ogl 530

from the SAGE Social Science Collections. All Rights Reserved.

Human Relations

[00 I 8-7267(200009)53:9]

Volume 53(9): I 125-1149:013426

Copyright © 2000

The Tavistock Institute ®

SAGE Publications

London, Thousand Oaks CA,

New Delhi

Varieties of discourse: On the study of organizations through discourse analysis Mats Alvesson and Dan Karreman

ABSTRACT

KEYWORDS

Discourse is a popular term used in a variety of ways, easily leading

to confusion. This article attempts to clarify the various meanings of

discourse in social studies, the term's relevance for organizational

analysis and some key theoretical positions in discourse analysis. It

also focuses on the methodological problem of the relationship

between: a) the level of discourse produced in interviews and in

everyday life observed as 'social texts' (in particular talk); b) other

kinds of phenomena, such as meanings, experiences, orientations,

events, material objects and social practices; and, c) discourses in the

sense of a large-scale, ordered, integrated way of reasoning/

constituting the social world. In particular, the relationship between

'micro and meso-level' discourse analysis (i.e. specific social texts

being the primary empirical material) and 'grand and mega-level'

discourse (i.e. large-scale orders) is investigated.

discourse ■ discourse analysis ■ methodology ■ organization

study

I 125

I I 2 6 Human Relations S3(9)

Recent developments in philosophy, sociology, social psychology and com­ munications theory (Deetz, 1992; Potter & Wetherell, 1987; Rosenau, 1992; Silverman, 1993; Steier, 1991) have demonstrated the ubiquity and import­ ance of language in social science. Indeed, it seems that language (and lan­ guage use) is increasingly being understood as the most important phenomenon, accessible for empirical investigation, in social and organiz­ ational research. To the extent social research is an empirical enterprise, most of it seems to be connected to how people use language - sometimes how language uses people - in particular situations. Interviews, for example, are obviously rich in linguistic interaction and most observations concern people engaged in conversations.

Arguably, the insight of the significance of language has contributed to an interest in discourses. This is part of the general turn in social science, including, among other disciplines, organization studies (Alvesson & Karre­ man, 2000). The question is what kind of discourse organizational analysts ought to pay attention to, and how much attention they ought to pay to dis­ course. The scope of this article is primarily theoretical and methodological (i.e. it concerns the interface between theory and method where we develop a standpoint on how we conceive and interpret empirical material). It aims to clarify the key analytical options available in discourse analysis and their consequences for the study of organizational and other social phenomena.

There is a wide array of ways of using the term discourse in social science and organization studies. It is often difficult to make sense of what people mean by discourse. In many texts, there are no definitions or discus­ sions of what discourse means. Authors treat the term as if the word has a clear, broadly agreed upon meaning. This is simply not the case. An import­ ant aim of this article is to reduce the confusion and facilitate more informed research into organizational discourse. The article proceeds from what seems to be two major and quite different approaches to 'discourse' in organization studies: the study of the social text (talk and written text in its social action contexts) and the study of social reality as discursively constructed and main­ tained (the shaping of social reality through language). The former approach highlights the 'talked' and 'textual' nature of everyday interaction in organiz­ ation. The latter focuses on the determination of social reality through his­ torically situated discursive moves.

Put differently, in the former approach discourses are viewed as local achievements, analytically distinct from other levels of social reality (such as the levels of meaning and practice), and with little or no general content. The latter approach, on the other hand, views discourses as general and preva­ lent systems for the formation and articulation of ideas in a particular period of time. In the latter approach - in viewing discourses as 'Discourses' - other

Alvesson & Karreman Varieties of discourse I I 2 7

levels of social reality are more or less shaped or even subordinated by the power-knowledge relations established in discourse. A common problem in organization analysis is, methodologically speaking, how to move beyond the specific empirical material, typically linguistic in its character (interview accounts, questionnaire responses, observed talk in 'natural settings' and written documents) and address discourses with a capital D - the stuff beyond the text functioning as a powerful ordering force. Traps and possi­ bilities will be discussed and the implications of moving from discourses to Grand Discourses are also considered.

The many meanings of discourse

The word discourse has, as indicated above, no agreed-upon definition, and confusingly many uses. As Potter and Wetherell remark, discourse and dis­ course analysis 'is a field in which it is perfectly possible to have two books [on the matter] with no overlap in content at all' (Potter & Wetherell, 1987: 6). Discourse may mean almost everything (Grant et al., 1998; Keenoy et al., 1997).

However, it is possible to identify some distinctive takes on the term among discourse analysts. Keenoy et al. (1997), for example, make two dis­ tinctions. The first is between authors using discourse as a device for making linguistic sense of organizations and organizational phenomena versus seeing discourses in the context of revealing the ambiguities of social constructions and the indeterminacy of organizational experiences. The second is between a position looking at discourse in a social context, including the social and political dimensions in addition to the discursive versus a more narrow focus on the text per se (treated as existing in a contextual vacuum). 1

In another review, Potter (1997) identifies five versions of discourse analysis. Three of these relate to linguistic and cognitive psychology and are of limited interest here. The fourth version is the standard Foucauldian pos­ ition as developed below. The fifth version is Potter and Wetherell's (1987) own. Here the task is to study discourse as texts and talk in social practices. Language is viewed as a medium for interaction, rather than as a system of differences (as in structuralism), or a set of rules for transforming statements (as in Foucauldian genealogies). Analysis of discourse becomes analysis of what people do with language in specific social settings (Potter, 1997: 146).

A well known and often copied take on discourse is usually attributed to Foucault. It is typically described to advance from the assumption that dis­ courses, or sets of statements, constitute objects and subjects. Language, put together as discourses, arranges and naturalizes the social world in a specific

I I 28 Human Relations 53(9)

way and thus informs social practices. These practices constitute particular forms of subjectivity in which human subjects are managed and given a certain form, viewed as self-evident and rational (Foucault, 1976, 1980).

Foucault proposed two ways of investigating discourses: archaeology and genealogy. Archaeology can be seen as directly related to the clarification of the history of the rules that regulate particular discourses. Genealogy looks after the forces and events that shape discursive practices into units, wholes and singularities. As Davidson (1986: 227) puts it:

Archaeology attempts to isolate the level of discursive practices and for­ mulate the rules of production and transformation for these practices. Genealogy, on the other hand, concentrates on the forces and relations of power connected to discursive practices; it does not insist on a separ­ ation of rules for production of discourse and relations of power. But genealogy does not so much displace archaeology as widen the kind of analysis to be pursued.

