Reflecting on Reflexivity: Reflexive Textual Practices in Organization and Management Theory

This paper identifies four sets of textual practices that researchers in the field of organization and management theory (OMT) have used in their attempts to be reflexive. We characterize them as multi&#8208;perspective, multi&#8208;voicing, positioning and destabilizing. We show how each set of practices can help to produce reflexive research, but also how each embodies limitations and paradoxes. Finally, we consider the interplay among these sets of practices to develop ideas for new avenues for reflexive practice by OMT researchers.


INTRODUCTION
Reflexive research has been attracting increasing attention in organization and management theory (OMT) in recent years, leading some to argue that theory construction has turned inward to become largely an 'exercise in disciplined reflexivity' (Weick, 1999, p. 803). While reflexivity has been a concern for some positivists and neo-empiricists (see Johnson and Duberley (2003) for discussion of the 'methodological' reflexivity employed by such researchers), we focus on critical, interpretive work that conceptualizes social reality as being constructed, rather than discovered, during research. Such work has defined reflexivity as research that turns back upon and takes account of itself (Clegg and Hardy, 1996a;Holland, 1999), to explore the situated nature of knowledge; the institutional, social and political processes whereby research is conducted and knowledge is produced; the dubious position of the researcher; and the constructive effects of language (Calás and Smircich, 1999).
Researchers in organization and management theory (OMT) have engaged in a range of practices in both conducting and writing up research, in their efforts to be reflexive. In the case of the former, reflexive practices are those embodied practices in which the researcher engages in relation to research subjects, practitioners and students (e.g. Boje and Rosile, 1994;Cunliffe, 2002aCunliffe, , 2002bCunliffe, , 2004. Researchers have therefore studied the ways in which they can act reflexively to foreground other participants' voices, especially when they conduct their empirical work. The emphasis is on embodiment and lived experience in addition to language use, in relation to others who are not necessarily other researchers (e.g. Barge, 2004;Cunliffe, 2002b). In the case of the latter (the ways in which researchers write up their research), language use is situated more narrowly (in relation to the academic community) and the 'other' is predominantly other researchers. This approach to reflexivity is 'conceptualized as an epistemological practice that emphasizes intellectual critique' (Barge, 2004, p. 70) as predominantly textual practices are used to invoke and present various forms of reflexive analysis. As a result -and other forms of intellectual engagement notwithstanding -the emphasis is on textual practices used by researchers to present their work reflexively (e.g. Clegg and Hardy, 1996a).
It has been argued that it is the textual practices underpinning intellectual critique which have attracted the most attention (Barge, 2004;Cunliffe, 2003). As a result, a reappraisal of these practices is warranted -to assess how reflexivity can contribute to research, and also to consider the ways in which it can be problematic (Lynch, 2000;Weick, 1999). In this way, we hope to contribute to the current debate about reflexivity in OMT in a number of ways. First, we show a range of ways in which reflexivity has been practiced in the way the OMT literature has been written. Second, we identify how these practices can slide into being unreflexive. Third, by drawing on four sets of textual practices that are commonly used in the literature, we show how, by combining and differentiating them, we can encourage new reflexive directions.
The paper is structured as follows. First, we provide a brief overview of four particular sets of textual reflexive practices used in OMT that we have identified from the literature and show their contributions. Second, we then explore some of the ways in which each may fail to meet its own reflexive aspirations. Third, we use our framework to show how a different reflexive approach might be achieved by combining and differentiating these practices. Finally, we conclude with a discussion of some of the implications of our approach to, and recognition of, multiple reflexive practices.

FORMS OF REFLEXIVE PRACTICE
Reflexive practices aim, to varying degrees, to cast doubt on the idea that 'competent observers' can 'with objectivity, clarity, and precision report on their own observations of the social world' (Denzin and Lincoln, 1994, p. 11). This literature owes a large intellectual debt to early concerns about the limits of objectivity and the provisional nature of knowledge. For example, Popper (1959) recognized that theory and data could not be neatly separated. Kuhn (1970), by highlighting the historically-situated nature of knowledge, played a pivotal role in the gradual shift towards a greater acceptance of indeterminism in the social sciences, setting the scene for constructivist and postmodernist approaches (Delanty and Strydom, 2003a). Post-empiricist scholars, notably Lakatos (1978) and Feyerabend (1978), challenged understandings of knowledge as being produced through purely rational and formal processes of theory testing, highlighting the role of political and other factors (Delanty and Strydom, 2003b). Writers such as these thus laid much of the groundwork for contemporary work in OMT that argues that interpretation-free, theory-neutral facts do not exist but are, rather, constructions (Alvesson and Sköldberg, 2000;Meriläinen et al., 2004), and which acknowledges that linguistic, social, political and theoretical elements are woven together in ways that shape the knowledge-development process (Calás and Smircich, 1999).
In relation to these processes, we are concerned with the practices of writing about research, rather than with the embodied practices involved in conducting empirical research, although this does not prevent us from recognizing their importance to reflexivity (e.g. Alvesson, 2003;Boje, 1998;Cunliffe, 2002b). We therefore examined textual practices used by researchers in their attempts to be reflexive. In focusing on reflexivity as textual practice, we argue that what members of a research community -in this case, OMT [1] -'know' to be reflexivity is shaped by and enacted in the textual practices of researchers. Accordingly, we selected texts in OMT that have explicitly addressed issues related to reflexivity, as well as texts that are frequently referred to in contemporary writings as being reflexive, based on our general familiarity with the literature as well as recommendations from colleagues, reviewers and the editor; although we acknowledge that our selection is illustrative rather than exhaustive.
From our analysis, we identified four relatively distinct sets of practices that commonly appear in the literature. Our four categories should not be seen as rigid boundaries: we acknowledge that some work expresses an interest in more than one practice (e.g. Boje et al., 1999;Cunliffe, 2002b). Nonetheless, our reading of the literature points to a range of textual practices -to reflexivities rather than reflexivity -that have been regularly used in work in OMT, where researchers have stated their intention to be reflexive. We believe that a comparison of these different textual practices offers a useful starting point for a discussion of reflexivity, although we fully expect that not all readers will agree with our particular categorization, a theme to which we will return in our discussion of self reflexivity.

