Accountability

Accounting, Auditing & Accountability Journal
Emerald Article: What counts as "theory" in qualitative management and accounting research? Introducing five levels of theorizing Sue Llewelyn

Article information:
To cite this document: Sue Llewelyn, (2003),"What counts as "theory" in qualitative management and accounting research? Introducing five levels of theorizing", Accounting, Auditing & Accountability Journal, Vol. 16 Iss: 4 pp. 662 - 708 Permanent link to this document: http://dx.doi.org/10.1108/09513570310492344 Downloaded on: 07-01-2013 References: This document contains references to 168 other documents Citations: This document has been cited by 41 other documents To copy this document: [e-mail] This document has been downloaded 4071 times since 2005. *

Users who downloaded this Article also downloaded: *
Sue Llewelyn, (2003),"What counts as "theory" in qualitative management and accounting research? Introducing five levels of theorizing", Accounting, Auditing & Accountability Journal, Vol. 16 Iss: 4 pp. 662 - 708 http://dx.doi.org/10.1108/09513570310492344 Sue Llewelyn, (2003),"What counts as "theory" in qualitative management and accounting research? Introducing five levels of theorizing", Accounting, Auditing & Accountability Journal, Vol. 16 Iss: 4 pp. 662 - 708 http://dx.doi.org/10.1108/09513570310492344 Sue Llewelyn, (2003),"What counts as "theory" in qualitative management and accounting research? Introducing five levels of theorizing", Accounting, Auditing & Accountability Journal, Vol. 16 Iss: 4 pp. 662 - 708 http://dx.doi.org/10.1108/09513570310492344

Access to this document was granted through an Emerald subscription provided by MONASH UNIVERSITY For Authors: If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service. Information about how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information. About Emerald www.emeraldinsight.com With over forty years ' experience, Emerald Group Publishing is a leading independent publisher of global research with impact in business, society, public policy and education. In total, Emerald publishes over 275 journals and more than 130 book series, as well as an extensive range of online products and services. Emerald is both COUNTER 3 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.

The Emerald Research Register for this journal is available at http://www.emeraldinsight.com/researchregister

The current issue and full text archive of this journal is available at http://www.emeraldinsight.com/0951-3574.htm

AAAJ 16,4

METHODOLOGICAL ISSUES

662
Received 11 November Revised 4 December 2002, 19 December 2002, 19 March 2003 Accepted 1 April 2003

What counts as ``theory ' ' in qualitative management and accounting research?
Introducing five levels of theorizing
Sue Llewelyn
The School of Management, The University of Edinburgh, Edinburgh, UK
Keywords Quality management, Quality concepts, Organizational theory, Research, Organizational analysis, Working practices Abstract The value of qualitative empirical research in the management and accounting disciplines lies in its ``conceptual framing ' ' of organizational actions, events, processes, and structures. Argues that the possibilities for conceptual framing extend beyond the highly abstract schema generally considered as ``theories ' ' by academics. In support of this argument, distinguishes five different forms of theorization. Explores the relationship between these theoretical ``levels ' ' and the different issues that empirical research explores, arguing that, as the ``level ' ' of theorizing ``rises ' ', issues of agency give way to a focus on practice and, in turn, to a concern with structure. As this happens, research aims directed towards abstraction and explanation supersede those for contextualization and understanding. Concludes that views on ``what counts as theory ' ' are, currently, too narrow to conceptualize agency, emergence and change adequately in organizational life and, hence, the full range of significant empirical phenomena that characterize the management and accounting areas are not being researched.

Introduction Qualitative research in the management and accounting disciplines[1] engages in empirical work to gather information on organizational actions, events, structures and processes. After this field research is over, researchers face some challenging questions. How should these actions and events be understood? How can organizational structures and processes be explained? The standard response to such questions is to ``incorporate some theory ' ' into accounts of these organizational issues. Theorization (or conceptual framing) is the ``value-added ' ' of qualitative academic research. Conceptual framing can offer a greater understanding of the empirical issues under discussion. Appropriate theorization can give fuller explanations of organizational structures and processes than those held by organization members. But what should this theorization consist of? One answer is to reach for
Accounting, Auditing & Accountability Journal Vol. 16 No. 4, 2003 pp. 662-708 # MCB UP Limited 0951-3574 DOI 10.1108/09513570310492344

This paper develops ideas first presented at a plenary session of the 2001 APIRA Conference, University of Adelaide, South Australia. Thanks are due to participants at APIRA for their helpful comments. The author is also indebted to two anonymous referees for constructive feedback that greatly aided the development of the paper. However, any conceptual misapprehensions remain the author 's own.

the work of influential philosophers or sociologists like Marx, Weber, Habermas, Foucault, Latour or Bourdieu on `` . . . the supermarket shelves of social theory . . . ' ' (Prichard and Willmott, 1997, p. 292). But do these abstract theorists always offer the right ``level ' ' of theorization for empirical work? Several authors interested in the relationship between theory and qualitative research think that they may not. Wodak (2001, p. 64) states that:
. . . the first question we have to address as researchers is not, ``Do we need grand theory? ' ' but rather, ``What conceptual tools are relevant for this or that problem and for this and that context? ' '.

What counts as ``theory ' '?

663

Mouzelis (1995, p. 9) argues that a pressing task for researchers is:
. . . to clarify current conceptual tools and to construct new ones by following criteria of utility rather than truth.

Alvesson and Skoldberg (2000, foreword) set out a rationale for a project: È
. . . to bring abstract philosophical (meta)theories, concepts and ideas down to a level where they become relevant and manageable in a qualitative methodological context.

These comments press a case for some re-thinking on the question of how theory can support empirical research; they also make reference to a bewildering array of theoretical forms: . grand theory; . conceptual tools; . philosophical (meta)theories; . concepts; and . ideas. This proliferation of terms raises a fundamental question that has not been resolved in social science ± what is ``theory ' '? Sayer (2000, p. 146) comments on the `` . . . extraordinarily unexamined character . . . ' ' both of the term ``theory ' ' and of its interconnections with empirical research. Furthermore, Weick (1989, p. 516) argues that:
Theory cannot be improved until we improve the theorizing process, and we cannot improve the theorizing process until we describe it more explicitly, operate it more self-consciously and decouple it from validation more deliberately.

This paper responds to this agenda by identifying and discussing five different ways of theorizing that are available to qualitative empirical researchers: (1) metaphor; (2) differentiation; (3) conceptualization; (4) context-bound theorizing of settings; and (5) context-free ``grand ' ' theorizing.

AAAJ 16,4

664

It is suggested that the first four ways of conceptually framing organizational issues have not been given enough prominence in discussions about theorizing in qualitative management and accounting research, whereas the last (``context-free grand theorizing ' ') has been rather over-emphasised. The structure of the paper is as follows: first, the question of ``What is theory? ' ' is addressed; second, the five levels of theory are discussed, in turn; third, these levels of theory are identified in published qualitative research; fourth, there is a discussion of how these levels of theory relate different empirical issues; fifth, links are forged between various methodologies and different levels of theory; and finally, some concluding comments are offered. What is theory? Before different ``levels ' ' of theorizing are introduced, an understanding of what is implied- in the context of qualitative research ± by the term ``theory ' ' is offered. It will already be evident that this article is proposing a broader understanding of ``theory ' ' than that held by many academics. As Sayer (2000, p. 147) has argued, there has been a widely held presumption in academe that theories are ``. . . few in number, monolithic and distinct . . . ' '. Furthermore, it can be asserted that what is presently thought to distinguish them is `` . . . their unfamiliar or esoteric character in relation to lay vocabularies ' ' (Sayer, 2000, p. 146). The understanding of what theory ``is ' ' in the social sciences has grown up in the shadow of what theory ``is ' ' in the natural sciences. Hence, in the Academy of Management Review and under the title of ``Theory construction as disciplined imagination ' ' Weick (1989, p. 517, citing Sutherland) was still presenting the following definition of ``theory ' ':
By theory we mean, ``an ordered set of assertions about a generic behaviour or structure assumed to hold throughout a significantly broad range of specific instances. ' ' (Sutherland, 1975, p. 9) . . . as the range of specific instances becomes broader, the resulting ideas are more deserving of the label theory.

This paper argues against this restricted view of ``theory ' ' for several reasons (these reasons are mentioned here briefly and will be picked up again in the later sections). First, it is not helpful for improving the relationship between theorizing and qualitative empirical research in social science disciplines such as management as it rests on assumptions from the natural sciences about ``what theory is ' '. Second, its focus on generic behaviours (and structures) and generalization excludes an interest in emergent, localized phenomena ± and in social science such ``instances ' ' are at least as significant, arguably more so, than patterned regularities. Third, it ignores the ``contextualization[2] ' ' of behaviours and structures that may be essential to understand them, by implicitly assuming the necessity of the ``abstraction ' ' that underpins their explanation. It is maintained here that ``theories ' ' reflect the contestation of meaning and significance in social and organizational life (Sayer, 1992, p. 83[3]). ``Meaning ' ' not only follows from the sense-making of individuals but is also concerned

with how something is connected or related to something else: how an event is connected into an episode or how social actors are related in a organizational structure (see Polkinghorne, 1988, p. 6). ``Meaning ' ' is relatively stable but assessments and attributions of ``significance ' ' differ more across cultural and historic contexts. Significance varies due to interaction effects and unintended consequences; a relatively minor and not very meaningful event may assume huge significance within a particular context[4]. Theories impose cohesion and stability (Czarniawska, 1997, p. 71). So that whenever ``life ' ' is ambiguous (which is most of the time!) people will work at confronting this ambiguity through ``theorizing ' '. Also, because ``life ' ' and situations commonly have multiple meanings and give rise to different assessments of significance, everyone has a need for ``theory ' ' to go about their everyday affairs. ``Theories ' ' do not just reside in libraries, waiting for academics to ``dust them down ' '; they are used whenever people address ambiguity, contradiction or paradox so that they can decide what to do (and think) next. Theories generate expectations about the world. They are also drawn on to ``gain theoretical leverage ' ' (Archer, 1996, p. xx) for a point of view in everyday discussion. In consequence, the terms ``theorizing ' ' or ``conceptual framing ' ' are generally used in this paper in preference to ``theory ' ', to distance the discussion from the rather grandiose assumptions that have, unfortunately, grown up around this word. For Archer (1996, p. 107) theories (along with beliefs, values and arguments) form part of culture. They form a part of the resource base that people draw on whenever they are working out what their opinion on a particular issue should be, how they should conduct themselves in relationships with other people and how they should act in uncertain situations. Archer (1996, p. 145) argues that different theories reproduce different agents and also take organizational cultures in different directions:
The logical properties of [agents '] theories or beliefs create entirely different situational logics for them. These effects mould the context of cultural action and, in turn, condition different patterns of ideational development.

What counts as ``theory ' '?

665

So theories are used for work and communication as well as reflection. In consequence, Sayer (1992, p. 59) argues for a more ``down to earth ' ' view of ``theory ' ' ± as a conceptual system:
We develop and use concepts not only through and for observing and representing the world but also for acting in it, for work and communicative interaction; for making and doing as well as speaking, writing, listening and reading, for running organizations and working in them, for programming computers, cooking meals, teaching children, sorting mail and so on . . . Conceptual systems concern not only what we (think we can) observe, but what we can do and how we do it. Again it may be wise to avoid thinking of knowledge as attempting to ``represent ' ' or ``mirror ' ' the world like a photograph. A better analogy may be that of a map or recipe or instruction manual, which provides means by which we can do things in the world or cope with events.