The reviews of Keenoy and Potter are valuable but we feel that research options and interests signalled by the word discourse are not exhausted by their reviews. And although Foucault is probably the single most influential author on how social scientists use the word discourse, it is clear that students of discourse feel that neither the archaeological, nor the genealogical approach exhaust their research possibilities. Within social science and organization studies 'discourse' generally signals a variety of interests, often difficult to grasp.

While recognizing the constraints of demands on rigour, we cannot help sometimes feeling that the word discourse is used to cover up muddled think­ ing or postponed decisions on vital analytical matter - as often is the case with popular words and areas. Discourse sometimes comes close to standing for everything, and thus nothing. One possible explanation is that researchers may want to avoid choosing between a language and a meaning focus - meaning here signalling a relatively stable way of relating to and making sense of something, a meaning being interrelated to an attitude, value, belief or idea. The motive may be to avoid the problems of assuming or investi­ gating a set of stable and connected meanings at the same time as a desire to withhold an interest in substantive social phenomena beyond the level of lan­ guage. This motive may be acceptable, but it calls for attempts to sort out the aspects involved: how language use is related to other issues (meanings, practices) must be dearly elaborated.

Sometimes discourse is used to signal a firm shift from traditional con­ cerns (inner worlds such as values and cognition as well as from external

Alvesson & Karreman Varieties of discourse I I 2 9

issues such as practices and events) to a strong interest in language in action/social context. Here the term is used to focus on language and lan­ guage use in social context, not meanings, the actor's point of view or some­ thing similar. For them, language represents another focus than cognition. There are, as critics of interviews referred to above point out, methodologi­ cal reasons for this. We will refer below to Potter and Wetherell (1987) as coherent proponents of such a view.

In many cases, however, discourse signals a less coherent, perhaps even more half-hearted - one might also say less reductionist - effort to get away from 'older' concerns about stable cognition, values, beliefs, ideologies and other meaning related issues. Discourse typically vaguely signals an interest in language and language use, but often also in these other issues. It may, for example, signal a set of representations and ways of structuring reality that put strong imprints on cognition and attitudes. Sometimes cognition and attitudes or a general way of relating to something become incorporated by the concept of discourse. Here the interest in discourse represents a started but not com­ pleted linguistic turn, a moderate version of it. Language per se is not strongly focused, as 'other' phenomena (such as ideas, meanings, etc.) get their share of attention. Quite often, discourse does not seem to signal any particular inter­ est in language, but refers to conceptions, a line of reasoning, a theoretical pos­ ition or something similar. Kerfoot and Knights (1998: 21), for example, refer to 'conceptions of the "New Man", the Men's Movement, Men's Rights .. .' and then, a few lines later, talk about 'these various discourses', and then again, 'this diverse range of activities', without any apparent distinctions.

Mapping versions of discourse analysis

It seems to us that many of the different versions of discourse analysis can be fruitfully analysed along two key dimensions. The first is the connection between discourse and meaning (broadly defined): does discourse precede and incorporate cultural meaning and subjectivity or is it best understood as referring to the level of talk (and other forms of social texts) loosely coupled to the level of meaning? The second is the formative range of discourse: is discourse best understood as a highly local, context-dependent phenomenon to be studied in detail or does it mean an interest in understanding broader, more generalized vocabularies/ways of structuring the social world? Below, we will develop and elaborate the meaning of each dimension. Combining the two dimensions used above - formative range and discourse/meaning­ relation - results in the following matrix for the analysis of discourse studies in social science (illustrated in Figure 1).

I I 3 0 Hilman Relations S3(S>)

Discourse determination

MUSCULAR

( discourse/meaning collapsed)

Close-range interest

(local-situational context) MYOPIC

GRANDIOSE

Long-range interest

(macro-system context)

Figure I Two core dimensions in discourse studies

Discourse autonomy

TRANSIENT

( discourse/meaning unrelated)

Discourse and meaning: overlapping, tightly coupled, loosely coupled or uncoupled

No language use is totally devoid of meaning. Language use calls for intelli­ gibility. Language use is distinct from other kinds of noise making. The dis­ tinction is, as always, a bit tricky as noise making without words may communicate intentionally. A cough, for example, may work as a warning or sanction.

However, one may distinguish between transient meaning, emerging from specific interaction, and durable meaning, existing so to speak 'beyond' specific linguistic interaction, in a more or less inert and stable manner. 2 This latter, broader meaning includes meaning-phenomena such as cultural and individual ideas, orientations, ways of sensemaking, and cognition. 3 We have more of a spectrum of opportunities and research positions in the relation­ ship between discourse in the sense of language and language use and mean­ ings 'existing' beyond - although hardly independent of - the temporal and specific use of language. These positions range from viewing discourse/ meaning as inseparable or strictly overlapping, to imagining various kinds of

Alvesson & Karreman Varieties of discourse I I 3 I

coupling, from tightly to loosely, and then to treat discourse and meaning as (almost) uncoupled - except from highly limited and temporal effects of dis­ course on meaning, fading away with the next moment of language use. We will start by illustrating the use of a concept of discourse including not only language but also other aspects, such as cognition.

Language is the central focus of all post-structuralism. In the broadest terms, language defines the possibilities of meaningful existence at the same time as it limits them. Through language, our sense of ourselves as distinct subjectivities is constituted. Subjectivity is constituted through a myriad of what post-structuralists term 'discursive practices': practices of talk, text, writing, cognition, argumentation, and rep­ resentation generally.

(Clegg, 1989: 151)

Discursive practices are here a collection of various stuff, including cog­ nition. Discourse drives subjectivity (our sense of ourselves, including our feelings, thoughts and orientations), presumably in an all-embracing and muscular fashion. This position is also expressed by Weedon (1987: 41) who views discourse 'as a structuring principle of society, in social institutions, modes of thought and individual subjectivity'. This 'structuring principle' determines meaning and subjectivity: 'subjectivity is itself an effect of dis­ course' (1987: 86).

Another version allows for some separation of discourse and meaning. The former affects (frames) the latter. Watson (1994: 113) for example, defines discourse as 'a connected set of statements, concepts, terms and expressions which constitutes a way of talking and writing about a particular issue, thus framing the way people understand and act with respect to that issue'. Here we find a relatively tight coupling between discourse which influences talking and writing which in their turn frame cognition and actions. However, dis­ course does not incorporate the latter phenomena. Such framing could, one may imagine, be strong or weak, even though the assumption seems to be that discourse typically is powerful in this respect. This formulation still gives some space for variation in the relationship between discourse and other 'extra­ discursive' social phenomena. The relationship is not solely a definitional matter. There is some space to investigate it empirically.