Reflexivity as Multi-Perspective Practices
One set of practices associated with reflexivity in OMT is the use of multiple perspectives (Holland, 1999;Lewis and Kelemen, 2002). Its original impetus came from the paradigm 'wars' over the incommensurability thesis, where some theorists argued that a multiparadigmatic view of a particular phenomenon or study could be used to provide a more comprehensive understanding (Gioia and Pitre, 1990). Similarly, Morgan (1983, p. 16) acknowledged 'the fallacy of trying to evaluate the different perspectives from a single perspective within the system' and argued in favour of a 'dialectic between a number of such points of view'. Morgan's subsequent work (1986) on metaphors complemented the use of multiple paradigms as a way of reflecting on knowledge production; an approach to reflexivity that has since expanded to include multiple vocabularies, theories, stories, interpretations, paradoxes and frames (Alvesson, 1996;Boje, 1995;Hassard, 1991;Poole and Van de Ven, 1989;Schultz and Hatch, 1996;Van de Ven and Poole, 2005;Willmott, 1993). While acknowledging the influence of the paradigm wars, we distinguish between the reflexive researcher and the paradigm 'warrior'. Instead of treating epistemological positions as manifestations of metaphysical principles, as paradigm warriors ( Jackson and Carter, 1993) do, this reflexive use of the paradigms involves seeking out anomalies among them: in a way that is mindful of the historically and politically situated quality of our reasoning. By becoming more practically reflexive about the conditions of theorizing, we move away from an external and seemingly authoritative form of analysis and towards an immanent, self-consciously situated form of critique. (Willmott, 1993, p. 708) Accordingly, researchers use tensions among different perspectives to expose different assumptions and open up new ways of thinking, much like Keenoy's (1999) hologram, where, by getting up and moving to another theoretical place, we can see things differently.
[H]olographic illusions of depth, contour, shade and shape and, sometimes, movement are entirely dependent on the relationship between the observer and the observed: they only come into being in the process of interaction. (Keenoy, 1999, p. 10) Drawing on Rorty's (1989) warning about the danger of believing in the superiority of a final vocabulary, the reflexive researcher uses a set of practices involving the juxtaposition of perspectives to draw attention to the limitations in using a single frame of reference and, in so doing, provide new insights. It is the accumulation of these perspectives that amounts to reflexivity: the use of different perspectives is enlightening in that it helps to complement otherwise 'incomplete' research. For example, Knights and McCabe (2002, p. 235; emphasis added) use rational managerialist, critical control, and processual interpretations 'to build on earlier approaches [to TQM] in the anticipation we might move beyond our present understanding'. These practices help the researcher to 'break the habits of routine thought' and 'see the world as though for the first time' (Cooper and Burrell, 1988, p. 101). They construct reflexivity as instructive and enlightening by helping researchers to answer the question: what are the different ways in which a phenomenon can be understood and how do they produce different knowledge(s)?

Reflexivity as Multi-Voicing Practices
A second set of practices focuses on the authorial identity of the field worker and their relation to the 'Other', i.e. the research subject, drawing from the work of sociology and anthropology, where the role of the researcher has increasingly been called into question (Van Maanen, 1988); and as more attention has focused on the role of the research subject in the construction of research texts (Clifford and Marcus, 1986). Linking research texts to Bakhtin's (1981) notion of a 'heteroglossia' -a 'dynamic multiplicity of voices, genres and social languages' (Maybin, 2001, p. 67) -it has been suggested that researcher and research subjects collectively negotiate the meaning of language, undermining the privileged position of researchers over research subjects and weakening the claims of the former to report reliably on the experiences of the latter (Denzin and Lincoln, 1994).
A number of specific textual practices have been used to question how OMT researchers can speak authentically of the experience of the Other, and in interrogating the relationship between the two. First, the researcher is recognized as part of the research project, a subject just like any other that is constructed in and through the research project: we do not simply 'bring the self to the field' so much as 'create the self in the field' Reflecting on Reflexivity 483 (Reinharz, 1997, p. 3). Second, it is incumbent on the researcher to declare the authorial personality -to present the details of their particular experiences and interests (Boje and Rosile, 1994) and to divulge the steps they have taken in order to present their work as meaningful (Golden-Biddle and Locke, 1993;Jeffcutt, 1994).
[I]nvestigators seek ways of demonstrating to their audiences their historical and geographic situatedness, their personal investments in the research, various biases they bring to the work, their surprises and 'undoings' in the process of the research endeavor, the ways in which their choices of literature tropes lend rhetorical force to the research report and/or the ways in which they have avoided or suppressed certain points of view. (Gergen andGergen, 2000, p. 1027) Third, various literary techniques have been employed to open up space for the Other in research accounts through the self-conscious use of writing techniques, e.g. reflexive ethnographies, literary autoethnographies, narratives of the self (Hayano, 1979) and applied in OMT through the use of fiction, drama, and narrative (Czarniawska, 1997;Hatch, 1996). By being more creative and experimental in writing, researchers can bridge 'the gulf between self and other by revealing both parties as vulnerable, experiencing subjects working to coproduce knowledge' (Tedlock, 2000, p. 467).