If theorizing is integral to ``working and doing ' ' (as well as abstract reflection), does this have an impact on what makes for ``good ' ' theory? Drawing on

AAAJ 16,4

666

Bhaskar 's (1979, p. 142) concept of ``practical adequacy ' ', Sayer (1992, p. 69) suggests that this may be a better criterion than ``truth ' ' for judging how well a phenomenon is being theorized, ``To be practically adequate, knowledge must generate expectations about the world and about the results of our actions that are actually realized ' '. Mouzelis (1995, p. 9) also endorses the idea of assessing theory on its practical adequacy; as discussed earlier, he sees a major project for theorists (interested in the relationship between theory and empirics) in the construction of theories that follow, `` . . . criteria of utility rather than truth ' '. He advocates a research agenda, which he terms ``conceptual pragmatism ' ' (Mouzelis, 1995, p. 8) to assist empirically orientated researchers to move from micro through meso to macro levels of analysis using conceptual tools that are flexible and open-ended. A criterion of utility rather than ``truth ' ' in assessing the success of any conceptual framing also makes more sense in the context of the social science disciplines. Theorizing in social science does not merely convey an external ``reality ' '; it enters into this ``reality ' '. Bhaskar (1979, p. 48) argues that:
. . . social structures exist only in virtue of the activities they govern, they do not exist independently of the conceptions that the agents possess of what they are doing in their activity, that is of some theory of these activities.

Sayer (1992, p. 30) concurs with Blaskar 's view and speaks of the ``conceptdependence ' ' of the social world, arguing that what social practices, institutions, roles and relationships are depends on what they mean to social actors[5]. The concept-dependence of the social world implies that `` . . . truth is always relative to a conceptual system . . . ' ' (Lakoff and Johnson, 1980, p. 159). Or as Burrell and Morgan (1979, p. 10) put it, knowledge of social phenomena is knowledge from a particular frame of reference (or paradigm). So conceptual framing is not an ``option ' ' that stands apart from empirical research, it enters into (and is a part of) the social phenomena under investigation. Moreover, as all social phenomena are concept-dependent, theorizing for qualitative work in the social sciences cannot only ``begin ' ' at the level of ``grand theory ' ', theorizing and conceptualizing must be conjoined. The conceptualization of empirical data is negotiated in ``theories ' '. Observed empirical ``data ' ' are always pre-theorized, the world is understood only through particular ``ways of seeing ' '. Indeed there is no perception without conceptual schemes within which to locate perceptions (see the discussion in Sayer, 1992, pp. 51-6). Observation, itself, is theory-laden (Hanson, 1958, p. 7). In consequence, new theories bring new objects into view and the ``same ' ' empirical object appears differently through different theoretical ``lens ' '. As the conceptual framework and the object that is being ``framed ' ' are interdependent in the social sciences, theories cannot be tested or filled out with ``data ' ' whose meaning is taken to be transparent. As Bhaskar (1979, p. 48) argues, what social structures ``are ' ' does not exist independently of either the practices they govern or the theories of the agents who engage in these practices. Hermeneutics recognizes this ± through insisting

that empirical researchers should draw on both their own pre-understandings and the theories of the agents that they study (see Llewellyn (1994) for a discussion of hermeneutics in management and accounting research). However, qualitative empirical research in the management and accounting disciplines has ignored the impact that the theories of organizational actors have on their accounts of organizational life. Qualitative researchers have also neglected to include agents ' theories in their own accounts of the research setting[6]. To summarize on the understanding of ``theory ' ' that underpins the discussion of qualitative research in this paper ± everyone theorizes! Theories are for working and doing as well as reflection ± theories are not only the preserve of academics. Theorizing expresses the meaning and significance of social phenomena, it negotiates peoples ' everyday experiences and it generates expectations about the social world. Theory and data are interdependent in social science. The meaning of data cannot be accessed without theory and theories cannot be validated (or discounted) in any straightforward way through ``data ' '. Criteria of utility rather than ``truth ' ' should be employed in making judgements on the adequacy of theory. Conceptual framing in qualitative empirical research encompasses both the theorizing of researchers and that of the organizational actors they study. With these presuppositions in place, the paper now turns to outline different ``levels ' ' of theory. Levels of theorization Rather than assuming that there is only one type (or form) of theorizing possible (and, therefore, available) for qualitative empirical research, this paper proposes that five ``levels[7] ' ' of theorization can be distinguished. These levels are briefly identified in Table I and then discussed, in turn, in detail. Level one: metaphor theories Peoples ' experience of the world is grounded in metaphor; as Lakoff and Johnson (1980) put it, metaphors are what we ``live by ' '. Yet the power of metaphor to theorize experience is not well understood. Consequently, the use of metaphor as a theoretical tool in qualitative empirical research is vastly underdeveloped. Morgan (1983, p. 601) argues that metaphor is:
. . . a basic structural form of experience through which human beings engage, organize, and understand their world. Level One Two Three Four Five Theory Metaphor theorizes Differentiation theorizes Concepts theorize Theorizing settings Theorizing structures Focus By image-ing and grounding experience By ``cutting the pie ' ' of experience By linking agency and structure through practice Explaining how contexts for practices are organized Explaining impersonal, large scale and enduring aspects of social life

What counts as ``theory ' '?

667

Table I. Five levels of theorization

AAAJ 16,4

668

He illustrates his argument in Morgan (1986) by showing the range of metaphors through which organizations are understood and experienced: machines, organisms, brains, cultures, political systems, psychic prisons, flux and transformation and instruments of domination. This understanding of how metaphor provides both a ``way of thinking ' ' about organizations and a ``way of seeing ' ' has been widely influential. For example, it has led Mangham (1986, p. 21) to state that:
Metaphor, far from being simply a poetic device or merely a matter of names and descriptive terms, is, above all else, a mode of thought.

This view of metaphor establishes its centrality in grounding the meaningfulness that human beings attribute to the world. However, despite the pivotal nature of metaphor, it has been under-utilized as a theoretical tool in the management and accounting disciplines. One example from the accounting field is Llewellyn 's (1994, 1998a) work on the metaphor of the boundary for understanding the ways in which accounting numbers create organizations. This ``basic-ness ' ' of metaphor to human experience implies that it forms the foundation of our conceptual systems; it also explains why new concepts tend to emerge through metaphor, as the intelligibility of new and unfamiliar concepts requires some prior acquaintance with familiar concepts ± often from a more ``basic ' ' context. For Lakoff and Turner (1989, pp. 38-9) metaphor maps understanding from a source domain (the familiar) to a target domain (the unfamiliar) and works through establishing a correspondence between the two. This `` . . . recognition of the essence . . . ' ' (Gadamar, 1975, p. 103) and ``Transposition from one sphere into another . . . ' ' (Gadamar, 1975, p. 390) is frequently built on ``picturing ' ' or ``image-ing ' ' the unfamiliar in terms of the familiar. Sayer (1992, p. 63) points out that even economics (the ``hard ' ' social science) is loaded with metaphorical expressions (e.g. inflation, co-operative game, free rider, supply chain and, classically, Adam Smith 's invisible hand). Once metaphor forges a connection, the new concept can begin to develop- based on the image established through its correspondence with a pre-existing and familiar idea. So, ``The essence of metaphor is understanding and experiencing one kind of thing in terms of another ' ' (Lakoff and Johnson, 1980, p. 5). The first example given by Lakoff and Johnson (1980, p. 4) is one particularly relevant to academic researchers ± the way that argument is understood and experienced in terms of war! They quote (among others) the following expressions and italize the metaphor:
He attacked every weak point in my argument. I demolished his argument. I 've never won an argument with him. He shot down all my arguments.

A crucial part of Lakoff and Johnson 's (1980, p. 4) analysis is to emphasize that metaphor does not just aid understanding, it also structures experience:
. . . we don 't just talk about arguments in terms of war. We can actually win or lose arguments. We see the person we are arguing with as an opponent. We attack his positions and we defend our own. We gain or lose ground. We plan and use strategies. If we find a position indefensible, we abandon it and take a new line of attack. Many of the things we do in arguing are partially structured by the concept of war. Although there is no physical battle, there is a verbal battle and the structure of an argument-attack, defense, counter-attack, etc.reflects this.

What counts as ``theory ' '?

669

This point raised by Lakoff and Johnson (that all concepts, including metaphor, enter into social experience) has been emphasised because it re-iterates the wider point made earlier ± that social life is concept-dependent. For example, a different understanding and experience of debate is provided by Habermas (1984, 1987) in his ``communicative rationality ' ' ± which he contrasts with ``strategic rationality ' ' (as the type of rationality likely to result in an argument!). In communicative rationality when people come together in debate they are motivated towards achieving a consensus ± so that further interaction may be undertaken. Instead of being infused with a metaphorical correspondence with war, debate is experienced a ``give and take ' ' affair that is directed towards agreement:
We entered into an exchange of opinions. Everyone was free to have a say. We each recognised the other 's point of view. We took turns at making the running in the issues.

Gentner (1989) points out the power of metaphor in establishing a primary understanding of any phenomenon. Any subsequent understandings have to acknowledge (or negotiate) the metaphorical foundations that have been established for understanding. In terms of the above examples, individuals ' understanding and experience of argument is going to be different dependent on whether they transfer relationships from the domain of war or the domain of free exchange. Likewise, any two individuals ' experience of an organization are going to vary dramatically if one takes the ``psychic prison ' ' as her/his basic metaphor and the other absorbs the ``culture ' ' metaphor. However, more commonly, people will employ several metaphors to understand, experience and theorize about a complex phenomenon like an organization. An exclusive focus on a single metaphor is likely to be unhelpful. For example, Ahrne (1990, p. 24) argues that the dominant ``system ' ' metaphor (be it ``closed ' ' or ``open ' ') for organization has distorted thinking by masking aspects of discontinuity and emergence in organizational life. Streatfield (2001, p. 126) also points out that the system metaphor conjures up a mistaken image of change- as movement of the whole organization into the future. This disguises a picture of change as a process that has multiple origins in different parts of the organization at various times (see later sections).

AAAJ 16,4

To sum up, metaphor theorizes through linking the unfamiliar to the familiar; it creates meaning and significance through ``picturing ' ' or ``image-ing ' ' the world. Level two: differentiation theories Peoples ' understanding and experience of the world is differentiated by a multitude of what come to be highly significant pairings, contrasts, dualities or dualisms. For example, thought and experience are highly structured by the following: . presence-absence; . up-down; . in-out; . finite-infinite; . mind-body; . public-private; . objective-subjective; . practical-theoretical, and so on . . . As with metaphor, the theoretical import of these dualities has not been thoroughly explored and, therefore, their potential in empirical work has not been fully exploited. Some of these dualities (or dualisms) are established in non-metaphorical ways. For example, the first three cited above are rooted in spatial orientation and bodily experience and are, therefore, not founded in metaphor (see Lakoff and Johnson, 1980, p. 56-60). However, once human endeavour moves beyond basic orientations in time and space, concepts with which to understand and express experience are developed metaphorically ± often, initially, by making a correspondence with basic contrasts that relate to space-time. For example, the concepts of public-private are partly structured in correspondence with out-in and external-internal. Concepts in social science are established metaphorically; they are developed through differentiation (i.e. through making contrasts with other concepts). Dualisms (or dualities) are the most basic way that this is done as the meanings of two inextricably interrelated terms develop alongside one another. These pairings are sometimes seen as in contradiction to each other, frequently with one term assuming dominance (as does ``objective ' ' in the objective/subjective contrast). But for concepts to be opposed they must be mutually intelligible (i.e. must have certain terms in common, over which they can contradict each other). So equally, it can be argued that these paired terms are interlocked and, although they refer to different aspects of reality, these aspects presuppose and derive their meaning from each other. An emphasis on the ``opposing ' ' nature of these key paired terms is captured by the label ``dualism ' '; a stress on their dependence is reflected in the descriptor ``duality ' '[8].