Some authors clearly distinguish between discourse and other elements. Van Dijk (1993: 251), for example, argues that in order to relate discourse and society, 'we need to examine in detail the role of social representations in the minds of social actors'. He also suggests that (critical) discourse analy­ sis, 'requires true multi-disciplinarity, and account of intricate relationships

I I 3 2 Human Relations 53(9)

between text, talk, social cognition, power, society and culture'. Here cogni­ tion and other non-linguistic elements are not just something that is 'run over', included or colonized by discourse, but are vital elements to examine.

Another possibility is to assume that the level of the discourse should, in terms of method, not be directly connected to another kind of phenom­ enon. We can study discourse and see what it accomplishes as a linguistic performance, but refrain from drawing any conclusions on, or making assumptions about the relationship to what is more strictly seen as other phenomena. One may argue that the level of discourse has some autonomy in relationship to other levels: language use follows its own dynamics. Talking in certain ways or reproducing a specific vocabulary does not imply any specific cognition, feelings or practices. One may talk about the weather without constituting one's subjectivity as a meteorologist. People may produce politically correct opinions in interviews or conversations without any particular feelings or convictions being involved. One may also take the position that the study of the discursive is (a) a major task and (b) may be carried out with a fairly high degree of rigour. Addressing what may go on in people's heads and hearts is another issue, disrupting discourse analysis, and may only be carried out in a speculative manner. This seems to be the position of discourse and conversational analysts and ethnomethodologists (e.g. Potter & Wetherell, 1987; Silverman, 1993).

We can here talk about a 'non-relation' or an agnostic view on the relationship between discourse and other elements (such as emotions, con­ victions and beliefs). Instead of embracing the increasingly popular view of discourse as constituting reality and subjectivity one may be more careful in one's assumptions. The idea of fragile subjects constituted by and/or within strong discourse may ascribe too much power to the latter (Newton, 1998), which is, of course not to say that discourse cannot be very powerful. There may be considerable variation. The ways in which subjects relate to discourse may be teflon-like; the language they are exposed to or use may not 'stick'. Rather than the discourse-driven subject, the subject may be a politically con­ scious language user, telling the right kind of stories to the right audiences at the right moment. Jackall (1988) suggests that the world of corporate management calls for sophisticated, politically conscious language use, in which the language user separates or loosely couples discourse and meaning:

The higher one goes in the corporate world, the more essential is the mastery of provisional language. In fact, advancement beyond the upper-levels depends greatly on one's ability to manipulate a whole variety of symbols without becoming tied to or identified with any of them.

(Jackall, 1988: 137)

Alvesson & Karreman Varieties of discourse I I 3 3

This is not to say that language use is without effects on the level of meaning (constituting subjectivity). The effects may, however, be uncertain, weak and temporal in terms of constituting subjectivity, defining the possi­ bilities of meaningful existence or something similarly powerful. We refer to the first position sketched above, in which discourse directly implies or incor­ porates social and psychological consequences, as discourse determination. We label the other end of the spectrum, where discourse, in principle, stands on its own or is loosely coupled to the social (individual), discourse auton­ omy. 4

The formative range of discourse: local-situated and macro­ systemic

The other key dimension concerns the formative range of discourse - assump­ tions on the scope and scale of discourse. One option is to take an interest in discourse at close range, considering and emphasizing local, situational context. Language use is here understood in relationship to the specific process and social context in which discourse is produced. At the other extreme we see discourse as a rather universal, if historically situated, set of vocabularies, standing loosely coupled to, referring to or constituting a particular phenomenon. We may talk about long-range, macro-systemic discourse. We may, for example, talk about a discourse on masculinity, on management in the Western world or on consumption in affluent society. In order to clarify the difference between discourse in these two senses, we refer to long-range discourse as Discourse.

There are, of course, also 'middle-range' conceptualizations of dis­ course/Discourse. Starting with Potter and Wetherell's (1987) concept of dis­ course and working ourselves 'upwards' it is possible to point out four versions of discourse analysis:

(1) micro-discourse approach - social texts, calling for the detailed study of language use in a specific micro-context;

(2) meso-discourse approach - being relatively sensitive to language use in context but interested in finding broader patterns and going beyond the details of the text and generalizing to similar local contexts;

(3) Grand Discourse approach - an assembly of discourses, ordered and presented as an integrated frame. A Grand Discourse may refer to/ constitute organizational reality, for example dominating language use about corporate culture or ideology;

(4) Mega-Discourse approach - an idea of a more or less universal connection of discourse material. Mega-Discourse typically addresses

I I 3 4 Human Relations 53(9)

more or less standardized ways of referring to/constituting a certain type of phenomenon, e.g. business re-engineering, diversity or global­ ization.

A Discourse, or rather indications of it, shows up at a large number of sites in more or less different ways and is, methodologically, treated as being of a more or less standardized nature. The idea is that it is possible to cut through the variation at the local levels through summaries and syntheses that identify over-arching themes operating in specific situations. Overall categories and standards then tend to be privileged in the treatment of em­ pirical material. As Miller (1997: 34) describes Foucault-inspired work: 'Whatever the form of the data, Foucauldian discourse studies involve treat­ ing the data as expressions of culturally standardised discourses that are associated with particular social settings'.

This is, of course, accomplished at the expense of paying attention to the complex social practices and variations at the local level. Close-range studies focus on these. They emphasize the need to take social context and interactions seriously. For those interested in dose-range studies, discourse is local business, not primarily expressions of imperialism. This does not mean that synthesis or connections to more general patterns are impossible; widely used cultural categories or resources may, for example, be identified in detailed studies of specific texts (e.g. Baker, 1997). Rather, it means that the richness of the material and considerations of uniqueness makes general pat­ terns less visible and also somewhat beside the point. In a similar manner, long-range conceptions of discourse do not deny local variation. The point is, however, to address big issues and not the delicacies and nuances that might embed them.

Sometimes researchers want to address both micro and macro levels. Fairclough (1993: 138), for example, is interested in discourse, in the sense of spoken or written language use at the level of discursive practice, as well as what he refers to as 'the order of discourse', 'the totality of discursive prac­ tice of an institution, and relationships between them'. We think that there is a tension between these two levels. Investigations of the local construction of discourse treat discourse as an emergent and locally constructed phenom­ enon, while the study of Discourses usually starts from well established a priori understandings of the phenomenon in question. It is not easy, we believe, to accurately account for both in the same study. This should not, however, discourage such efforts. Rigour should sometimes be downplayed for the benefit of social relevance.