Reflexivity as Positioning Practices
A third set of practices goes beyond a focus on authorial identity and emphasizes the fact that 'knowledge is not something that people possess in their heads, but rather, it is something that people do together' (Gergen, 1991, p. 270). What distinguishes this approach from the previous construction of reflexivity is that it is not simply concerned with the relationship between researcher and research subject, but also with the way that the author's research takes place within a broader network or field. These broader social processes shape knowledge, meaning that the researcher can construct 'knowledge' only in the context of a particular research community and society (Callon, 1986).
Work in the sociology of scientific knowledge and science and technology studies has been particularly important in developing this concept of reflexivity. Early researchers (Barnes, 1974) argued that scientific knowledge could be understood in the same way as any other area of culture. Subsequent researchers incorporated social constructionism (Woolgar, 1988) and drew attention to the degree to which politics is found in scientific workplaces (Latour, 1987;Latour and Woolgar, 1979). As a result, the burgeoning literature featured empirical studies of how scientific claims were secured, helping to identify the political and rhetorical processes by which knowledge claims were accepted as true or false in the particular institutional setting (Shapin, 1995). From a somewhat different perspective, Bourdieu also emphasized research as a collective enterprise (Bourdieu and Wacquant, 1992, p. 40), arguing that 'the subject of reflexivity must ultimately be the social scientific field in toto' that takes into consideration 'the social organization of social science, as an institution inscribed in both objective and mental mechanisms' (p. 41).
Researchers interested in how the larger network of practices and interests produces particular interpretations of knowledge have drawn on these ideas to explore research on organizations (Hardy et al., 2001;Oakes et al., 1998). Reflexive practices thus explore the broader social landscape within which research and researchers are positioned: 'the networks of beliefs, practices, and interests that favor one interpretation over another; and, ideally, the way that one interpretation rather than another comes to predominate' (Collins, 1998, p. 297). They are used to examine the fates of competing claims made by actors, how technical and discursive resources are used to legitimate claims in keeping with broader institutionalized norms, and how context, power, and historical circumstances combine to produce knowledge.

Reflexivity as Destabilizing Practices
A fourth set of practices has been influenced by the writings of Derrida and Foucault. This literature is somewhat different from the work described above in that researchers do not reflect on their own theorizing so much as they target the unreflexive research of others. We include an examination of these practices because they make reflexive claims through the way in which they interrogate claims to knowledge; and they have been considered under the rubric of reflexivity by others (e.g. Calás and Smircich, 1999;Clegg and Hardy, 1996a). According to this view, the production of knowledge, particularly positivist versions that try to establish 'the truth', lead to a certain version of the social world, with associated power effects. All knowledge projects are thus 'dangerous', insofar as any version of truth carries with it a particular freezing of the social world and a configuration of political privileges and should, therefore, be closely interrogated and cross-examined. The means to do so lie with postmodern theoretical and epistemological assumptions that undermine the idea that research is ultimately a progressive (or even meandering and circuitous) path towards universal 'truths'. Knights (1992), for example, used a Foucauldian methodology to critique the work of Michael Porter on strategy and to show that strategic discourse and practice represent a set of power-knowledge relations that constitutes the subjectivity of managers and employees. Archaeological investigations help to uncover 'the philosophical, political, social, and economic rules of formation that underlie the development of specific management theories'; while genealogical analyses can be used 'to examine the conditions of possibility for such knowledge to be drawn upon in the exercise of power' (Knights, 1992, p. 530). In this way, the pretensions of Porter's claim to universal truth are exposed. Foucauldian insights have also been employed to interrogate Total Quality Management (Knights and McCabe, 2002), Human Resource Management (Townley, 1993) and accounting (Hopwood, 1987).
Other practices draw on the concept of deconstruction associated with Derrida. They have been used to challenge claims in accounting that 'external metaphysical grounding' exists outside the text and to suggest that, rather, such grounding is provided by 'the linguistic and rhetorical strategies' of the author (Arrington and Francis, 1989, p. 3). Linstead (1993) has proposed deconstructive ethnographies of organizations to call into question the authority of any particular account. Calás and Smircich (1991) have deconstructed classic organizational texts using Foucault, Derrida and feminist poststructural theory to show how the success of leadership can be explained by its seductive nature, which is hidden behind knowledge claims.
In questioning the conditions and consequences of the construction of a theory, these practices are intended to destabilize the epistemological assumptions of other forms of theorizing. Research is interrogated through 'critical attention to presuppositions about both its status as science and its claim to produce empirical evidence about "the way the world is" independently of its own values about the "the way the world should be" ' (Arrington and Francis, 1989, p. 2). Destabilizing practices hold theories intellectually accountable by problematizing the conditions and consequences of their formation: the rationality, truth, and progress claimed by particular theories are exposed as myths.

LIMITATIONS OF REFLEXIVE PRACTICES
In this section, having provided an overview of the four sets of practices (Table I), we point out some of the paradoxes and limitations associated with them.