670

In accounting, the relevance lost/relevance regained (Johnson and Kaplan, 1991) differentiation was pivotal in theorizing the demise and rebirth of management accounting. McGregor 's (1960) Theory X and Theory Y differentiation has been an influential understanding of how managers conceptualize employee motivation. Brunsson 's (1994) political organization/ action organization duality is a more contemporary example of theorizing through a key differentiation in the organizational theory area. This duality theorizes the nature of public sector organizations as basically political and concerned with the enabling of debate, while private sector organizations are conceptualized as oriented towards consensus and action. Creating dualisms (or dualities) may be the most basic but it is not the only way that `` . . . people cut the pie of experience ' ' (Triandis, 2001, p. 20). Carving up the world through categorization underlies peoples ' understanding and experience of social life. As Potter and Wetherell (1987, p. 116) point out:
Categorization is fundamental to the social scientist [For example] Much of social psychology is concerned with the attributes of social groups: males, political extremists, working-class adolescents . . . Social categories are, in one way or another, the principal building blocks in many areas of social research.

What counts as ``theory ' '?

671

As with metaphor and other conceptual tools, categorization does not only support reflection, it sustains acting and doing ± as categories are constructed to further certain goals and used as justifications for particular ways of acting. Maslow 's (1943) hierarchy of needs is a layered categorization of human motivation that has been mobilized to support particular ways of securing the active commitment of the workforce. Ouchi 's (1980) work on the markets/ bureaucracies/clans categorization has been very widely discussed in the governance literature, with several authors offering extensions or modifications (see, for example, Thompson 's (1991) addition of ``networks ' '). Challenge to key dualities and categories (and other levels of theorization) is an important component of theoretical development as they can impede as well as enable new (and possibly more productive) ways of both thinking and doing. For example, Lakoff and Johnson (1980, p. 185) argue that the objective/ subjective dualism has stalled debate on the epistemology of knowledge in the social sciences:
To many people raised in the culture of science or in other sub-cultures where absolute truth is taken for granted, this [the acceptance that truth is relative to a conceptual system] will be seen as a surrender to subjectivity and arbitrariness ± to the Humpty-Dumpty notion that something means ``just what I choose it to mean ± neither more nor less ' '.

The implication of their argument is that the objective/subjective categorization is preventing epistemological progress in the social sciences and requires some re-thinking. Sometimes change can be accomplished through the introduction of a new ``bridging ' ' concept that crosses a key dualism. For instance, Berger and Luckmann 's (1966) concept of ``social construction ' ' can be seen as an attempt to ``bridge ' ' the objective/subjective dualism. Social constructivism posits that the concept of gender, for example, is produced

AAAJ 16,4

672

through ``inter-subjective ' ' agreement on what constitutes the ``correct ' ' attitudes and behaviours for each of the sexes but, once accomplished, this category confronts any individual as an ``objectified reality ' ' ± or a rigid gender stereotype that may be, personally, extremely constraining. Such was the success of ``constructionism ' ' in the social sciences that Hacking (1999) argued that it is becoming hard to think of anything that cannot now be described as ``socially constructed ' ' (e.g. facts, people, ideas, and theories all seem to qualify). Metaphor may also be used to ``bridge ' ' a dualism. For example, the idea of a ``conceptual tool ' ' attempts to attenuate some of the negative consequences of the practical/theoretical duality in its metaphorical emphasis on the usefulness of concepts as practical implements. Delanty (2003) has argued that categorization is a hallmark of modernity; in contrast, post-modern approaches to theorizing seek to bridge, blur or merge analytical categories. To sum up, differentiations theorize through ``cutting up ' ' experience; they create meaning and significance through setting up contrasts and categories that order the world. Level three: concepts theories The theoretical significance of concepts in the social sciences has been recognized. As the earlier section on theory points out, concepts are the fundamental tools used in social practice (and in social science) both to observe and represent the world and to act and work in it. But concepts have been thought of as part of theories rather than theoretical tools in themselves. It is argued here, first, that concepts are essential to praxis and, second, that concepts constitute theories of practice. The introduction of new concepts (so that practical developments in the world can be adequately discussed) and the refinement of the senses of existing ones both happen through ``level three ' ' theorizing. New concepts reflect different ways of thinking and acting in the world; Giddens (1987, p. 43) comments on conceptual innovation:
Conceptual innovation (coupled to empirical research) is at least as important in social science as is the formulating of novel generalisations. For such innovation opens up ``ways of seeing ' ' that do not exist within the perspectives of lay actors, disclosing unsuspected aspects of, and potentialities within, a given set of institutions.

For example, ``feminism ' ' was a conceptual innovation. This concept allowed women to assess their experience from a new perspective, it enabled them to act differently and it drove the emergence of new political movements that challenged the established order of things. The refinement of concepts sharpens and/or changes their meaning either by exploring new internal differentiations (using level two distinctions) or by re-working relationships with other related concepts. A concept is internally differentiated through identifying the various distinctions it encompasses. For example, Lindkvist and Llewellyn (in press, p. 16) explicate the several ways in which people experience a ``sense ' ' of accountability. They argue that, broadly speaking, within organizations, individuals ' senses of accountability are for

either ``results ' ' or ``values ' ' and that the locus of accountability may be held at either an individual or group level. Sayers (1992, p. 81) comments, as follows, on ``re-working ' ' the meaning of concepts through repositioning them in a network:
These alterations require us to ``explicate ' ' problematic concepts; that is, give concise definitions to important but vaguely understood terms through re-working their relations with other terms in the network.

What counts as ``theory ' '?

673

Re-working is also called for when concepts ``lose ' ' their meaning:
There are also cases where there are so many competing explications of particularly difficult concepts that it becomes uncertain whether we are still talking about the same thing. So many sense-relations may be brought into question and suspended that the term loses its meaning; possible examples are ``ideology ', ``class ' ', ``value ' ' (in economics) and ``urban ' ' (Sayers 1992, p. 82).

The conceptual is the ``highest ' ' level of theorization that can still take the agent as its unit of analysis. Giddens (1984) developed structuration theory as a way of explaining how agents engage with structure in social practice. The ``duality of structure ' ' in structuration theory can be seen as a level three concept because Giddens uses the idea of ``structure ' ' in a particular and rather idiosyncratic way ± as the rules and resources that agents mobilize in their everyday activities. For Giddens, ``structure ' ' is not something that goes on ``behind the backs of people ' ' (Layder, 1994, p. 135), he sees ``structure ' ' as the medium and outcome of social activities and therefore, in Giddens ' work, structures and institutions have no existence outside of the practices they embody. Drawing on Giddens (1979), Layder (1994, p. 28) argues that:
. . . practices . . . are the primary point of connection between people and social structures. Practices are situated mainly at the meso (or organizational) level of analysis and ``bridge the gap ' ' between the micro and macro levels.

Concepts are the primary means of theorizing practices and practices are sites of struggle (Bourdieu, 1988). Organizational practices such as ``accountability ' ', ``politics ' ', ``decision-making ' ', ``financial reporting ' ', ``human resource management ' ' and so on are places where ``agency ' ' meets ``structure ' ' and are the primary point at which people in their everyday lives struggle to make a difference in the world ± to make effective interventions in the course of history. Accomplishing more adequate conceptualizations of practice is part of this struggle. As Gellner (1970) points out, the hypocrisy and deceptiveness inherent in many concepts supports and sustains many repressive power relationships. Level three theorizing can be thought of as linking lower levels of theorization (metaphor and differentiation) to higher ones (context-bound and context-free theorizing). Metaphor is point of origin of a concept; indeed Gasch_ e (1986) argued that all concepts are worn-out metaphors. Level two dualities and categorizations can be seen as internal differentiated concepts. Levels four and five are extended networks of concepts.

AAAJ 16,4

674

Another related reason for seeing concepts as a ``link ' ' between the micro and macro levels of analysis is what Layder (1993, p. 130) describes as their ``two-sided ' ' nature or ``duality of reference ' ' character in relation to the subjective/objective dualism. As practices, they reflect both the subjective and objective realms of experience. The concept of ``accountable ' ', for example, can describe both a feeling of responsibility to and for others and a structural aspect of organizations that sets up a system through which people are called to account. Similarly, the concept of power can encompass both a subjective sense of control over the organizational environment and an objective reality of subjugation to others in a hierarchical organizational structure. In the management and accounting disciplines where the central concern is with practices, the power of concepts in theorizing is enormous. Examples are Roberts ' (1991) theorization of accountability and Mintzberg (1989) on theorizing ``management ' '. To take just one other, Child (1972) developed the concept of ``strategic choice ' ', in part to counteract the deterministic tendencies of contingency and institutional theories. Although concepts that reflect specific practices (``accountability ' ', ``power ' ' and so on) have not generally been thought of as ``theories ' ', certain more philosophical concepts ± the ``isms ' ' ± have. Functionalism, positivism, structuralism and, more latterly, interpretivism, post-structuralism and post-modernism all ``fly ' ' under the banner of ``theory ' '. Presumably, the practical-theoretical dualism is impeding any understanding of the ``theorystatus ' ' of any concepts that are firmly rooted in practice. To sum up, concepts theorize through explicating practice; they create meaning and significance through linking the subjective and objective realms of experience. Level four: theorizing settings Level four explains specific social, organizational or individual phenomena in their settings. A range of differentiations and concepts will be tied together to construct a broader schema. Such schemas may either focus on specific issues at a particular level of analysis (e.g. conflict in organizations or how markets work in the public sector) or explicate relationships between different levels of analysis (e.g. how the organization conforms to wider social expectations or how organizations seek to enact their environments). The aim is to understand or explain how wider settings or contexts for activity are socially organized. In terms of grasping level four theorizing as the social organization of the settings for human activity, Ahrne (1990, p. 36) uses the metaphor of gardening:
. . . organizing may be likened to gardening, to cultivation and arrangement of the soil for the plants to grow in a controlled way. Gardening also presupposes constructing shelters against undesirable influences from the environing landscape in the form of rain, wind, sunshine or insects. It also involves regulating nature through water and nutrient supplies. Gardening is a constant struggle to manipulate and control the environment.

So, although level four theory may encompass practices, it is more concerned with the social conditions under which these practices are reproduced. Also, in social ``gardens ' ' any individual 's activities within an organization must negotiate that of her/his colleagues along with the `` . . . objectified praxis from the past activities of other people ' ' (Ahrne, 1990, p. 136). The social conditions for action encompass a wide variety of institutional forms ± for example, the family, the school, the workplace, the church and the government are all potential ``settings ' ' for level four theorization. At this level theory begins to move away from a focus on directly observable social or organizational activities, processes and events. The level of abstraction from the empirical world is greater so that non-observable structural or institutional issues may be included; theory begins to be, at least partially, independent of empirical data. In the accounting discipline, the seminal work of Hopwood (1983) launched a new genre of research that aimed to study accounting not as a technical abstraction but as a practice embedded in its organizational setting. Initially, his work was seen as a call to theorize the relationship between accounting and organizing ± at what would be characterized here as level four. But, latterly, his approach has also been taken as the starting point for theorizing accounting as a social practice ± at level three (see, for example, Miller, 2003). Theories of organizing and organization are couched mostly at level four. Research that focuses on the internal differentiation of organizations includes the very dominant ± until relatively recently ± varieties of systems theory. Systems theory relies on either organic or machine metaphors to argue that organizations are characterised by equilibrium and the interdependence of their ``parts ' '. As closed system thinking gave way to more open systems theories, ideas concerned primarily with the relationship between organizations and their environment emerged, for example, contingency theory, legitimacy theory, resource dependency theory and institutional theory. Much work in the accounting area has drawn on these ``open ' ' theoretical frameworks (see, for example, Otley (1980) through to Chapman (1997) on contingency theory). Latterly, work has also looked at the relationship between organizations (see Brunsson and Olsen (1998) on organizing organizations and Ahrne (1990) on the ``terrain ' ' between organizations). Influential theories at level four that focus on the inter-relationships between the professions that work in organizations include Abbott (1988). Braverman 's (1974) labour process theory also comes under level four ± as it is more concerned with the wider settings of power that structure the labour force than the agency of labour. As argued in the last section, conceptualization is the highest level of theory that can be concerned with agency. At level four the focus has shifted to the settings in which action takes place or the conditions under which agents pursue their projects. If the danger at levels one and two is theorizing that privileges the subjectivity of the agent and indulges in romanticism, then the risk at levels four and five is one of reification. Whenever it is assumed that organizations ``act ' ' or that institutions somehow come into being through social ``forces ' ' (and without active human endeavour) then social

What counts as ``theory ' '?