From now on we will illustrate the framework from an empirical point of view. This will enable us to elaborate on the typology offered in Figure 2.

Alvesson & Karreman Varieties of discourse I I 3 5

Close-range interest

(local-situational context)

t Micro-discourse

CLOSE-RANGFJDETERMINATION CLOSE-RANGFJAUTONOMY

Meso-discourse

Discourse

-+-- collapsed - tightly coupled

determination

Discourse

loosely coupled -- unrelated ---+­ autonomy

Grand Discourse

LONG-RANGFJDETERMINATION LONG-RANGFJAUTONOMY

Long-range interest (macro-system context}

Figure 2 Elaboration of core dimensions and summary of positions in discourse studies

We start by indicating how a piece of empirical material (an interview account) may be handled. In a later section we will discuss and re-interpret parts of Fournier's (1998) full empirical study, examining how her con­ ceptualization of discourse differs from those presented in Figure 2.

The analytic options in discourse analysis: an empirical illustration

The differences between the various perspectives is probably most effectively demonstrated through showing how each perspective would deal with a dis­ cursive fragment generated in an ongoing empirical study. The study con­ cerns journalists, leadership and ethics in a newspaper. One of our informants

I 136 Human Relations 53(9)

communicated the following piece of discourse in an interview with one of us:

Leadership involving reporters should ideally be an interplay tuned towards the optimal story, but in reality - due to deadlines and things like that - stories are often less than optimal. The normal case is that you assign a reporter to a loosely defined task, who goes out and gives the story his or her best shot, comes back and writes up the piece, and you skim through the piece, and put it in one of your drawers and that's really how close you come to a discussion, unfortunately. That's the bitter truth.

(Editor-in-chief)

What makes this piece of discourse interesting from a close­ range/autonomous point of view is how it is located in the stream of dis­ course, rather than what it may say extracted from the specific textual interaction that produced the statement. Thus, from this perspective, its meaning or significance cannot be decided in isolation from the interaction that shapes the context. The account is produced in interaction with a researcher from a business school, asking questions about 'leadership'. This leads the interviewee to address this topic. In the account, the importance of the topic is acknowledged. Leadership is described as involving interplay tuned towards making articles better, even optimal. Whether the statement says anything substantial about the values, conceptions and practices of the newsmakers is a non-issue. 'Leadership', 'reality' and 'deadline', etc. are viewed as purely textual phenomena, meaning that what is of interest is what they do in the account (make it, for example, legitimate, innovative and/or trustworthy or make the social interaction run smoothly), and not whether they reflect 'true' conditions or not. Generally, or in other specific situations within the site of study, this discourse on 'leadership' may not be visible at all. In the account, 'leadership' is viewed as targeted towards the optimiza­ tion of the journalist's story. But 'reality', not human agency, means that not much 'leadership' in this sense can be produced. Deadlines 'and things like that' being part of 'reality' means that the optimal performance is not accom­ plished. The interviewee presents himself as an honest, realistic, down­ to-earth person, not wanting to idealize things. He works and adapts to reality, and speaks the 'bitter truth'.

The account is brief and further research proceeding from a close­ autonomous approach to discourse may benefit from more material allow­ ing exploration of questions like: To what extent and with what means is the leadership theme sustained, varied and contradicted? What particular social

Alvesson & Karreman Varieties of discourse I I 3 7

reality is invoked and maintained? To what extent and how is a leadership discourse part of the interpretive repertoire of the people in the study (Potter & Wetherell, 1987)? It is these kinds of question - and only these - a researcher operating within a close-range/autonomous approach would ask and seek to answer.

The editor-in-chief's account of leadership and work practices is inter­ esting from a long-range/autonomous position to the extent that it displays standardized forms of speech on the subject matter. The relevant question is thus whether the utterance in question can be related to other, similar utter­ ances on the topic of leadership and newsmaker work practices. Again, whether the statement (or the statements) reflects 'true' conceptions of leader­ ship, and leadership and work practices, or not, is a non-issue. 5 If the editor­ in-chief's utterance is broadly coherent with other examples of discourse - other people talking on the subject in other places and settings - then it is used, in this approach, both as part of and as evidence of a particular stan­ dardized and trans-local discourse on leadership. Such a discourse may for example be located in a Swedish cultural context, where 'leadership' nation­ ally may be coupled with words like participation and dialogue rather than language use emphasizing the control and authority of the great leader. The macro-level discourse, of which the account is seen as an example, may also be interpreted as a professional discourse, perhaps independent of national cultural orientations. Professional discourses emphasize the competence and discretion of the professionals and this may affect leadership discourse. Another possibility is to interpret the account as saying something about organizational discourse on the subject matter. The particular organization, rather than the country or the profession, would be seen as the site or context of the discourse of which the account may be seen as an example. Words like 'interplay' and 'discussion' rather than leading or controlling may thus be seen as connecting to either a Swedish, a professional or an organizational (company-specific) discourse on leadership. Of course, more empirical material is called for before any conclusions may be drawn. If it cannot be connected to other statements it is ignored or regarded as an interesting deviant case, depending on how it relates to the conventional wisdom (and, perhaps, the researcher's eye for detail).

Researchers adopting the close-range/determination approach or inter­ est assume that discourses offer important clues to other kinds of practices than pure language use, which of course stands in stark contrast to what 'discourse autonomy' advocates believe. This does not mean that they nai'vely assume that what the editor-in-chief claims in the statement must be true in any representational sense of the word. Rather, it means that the account above may say something instructive on the local construction of leadership,

I I 3 8 Human Relations 53(9)

including feelings and norms toward the subject matter. This means that the statement enables the researcher to further specify the domain of social reality of interest. Thus, when the editor-in-chief talks about leadership as tuning stories towards optimality, researchers acting from the close-range/determi­ nation perspective treat this as clues on a normative ideal, constructing the subjectivity of the person and framing his action. The ideal of leadership and the perceived failure to live up to it - discursively brought about by the self­ scrutiny contingent upon the presence of this discourse - then invokes the notion of the subject confessing a less than optimal way of being. The tem­ poral operation of the leadership discourse then would imply a temporal re-constitution of the editor-in-chief as a 'leader-interplay-tuned-towards-the­ optimal-story'.

