dialogues between discourse theories and the sociology of knowledge. <back>

3) This is close to a definition given by Stuart HALL: "Discourses are ways of referring to or constructing knowledge about a particular topic of practice: a cluster (or formation) of ideas, images and practices, which provide ways of talking about, forms of knowledge and conduct associated with a particular topic, social activity or institutional site in society" (HALL 1997, p.4). <back>

4) GIDDENS is neither the first nor the only one to propose this understanding of the structure/action (practice) relationship, but it was through his work that this argument became a focal point in social theory. <back>

5) See KELLER (2004) for an extended discussion of the discourse research process. <back>

6) Recently I was told that in French class rooms one is supposed to accept the one true and authoritative meaning of texts such as poems or novels, whereas in German schools one is taught to question critically what is presented as the only possible interpretation. <back>

7) See LAW (1994) for the procedural character of symbolic order which is never fully accomplished. <back>

8) This concept is close to the concept of frame as used recently in interactionist social movement research (e.g. by William GAMSON 1988). <back>

9) For example, one considers human bodies as machines, and health as the functioning of the machine, in case of illness one will conceive of physicians as human engineers. <back>

12) See KELLER (2004, 2005) and KELLER, HIRSELAND, SCHNEIDER and VIEHÖVER (2005). <back>

Angermüller, Johannes (2004). Michel Foucault – auf dem Weg zum soziologischen Klassiker? Soziologische Revue, 27(4), 385-394.

Berger, Peter L. & Luckmann, Thomas (1980). Die gesellschaftliche Konstruktion der Wirklichkeit. Eine Theorie der Wissenssoziologie. Frankfurt/Main: Fischer. [The Social Construction of Reality. New York 1966]

Bowker, Geoffrey C. & Leigh Star, Susan (2000). Sorting Things Out. Classification and Its Consequences. Cambridge, Mass.: The MIT Press.

Dreyfus, Hubert L. & Rabinow, Paul (1987). Michel Foucault. Jenseits von Strukturalismus und Hermeneutik. Frankfurt/Main: athenaeum. [Michel Foucault. Beyond structuralism and hermeneutics. Chicago 1982]

Foucault, Michel (1972). Die Geburt der Klinik. Eine Archäologie des ärztlichen Blicks. Frankfurt/Main: Ullstein. [Naissance de la clinique: une archéologie du regard medical. Paris 1963]

Foucault, Michel (1973). Wahnsinn und Gesellschaft. Frankfurt/Main: Suhrkamp. [Abridged German translation of: Folie et déraison: Histoire de la Folie à l'âge classique. Paris 1961]

Foucault, Michel (1974a). Die Ordnung der Dinge. Eine Archäologie der Humanwissenschaften. Frankfurt/Main: Suhrkamp. [Les mots et les choses. Une archéologie des sciences humaines. Paris 1966]

Foucault, Michel (1974b). Die Ordnung des Diskurses. München: Hanser. [L'ordre du discourse. Paris 1971]

Foucault, Michel (1988). Archäologie des Wissens. Frankfurt/Main: Suhrkamp. [Archéologie du savoir. Paris 1969]

Foucault, Michel (1992). Was ist Kritik? Berlin: Merve. [Qu'est-ce que la critique? Bulletin de la Société française de philosophie 1990 (Original oral communication 1978)]

Foucault, Michel (2002). Die Wahrheit und die juristischen Formen. In Michel Foucault, Schriften in vier Bänden (Dits et Écrits. edited by Daniel Defert &. François Ewald, vol. 2: 1970-1975) (pp.669-792). Frankfurt/Main: Suhrkamp. [A verdade e as formas juridicas. Cadernos da P.U.C Nr. 16, 1974, pp 5-133; Communications at the Catholic University of Rio de Janeiro, May 21-25, 1973]

Foucault, Michel (2004a). Geschichte der Gouvernementalität I. Sicherheit, Territorium, Bevölkerung. Vorlesungen am Collège de France 1977/1978. Frankfurt/Main: Suhrkamp. [Histoire de la gouvernementalité I. Sécurité, territoire et population. Paris 2004]

Foucault, Michel (2004b). Geschichte der Gouvernementalität II. Geburt der Biopolitik. Vorlesungen am Collège de France 1978/1979. Frankfurt/Main: Suhrkamp. [Histoire de la gouvernementalité II. Naissance de la biopolitique. Paris 2004]

Foucault, Michel (2005a). Der Staub und die Wolke. In Michel Foucault, Schriften in vier Bänden (Dits et Écrits. edited by Daniel Defert & François Ewald, vol. 4: 1980-1988). Frankfurt/Main: Suhrkamp. [1994, pp.12-24, La poussière et le nuage. In: Perrot, Michelle (Ed.): L'Impossible Prison. Paris 1980, pp.29-29]