Multi-Perspective Practices
This reflexive researcher is a traveller, periodically moving from place to place so that he or she may see things differently. Thus, writers tend to speak in terms of a journey through the paradigms (Hassard, 1991), bridging paradigms (Gioia and Pitre, 1990), and crossing paradigms (Schultz and Hatch, 1996). He or she is also a builder or 'bricoleur', piecing together a richer, more varied picture by viewing research -their own or others' -from different angles. These practices guard against theorizing that presents an unambiguous view of reality represented in the form of a single 'grand narrative', not by dismissing the foundational claim of any single perspective that it offers a better understanding of reality (as destabilizing practices tend to do), but by showing how other perspectives provide different understandings and, by combining them, greater insight might be achieved.
This approach nonetheless raises a number of questions. First, researchers recognize that no single paradigm, metaphor, or theory can account for the language games in which it is embedded (Calás and Smircich, 1999). How is it, then, that rotating among a selection of them can do so, when each is individually flawed? Second, how does one combine the different perspectives? Schultz and Hatch (1996) note three different ways in which researchers link the paradigms: incommensurability, where researchers focus on the differences; integration, where researchers ignore the differences; and crossing, where researchers engage with the differences in one of four different ways (sequentially, in a parallel manner, through second order concepts and by interplay). The result is a complex set of choices for the reflexive researcher. Third, how can researchers apply all paradigms equally when they inevitably have a preferred position (Parker and McHugh, 1991)? Some have argued that it would require a 'quasi-religious conversion' (Lewis and Kelemen, 2002, p. 265) to do so. Multi-perspective reflexive practices thus evoke pantheism -in worshipping multiple paradigms, metaphors or theories, researchers assume that the problems associated with any individual theory are overcome. The paradox is  Rorty, 1989;Morgan, 1979 Hayano, 1979;Clifford andMarcus, 1986 Latour andWoolgar, 1979;Bourdieu andWacquant, 1992 Foucault, 1980;Derrida, 1982 OMT examples Hassard, 1991;Alvesson, 1996;Schultz andHatch, 1996 Jeffcutt, 1994;Rosile, 1994 Oakes et al., 1998;Hardy et al., 2001 Calás andSmircich, 1991;Knights, 1992 Means used Reflecting on Reflexivity 487 that, having abandoned the idea that a particular paradigm or metaphor is 'correct' or 'superior', this form of reflexivity is limited in terms of providing grounds for selecting from among the countless other paradigms, metaphors or theories that are available for worship. The main grounds for choosing are found with reference to the academic community, not the paradigms themselves, and rely on academic norms or fashions which are inevitably politicized, socialized, and institutionalized as, for example, an interest in paradigms has been replaced by a concern with multiple frames, then metaphors, and so on.

Multi-Voicing Practices
Multi-voicing practices force the researcher to ask questions about the relationship between the author and the Other and to consider whether, and how, the researcher can speak authentically of the research subject. Reflexive researchers open up texts to 'multiple readings; to decentre authors as authority figures; and to involve participants, readers and audiences in the production of research' (Putnam, 1996, p. 386). In so doing, the intention is to undermine the authority of the research account, which is merely one representation among many, and the privileged power position of the researcher by giving the reader and/or research subject a more active role in interpreting meaning (Marcus, 1994). By engaging in these practices, the reflexive researcher acknowledges his or her identity as a participant in the research and confesses any 'sins' in terms of personal interests or rhetorical manoeuvres. The reflexive researcher is 'just' another subject, albeit one with the artistic and literary skill necessary to carry out these practices successfully (Vidich and Lyman, 2000). Paradoxically, practices to 'downplay' the researcher and give greater space to the subject often end up drawing considerable attention towards the researcher (Clegg and Hardy, 1996a). New forms of writing have often placed the researcher's personal experience centre-stage (Van Maanen, 1988). In this way, these practices have been criticized for turning the self (of the researcher) into a fieldsite (Robertson, 2002) at the expense of the empirical work in question (Fournier and Grey, 2000) because researchers 'tend to be more interested in our practices than in those of anybody else' (Weick, 2002, p. 898). It is this narcissism that shows the limitations of this approach: the impossibility of giving everyone associated with the research -researcher, research subject and reader -a voice, let alone an equal voice, even with the best intentions.

Positioning Practices
Positioning practices draw attention to the various political, cultural, and institutional constraints embedded in the academic community. They locate the reflexive researcher in this landscape: subjected to and resistant against the controls embedded in professional networks of individuals and institutions (Clegg and Hardy, 1996a). He or she recognizes the way in which research and researcher are influenced by the shared orientations of a particular research community, whose conventions 'not only conceal but actively misrepresent the complex and diverse processes involved in the production and legitimation of scientific findings' (Mulkay, 1992, p. 69), as well as the political interests from which these conventions arise. The reflexive researcher is thus a networker and politician (Collins, 1998;Deetz, 1996), able to identify conventions, fashions, and conformist pressures embedded in publication outlets, journal formats, conferences, and funding arrangements, as well as an adventurer-explorer who navigates them through the judicious use of power (Putnam, 1996), artful deployment of rhetoric (Shapin, 1995) and knowledge of the rules of the game (Mauws and Phillips, 1995).
Ironically, in showing how reflexive researchers can navigate supposedly inescapable social forces, these practices help to construct the heroic -if somewhat cynical and jaded -researcher that they are trying to repudiate. For example, Hardy et al. (2001), in trying to be reflexive about their own research, acknowledge that there is no way that an individual can produce science; only a community can produce science. However, they go on to show how they were nonetheless able successfully to represent the many silent actors of the social world that they mobilized through their research study and, in particular, become 'indispensable' authors by ensuring their form of analysis became an obligatory passage point -and hence publishable. Thus, the reflexive researcher is supposedly able to see constraints in a way that others do not and, while he or she may not be able to dismantle them, he or she can nonetheless work around them. Accordingly, the paradox associated with this construction of reflexivity relates to the highly individualized portrayal of the researcher as a superior navigator or negotiator of social forces that should not be navigable or negotiable at all. Consequently, these reflexive practices are problematic because the emphasis on the limitations of agency, which sets it apart from other practices that tend to focus on the individual researcher, is subsequently undermined by highly individualistic explanations of knowledge production.