675

AAAJ 16,4

676

structures and systems are ``reified ' '. Layder (1994, p. 129) argues that many varieties of systems theory exhibit this type of error, for example, by assuming that systems have ``needs ' ' that they ``pursue ' ' ± such as the ``need ' ' for integration, balance or adaptation. Organization theory (with some exceptions ± for example, labour process theory) and much theory primarily focussed on explicating relationships at the level of institutions ± has been condemned for its neglect of critique (see for example Thompson and McHugh, 2002, pp. 363-87). A concern with the functionality of particular settings (for example, the prime focus of organization theory on efficiency and value consensus) has deflected attention from exploring the exploitative and repressive aspects of ``settings ' ' on their human inhabitants. To a limited extent, authors such as Silverman (1970) sought to produce more critical theories of practice within organizations (see level three theories) but, largely, critical work has drawn on social theory (and, therefore, turned to level five type theorization, see below). To sum up, context-bound theories offer an understanding of the settings for experience; they create meaning and significance through explaining relationships between phenomena. Level five: grand theorizing Grand theorizing is done in the ``world of ideas ' ' rather than the ``world of practice ' '. Popper (1972) calls the world of ideas ``world three ' '; Archer (1996, p. xvii) equates this world to ``the contents of libraries ' '. These meta-narratives are formulated at a high level of generality and reflect ideas that have been arrived at by thinking through issues and relationships in an abstract way ± rather than being derived from empirical research. Alvesson and Skoldberg È (2000, p. 160) dub grand theory, ``cathedrals of thought ' '. Hassard (1988) sees grand theories as technical social science language games that academics, in particular, are trained in. Latterly, the grand theory approach to theorizing has been under threat from the post-modern position, Lyotard (1984, p. xxiv) comments, ``Simplifying to the extreme, I define post-modern as incredulity toward metanarratives ' '. Grand theory is concerned with structural, impersonal, large-scale and enduring aspects of the social realm (e.g. social institutions, culture, class hierarchies and the distribution of power and resources). Social institutions, culture, class relations and the distribution of ``goods ' ' constitute regular and patterned social arrangements that individuals are born into and which will last beyond their lifetimes. Level five theorists explicate patterned relationships and their theories tend to be about necessity in the world ± as they set out the causal antecedents of action or focus on enduring structural conditions. At the extreme, such theorizing may aim for universal explanations that are beyond history and society. The modernist work of grand theorists can be traced to the Enlightenment project ± an endeavour that sought to apply rationality to the dilemmas that affect mankind. The hope was that, through mobilizing a comprehensive

framework of ideas and relying on enlightened (rather than instrumental) reason to achieve progress, people would be liberated from illusion, exploitation and domination. Such thinking imagines a utopian model of a just social order ± ``the solution ' ' (Alvesson and Skoldberg, 2000, p. 162). Grand È theory that is critical in this sense is usually associated with a leftist political stance that regards technological-capitalist organizing as repressive. Layder (1994, p. 7) calls this type of theorizing the ``view from on high ' ' for two reasons. First, because writers in this tradition `` . . . stress the idea that the external (macro) social conditions, to varying degrees, influence the form of social action . . . ' ', and, second, because grand theorizing aims for `` . . . a theory that tries to explain everything in terms of a comprehensive framework of concepts ' ' (Layder, 1994, p. 104). These insights into the key assumptions of grand theorizing (that ``agency ' ' is largely understood in relation to ``structure ' ' and that to earn the label ``theory ' ' conceptual framing should be totalizing) explain both the relative lack of importance accorded to empirical work by academics who espouse ``views from on high ' ' and the focus of any empirical research that they do engage in. Alvesson and Skoldberg (2000, p. 130) draw on Bourdieu (1977) to point out È that critical theory is not oriented towards empirics and, indeed, could be seen as failing researchers in this respect. In this research tradition, the actions of organization members are explained by reference to the external social ``forces ' ' that they are exposed to. Individual motivations, intentions and reasons for action are either not explored or not thought to be important. Likewise, local organizational conditions and issues as a context for action are not highlighted. Moreover the meanings attributed to local conditions and issues by organizational actors are not seen to be particularly significant. What is required is the identification the totality of the external social conditions that ``tower above ' ' individual motivations, intentions and endeavours, localized issues, actions, and events and inter-subjective meanings, group interactions and collective accomplishments. This orientation introduces a paradox concerning the emancipatory intent of critical theory, if structure largely determines agency then any overturning of repressive technological-capitalist forms of organizing does not imply that agents will be emancipated ± only that they will become subject to different structures. In grand theories structural conditions largely determine the nature of all the individual and localized action taking place ``on the ground ' '. However, structural phenomena such as social institutions, culture and the distribution of power and resources are not directly observable from the standard methods of qualitative research (i.e. interviews, focus groups and participant observation), hence the greater reliance on rational argumentation (rather than empirical research) when the view is from ``on high ' '. Moreover, grand theory is unlikely to be challenged, modified or revised following an encounter with empirical ``reality ' '. This is partly due to its self-referential nature (all concepts are internally related to each other in a comprehensive framework) and partly

What counts as ``theory ' '?

677

AAAJ 16,4

678

follows from its associated ``systems logic ' ' that sees social developmental processes as largely pre-determined. The work of both Marx and Habermas is positioned as ``grand theory ' ', according to the criteria used here. Both are concerned with repressive social structures that endure across large space/time dimensions, both see the possibility for progress through reasoned analysis and both couch their ideas at a highly abstract level. Others have concurred with this. Donaldson (1984, p. 125) comments that Marxism is a theory of society and, therefore, should not be seen as a theory of organization. (In this paper a theory of organization is a level four theory). Silverman (1985, p. 36) writes that `` . . . Habermas is known (and properly criticized) for producing vast, abstract theoretical schemes with little apparent empirical direction ' '. However, this positioning of Marx and Habermas should not be taken to imply that categorizing the body of work of a particular writer into a particular ``level ' ' is always unambiguous ± Foucault is a good example[9]. His project could be seen as a highly elaborated theory of the concept of the subject (or how people are drained of agency) at level three or as level four theory that is concerned with how modes of domination are exercised through a network of relational practices that include power, knowledge, discourse, technology and surveillance. Rabinow (1984, pp. 5-6) argues that Foucault thought that ``theory ' ' or the ``will to knowledge ' ' actually obscures the operation of power in society and so ``theory ' ' (as abstractions, first principles and utopian schemes) should be cast aside. It seems that Foucault would certainly not have seen himself as a level five theorist. On the other hand, Foucault 's analysis was not derived from empirical work either. As Sutton and Shaw (1995, p. 383) point out it is instructive to reflect as to whether the evidence provided by such theorists as:
. . . Freud, Marx, or Darwin would meet the empirical standards of the top journals on organizational research. Would their work be rejected outright, or would they be given the opportunity to go thorough several rounds of revision?

It is perhaps wise at this point to emphasise that this article does not seek to play down the undoubtedly highly significant contribution to social science of the ``grand theorists ' '. Indeed, the world was not the same place after it had absorbed the ideas of Freud, Marx and Darwin. It is difficult to imagine how any empirical research project could possibly have such an impact. Nevertheless, the starting point for this article is that empirical research has value and this paper does question the utility of ``grand theory ' ' across all areas of empirical investigation in management and accounting research. To sum up, grand theories offer understanding of (and explanations for) enduring structural aspects of experience; they create meaning and significance through expressing hope that the world can be changed for the better. How recent work relates to these levels of theorizing Recently published qualitative empirical research in the management and accounting field covers a very wide area; a comprehensive review relating all of

this work to the levels of theorization introduced here lies outside of the scope of this paper. However, some insight into the usefulness of these ``levels of theory ' ' may be gained by assessing how well the framework proposed here ``fits ' ' with recent work. This review of ``fit ' ' is undertaken through using examples of qualitative empirical research published within the last five years in Accounting, Auditing & Accountability Journal (AAAJ)[10]. Level one: metaphor An example in AAAJ of a paper that relied primarily on a central metaphor to theorize empirical understandings and events; this was the ``tree ' ' metaphor in Mouritsen et al. (2001) ± where idea of the ``tree ' ' conveys the desired associations for the emerging concept of ``intellectual capital ' ' at the research site of Skandia (i.e. ``green ' ', ``growth ' ', ``survival ' ', ``fruition ' ' and the necessity of regular ``pruning ' '). For other examples of empirical research that employ a central metaphor rather than higher levels of theorizing, see Guthrie and Parker (1999) or Llewellyn (2001). However, in AAAJ there were also articles where metaphor was being used to frame an understanding of some of the pertinent issues. For example, Parker et al. (1998), in their paper in research in universities use the metaphor of ``gatekeeping ' ' to convey the impossibility of ``bypassing ' ' the role of senior academics in making judgments on research quality. Boden et al. (1998) use the metaphorical imagery of ``Men in white coats . . . men in grey suits ' ' to create a picture of two worlds (science and management) linked through the values of impersonality, technology and bureaucratic rationality. Fogarty and Radcliffe (1999) write of accountants ' ``travel ' ' as a metaphor for the spread of the accounting jurisdiction by a mature profession seeking to extend the market for its services. Drawing on the Lakoff and Johnson (1980) work discussed earlier Reiter (1997)[11] argues that an ``independence as separation ' ' metaphor underlies discussions of auditor independence. Page and Spira (1999)[12] substitute ``underwear ' ' for ``framework ' ' in a discussion of financial reporting and accounting standard setting and demonstrate how the new metaphor of ``underwear ' ' unsettles some of the taken-for-granted understandings (around solidarity, permanence, symmetry and balance) conveyed by the dead metaphor of ``framework ' '. These two papers echo the point made earlier that the understandings conveyed through metaphor are all the more powerful by being unexamined. Despite the recognition in these papers that metaphor drives basic understanding of the social realm, qualitative researchers are still not really cognisant of the power of metaphor in the research settings they investigate nor do they fully exploit the potential of metaphor in their research accounts. Level two: differentiation Several papers in AAAJ relied on dualities or categories for theorization. Jacobs and Kemp (2002) use a very basic distinction to theorize their study of

What counts as ``theory ' '?