Foucault, Michel (2005b). Diskussion vom 20. Mai 1978. In Michel Foucault, Schriften in vier Bänden (Dits et Écrits, edited by Daniel Defert & François Ewald, vol. 4: 1980-1988). Frankfurt/Main: Suhrkamp. [1994, pp.25-43, Table ronde du 20 mai 1978. In: Perrot, Michelle (Ed.): L'Impossible Prison. Paris 1980, pp.40-56]

Gamson, Willliam A. (1988): The 1987 Distinguished Lecture: A Constructionist Approach to Mass Media and Public Opinion. Symbolic Interaction, 2, 161-174

Giddens, Anthony (1992). Die Konstitution der Gesellschaft. Entwurf einer Theorie der Strukturierung. Frankfurt/Main: Campus. [The Constitution of Society. Outline of a Theory of Structuration. Cambridge 1984]

Gusfield, Joseph (1981). The Culture of Public Problems: Drinking-Driving and the Symbolic Order. Chicago: University of Chicago Press

Hall, Stuart (1997). Introduction. In Stuart Hall (Ed.), Representation: Cultural Representations and Signifying Practices (pp.1-12). London: Sage.

Hitzler, Ronald & Honer, Anne (Eds.) (1997). Sozialwissenschaftliche Hermeneutik. Eine Einführung. Opladen: Leske + Budrich (UTB).

Hitzler, Ronald, Reichertz, Jo & Schröer, Norbert (Eds.) (1999). Hermeneutische Wissenssoziologie. Standpunkte zur Theorie der Interpretation. Konstanz: UVK.

Jäger, Siegfried. (1999). Kritische Diskursanalyse. Eine Einführung (2nd edition). Duisburg: Diss.

Jørgensen Marianne W. & Philipps, Louise J. (2002). Discourse Analysis as Theory and Method. London: Sage.

Keller, Reiner (1997). Diskursanalyse. In: Ronald Hitzler & Anne Honer (Eds.), Sozialwissenschaftliche Hermeneutik. Eine Einführung (pp.309-334). Opladen: Leske + Budrich (UTB).

Keller, Reiner (1998). Müll – Die gesellschaftliche Konstruktion des Wertvollen. Opladen: Westdeutscher Verlag.

Keller, Reiner (2000). Der Müll in der Öffentlichkeit. Reflexive Modernisierung als kulturelle Transformation. Ein deutsch-französischer Vergleich. Soziale Welt, 51(3), 245-266.

Keller, Reiner (2001). Wissenssoziologische Diskursanalyse. In Reiner Keller, Andreas Hirseland, Werner Schneider & Willy Viehöver (Eds.), Handbuch Sozialwissenschaftliche Diskursanalyse Bd. 1: Theorien und Methoden (pp.113-145). Opladen: Leske + Budrich.

Keller, Reiner (2003). Distanziertes Mitleiden. Katastrophische Ereignisse, Massenmedien und kulturelle Transformation. Berliner Journal für Soziologie, 3, 395-414.

Keller, Reiner (2004). Diskursforschung. Eine Einführung für SozialwissenschaftlerInnen (2nd edition). Wiesbaden: VS-Verlag.

Keller, Reiner (2005). Wissenssoziologische Diskursanalyse. Grundlegung eines Forschungsprogramms. Wiesbaden: VS-Verlag.

Keller, Reiner, Hirseland, Andreas, Schneider, Werner & Viehöver, Willy (Eds.) (2001). Handbuch Sozialwissenschaftliche Diskursanalyse Bd. 1: Theorien und Methoden. Opladen: Leske + Budrich.

Keller, Reiner, Hirseland, Andreas, Schneider, Werner & Viehöver, Willy (Eds.) (2003). Handbuch Sozialwissenschaftliche Diskursanalyse Bd. 2: Forschungspraxis. Opladen: Leske + Budrich. [2nd edition 2004 Wiesbaden: VS-Verlag]

Keller, Reiner, Hirseland, Andreas, Schneider, Werner, Viehöver, Willy (Eds.). (2005). Die diskursive Konstruktion von Wirklichkeit. Zum Verhältnis von Wissenssoziologie und Diskursforschung. Konstanz: UVK.

Knoblauch, Hubert (1995). Kommunikationskultur. Die kommunikative Konstruktion kultureller Kontexte. Berlin: de Gruyter.

Laclau, Ernesto & Mouffe, Chantal (1991). Hegemonie und radikale Demokratie. Zur Dekonstruktion des Marxismus. Wien: Passagen. [Hegemony and Socialist Strategy. London 1985]

Luckmann, Thomas (2002). Das kommunikative Paradigma der "neuen" Wissenssoziologie. In Thomas Luckmann, Wissen und Gesellschaft. Ausgewählte Aufsätze 1981-2002 (edited by Hubert Knoblauch, Jürgen Raab & Bernt Schnettler) (pp.201-210). Konstanz: UVK.