In order to qualify for a close-range/determination notion of discourse, the account must be evaluated as having structuring effects, either on the sub­ jectivity of the interviewee (and other people in the site in question) and/or in terms of framing action. This would imply that the researcher would try to investigate the extent normative ideals are translated into practices in the setting under study or getting other indications on the subjectivity of the inter­ viewee (or other people) in the kind of situation/context the interview account is located within. 6 Addressing the account as a discourse having effects is something other than addressing it as pure talk (without the determining power of a discourse). From the close-range/determination perspective, dis­ course can reveal information on non-discursive phenomena but it can only do that for the particular, highly local domain of social reality under study.

From a long-range/determination position the account of the editor­ in-chief provides discursive fragments that illustrate the rules that decide how we can talk about and experience leadership and work practices. Thus, in this perspective, the statement made by the editor-in-chief is produced, not by a concrete individual in specific circumstances, but rather by the rules that decide how individuals can articulate the leadership phenomenon, both in discourse (in the narrow sense of language use) and as an experiential phenomenon. The long-range/determination interest in discourse assumes that discourse, subjectivity and practice are densely interwoven, and that dis­ course is primary to subjectivity/practice through its constituting or framing powers. This means that dominant and widespread discourse shapes both how to talk about a subject matter and the meanings that we develop about it. The broad, long-range interest in discourse means that the empirical material is treated in a standardized way. It is not the details of the account and its context as much as a perceived general tendency that is deemed sig­ nificant to use, together with other material broadly pointing in a similar direction.

Alvesson & Karreman Varieties of discourse I I 3 9

In the particular piece of (what may be read as) discourse communi­ cated by the editor-in-chief, researchers interested in a long-range/determi­ nation notion of discourse may view an interesting illustration of discursive formation of leadership issues and various processes of normalization. One could then interpret the material as implying that it is 'normal' to avoid or minimize discussion on ongoing work, to not do much 'leadership', to present it as a question of dialogue and communication, and so on. The account would then imply the salience of a sort of laissez-faire, discussion oriented discourse on leadership in, for example, the editorial parts of news­ paper organizations (in Sweden). This discourse would then constitute editors' subjectivity or at least frame their way of thinking and orienting themselves in interacting with and influencing journalists. Of course, the account cannot stand on its own; it needs to be supplemented with other accounts before it is possible to say something about discourse on this macro­ systemic level.

This example primarily illustrates methodological issues in dealing with discursive material. We have proceeded from an interview extract and, proceeding from the four positions indicated by the crossing of the two major dimensions in discourse studies, discussed what can be done with it, what additional empirical material may be needed and how different lines of interpretation may be chosen. Our second example will address a full study, in which a 'modest version' of a middle-to-long-range/determination view on discourse (Discourse), will be targeted for re-interpretations from a more close-range/autonomous view on discourse.

Climbing the ladder of discourse - following Fournier from 'discourse' to 'Discourse' and back again

This section investigates an example of discursive study addressing Discourse in detail, with particular focus on how the researcher arrives at this level. Since most organizational phenomena generally are believed to exist beyond the text, this route is most likely to be travelled by organizational researchers adopting a 'discourse' perspective. Rather few studies have focused on Dis­ course in a detailed and context-sensitive manner. We have chosen Valerie Fournier's (1998) study on the new career model celebrated in managerialist discourse. The study is chosen because we think it is a good study that illus­ trates the fruitfulness of the idea of Discourse(s), as well as revealing some of its pitfalls. For our purposes it has the advantage that it presents more empirical material than is common, thus making re-interpretations less difficult.

I I 4 o Human Relations 53(9)

The new career model, as elaborated by Fournier, portrays career as a self-managed project through which individuals are empowered to cultivate their selves, to realize their dreams and at the same time to contribute to organizational excellence. This is partly accomplished through initiative and lateral moves and not just vertical promotions. The study, of graduates employed in their first job in a UK company, indicates that most of the graduates have 'bought' the career discourse, which according to the author, operates as a power/knowledge regime constituting subjectivity. Thus a high­ powered or muscular view on the capacity of discourse is expressed. As with a lot of discourse talk, it is sometimes rather unclear what 'discourse' refers to. Mainly it refers to the career discourse promoted by the company - i.e. a set of statements about the subject matter - but also other expressions emerge in the text:

Thus people in Teamco are discursively imagined .... The enterprise discourse maps and orders subject positions by defining what counts ... 'being close to a customer' is not given but is an achievement, it has to be performed discursively ...

(Fournier, 1998: 66-7)

Here one can note the powers ascribed to discourse and a certain all­ embracing quality of discourse. If performing discursively (in the sense of using the right vocabulary) is what makes the achievement, it should not be too difficult to accomplish. If it also calls for a specific social interaction and establishment of a relation, it may be a more difficult enterprise, at least for people in administrative positions. Presumably, something 'extra-linguistic' is also included in this achievement; 'to be performed discursively' then goes beyond the use of the correct vocabulary.

In the study, graduates in finance, marketing and R&D used the 'career discourse' promoted by the company, while those in CIS (computing-related graduates) drew upon what Fournier describes as a more 'militant register'. The CIS graduates were doing less well than the others, they generally had much less inviting material work environments and while all but three of the other graduates had been promoted to level 2 in the management and professional grade structure, all the CIS people remained on level 1. Earlier in the article, interview statements are described as 'stories', but later these are referred to as the 'careering and militant discourses in Teamco'. The first is basically the same as the corporate version, the second is intended to distance oneself from it, for example through disclaiming any ownership or control over their careers, negative opinions of review and development techniques and 'fatalistic' rather than 'enterprising' accounts of the self. Rather than expressing any effective

Alvesson & Karrernan Yarietles of discourse I I 4 I

resistance against the dominant discourse, the militants become constructed/ construct themselves as the Other, the marginalized, backward militant, unable to live up to the norm of engaging in a career. With all the interview statements it appears to be possible to pigeon-hole them into one of these discourses. Some of the interview statements of 'the militants' imply a self-image of an honest and responsible individual not compromising with the corrupting world of business. This discourse does not solely submit to the enterprise by reproduc­ ing its dominant discourse or expressing the marginalization of those oppos­ ing it, but indicates a space of its own, new positions to occupy. One CIS graduate expresses him/herself as follows:

I want to stay in a technical role. I don't want to move into general or senior management; you have to be hard and tough as a manager, to manipulate people, I am not like that. I thought I'd go straight to the top but I realised what it takes to get there and I don't want to change like that. They are not looking for people with technical skills but for slave masters. I'm sensitive, I can't treat people like that.