679

AAAJ 16,4

680

accounting practices in the daily lives of small traders in Bangladesh ± that of ``presence ' ' and ``absence ' '. The yin-yang duality is mobilized by Greer and Patel (2000) to understand core Australian indigenous yin cultural values and their colonisation by the yang orientation of accounting and finance. Funnell (1998) seeks to understand the trajectory of the new accounting history through the distinction between the narrative and the counter-narrative. Several key differences drive Broadbent and Laughlin (1998). Change is categorized as ``first ' ' or ``second ' ' order. In turn, first order change leads to ``re-orientation ' ' and second order to ``colonization ' '. In a typology that is derived from the research data, groups are differentiated as: . fully committed absorbing group; . agnostic absorbing group; and . absorbing group with colonizing intentions. Another paper that draws its key distinctions from the research setting (rather than taking them from prior theorizing) includes Lightbody 's (2000) paper that employs the categories of ``storing ' ' and ``shielding ' ' to understand the behaviour of financial managers in a church organization. Also Jones (1999) engages in another level two-type theorization when exploring alternative forms of governance and management accounting in UK National Health Service hospitals through the categories of markets, hierarchies, networks and clans. Macintosh and Baker (2002) is driven by level two-type theorizing and understands change in financial reporting through a five part typology: (1) expressive realism; (2) the new criticism; (3) structuralism; (4) deconstructionism; and (5) the heteroglossic form. As with metaphor, recognition that, in the social sciences, theorization proceeds through the creation of dualities and categories has been slow. Yet ``how the pie of experience is cut ' ' (Triandis, 2001, p. 20) can differ profoundly between individuals and between cultures, understanding and expressing this is essential in social research. Level three: concepts As might be expected for a journal named for the concept of ``accountability ' ', several articles over the last five years have sought to apply, extend, or refine this concept through empirical research that charts how ``accountabilities ' ' are being produced and reproduced in practice. Three papers have explored the relationship between ``politics ' ' and ``accountability ' '. Coy and Pratt (1998) use a case study to track political

influences on the changing nature of the concept of accountability in universities. In a similar vein, Klumpes (2001) also presents a case study that explores political interactions and the changing accountability and financial management roles of the State Superannuation Authority of New South Wales. Grendron et al. (2001) argue that state auditors in Alberta, Canada are actively involved in producing new forms of accountability that resonate directly with ``new public management ' '. Moving away from ``politics ' ' but staying with the approach that reworks the concept of accountability through exploring its interface with other concepts, Llewellyn (1998b) seeks to sharpen understanding of the concept of accountability through exploring how it fits with the related idea of ``responsibility ' ' in the context of financial reform in the UK social services. Through exploring the relationship between the concept of ``stakeholders ' ' and that of ``accountability ' ', Burritt and Welch (1997) adopt a more applied approach and develop an environmental accounting system for the public sector. Similarly, Jacobs (2000) explicates the articulation between accountability and visibility and argues that patterns of visibility were not symmetrical between government departments in New Zealand and the indigenous Maori population. Neu (2000) is also concerned with accountability and focuses on settler-indigenous peoples ' relations through mobilizing Foucault 's (1991, p. 269) concept of governmentality, `` . . . not as an overarching theory but as a `field of investigation ' . . . ' '. This description is particularly significant in the context of this paper as an example of an author explicitly adopting the ``conceptual tool ' ' rather than the ``view from on high ' ' approach. Also several papers explored concepts other than ``accountability ' '. Davie (2000) introduces the concept of ``indigenous collaboration ' ' to theorize the involvement of accounting in imperial expansion in the South Pacific. Susela (1999) theorizes standard-setting through the central concept of ``interests ' '. The relationship between the concept of ``trust ' ' and accounting is explored by Seal and Vincent-Jones (1997). Rahman (1998) seeks to extend a pluralist, decisionoriented concept of power better to inform his study of international accounting regulation at the United Nations. The concept of ``negotiated order ' ' is mobilized by Rahaman and Lawrence (2001) to theorize how agents reproduce structures. Walker and Shackleton (1998) theorize the development of the accounting profession between 1957 and 1970 through the concept of closure (while offering an understanding of ``closure ' ' through the metaphor of ``ring fence ' '). Llewellyn (1997) extends the concept of ``purchasing power ' ' and introduces the concept of ``polarized professionalism ' ' in a study of the incentives inherent in general practitioner fundholding in the UK. Finally, in terms of theorizing through a central concept, a special edition of AAAJ, ``Accounting at home ' ' (Llewellyn and Walker, 2000) was devoted to extending the meaning of the concept of money within the private world of the household. Two empirical studies (Northcott and Doolin, 2000; Pahl, 2000) took the neglected (in the context of money) ``private ' ' aspect of the public-private

What counts as ``theory ' '?

681

AAAJ 16,4

682

duality and opened it up to an investigation of how financial management operates within the home. Appropriate conceptual theorization is fundamental to research in organizations ± as organizations exist to embody practices. Theorizing through concepts is already underway in the management and accounting disciplines but more open discussion on its significance is still called for. Level four: theorizing settings As organizational theory is at level four it is not surprising that much theorization in AAAJ is concerned with organizational settings. Hines et al. (2001) follow a level four theory in their institutional analysis of exploring how a financial review board functions to meet social expectations, whenever its operations are publicly visible. Hussain and Hoque (2002) also theorize under the ``institutional banner ' '; their ``new institutional sociology ' ' perspective focuses on the relationship between organizational structures and processes and the wider social environment. Burns (2002, p. 571) distinguishes ``old institutional economics ' ' from ``new institutional sociology ' ' and argues that the former, `` . . . deals more directly with the emergence, continuity and change of institutions through time . . . ' '. Also, Potter (2002) employs another ``level ' ' of institutional theory ± ``Institutional thinking ' ' (Douglas, 1986), a theory that explores the impact of organizational constraint on individuals. Legitimacy theory was showcased in the Deegan (2002a) special edition of AAAJ, where two qualitative empirical papers (O 'Donovan, 2002; O 'Dwyer, 2002) adopt this perspective. Legitimacy theory can be seen as a variant (or a derivative) of institutional theory in its emphasis on how organizations are dependent on (and, hence, respond and adapt to) their environments. Deegan (2002b) argues that if societal expectations on the legitimacy of an organization 's operations are not met then its ``contract ' ' to continue to function is effectively revoked. Hence he reasons that legitimacy theory directly relies on the concept of the ``social contract ' '. This argument is significant in the context of this paper as it reflects the idea that higher ``levels ' ' of theorization are developed from lower ones. Level four theorization is certainly prevalent in the management and accounting areas, yet the range of theories employed is rather limited, the dominant focus is on how the environment impacts on the organization. With the exception of Potter (2002) there was no work on how organizational settings influence individuals. Nor was there any research on how organizations enact their environments or how they interact with each other on what Ahrne (1990, p. 94) dubs the inter-organizational terrain. Level five: theorizing structures Level five theorization entered into research in AAAJ in rather complex ways. For example, Latour 's (1979, 1987, 1993a, 1996) work was developed to analyse the nature and processes of science discovery and, in particular, how scientific ``facts ' ' are established. Within this specific context, it would be considered as a

``level ' ' four theory. But Latour 's ideas have been applied to theorize various kinds of organizational change across time and space ± so in its application by other empirical researchers Latour 's theory may ``jump ' ' to level five. For example, Lowe (2000, p. 84) uses Latour to provide an `` . . . understanding of the mobilisation of the casemix and related information systems at a large regional hospital . . . ' ' and, in doing so, extends the intended explanatory power of the theory to a realm outside of its initial frame of reference. Several papers isolate specific differentiations or concepts from level five type theory and ``bring them down ' ' to lower levels of theorization (see the earlier recommendation by Alvesson and Skoldberg (2000)). Broadbent et al. È (2001) take the work of Habermas down to ``level ' ' four when they theorize resistance to accounting and finance changes in general medical practice. Dyball and Valcarel (1999) draw a level two typology from Weber (the distinction between ``rational ' ' and ``traditional ' ' sources of authority) to understand accounting regulation in the Philippines. Tinker and Koutsoumadi (1997) take the level three concept of a ``commodity ' ' from Marxism and apply it to explain the present pressures on self-understanding for North American accounting students in a highly commercialized job market. From the limited review carried out in this section, it appears that the five levels of theory identified can be used to categorize the different ways that empirical researchers conceptually frame their work. Preliminary evidence indicates a primary reliance in the management and accounting areas on theorization at level three. In AAAJ, over this five-year time period, theorizing was mainly through five central concepts: accountability, politics, stakeholders, power and money. This reflects the meso nature of the work, which is mostly about practices in organizations. Research that relies on either level one or level five theorization is limited. In the case of metaphor this may be due to a relative lack of familiarity with this theoretical approach. Where empirical research draws on level five type theorizing, the tendency is to ``take the theory down ' ' to a lower level. Over the last five years, where AAAJ articles have extended discussion of a particular ``level five ' ' theorist their work is abstract rather than applied (see for example Thompson (1999) on Bourdieu; Lowe (2001) on Latour; Tinker (1999) on Hegel). The implications of defining different ``levels ' ' of theory These five ways of theorizing are being offered as conceptual tools that will be more or less useful in relation to different research agenda. Before looking at these varying research problems in more detail, this section explores some general issues that are involved when defining ``levels ' ' of theory. These levels can be seen as constituting progressively higher ``levels ' ' of conceptual framing. As discussed above, Layder (1994, pp. 13) labels ``grand theory ' ' as ``the view from on high ' ' and this paper concurs ± positioning grand theory as the most abstract (in the sense of abstracted from empirical conditions) form of theorization available to researchers ± Silverman (2001, p. 53) speaks of `` . . . data-free grand theories . . . ' '. Grand theory is also more

What counts as ``theory ' '?

683

AAAJ 16,4

684

ambitious in scope than ``lower ' ' levels of theorization, offering a meta-narrative that spans space and time. Metaphor is described here as the ``lowest ' ' level but this positioning should not be taken as indicative of its relative significance. Metaphor is the lowest as it is the most fundamental and ubiquitous form of theorizing as it ``grounds ' ' experience. Arguably, it is also the most influential. Lakoff and Johnson (1980, p. 3) point out the pervasiveness of metaphor in peoples ' ordinary conceptual systems. And in a similar vein, Lakoff and Turner (1989, pp. 62-3) argue:
. . . once we learn a conceptual metaphor, it . . . is just there, conventionalized, a ready and powerful tool-automatic, effortless and largely unconscious. The things most alive in our conceptual system are those things that we use constantly, unconsciously and automatically . . . metaphors have persuasive power over us.

So, although it is possible to see these ``levels ' ' of theorizing as becoming increasingly sophisticated and as stages on the ``ascent ' ' to grand theory, it also makes sense to see a progression from grand theory to metaphor as conceptual framing becomes more and more omnipresent in thought and language (Lakoff and Turner, 1989, p. xi). Grand theory can be described as the most highly developed type of conceptual framing, but metaphor can be seen as the primary form. This seeming paradox can be understood in the context of the relative significance accorded to ``high ' ' and ``low ' ' levels of theorization by academics and lay people. Academics, `` . . . the self-appointed guardians of theory . . . ' ' (Sayer, 1992, p. 263) tend, in their work, to privilege theorization that ``starts ' ' no ``lower ' ' than level three (conceptualisation) and may deem levels four and five only to be worthy of the ``theory ' ' label. Practitioners are unlikely to understand or explain their experience in theories that go above level three and rely primarily on metaphor, differentiation and categorization to theorize their worlds. Mouzelis (1995) argues, first, that theorization is in disarray in the social sciences and, second, that theory and empirical research are disconnected. ``Conceptual pragmatism ' ' (Mouzelis, 1995, p. 8) is proposed as a way out of this dilemma. In support of a project to achieve conceptual pragmatism he draws on the Althusserian distinction between theory as a tool/means (labelled generalities II by Althusser) and theory as an end product (termed generalities III by Althusser). With this dichotomy, what is considered as level five theory in this paper could not sensibly be used in support of empirical research. As such theorization is already an ``end-product ' ', any empirical research would only be mobilized as illustrations of what are already `` . . . law-like, universal and transhistorical social theories . . . ' ' (Mouzelis, 1995, p. 2). Although this article would judge this position to be too extreme, it is argued that in the journey from metaphor to ``context-free ' ' theorizing, theory becomes more and more a ``perfected product ' ' (Layder, 1993, p. 45) and its relationship with empirical research becomes increasingly tenuous. For although researchers can never be tabula rasa (there is no view from nowhere (Nagel, 1986)), as Alvesson and Skoldberg (2000, p. 68) point out: È

. . . too far-reaching an emphasis on pre-understanding and given frames of reference amounts to the disappearance of the object and its absorption by the subject . . .