Lüders, Christian & Meuser, Michael (1997). Deutungsmusteranalyse. In Ronald Hitzler & Anne Honer (Eds.), Sozialwissenschaftliche Hermeneutik. Eine Einführung (pp.57-80). Opladen: Leske + Budrich (UTB).

Potter, Jonathan & Wetherell, Margaret (1995). Soziale Repräsentationen, Diskursanalyse und Rassismus. In Uwe Flick (Ed.), Psychologie des Sozialen. Repräsentationen in Wissen und Sprache (pp.177-199). Reinbek: Rowohlt. [Social Representations, Discourse Analysis, and Racism. In Uwe Flick (Ed.), The Psychology of the Social, Cambridge 1998]

Reichertz, Jo (2002). Die Abduktion in der qualitativen Sozialforschung. Opladen: Leske + Budrich.

Reichertz, Jo (2004, September). Das Handlungsrepertoire von Gesellschaften erweitern. Hans-Georg Soeffner im Gespräch mit Jo Reichertz [65 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 5(3), Art. 29. Available at: http://www.qualitative-research.net/fqs-texte/3-04/04-3-29-d.htm [Date of Access: July 21, 2005].

Ricoeur, Paul (1988). Zeit und Erzählung. Bd. 1: Zeit und historische Erzählung. München: Fink. [Temps et récit. Vol. 1.: L'intrigue et le récit historique. Paris 1983]

Schetsche, Michael (2000). Wissenssoziologie sozialer Probleme. Begründung einer relativistischen Problemtheorie. Opladen: Westdeutscher Verlag.

Schröer, Norbert (1997). Wissenssoziologische Hermeneutik. In Ronald Hitzler & Anne Honer (Eds.), Sozialwissenschaftliche Hermeneutik. Eine Einführung (pp.109-132). Opladen: Leske + Budrich (UTB).

Schütz, Alfred & Luckmann, Thomas (1979). Strukturen der Lebenswelt, vol. 1. Frankfurt/Main: Suhrkamp.

Soeffner, Hans-Georg (1989). Auslegung des Alltags – Der Alltag der Auslegung. Zur wissenssoziologischen Konzeption einer sozialwissenschaftlichen Hermeneutik. Frankfurt/Main: Suhrkamp.

Soeffner, Hans-Georg & Ronald Hitzler (1994). Hermeneutik als Haltung und Handlung. Über methodisch kontrolliertes Verstehen. In Norbert Schröer (Ed.), Interpretative Sozialforschung. Auf dem Wege zu einer hermeneutischen Wissenssoziologie (pp.28-55). Opladen: Westdeutscher Verlag.

Strauss, Anselm (1998). Grundlagen qualitativer Sozialforschung. München: Fink (UTB). [Qualitative Analysis for Social Scientists. Cambridge 1987]

Viehöver, Willy (2001). Diskurse als Narrationen. In Reiner Keller, Andreas Hirseland, Werner Schneider & Willy Viehöver (Eds.), Handbuch Sozialwissenschaftliche Diskursanalyse Bd. 1: Theorien und Methoden (pp.177-206). Opladen: Leske + Budrich.

Viehöver, Willy (2003). Die Wissenschaft und die Wiederverzauberung des sublunaren Raumes. Der Klimadiskurs im Licht der narrativen Diskursanalyse. In Reiner Keller, Andreas Hirseland, Werner Schneider & Willy Viehöver (Eds.), Handbuch Sozialwissenschaftliche Diskursanalyse Bd. 2: Forschungspraxis (pp.233-270). Opladen: Leske + Budrich.

Wetherell, Margaret (1998). Positioning and interpretative repertoires: conversation analysis and post-structuralism in dialogue. Discourse & Society, 9(3), 387-412.

Dr. Reiner KELLER, born in 1962, studied sociology at the University of Saarbrücken (Germany), Rennes (France), and Bamberg (Germany). In the early nineties he did empirical research on questions of the social use of psychotherapy and on environmental discourse in different European countries. His PhD thesis analysed the symbolic-cultural dimension of German and French public discourses on waste policies. Since 1997 he has been teaching sociology at the University of Augsburg (Germany). In 2004 he completed his postdoctorate research on the subject of the Wissenssoziologische Diskursanalyse. His main research interests include the development of discourse theory and research, the sociology of knowledge and culture, the theory of reflexive modernisation and the boundary management between nature and society inherent to institutional practices. He has published extensively on these issues.

Universität Augsburg
PhilSo-Fakultät (Lehrstuhl für Soziologie)
Postfach
86135 Augsburg, Gerrmany

Keller, Reiner (2005). Analysing Discourse. An Approach From the Sociology of Knowledge [33 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(3), Art. 32, http://nbn-resolving.de/urn:nbn:de:0114-fqs0503327.

Forum Qualitative Sozialforschung / Forum: Qualitative Social Research (FQS)

ISSN 1438-5627

Creative Commons Attribution 4.0 International License