Destabilizing Practices
Destabilizing practices are powerful in their ability to call theorizing to account for itself and to point out a lack of reflexivity, usually on the part of others. The reflexive researcher is disruptive, willing to unsettle the academic community, and to make trouble, especially for research that is readily accepted in the wider academic community. As an insurgent, the reflexive researcher challenges research by taking up a place 'outside' the target project, which is usually undertaken by other researchers, and then infiltrates it in order to undermine its very foundations -its claims to knowledge and progress -by asking fundamental questions about the conditions and consequences of its construction. In this way, the reflexive researcher makes regular incursions over a metaphorical border between different epistemologies, bringing the preferred one (usually Foucauldian or Derridean) to bear on and undermine the other (usually a mainstay of more orthodox work).
These reflexive practices nonetheless embody a paradox in that they are used to produce an authoritative text, while relying on a set of assumptions that stresses that there can be no such thing. Researchers assert that, by engaging in these reflexive practices, they can create sufficient distance to reflect on the assumptions, reasoning, and knowledge inherent in someone else's research project, and to see something that others do not. For example, Knights (1992, p. 516) points to how he has been able to identify the philosophical, political, social, and economic rules of formation that underlie the development of specific management practices and discourses, which 'would ordinarily elude the conscious awareness' of the researcher. In this way, the practices ironically evoke a degree of omnipotence in that the reflexive researcher claims to possess insights that others do not. Since these practices tend to be used to undermine the work of others, rather than applied to the reflexive researchers' own research, researchers are able, to a certain extent, to avoid making the truth claims that they would dismiss in the case of others. But, in offering warnings rather than guidelines, these practices are limited in their ability to generate new knowledge.

REFLECTING ON REFLEXIVE PRACTICES
By drawing attention to some of the limitations associated with reflexive practices (Table II), we do not intend to imply they are not worthwhile; merely that each is limited in what it can achieve and it is helpful to reflect upon them. Furthermore, by reflecting on the framework that we have constructed, we can identify additional ways of being reflexive.