(1998: 75)

This is interpreted as expressing a logic of integrity and authenticity, which is not treated positively by the Foucauldian/anti-modernist author, referring to 'the oppressive nature of the search for authenticity' (1998: 76). Fournier continues by repeating the critique of the career discourse claims:

And the supposedly emancipatory prospects of the enterprising self are hardly more promising: the freedom to continually re-invent ourselves has the bitter taste of compulsion. So any attempt to make a case for enterprise or for integrity/authenticity is doomed to failure. What seems more important is to give voice to a plurality of discourses for articu­ lating one's experiences.

(1998: 76-7)

It is a denial of plurality that is the major problem with enterprise discourses such as the one of careering:

... enterprise flattens difference, threatens plurality, and deletes possi­ bilities for writing ourselves into scripts other than the ones in its cast. It denies us alternative worlds of experience ....

(1998: 77)

I 142 Human Relations 53(9)

Arguably, and slightly ironically, this is also what Grand Discourse studies are doing. Typically, as in Fournier's own case, they reduce plurality to one or two Discourses. One could, of course, argue that 'facts' point in this direction, but this is not an effective defence these days. Empirical material may be interpreted in many ways. While recognizing that the empiri­ cal material lends itself relatively nicely to the two Discourses in which Fournier orders her material, there certainly is the option to take variation of the empirical content even more seriously. Let us follow such a line and take a more careful look at the quotation of the CIS interviewee on page 1141. We will then primarily concentrate on close-range analysis, but also move up the discursive ladder. In addition to addressing the formative dimen­ sion, we wish to also comment upon the autonomy-determination aspects.

In the account the interviewee subject presents him/herself as a technical person, as too sensitive and human to be able to work as a (senior) manager. The subject is not hard, tough, a slave master or a manipulator and does not want to change in this direction. Being in a technical role is constructed as being morally superior to being a manager. Becoming a manager is then for the 'rude bastard' or somebody willing to become one. The interview account is a good example of what may be referred to as moral story-telling (Silverman, 1985). In interviews subjects present themselves in a positive light.

The specific (micro) context of the account is not revealed. We only know that it is an interview situation. The interviewee of course notices that the interviewer is a woman, and most likely that she comes from a business school and that she is interested in the career theme. Whether the researcher's slight scepticism towards managerialist ideologies - visible in the article - come through in the social interaction within which the interview account is produced is difficult to tell. (The majority of the accounts - being positive to the career discourse - do not reflect any such scepticism.) Nevertheless, it is likely that the account may be understood in the light of some of these cir­ cumstances. The interviewee's self-presentation as a feminine person (this construction is not necessary contingent upon sex) may be facilitated by the female interviewer seemingly being interested in people. The strong anti­ managerialism in the story may, to some extent, be seen as a dynamic effect of the situation of a temporary engagement in a story line. Questions about career may also trigger a kind of distancing from what is implied in the notion of career (promotion) by a person that may feel that s/he is, or anticipates being perceived as, doing less well.

In the account, the firm identity of the interviewee is the source of motives, wants and options. But this identity is an outcome partly of the com­ parison with the brutal nature of the management world. It is against this somewhat mythical construction that being and remaining in a technical role

Alvesson & Karreman Varieties of discourse I I 4 l

makes sense for a sensitive person. Apart from the moral problems of man­ agers, there is nothing in the account or self-construction that motivates staying in a technical role - apart from hints about a (general) unwillingness to change and the company's lack of interest for people with technical skills. A sensitive person wanting to treat people in a human, soft, non­ manipulative way would seem to fit better in a feminine occupation in a more humanistic organization than being a computer specialist in a company run by 'slave masters'. The idea of this discourse 'opening up new space [for CIS graduates] to occupy' (1998: 75) should then imply that the new space in this case is outside the organization. We here want to make two points. The first is that the self-identity produced in relationship to the negative Other of management, leads to a defensive 'staying in a technical role'. The second is that voluntary staying in this role implies being subordinated to slave masters (i.e. a compliance with a slave-like position).

This is not directly at odds with Fournier's position; a person accepting being subordinated to the manager-brutes may preserve integrity and authen­ ticity. The point is that the account expresses considerable complexity and variation, suppressed by classifying it as a matter of integrity and authenticity.

The interpretation here is conducted at the micro-discursive level: we read the account as a text (a story, not a truthful testimony of a personal con­ viction) and look at the claims and logic that it expresses. No assumptions are made regarding the constituting of subjectivity or expressions of mean­ ings (intentions, beliefs, standpoints) outside the situation of language use. From a discourse point of view, the interviewee may well talk and act in ways inconsistent with the account.

One may, however, emphasize the subjectivity and expression of meaning in connection to the account. The account triggers the constitution of a way of relating to the world, if only a temporary one, tied to the dis­ cursive act. This invoking of subjectivity/meaning then means that we move from the autonomy, emphasized above in this section, to extreme determi­ nation, thus ascribing to discourse plenty of muscles, while still remaining within the dose-range approach to discourse. A mesa-discourse analysis would be somewhat more inclined to look for slightly broader and more general themes while still being careful to avoid gross categorizations. The construction of management as morally problematic and of those being or staying outside as more human then is viewed as an inclination to sometimes talk about the moral quality of management and managers in a pejorative way. Such talk may be more or less significant in the constitution of relations between (some) subordinates and senior managers or of people's orientation towards a managerial career. The discourse on the moral problems of management may be seen as comparatively autonomous or it may be seen as

I I 4 4 Human Relations S3(9)

active m framing people's ways of being and becoming. If temporary subjectivity-constituting and meaning-expressing qualities of discourses were to be targeted, the position embraced would be located at the determination pole. This (kind of) talk reflects or brings about sentiments of negativity to management and managers, not so much as a fixed or permanent stance, but as a temporary orientation that occasionally, and in connection to this kind of talk, becomes espoused. Being occasionally in operation, it sometimes has short-term action consequences.

We classify Fournier's own interpretation as a Grand Discourse

approach - although she clearly uses a different vocabulary. We thus leave

this level and briefly relate it to mega-Discourse. In a society more and more heavily characterized by managerialism, there are also increasing signs of anti-managerialism, discontent and subtle protests against its domination and moral problems. In this sense, the account could be treated as material synthesized with many other indications on an anti-managerial discourse.