This danger becomes more acute as one reaches ``the view from on high ' '. Empirical case studies using grand theorisation can become merely carriers of an identical prior understanding and, hence, any distinctive new themes (see ``emergence ' ' below) in empirical work can be missed, as they are pre-emptively absorbed into the pre-existing theoretical framework. This absorption of the supposed object of the empirical research is consequent on the reductionist character of ``grand theory ' '. Although grand theories construct an extended network of ideas, these are linked back to a central concept or ``entry point ' '. Amarglio et al. (1993, p. 164-5) define an ``entry point ' ' as:
. . . a concept or concepts which a theorist uses to enter into, to begin, discourse about some object of analysis. This entry can be, but need not be, an essence ± the primary truth or the primary determinant cause ± in the discourse that results . . . An entry point is a concept which will distinctively shape the asking of all questions and which will condition (and be conditioned by) all other concepts within a discourse . . . From the point of view of one school or another, the content and use of the entry point may construct this school 's attempts to define the limits of its discipline.

What counts as ``theory ' '?

685

Such a definition posits an entry point is a central organizing concept from which a discourse flows. Although level five theories set out an extended network of metaphors, differentiations and concepts, they also have a centre. For example, Layder (1993, p. 72) argues:
In the case of Marxism the centre is production and its social relations. Other structures and phenomena may be admitted but they are seen in terms of their relations to the centre.

Similarly the life world/system duality in Habermas is the nexus around which other concepts are developed and refined. At the centre, the theory can negotiate empirics purely on its own terms, hence the tendency for empirical data to ``disappear ' ' into the vortex of any grand theory. Another critique of ``grand theory ' ' has come from questioning its roots in a societal ``system ' ' level of analysis. Nowadays, human purposes are fulfilled through organizations rather then societies (Lindkvist and Llewellyn, in press). Ahrne (1990, pp. 141-2) argues that, contemporarily, an ``organizational view of society ' ' is more practically adequate than a societal perspective on organizations:
The implication of an organizational theory of society is that social processes and social change must first of all be comprehended in their organizational settings and as the effects of interactions among organizations of various constellations. The relations between individuals and society can best be understood in terms of forms of organizational affiliationmembership, ownership, citizenship, kinship and employment . . . Everyday life takes place in the realms of organizations. It does not occupy a sphere of its own . . . One cannot communicate with a system. Concepts like power, democracy, social order, are meaningless and lose their content outside particular organizational settings. Processes such as rationalization, modernization, bureaucratization, specialization, professionalization, must be theorized and investigated in organizational terms . . . There is no higher order above or beyond these organizations . . . there is only disorder.

AAAJ 16,4

686

The implication of Ahrne 's arguments is that active human projects (rationalization, modernization, bureaucratization, specialization, professionalization) cannot be analysed through theories above level four, as above this level there is only ``disorder ' '. Disenchantment with ``the view from on high ' ' has led some empirical researchers to advocate ``middle range ' ' thinking. Merton developed the idea of ``middle range ' ' theory (Layder, 1993, p. 3). And Merton (1968) used the term ``middle range ' ' in two ways. First, to indicate that theory should be developed at a level that falls between ``minor hypotheses ' ' and ``grand theories ' ', and, second, to direct the researcher towards the type of social phenomena he thought to be suitable for empirical research (e.g. recurrent aspects of social life but ones which do not entail issues of long-term social development). There is a similarity between middle range theory and the approach to theorizing taken in this paper ± in so far as ``middle range ' ' theorizing does not generally adopt an encompassing ``grand theory ' ' in advance of beginning an empirical research project. But there are also many significant points of departure. If ``hunches or insights ' ' are taken to equate to the metaphors, dualities and categorizations that structure thought and experience, it is argued here that these are actually fundamental levels of theory (see earlier discussion). Also Merton adopts a ``theory-testing through data ' ' approach. The strict separation between ``theory ' ' and ``data ' ' that this approach entails does not hold in qualitative research because of the concept-dependent character of social phenomena (see earlier section). Hence, ``theory ' ' cannot be tested or ``filled out ' ' with ``data ' ' whose meaning is non-problematic. Moreover, Merton is concerned only with ``recurrent aspects of social life ' ' and, excludes the exploration of unique events ± yet there can be much to learn from things that only happen once. The importance of analysing unique events or ``spaciotemporal specifics ' ' is well accepted in history (Parker, 2002) (perhaps because it is not so self-consciously theoretical), but not in the other social science disciplines. Also, in Merton 's approach the phenomenon of ``emergence ' ' is effectively marginalized in empirical research. In contrast, it is argued here that emergence is a significant issue in empirical research (see the discussion above and in the next section). Laughlin (1995, 1997) brought the idea of ``middle range ' ' theory into management and accounting research. And, as this paper is concerned with ``levels ' ' of theory in the same disciplines, it seems appropriate to compare briefly the two approaches. Laughlin (1995, pp. 78-9) indicates that he is using the term ``middle range ' ' in a rather different way than Merton:
The following . . . presents an argument for what can [be] loosely referred to as ``middle range ' ' thinking. It is ``middle range ' ' since it argues a case for taking a mid-point on each of . . . three continuums (theory, methodology and change) . . . Merton 's ``middle range ' ' thinking refers more to an initial limitation of the social focus for the discovery process, not to the nature of theory, methodology and change.

The approach taken in this paper (see above and below) is broadly to classify theorizing into five levels and to argue that each ``level ' ' has particular strengths

with respect to empirical research issues. For example, change through emergence would be likely to be theorized at levels one, two or three, whereas structural phenomena are usually analysed at levels four and five (albeit that ``structures ' ' exist at micro, meso and macro levels ± see next section). The ``middle point ' ' on the levels presented here is ``conceptualization ' ' ± this is not thought to be the ``best ' ' level of theorizing for all phenomena. However, in so far as management is conducted in organizations (and, therefore, at the ``meso ' ' level of analysis) and concepts relate closely to practices (and management is predominantly about practices) then, often, conceptualization is the most appropriate form of theorizing in the management and accounting disciplines. Having considered some of the implications of these five levels of theory, the next and the final sections now link them more comprehensively to different empirical issues and research methodologies. How ``levels ' ' of theory relate to different empirical issues Different theories (and ``levels ' ' of theory) in empirical research reflect the contestation of meaning and significance in the social sciences. In consequence, some academics may pursue their entire research careers with an ideological allegiance to one particular position ± usually a level five type (meta)theory. But where a ``choice ' ' is made between the different levels of theory proposed here, that choice is likely to be conditioned by the ``type ' ' of phenomena being researched. In general, as ``lower ' ' levels of theorization give way to ``higher ' ' ones, changes in emphasis can be discerned as different kinds of empirical issues and problems are addressed. Table II gives an indication of these issues. Reasons, actions, agency, social processes and social production are better understood through metaphor, differentiation, categorization and conceptualization, while causes, structures, social ``forces ' ' and social reproduction become the focus of level four and level five theories. Similarly, ``micro ' ' phenomena such as motivation, intention, social interaction and the presentation of self tend not to be well captured in any theorization above level three, while macro phenomena such as class, gender and power relations, culture and the distribution of resources, are not generally theorized below level
Level One Two Three Four Five Theory Metaphor Differentiation Concepts Settings Structures Empirical issues ``Micro ' ' reasons, actions; social productiona ``Micro ' ' social processesa ``Meso ' ' agency ± how individuals make things happen through resources The social organization of relationships between individuals, organizations and environments Class, gender, power relations and the distribution of resources

What counts as ``theory ' '?

687

Note: a Metaphor and differentiation also permeate the understanding of human experience across the levels of theorization outlined here

Table II. The links between different levels of theorization and empirical issues

AAAJ 16,4

688

four (although the concepts underlying these relations may be theorized at level three). The theorization of meso phenomena such as practices (or how people make things happen in particular contexts) is generally at the level of concepts. The unit of analysis also differs between the different levels of theory. The individual and his/her subjective experience are generally theorized through metaphor and differentiation. Categorization is most usually focussed on social groups or collective phenomena. The analysis of organizations rarely begins below level three (although Morgan 's work is a notable exception as it is concerned with how people experience organizations). Level four theorization is most commonly concerned with the internal differentiation of organizations or their relationship with their societal environment. Level five theories are generally at a societal level but, may, as argued earlier, claim to transcend particular societies and periods of history. The ``claims ' ' made by theorists at these various levels of conceptual framing also differ. Table III gives an outline of these claims. Level five theorists tend to assert universal a-historical applicability across large time-space dimensions. Their claims are couched in terms of explanation rather than understanding and hence rely on modes of abstraction rather than contextualization. Having distinguished between explanation and understanding, it should be pointed out that ``explanation ' ' takes on a different meaning in social science than natural science. As Czarniawska (1997, p. 18-19) argues, in the natural sciences explanation is achieved when a phenomenon is located as an instance of a general law, but in the social sciences this ``law ' ' has arisen because of the existence of a human project[13] (i.e. rationalization, modernization, bureaucratization, specialization and professionalization). Nevertheless, despite this difference in the meaning of the ``law ' ' in question, level five theorists still comprehend themselves as aiming for something closer to the natural science ``law-seeking ' ' (Sayers, 2000, p. 137) model of what theory ``is ' ' (see earlier discussion). Level four modes of theorization are more embedded in time-space settings than level five but they, generally, still aim for explanation rather than understanding. Level three concepts (because of the link to practices) are rooted in localities that provide the contextualization necessary for understanding (much empirical research theorized through a concept such as accountability, for example, is concerned with the localized
Level One Two Three Four Theory Metaphor Differentiation Concepts Theorizing settings Theorizing structures Claim ``Grounds ' ' experience ``Cuts up ' ' experience Explicate practices Explains relationships between social phenomena in context Explains universal, a-historical and large scalea dimensions of social life