Combinations
One conclusion that follows from the identification of four separate sets of practices is that, by combining them, we might generate additional questions for researchers to consider.
For researchers employing multi-perspective practices, the addition of multi-voicing practices might help them explore how their use of multiple perspectives constructs the author in different ways. For example, does the author emerge out of one particular perspective and then apply other perspectives? Does a 'home' position provide a richer, more informed reading, while other perspectives are used in a more selective, pragmatic way? Or does the author (or different authors) emerge from where the perspectives overlap or, even perhaps, from gaps where the perspectives do not connect? (Is there, in fact, such a space and, if so, how might we conceptualize it?) How do the relationships between research subject and researcher-as-subject change according to the different perspectives that are used? Are some voices always left out, regardless of the choices made regarding perspectives; or does the 'discovery' of new voices produce new perspectives that embody different relationships among self and (the search for) Others?
Positioning practices help to interrogate how the choice of perspectives is influenced by time and context. Arguably, there are structured sets of positions or templates for the use of multi-perspective practices that are institutionally supported and rewarded. For some time, multi-perspective studies have typically used Burrell and Morgan's (1979) original paradigms. What is the political agenda associated with particular choices? How do new choices emerge and become legitimated? How are particular paradigms produced out of the relationships among researcher, research community, and the available resources, i.e. accepted conventions related to theoretical developments in the field?
Finally, the application of destabilizing practices might lead to a different dynamic: instead of the sequential or parallel use of perspectives to build on each other, researchers might use them to undermine each other and to interrogate the relationships among them more fundamentally. Rather than positing perspectives as complementary and, therefore by changing interpretation, adding more positive knowledge to produce a more holistic picture, different perspectives might be used to challenge each other and to illuminate potential problems with one particular perspective, as in Alvesson's (1996) juxtaposition of Foucault and Habermas. When multi-voicing practices are combined with multi-perspective practices, each research subject could be seen from multiple perspectives -a specific subject is not just, for example, a female, but may be viewed as gendered (and non-gendered) and located in varied ways (e.g. victim, emancipated, dominant) made possible through the use of specific theoretical perspectives. Reflexive researchers could examine the relationships among these multiple subjects brought into focus by alternative readings encouraged by the use of different perspectives.
Positioning practices might raise questions about the relationships between the use of multi-voicing practices and the particular research community, in terms of how the latter constructs voices that are deemed legitimate subjects for 'discovery'. For example, why are these practices so much less common in OMT than in anthropology? Why are some subjects less likely to be given a voice? Why is giving managers a voice -in a sensitive way -a relatively rare practice in critical management studies? How does this relate to the way in which critical management studies has been constructed from particular voices, such as Marx, Gramsci and Habermas? What imprint has been left by these ancestral voices and how does it shape the political agenda of reflexive research?
The use of destabilizing practices might help reflexive researchers deconstruct the multiplicity of voices that appear in researchers' accounts. If the aim of multi-voicing practices is to give voice to the marginalized, is it not possible to use destabilizing practices to subvert that 'worthy' cause, regardless of how 'politically incorrect' it may appear? And what is the relation of liberated voices to understandings of political correctness? How does the postmodern and critical project shape the choices of the reflexive researcher in silencing or amplifying particular voices? In other words, in what other ways can the relationship between the author, the research subject, and the reflexive researcher be destabilized?
Positioning practices may be complemented through the employment of multiperspective practices that show how the reflexive researcher selects a particular representation of the social terrain and the social forces acting upon the research process. Different theories may be helpful in moving beyond the research community -or a particular version of it -to consider how broader societal and cultural trends influence research. For example, Lasch's (1978) conception of a narcissistic culture or Baudrillard's (1983) conceptions of a postmodern society may provide theoretical resources for alternative constructions of the social logics that influence accounts of positioning, as well as providing a different portfolio of perspectives than is usually used.
Multi-voicing practices can be used to trigger questions around who puts imprints on the final text and how the researcher constructs him/herself and the role of others in it. One could imagine a research text as an outcome of a multitude of different voices, all trying to give as much input as possible to the process and the outcome. The researcher, gatekeepers and guardians in the form of authors of key references and, more directly, the editor, reviewers and anticipated readers can be seen as 'ghostwriters' of the text, providing limited space for author originality. Informants and other representatives of the business community also play a role in shaping the work. Making sense of and reflecting upon how these voices are being emphasized, marginalized, repressed, channelled, and translated would be a useful exercise in reflexivity Positioning practices can also be interrogated by the use of destabilizing practices that draw attention to the relationship between the researcher and the research community. In the same way that the practices of particular authors have been deconstructed for how they make knowledge claims, the practices of reflexive authors might be destabilized to show how they make reflexivity claims. The subtleties and rhetorical moves in accounts of how the researcher locates him/herself in a particular academic or societal context could be explored.
Destabilizing practices can be combined with multi-perspective practices. For example, Calás and Smircich (1992) demonstrate that examining organization studies from different positions generates different understandings of the field (multi-perspective practice), and simultaneously challenges the gendered epistemological basis of the discipline (destabilizing practice). Similarly, Knights and McCabe (2002) use multi-perspective practices to interrogate TQM, before moving to a Foucauldian analysis aimed at challenging the epistemological foundations of the perspectives they have just used.
Those who rely on destabilizing practices tend to construct one particular reading of other theorists. For example, Porter has been singled out for his work on strategy (Knights, 1992) and Weick on organizing (Calás and Smircich, 1991), and their work deconstructed, typically using one text or segment of a text to produce a finely grained, but also highly constrained, analysis of an individual reading. Applying multi-voicing practices would lead those who use destabilizing practices to target more fragmented and diverse voices. What if, for example, multiple texts written by Porter or Weick were explored? Have their voices changed over time: can vintage Porter and Weick be juxtaposed with Porter-nouveau or Weick-au-courant? How have the voices of these authors changed in response to their destabilization? Have they been muted in deference or become stridently defensive? Targeting multiple voices and using practices to destabilize the relationships among them, would provide insights to complement the destabilization of a single, disconnected voice.
By applying positioning practices these researchers might ask questions about their own reflexive practices. For example, where do these deconstructions sit in the academic community? How does destabilization produce the reflexive researcher as an obligatory passage point, as opposed to other sets of practices? What are the political benefits of deconstructing the work of others? How does marginalizing the Other benefit one's own career prospects (cf. Sangren, 1992)? Another set of questions might involve considering the relationships between the destabilizing and destabilized subjects? How might the destabilized subject destabilize the destabilizer? How can Weick or Porter reply within the strictures laid down by the tenets of postmodernism; can they resist in a way that would be acknowledged as epistemologically valid by the reflexive researcher? The combination of these sets of practices provides an endless space of potentially useful interplays. The questions that we raise are not intended to be exhaustive; nor do we suggest that our four sets of practices offer a complete or final summary of reflexive practices. Nonetheless, by exploring the different ways in which reflexivity is practiced in OMT and combining them, we are in a position to learn more about how academic knowledge is produced -the emphasis is less about how an individual researcher conducts a particular study and more about the ways in which knowledge is constructed by and in a community.