In terms of the autonomy-determination dimension, Fournier assumes, with Foucault, considerable constitutive powers of discourse. Fournier inter­ prets this account as an example of integrity versus corruption. It thus becomes part of a militant, organizational-level Discourse, capturing the pos­ itioning and available space for those expressing this discourse and, hence, stand. In relation to the earlier matrix (see Figure 2), it thus proceeds from a (moderate, rather than extreme) long-range/determination stance. This is

not unreasonable, but it illustrates gross categorization - the suppression of variation and nuance through matching 'data' to a pre-ordered framework and the wish to structure empirical material and sort it into a few fixed cat­ egories (Potter & Wetherell, 1987) - rather than taking the close-range level of discourse seriously.

Having said that, we do not want to claim that our analysis/reinter­ pretations are superior to Fournier's own interpretation. Her paper is, as we pointed out above, of high quality and its use of the empirical material is fully acceptable. Our point is mainly that concentrating somewhat more specific­ ally on the details and variations of the discourse may be an alternative to a broader approach. Moves to the latter may benefit from carefully thinking through discursive material in relation to its production before turning dis­ courses into Discourse.

Conclusion: 'discourse' and 'Discourse'

In this article we have tried to contribute to clearer and more reflective con­ ceptualizations of discourse in social and organizational research. We have a feeling that discourse is too frequently used in a vague and incoherent way

Alvesson & Karreman Varieties of discourse I I 4 5

and functions as a smokescreen for an unclear and ambivalent view on language. Two key dimensions have been identified and explored in order to unpack the confusion around discourse: a) the relationship between discourse and meaning (broadly defined) and b) the attentiveness to detail and specific context versus an interest in more standardized forms of language use. In this article, we deal with the problematic relationship between language use in an institutional context and 'other' - extra-linguistic - phenomena from two alternative points of departures: ( 1) assuming that discourse in this sense stands in a relatively 'free', autonomous or loosely coupled relationship to these other phenomena; and (2) assuming that discourse works as a struc­ turing, constituting force, directly implying or tightly framing subjectivity, practice and meaning. We have also identified some ideal-typical positions in terms of the formative range of what 'discourse' refers to: from micro­ discourse to mega-Discourse.

We have, from a methodological point of view, expressed some sym­ pathy for reducing the range in the study of discourses, thus being more atten­ tive to the local social context of language use in organizations. We have also, although to a minor extent, addressed a tendency to ascribe too much power to discourse, over for example fragile subjects and a discourse-driven social reality. We thus want to highlight problems with the tendency to work with a too grandiose and too muscular view on discourse. This leads to the ques­ tion about rigour versus significance. It may be argued that a great interest in the details of the social texts is rather myopic and neglects broader and more vital patterns. As Hardy et al. (1998) argue, the study of talk also needs to consider the social context and the participants: to just hear the story - focus on the talk - is insufficient. There is the trap of linguistic reductionism and/or a rather narrow focus on details of language use that may lead dis­ course studies to a somewhat peripheral position seen as esoteric by organiz­ ational participants. One may, however, also argue that a preoccupation with aggregated patterns means that one glosses over the operations of discourse. The degree of language-sensitivity - including language use in its specific con­ texts - may be low. One may also risk imposing a discursive macro order: diversity is neglected at the expense of broader entities (such as corporate culture, marketization, etc.). Inclinations to 'jump over' language use in a social context and make broader statements about discourse at an aggregate level (Grand or mega-Discourse) may reproduce the somewhat careless atti­ tude to language that the very idea of discourse studies, according to advo­ cates of a linguistic turn in social studies, should counter. This would imply some care in using the label discourse. In many cases, employing this label does not add anything new and simply brings confusion to the study of topics that can be addressed through the use of other, although perhaps less fashion­ able, concepts like, for example, ideology.

I I 4 6 Human Relations 53(9)

A possible response to the complexity of interview statements and other accounts is to provide interpretations of the various ways the accounts may be used for research purposes (Alvesson & Deetz, 2000). Accounts provide uncertain, but often interesting clues for the understanding of social reality and ideas, beliefs, values and other aspects of 'subjectivities'. They may also inform the researcher about language use - an interesting topic in itself. Nevertheless, investigating talk (and other social texts) is difficult as talk varies according to setting and the variety of discourses available as well as different interviewees' verbal skills and creativity in producing accounts.

In order to conduct research that goes 'beyond' language the researcher must systematically consider all empirical material before deciding what it can be used for. The researcher must critically evaluate the empirical material in terms of situated meaning versus meaning that is stable enough to allow transportation beyond the local context (e.g. an interview conversation) and thus comparison. Conventionally, three interpretations are possible: (1) state­ ments say something about social reality (e.g. leadership behaviour, events); (2) statements say something about individual or socially shared 'subjective reality' (experience, beliefs, stereotypes, cognition, values, feelings or ideas); and (3) statements say something about norms of expression, ways of pro­ ducing effects (e.g. impressions, identity work, legitimacy) or something else where accounts must be interpreted in terms of what they accomplish rather than what they mirror - as action rather than in terms of true/false (Alves­ son & Skoldberg, 2000).

A fourth level, overlapping the other three (in particular the second and, even more so, the third) is the one of Discourse. One option is, as we have shown above, to use empirical material about social texts to move up on 'a discursive ladder' and build a case for the Discourse(s). This is, however, not a self-evident or easy move and the problem of going from the specific empirical materials to address Discourses is not frequently dealt with in the literature. Quite often Discourses are addressed directly or inspired by read­ ings of textbooks and popular management texts.7 Here texts are typically seen as exhibiting one Discourse. In organization studies there are one or possibly two Discourses in which the language use forming the empirical material may be 'plugged into'. A vital question then is: How does one in empirical work proceed from encounters with texts (documents, interview talk, observed talk) to make summaries and interpretations of wider sets of discourses including aggregations of a variety of elements, an integrated framework of vocabularies, ideas, cognition and, interrelated with these, practices of various kinds? In short: To what extent - and if so, when and how - can we move from discourses to Discourse(s)?