Table III. The claims made at different levels of theorization

Five

Note: a Structures also feature at the interactional and organizational levels of analysis

meaning of the concept in question). Metaphors and differentiations contextualize personal experience. Contextualization is one way of understanding experience; another way is to distinguish between ``local ' ' and ``global ' ' explanations (Fairclough, 1995, p. 7). A ``local ' ' explanation makes reference to a personal project (obtaining a job); whereas a ``global ' ' explanation extends to a human project (specialization in the labour market). In terms of how theory changes and develops at these different levels, certain level one metaphor and level two differentiations can assume the endurance claimed for level five theories. For example, the metaphor ``organizations are cultures ' ' and the duality of ``objective/subjective ' ' have had considerable stability across time and place. This observation points to the enduring structural aspects of interpersonal and interactional experiences. As Sayer (1992, p. 249) argues, structures do not exist only at organizational and societal levels but also permeate relationships between individuals. Structures (as roles, rules and regulations) exist throughout the levels of theory analysed here, but the opportunity of changing them lies in practice. This observation indicates that these five ways of theorizing may be better thought of as ``points on a circle ' ' (that juxtapose metaphor and ``the view from on high ' ') rather than ``rungs on a ladder ' ' (that position level one and level five as far apart as possible). The circle image would position level three concepts ``away ' ' from levels one and five, and this positioning would capture the more dynamic nature of concepts. As concepts theorize practices their meaning is more fluid. There are also perspectives from which it may be more helpful to picture the theoretical ``levels ' ' as points on a ``spiral ' '. The spiral image brings out the ideas of continuity and growth between metaphor-differentiation-concepts-theorizing settings-theorizing structures, whereas the idea of ``levels ' ' suggests disjunctions. The relationship between ``levels ' ' is often recursive ± in the sense that higher levels incorporate the levels ``beneath ' ' them ± and the spiral metaphor also helps bring out this recursive element. If theorization at levels four and five is more usually appropriate for patterned regularities (excepting the structural aspects of personal experience), lower levels of theorization are more helpful when empirical research is concerned with ``emergent ' ' and ``discontinuous ' ' phenomena. The rest of this section will explore ``emergence ' ' and ``discontinuity ' ' in some detail, as they have not, so far, received much attention in the management and accounting disciplines. A metaphor that assists in understanding ``emergence ' ' and ``discontinuity ' ' is the social ``landscape ' ' (Ahrne, 1990, pp. 22-7). Ahrne 's (1990, p. 23) metaphor suggests that the world is not only characterized by enduring structural conditions, patterned relationships and interdependencies but also by processes that are ending (and beginning) and phenomena that co-exist but are independent:
The social world is . . . like a rough landscape which is the result of long chains of development: a mountain crumbling away, a forest growing, a desert spreading, a motorway under construction. A landscape is the co-incidence of several phenomena with different origins and nature and, in fact, with little in common with each other except that they happen

What counts as ``theory ' '?

689

AAAJ 16,4

690

to be near each other at a certain point in time . . . A picturesque scene looks perfectly integrated: a few houses, a river with a bridge over it, a meadow, a forest, a mountain, flowers, a horse, some clouds, an old rusty tractor. But this apparent integration is illusory. Basically these objects function independently of each other. There are interdependencies among them, however. The flowers need soil to grow in . . . The bridge is there because of the river. But it is impossible to understand the existence of the house with the help of other objects in the landscape . . . The construction of the tractor has nothing to do with the horse. The landscape is not a unit.

It is just as important for qualitative management research to track ``emergence ' ' (the forest growing and the motorway under construction), ``endings ' ' (the mountain crumbling away) and discontinuities (the tractor and the horse) as it is to identify enduring conditions and patterned relationships. As Sayer (1992, p. 144) arues, many matters in social science are contingent, i.e. two or more phenomena may co-exist at any point in time with a potential connection between them that is neither necessary nor impossible. It is possible that the horse and the tractor can be ``understood ' ' as phenomena belonging to the activity of farming, equally the horse could be wild (and separated from its herd) and have no connection with the tractor. If this landscape were the focus of a case study it would be important to understand which features in the landscape were related and which were not. Also, as Ahrne (1990, pp. 23-4) points out, the landscape features several phenomena that just happen to co-exist (e.g. the design of the house, the growth of the flowers, the cloud formations and the rust on the tractor). Explaining the landscape in a case study would necessitate insights from theories across a broad spectrum (e.g. architecture, botany, geology and mechanics). Adding people to a natural landscape creates a social one. Social landscapes also encompass diverse structures and processes with different natures and originsat individual, group, organizational, and societal levels of analysis (e.g. nation states, capitalist enterprises, cultural groups and individual motivations). Moreover, these social elements are played out in a material time-space landscape of buildings, machines, ``nine-to-five ' ' time periods and communicational networks. Attempting to capture the totality of this diversity with reference to a single monolithic ``theory ' ' is usually misguided. Insights from several theories at different levels of analysis are appropriate for social landscapes. The phenomenon of emergence points to ``conditions of possibility ' ' (Foucault, 1980, p. 243) in social landscapes. As Markovic (1974, pp. 10-11) points out, although the ``seeds ' ' of the future are living in the present, these ``seeds ' ' may or may not emerge ± they require active cultivation:
Past and future are living in the present. Whatever human beings do in the present is decisively influenced by the past and by the future . . . the future is not something that will come later, independently of our will. There are several possible futures and one of them has to be made.

Awareness of emergent powers enables the recognition that `` . . . we could become many things which currently we are not . . . ' ' (Sayer, 2000, p. 12).

Consideration of the current conditions of possibility should enable a more informed position on the advocacy of change. If the seeds of a desired possible future are already in the present landscape then success in working towards this change is much more likely. Beeson and Davis (2000, p. 182) define emergence in the context of change, `` . . . more dynamically as a property of novel behaviour produced at any point in the system . . . behaviour which is not built-in, pre-specified or predictable '. ' Layder (1994, p. 157) sees emergence as a ``property ' ' that cannot be read off from either the novel behaviours of individuals or structural resources; he views it as created through `` . . . the dynamics of encounters which draw together the inputs of several actors, producing an unanticipated result ' '. Elements that have been discontinuous (e.g. the tractor and the horse on the social landscape-see above) may come together at a certain point of time in a common venture (e.g. farming ± see above). Moving back from emergence ± as manifested in behaviour or interaction ± to the emergent powers that may be latent in a system, Sayers (2000, p. 12) defines such powers as follows:
. . . powers which can be either activated or remain dormant . . . the world is characterised by emergence, that is the situations in which the conjunction of two or more features or aspects gives rise to new phenomena, which have properties which are irreducible to those of their constituents, even though the latter are necessary for their existence. The standard physical example is the emergent properties of water which are quite different from those of its constituents, hydrogen and oxygen.

What counts as ``theory ' '?

691

Polkinghorne (1988, p. 2) argues that new social realities emerge through the integration of previously isolated elements and these integrated parts perform differently dependent on their specific organization, structure and configuration. An example of an historical study of emergence from AAAJ is Davie (2000). As discussed above, the level of theorization here is at level three through the central concept of indigenous collaboration between the Fijian chieftaincy and the British imperialists. This collaboration was not predictable in advance; its dynamics, forms and modes of engagement were accomplished through active agency. Specifically, the paper brings out the contingent role of accounting in the emergence of the accommodation between the representatives of the British state and their indigenous allies. Another instance of a paper that focuses on ``emergence ' ' is the Mouritsen et al. (2001) paper (referred to above). This article shows how management at Skandia actively mould the emergence of peoples ' understanding and experience of intellectual capital concept through making it in the image of a tree. In summary for this section, as the type of empirical phenomena changes, so does the ``level ' ' of theorization. Change and emergence are better understood through metaphor and differentiation, practices are most adequately theorized through concepts and patterned regularities through the highest levels of theory. At higher levels of theory, contextualization and understanding tend to be replaced by abstraction and explanation.

AAAJ 16,4

692

How ``levels ' ' of theory relate to different methodologies Just as the ``type ' ' of phenomena being researched will condition ``choices ' ' between different levels of theory, there are also links between different theories and methodologies. These connections may be quite close; methodologies (and espoused theories) reflect the epistemological and ontological assumptions of researchers (Llewellyn, 1992). Hence research methodologies and levels of theory are intertwined[14]. Beginning at ``lower ' ' levels, ethnographic studies aim to capture the thoughts, beliefs, values and motivations of organization members through lengthy ``immersion ' ' in their worlds. The method was developed within anthropology and seeks the ``native 's point of view ' '. When the ``meaningful worlds ' ' (Rock, 1979) of social groups (rather than individuals) are the object of the research, ethnographies may be conducted under the banner of symbolic interactionism. As such, and whether focused on individuals or groups, ethnographies are best theorized through metaphor and differentiation. However, many ethnographers also seek to understand how beliefs and values of individuals or groups relate to action and how such agency shapes organizational practices; with this agenda, theorization through concepts will be most appropriate. The assumption that ethnographies offer ``thick descriptions ' ' (Geertz, 1973) of everyday life can be understood to imply that they are a-theoretical[15]. This assumption is mistaken on several counts. First, as argued earlier, everyone needs ``theory ' ' to go about his or her everyday lives. Therefore, ethnographies will be informed by the theories that organizational members use to address their problems. Second, a ``description ' ' of any phenomenon already harnesses certain theories and marginalizes others ± as Czarniawska (1997) points out, if, for example, individual decision making is described, new institutionalism is hardly likely to be drawn on to explain. Third, ethnographies expose the inner workings of aspects of everyday life that everyone assumes and, therefore, usually ignores (Strong and Robinson, 1990, p. 8). To ``expose the inner workings ' ' is to theorize. Much famous ethnography has been highly theoretically informative; Goffman 's (1961) concept of the ``total institution ' ' was established through this method. Ethnomethodology, like ethnography, is also concerned with everyday life and a strict ``line of demarcation ' ' cannot be drawn between them. If there is a difference in emphasis, it is that ``lived experience ' ' (rather than the thoughts, beliefs, values, motivations and meanings that permeate experience) is of primary interest in this research tradition. ``Doing ' ' rather than ``thinking ' ' is the domain of ethnomethodogy. Ethnomethodologists aim to uncover the underlying expectations and implicit rules that govern everyday life; they are also concerned with local knowledges, social orders, membership and accountability (see Alvesson and Skolberg, 2000, pp. 39-40). Incongruity and È deviance (understood as departures from commonsense ideas, implicit rules, local practices and social conventions) has usually been analysed from an ethnomethodological standpoint ± as it is often only through incongruent or

deviant behaviour that taken-for-granted assumptions are uncovered (see Goffman, 1959). This understanding of ethnomethodology brings out its links to practices ± albeit that the ``practices ' ' under consideration are usually located in the ``micro ' ' world. (For example, ethnomethodologists ' interest in accountability is focussed on how individuals account to each other for their actions on a face-to-face, day-to-day basis rather than, for example, how organizations institutionalise systems of accountability). Because of the focus on ``doing ' ' (including links to the rules and conventions that govern ``doing ' '), the theoretical framework for ethnomethodological studies is usually at level three ± conceptualization. Perhaps the most common starting point for a qualitative researcher in the management and accounting disciplines is to conceive of her or his project as a case study ± where this case study involves field research that brings the researcher into direct contact with organizational participants (Ferreira and Merchant, 1992). Case studies overlap considerably with ethnographies (and ethnomethodologies) but the work of Yin (1991, p. 13) has popularized an understanding of the case study as appropriate `` . . . when the focus is on a contemporary phenomenon within some real-life context ' ' (emphasis added). Contextualization (rather than abstraction) is also characteristic of ethnographies, but a particular concern with ``settings ' ' (as the conditions and constraints under which social action is organized) implies that case studies[16] are most likely to be theorized at level four. Case studies of ``settings ' ' are usually carried out through field research. But, as Czarniawska (1997, p. 64) points out, case studies do not always equate to field studies ± indeed, the case study method has prototypes in history, medicine and law. Where case studies focus on particular individuals or events, theorization will be at the lower levels one, two or, possibly, three. Scapens (1990) called for more use of the case study method to understand management and accounting practice but case studies are still an under-utilized methodology for exploring and deriving theories of agency. As discussed above, ethnomethodology is about ``doing ' ' but its emphasis tends to be on faceto-face encounters rather than organizationally embedded practices. In terms of the above example on ``accountability ' ', ethnomethodology is about accounting to each other; case studies are more appropriate vehicles for exploring accountability practices in organizations. Because ``theory ' ' has been equated with levels four and five only (and these levels deal with settings, relationships and structures), social science lacks theories of practice. Practice is seen here as equivalent to agency. Agency is about how things are accomplished ± as opposed to action, which is about what is done. Child 's (1972) concept of strategic choice can be seen as a theory of agency in this sense, as can Giddens ' (1984) structuration theory (both discussed above). Bourdieu 's (1977) theory of practice also seems to be about agency and there are similarities between Bourdieu 's concept of habitus and Giddens 's of structure but, as Layder (1994, p. 144) points out:

What counts as ``theory ' '?