Dialectical Differentiation
In this section, we show how the four sets of reflexive practices might usefully differentiate between those that emphasize avoiding problematic or 'dangerous' things -intellectually, politically or ethically -and those that try to produce new insights. We refer to the former as D-reflexivity: D stands for deconstruction, defence, declaiming, destabilizing and danger-warning. We call the latter R-reflexivity: R refers to reconstruction, reframing, reclaiming, re-presentation. Of the four sets of practices above, destabilizing and positioning practices are mainly concerned with D-reflexivity, as their aim is to undermine positive claims to results and contributions, while multi-perspective and multi-voicing practices are related to R-reflexivity in that they encourage consideration of alternative views. [2] D-reflexivity practices challenge orthodox understandings by pointing out the limitations of, and uncertainties behind, the manufactured unity and coherence of texts, as well as the way in which conformism, institutional domination and academic and business fashion may account for the production of particular knowledge. It engages with the problems, uncertainties and social contingencies of knowledge claims -whether empirical claims, concepts or theoretical propositions. By emphasizing how social science orders the world in a particular way, power/knowledge connections are illuminated and truth-creating effects are disarmed. These practices are conducted in attempts to counteract harm -to challenge efforts to stabilize the view of the world in a particular way and expose the unreflexive reproduction of dominant vocabularies, rules or conventions in social research. More radical practices -deconstructive or Foucauldian -emphasize the arbitrary and subjectivity-shaping character of knowledge, while weaker practices encourage moderate scepticism around interpretive and textual moves to convey legitimacy, certainty and closure. In both cases, the aim is to challenge the text in fundamental ways -to question the chosen elements of the logic of the research project and its outcomes, showing them to be associated with particular paradigmatic roots and perspectives; with forms and politics of representation; and the socio-political forces of research.
R-reflexivity is about developing and adding something; the R-reflexivist is in the construction rather than demolition industry. It means bringing in issues of alternative paradigms, root metaphors, perspectives, vocabularies, lines of interpretation, political values, and representations; re-balancing and reframing voices independently of data in order to interrogate these data in a more fundamental way. Instances of alternative constructions and reconstruction of fundamental elements of the research project are central to these reflexive practices. R-reflexive practices are employed to illuminate what is left out and marginalized: the (almost) missed opportunity, premature framing, reproduction of received wisdom, re-enforcement of power relations and unimaginative labelling. They provide alternative descriptions, interpretations, results, vocabularies, voices, and points of departures that could be taken into account, and show some of the differences that they would make. R-reflexivity aims to open up new avenues, paths, and lines of interpretation to produce 'better' research ethically, politically, empirically, and theoretically.
We suggest that reflexive researchers might engage in practices that create a dialectic between D-reflexivity and R-reflexivity. Moving between tearing down -pointing at the weaknesses in the text and disarming truth claims -and then developing something new or different, where the anxieties of offering positive knowledge do not hold the researcher back. Martin (1990) shows how one can use D-reflexive practices to demolish the assumptions of a text, thereby creating space to engage in R-reflexivity and construct an alternative and emancipatory text. Martin engages in D-reflexivity when she deconstructs the story of a female employee having a caesarean, as told from the perspective of her employer. Deconstruction is used not to generate new knowledge but, to destabilize the text and to challenge its gendered and managerialist assumptions, which then enables the use of R-reflexivity to introduce new assumptions that construct a different and potentially emancipatory text, providing a new understanding of gender and organizational life.

Self-Reflexive Practices
In bringing the understanding of reflexivity to bear on ourselves, we acknowledge that we take the focus away from what other researchers are doing with regard to reflexivity and turn the spotlight on ourselves, possibly earning the criticism of narcissism as a result. In trying to be reflexive we naturally face the sorts of problems that others face and we do not intend to set ourselves up as somehow being immune from the problems that beset other scholars. We hope, however, that we can fruitfully bring the range of reflexive practices that we have identified to bear on our own efforts at knowledge production.
We start by confessing our criterion for successful reflexivity; that is, whether it makes a productive difference. We believe some kind of tangible result should be demonstrated, such as ideas, concepts, challenges to conventional thinking, or suggestions for new research. Being productive does not necessarily mean being positive -negating or deconstructing ideas is also a productive outcome. Going through the intimate relation between the researcher and their knowledge in a reflexive loop should, we believe, lead to some novel (re)descriptions, (re)interpretations or (re)problematizations that add some quality to the text and the results it communicates. We also believe that another purpose of reflexivity is to 'improve' research and theorizing -producing fieldwork, texts or theoretical results that are 'better' in some distinctive way than they would be without reflexivity. The meaning of 'better' is not self evident -it may be more creative, offering a broader set of ideas/interpretations, more ethically informed or sensitive or avoiding getting caught by the social conventions or fashions. Nonetheless, it seems clear that we favour instrumental reflexivity (Weick, 1999): for us, reflexivity is not primarily an end in itself, but a means to improve research in some way.
We therefore present a classification which, we tell readers, constitutes a new way of thinking -about reflexivities rather than reflexivity; and about practices that come to be known as various forms of reflexivity. Thus we argue that our categorization constitutes a contribution, not only informing theoretical reflection but also research design. But what exactly is our contribution? Have we identified something? Or have we constructed a particular version of reflexivity? Have we simply mapped a body of literature? Or have we ordered and domesticated the field, using our powers as established researchers to normalize how reflexivity is to be understood? Have we reduced 'reflexivity' to a set of four packages? Or have we provided a catalyst for further discussions on reflexivity? And, if the latter, what does our classification mean for the construction of knowledge about reflexivity?
Multi-perspective and multi-voicing practices help us address these questions. One observation is that different consumers might use our work differently. For example, functionalist readers might see our framework as a more or less accurate representation of a body of work. Interpretivist readers, on the other hand, might note that a map does not need to be accurate to be useful: the accuracy of a map may be less important than whether and how it is consumed. Postmodern readers may resist the production of the map altogether, arguing that it simply reifies knowledge and represents modernist claims to authorial privilege. Foucauldians may say that we arbitrarily order what is perhaps better seen as an open space for 'reflexivity' and thereby freeze understandings. Either way, whether or not our categorization ever serves as a catalyst for discussions about reflexivity depends upon voices of readers who will ultimately determine the 'success' of this categorization (e.g. good map; poor map; catalyst, or not) through the way in which they engage with it.
So far, this discussion is very much informed by R-reflexivity. Drawing on a range of reflexive practices, we emphasize our contribution: the production of new insights by reconceptualizing reflexivity as practice; and an account with alternative descriptions and points of departure to produce 'better' research. We acknowledge that our classification is but one reading for which we attempt to make a persuasive case, using the usual rhetoric associated with our academic community, such as acknowledging that the questions we raise are not intended to be exhaustive nor are the four sets of practices a complete account of reflexive practices.
To explore the dangers of our contribution, we must introduce D-reflexivity. Despite our protestations that there is an open and diverse world of reflexivities, there is always a risk in ordering and fixing it, as we do with our categorization: some potential candidates for inclusion are left out or packed together with someone else; while the labels that we use help to construct the phenomena in a particular way. The work that fits neatly into one box or another is inevitably privileged and, quite possibly, reified. Work that straddles classifications is either forced into one of them or left in limbo. So, for example, we combine two different approaches under the heading of multiperspective practices -those that use different perspectives to build a more complete picture, and those that use them to show that a complete picture is impossible. We must therefore acknowledge the dangers of instrumental reflexivity, which can domesticate the unruly, worrying, and unpredictable by forcing it into boxes and categories, which may then be appropriated by other researchers as a convenient tool that can be used in ways that may be antithetical to many reflexive researchers. D-reflexivity also draws our attention to some of the things absent from our classification. First, while we present a classification reminiscent of the paradigms but, by focusing on practice, we avoid the epistemological and ontological debates that characterized the paradigm wars. Second, in focusing on textual practice, we avoid a discussion of the practice of doing research, in the sense of fieldwork. Our emphasis on the practice of writing (and of generally thinking about) research allows us to ignore the myriad research subjects that co-construct research through their relations with researchers. To be sure, we acknowledge these research subjects in our discussion of multi-voicing, for example, but we never engage with them. They are absent from our discussion. Of course, we have answers for these self criticisms. For example, the paradigm debate is irresolvable (cf. Clegg and Hardy, 1996b). What comes to be constructed as reflexivity are the practices in published work, not purist theoretical discussions. Space constraints and demands for a clear focus are imposed on journal articles, which makes it impossible for us to include everything and, therefore, reflexivity in research design practices is something that should be taken up elsewhere. And, ultimately, we argue that is in the ongoing closing down of ideas and opening them up through debate that knowledge -albeit partial -is produced. Thus we return to R-reflexivity.
Why do we advocate this particular dynamic between D-and R-reflexivity, ending with the latter? Part of the answer probably lies with the fact that we do empirical research and, some reflexive practices make it almost impossible for us to write up our empirical work. Furthermore, it is difficult emotionally to engage in one's own research and, then, simply to dismiss it as dangerous knowledge. Thus we prefer a pragmatic rather than an idealistic engagement with reflexivity. At the same time, we conduct critical research to which D-reflexivity is central. As a result, it also profits us to acknowledge D-reflexivity. Equally, it is perhaps not surprising that we then suggest moves to negate it by drawing attention back to some of the practices that we believe facilitate productive responses. As a result, we counteract our R-reflexive tendencies with D-reflexivity in a dialectic that is professional and personal; pragmatic and idealistic and which, of course, draws on many of the practices that we have already identified to make a persuasive, credible case.