There is of course no final answer to this question. It calls for

Alvesson & Karreman Varieties of discourse I I 4 7

continuous reflection. It is quite clear that the move from discourse to Dis­

course includes a shift in perspective. Moving from discourse to Discourse is not only about aggregation. The analytical framework is already set in the decision to view utterances as potential Discourse material. The choice of being inattentive to local context and variation is made a priori (or at least in an early phase of the process). Since most organizational phenomena are

believed to include something extra-discursive - beyond the text, so to speak - there is a strong temptation to attempt quickly to climb the discursive ladder in organizational research and to ascribe to discourse determining capacities. This makes it possible to address important social issues. However, there are good reasons to sometimes resist the temptation and engage in further contemplation at the level of the text and perhaps make more of it (Alvesson & Karreman, 2000). Grandiosation and musculariza­ tion of discourse should be grounded and shown - rather than, as in some Foucauldian and poststructuralist writings, be postulated (see Newton,

1998). As we have shown, there are other ways of looking at discourse material - analytical options that might do the material at hand better

justice.

Notes

1 We will return to the issue of context, arguing that it is important to dis­ tinguish between different levels of contexts in relationship to discourse.

2 One may talk about discursive meaning, meaning being tightly but tem­ porarily connected to specific instances of language use, and cultural or subject meaning, which may call for going beyond the details of language use and attentiveness to local context in order to interpret meaning.

3 To repeat, meaning is here broadly defined as a (collectivity of) subjects' way of relating to - making sense of, interpreting, valuing, thinking and feeling about - a specific issue. Meaning thus touches upon cognition and/or emotions.

4 Discourse determination may also be described as a thick concept of dis­ course while autonomy means a thinner notion, referring to a narrower set of phenomena.

5 If one wants to go outside discourse analysis and relate the account to signs on practice or whatever, this calls for specific investigations, i.e. a lengthy period of participant observation. Discourse analysis in the version here discussed cannot say anything about conditions outside language use.

6 Of course, if one has the conviction sometimes espoused in poststructural­ ist writings - that discourse runs the world and is sceptical to empirical inquiries, then efforts to get some empirical clues on the connections between the discourse in the account and levels of subjectivity and practice would be somewhat misguided or futile. It is assumed (or taken for granted) that discourse constitutes subjectivity.

I 148 Human Relations 53(9)

7 There are for example within the rather extensive Foucault-inspired work in management studies not many scholars addressing discourses directly interplaying with corporate practices through first-hand empirical studies.

References

Alvesson, M. & Deetz, S. Doing critical management research. London: Sage, 2000.

Alvesson, M. & Karreman, D. Taking the linguistic turn in organizational research - Challenges, responses, consequences. Journal of Applied Behav­ ioral Science, 2000, forthcoming.

Alvesson, M. & Skoldberg, K. Towards a reflexive methodology. London: Sage, 2000.

Baker, C. Membership categorizations and interview accounts. In D. Silverman (Ed.), Qualitative research. London: Sage, 1997, pp. 130-43.

Clegg, S.R. Frameworks of power. London: Sage, 1989. Davidson, A.I. Archaeology, geneaology, ethics. In D. Couzens Hoy (Ed.),

Foucault: A critical reader. Oxford: Blackwell, 1986, pp. 221-33. Deetz, S. Democracy in an age of corporate colonization: Developments in com­

munication and the politics of everyday life. Albany: State University of New York Press, 1992.

Fairclough, N. Critical discourse analysis and the marketization of public dis­ course. Discourse and Society, 1993, 4, 133-59.

Foucault, M. The history of sexuality: Volume 1. New York: Pantheon, 1976. Foucault, M. Power/knowledge. Selected interviews and other writings

1972-1977. New York: Pantheon, 1980. Fournier, V. Stories of development and exploitation: Militant voices in enterprise

culture. Organization, 1998, 5, 55-80. Grant, D., Keenoy, T. & Oswick, C. Organizational discourse: Of diversity,

dichotomy and multi-disciplinarity. In D. Grant, T. Keenoy & C. Oswick (Eds), Discourse and organization. London: Sage, 1998, pp. 1-13.

Hardy, C., Lawrence, T. & Phillips, N. Talking action: Conversations, narrative and action in interorganizational collaboration. In D. Grant, T. Keenoy and C. Oswick (Eds), Discourse and organization. London: Sage, 1998, pp. 65-83.

Jackall, R. Moral mazes: The world of corporate managers. New York: Oxford University Press, 1988.

Keenoy, T., Oswick, C. & Grant, D. Organizational discourses: Text and context. Organization, 1997, 4, 147-57.

Kerfoot, D. & Knights, D. Managing masculinity in contemporary organizational life: A 'man'agerial project. Organization, 1998, 5, 7-26.

Miller, G. Building bridges: The possibilities of analytic dialogue between ethnog­ raphy, conversation analysis and Foucault. In D. Silverman (Ed.), Quali­ tative research. London: Sage, 1997, pp. 24-44.

Newton, T.J. Theorizing subjectivity in organizations: The failure of Foucauldian studies. Organization Studies, 1998, 19, 415-47.

Alvesson & Karreman Varieties of discourse I I 4 9

Potter, J. Discourse analysis as a way of analysing naturally occuring talk. In D. Silverman (Ed.), Qualitative research. London: Sage, 1997, pp. 144-60.

Potter, J. & Wetherell, M. Discourse and social psychology: Beyond attitudes and behaviour. London: Sage, 1987.

Rosenau, P.M. Post-modernism and the social sciences: Insights, inroads, and intrusions. Princeton, NJ: Princeton University Press, 1992.

Silverman, D. Qualitative methodology & sociology. Aldershot: Gower, 1985. Silverman, D. Interpreting qualitative data. Methods for analysing talk, text and

interaction. London: Sage, 1993. Steier, F. Research as self-reflexivity, self-reflexivity as social process. In F. Steier

(Ed.), Research and reflexivity. London: Sage, 1991, pp. 1-11. Van Dijk, T. Principles of critical discourse analysis. Discourse & Society, 1993,

4, 249-83. Watson, T. In search of management. London: Routledge, 1994. Weedon, C. Feminist practice and poststructuralist theory. Oxford: Basil Black­

well, 1987.

ttats Alvessonworks: at the University of LUAd, Sweden. He was Arthur ~derten V1$iting Scholar at,the Jucfie I~ of Management Studies,

u~, 'of• Cambr'tdge. •~ 2000. His research interests indude critical ~cy. o~ ci.tlture, the management of know!~ ifltenslve work and qualitadve metmxlology. 1-M has recentJy published

Reflexive ~ •CS., lOfJO, with K. Sk<>lcljJerg) and Doing critical ~ research (Sage, 2000, .with S. Deetz). [E-mail! [email protected]]

Dan Karreman is a researcher and teacher at Lund University, Sweden.

He is currently engaged in research on discourse and organization, know­

kldge work and i<nowledge-intensive fums.

[E-mail: [email protected] J