CONCLUSIONS
We believe that reflexivity is important to the understanding of what happens in research. Reflexivity means thinking through what one is doing to encourage insights about the nature of social science and, especially, the role that language, power/ knowledge connections, social interests and ideologies, rhetorical moves and manoeuvring in the socio-political field play in producing particular accounts. It may also inspire creativity through opening up for new perspectives and providing reference points for what one is doing and to avoid or minimize certain 'harmful' aspects of research that follow from lack of reflexivity. To this end, we have analysed the existing work on reflexivity and then categorized and conceptualized it in terms of four common sets of practices: as multi-perspective, multi-voicing, positioning and destabilizing. We have also introduced a framework of reflexivity around a dialectic between 'positive' and 'negative' positions: R-and D-reflexivity. This allows us to understand in more depth the different ways in which reflexivity can be achieved. We also believe that these conceptualizations are useful in adopting a more critical stance on reflexivity; we cannot judge whether and how reflexive research is useful unless we understand the practices it produces and their effects.
It is, in our opinion, important not to mystify reflexivity. Without critical interrogation, reflexivity risks becoming a dead end rather than a route to more thoughtful and interesting research. It may be used as a cynical rhetorical device designed to demonstrate researcher credentials in critical or postmodern circles in the same way that conventions of rigour and replicability apply in positivist circles -an academic hoop through which those wishing to publish must jump (Alvesson and Sköldberg, 2000). As such it may primarily fulfil ceremonial purposes of legitimation -similar to the methods section in academic papers where quantitative and qualitative research is disciplined by neo-positivist templates.
In reviewing reflexivity as textual practice, we are interested in how research is written up in ways that have implications for reflexivity. If knowledge more generally is a product of linguistic, political, and institutional influences, so too is reflexivity: it is a construction of communities of researchers whose work is informed by particular theoretical influences; who are subject to the demands of particular university systems, journals, and granting agencies; who operate within discourses of science, education, management, and progress; and who use language to promote particular versions of 'truth' or claims to superior insights. Reflexivity is not a fixed 'thing': what we -as members of a research community -know to be reflexivity is shaped by practices carried out by researchers in producing texts which are accepted as being reflexive. As papers are published in reputable journals, the practices described or enacted by researchers play an important role in constructing the meaning of reflexivity; and, as these meanings come to be widely shared, they become institutionalized. Thus the practices outlined here collectively construct the meaning of reflexivity and, as such, warrant closer examination.

NOTES
[1] For a review of reflexivity in other disciplines, see Lynch (2000).
[2] This is not to deny that the sets of practices include some elements of both: destabilizing practices offer some sort of alternative understanding -even Foucauldian ideas on how knowledge produces rather than reveals truth says something about how subjects are created; in the case of multi-perspectivist practices, there is frequently a partial or minimalist deconstruction when one perspective is used to disturb another (Alvesson, 1996).