693

AAAJ 16,4

. . . Bourdieu views human behaviour in a more mechanical and ``determined ' ' manner than Giddens, insofar as it is always conditioned by the habitus . . . In Giddens 's work the human actor is allowed much more freedom to be transformative and creative within the social environment.

694

Bourdieu focuses too much on constraint for his theory to be truly explanatory on agency. Generally, understanding of agency is rather embryonic ± as compared to that of structure. Archer (1996, p. xii) captures the necessity of an understanding of both structure and agency both in social theory and social life:
. . . it is part and parcel of our everyday life to feel both free and enchained, capable of shaping our own future and yet confronted by towering, seemingly impersonal constraints. Those whose reflection leads them to reject the grandiose delusion of being puppet-masters but also to resist the supine conclusion that they are mere marionettes then have the same task of reconciling this experiential bivalence, and must do so if their moral choice is not to become inert or their ``political ' ' action ineffectual. Consequently in facing up to the problem of structure and agency social theorists are not just addressing crucial technical problems in the study of society, they are also confronting the most pressing social problem of the human condition.

Action research appears to be a promising research methodology for work on agency but its potential in this respect has not been fully realised. Reason (1988, p. 1) describes the ``action ' ' approach as a form of co-operative inquiry where the research is `` . . . with and for people rather than on people ' ' (emphasis in the original). The distinction between researcher(s) and organizational member(s) is blurred in action research. The intention is that all involved should contribute to deciding the research agenda (usually an organizational problem), conducting the inquiry, and interpreting the results. Argyris et al. (1985, p. 82) make it clear that the action researcher (or in Argyris 's terminology, the action ``scientist ' ') should be fully aware of the theories that guide practice and should aim to improve these:
Becoming an action scientist involves learning to reflect on reflection in action, making explicit the theories-in-use that inform it and learning to design and produce new theories-inuse for reflection and action.

Arygris 's ideas on action research imply a ``grounded ' ' research methodology (see below) with theorization at levels one, two and three; such an approach should be fruitful in generating theories of agency. However, the action research approach has been criticised both for adopting positivistic methods and for being closer to consultancy (or journalism) than to academic work (Gummesson, 1988, p. 99). Perceptions of a correspondence between action research and consultancy, however, do not follow from tracing the origins of the approach. When Lewin (1946) developed the methodology he saw it as triggering change through public inquiry into the pressing social problems consequent on the Second World War; he also conceived of the approach as creating democratic communities who worked towards formulating theory that addressed practical issues (Dickens and Watkins, 1999). Dickens and Watkins (1999, p. 131) point

to how action research can be revived as a methodology, `` . . . to improve and to involve . . . ' ' and, in this context, the methodology can reconnect with its roots in social inquiry. Up until relatively recently it could be argued that the study of language and discourse (or talk and texts) had been neglected in social (and management) research. This is no longer the case. Interest in semiotics and discourse analysis (see below) is now firmly established. Semiotics is the study of signs; Pierce (1955), among others, drove its development. Pierce differentiated three ways in which signs can ``stand for ' ' referents. First, an icon ``stands for ' ' its referent by resembling it (for example, a statue, painting or photograph represents a person). Second, an index ``stands for ' ' its referent by indicating it (for example, a thermometer shows temperature, smoke is a precursor of fire, music is used to evoke particular emotions and the leather jacket signifies youth and rebellion). But in the third case of symbols, the ``standing for ' ' relationship is purely arbitrary. In language, neither the written word nor the spoken sound of the symbol ``cat ' ' resembles (or even indicates) its referent. Saussure (1974) has been influential in arguing that a word (or signifier) derive its meaning not from its referent (or signified) in the world but from its relationship with other words in a language; he uses the term ``sign ' ' in an equivalent way to Pierce 's ``symbol ' ' (i.e. Saussure is concerned with language as a series of arbitrary signs that become meaningful in relation to each other). In Alvesson and Skoldberg 's È (2000, p. 271) words, semiotics studies, `` . . . the autonomy and ambiguity of language in relation to a non-linguistic reality . . . ' '. Work that explores the significance of signs has grown consequent on the increasing contemporary dependence on signs for accessing experience. In the main, people no longer encounter the world directly; instead the world is mediated through signs. Representations of the world in television, computers, newspapers, and telecommunications now substitute for direct experience. Fairclough (1995, p. 4) argues that texts[17] in contemporary society are multi-semiotic (in terms of the Pierce 's differentiations, discussed above); he uses the example of television ± a medium that primarily uses language but combines it with visual images, music and sound effects. Ahrne (1990, pp. 1-2) captures the power of television ± as a multi-semiotic text that (re)presents the world:
In your living room you can follow in full detail hour by hour the unfolding of a hijacking drama in an airport of some unknown city, but you do not know what your next-door neighbour is doing . . . You are brought nearer to what seems to be the centre of events but this only heightens your feeling of powerlessness. Your position as an onlooker is even more pronounced. With your remote control you can chose among a number of TV channels but you have no way of interfering with what is going on. The ratio between what you know and watch on the one hand and what you are actually able to do runs higher and higher.

What counts as ``theory ' '?

695

When Ahrne wrote this in 1990 he was expressing the ``one-way ' ' impact of signs on the passive, isolated and somewhat impotent observer (who can watch far-off events but may not connect with her/his neighbour). But, since this time, the Internet has introduced an interactive capacity that is further binding global

AAAJ 16,4

696

space/time together through a two-way dialogue based on signs (Featherstone, 1997, p. 7) Because of the present importance of signs (particularly language) in the construction of meaning and experience, semiotics is now a prominent research methodology. Semiotics differs from much qualitative research in that it is not primarily concerned with the substance of what is being conveyed through language but with its nature of expression through the re-articulation of signs (Bryman, 1989, p. 223). Research in the semiotic tradition studies how language is used to effect particular objectives. New labels may be attached to entities to re-position them, for example, in the UK the ``Open ' ' University was named to connote freedom of access and this naming created a new understanding of other universities (i.e. that they may be relatively ``closed ' '). Equally the ``lack ' ' of a language label can render action impossible, for example until ``racially motivated crime ' ' entered the vocabulary of the police, crimes of this nature could not be investigated. Particular modes of expression may be adopted to promote certain interpretations, for example, to argue ``a hostile political climate is impelling change ' ' implies that change is inevitable and externallydriven. Latour (1993b, p. 131) declared that, ``Semiotics is the ethnomethodology of texts ' '; implying that ``doing ' ' is through language. Semiotic research has ``moved up ' ' the levels of theory outlined here. Early work focussed mostly on micro ``talk ' ' or conversation analysis (i.e. focussing on how ``turn-taking ' ' proceeds, on who controls the introduction of new points of argument, on who interrupts who, on who hesitates and so on, see Reissman, (1993) for the narrative analysis of conversation). This early research was on how language (as words and sentences) is used in the ``interaction order ' ' (Goffman, 1983) ± and as such the level of theorization was mostly at levels one or two. Discourse analysis moves from linguistic units (words and sentences) to how larger texts are mobilized in social practices. Potter and Wetherell (1987, p. 9) analysed how language (as discourse) is a medium for action. They were concerned with both consistency and discontinuity, aiming to understand how the very different accounts given by people of the ``same ' ' phenomenon were linked to the very varied things that they were attempting to accomplish. In other words, ``accounts ' ' were understood as constructed from various action orientations (Potter and Wetherell, 1987, p. 183). As the discourse analysis of Potter and Wetherell is primarily concerned with the relationship between language and practice, level three theorization is most appropriate. Fairclough 's (1995, pp. 36-7) critical discourse analysis, however, is clearly formulated to answer macro questions through theorizing at levels four and five:
. . . how can it be that people are standardly unaware of how their ways of speaking are socially determined, and of what social effects they may cumulatively lead to? . . . What I suggest . . . is that we can begin to formulate answers to these and other questions, and to develop a theoretical framework which will facilitate researching them, by focusing attention on the ``social institution ' ' and on discourses which are clearly associable with particular institutions rather than on casual conversation, as has been the fashion . . .

Semiotics has largely distanced itself from the representativity of language ± in the sense of analysing the content of language for the insights, experiences and factual information it conveys (Alvesson and Skoldberg, 2000, p. 208). È However, other contemporary research methodologies are still very concerned with authenticity. Historiography, for example, seeks to set up criteria for the evaluation of data. Alvesson and Skoldberg (2000, p. 69) refer to these criteria as ``source È criticism ' '[18]. Traditionally, historians have focussed on the credibility of their sources of evidence and they have been concerned with unique events, ``lived experience ' ', individual leaders of change and the beginnings and endings of series of events (Funnell, 1996, 1998). Historians are also concerned with linkages between events and, in assessing the significance (or otherwise) of these connections, they engage in counterfactual analysis (Carson and Carson, 1998). If a particular event had not occurred, what would have been the impact on subsequent ones? If the answer to this counterfactual seems to be ``very little ' ', then the event under consideration is not thought to be causal ± or of much significance in the course of history. Perhaps because of these issues around the evaluation of data, history has been the least theoretical of the social sciences, employing narrative description and interpretation (Polkinghorne, 1988, p. 37) rather than explicit conceptual framing. Histories are concerned with the relationship between the present and the past; they can either trace the origins of the present in the past or identify the outcomes of the past that survive into the present (Parker, 2002). But if historians had a methodological imperative it was historicism ± to see the past `` . . . from the standpoint of its own time-bound context and beliefs ' ' (Parker, 1997, p. 124). This injunction implies theorization at levels one and two but, traditionally, methodologies in history have been ± at least overtly ± rather divorced from theory. Indeed, Funnell (1996, p. 45) charges traditional history as being `` . . . obsessively empirical with an overwhelming concern for the `facts ' . . . ' '. Latterly, some ``new historians ' ' have challenged history 's presumed distance from ``theory ' '. They argue against any claims to factual and disinterested accounts of the past and advocate histories theorized through critical social theory (see Funnell (1998) and Parker (1999) for reviews of this trend). If historians assume this particular theoretical position they are, usually, ``viewing from on high ' ' ± level five. In contrast, Sharpe (1991) supports ``history from below ' ' as an investigation of the everyday lives of ordinary working people ± pointing to theorizing at levels one, two and three. Others have argued for plural, interdisciplinary approaches to the study of history, questioned the functionalist assumptions of historical progress or called for more analysis of the temporal underpinnings of historical experience (see for example Miller et al. (1991), Napier (2001) and Parker (2002) respectively). These campaigns clearly indicate desires for more theoretical input into histories but leave the appropriate level of theorization open and dependent on the problem under consideration.

What counts as ``theory ' '?

697

AAAJ 16,4

698

At whatever level the theory is at, any research project will be theoreticallyinformed; this does not mean that it cannot also be theoretically informative (Sayer, 1992, p. 144). This is particularly the case in management and accounting ± as they are practical disciplines. Practice may lead theory and theory may be developed from practice. As Ferreira and Merchant (1992) point out, as workers strive to solve immediate problems, theories are worked out ``on the ground ' '. Partly for this reason, the theoretical framework that drove a research project initially may differ from the theory that emerges at the end of the work, after practice has been conceptualized. The classic work of Glaser and Strauss (1967) argued the case for theory that is ``grounded ' ' in data (see also Parker and Roffey (1997) for a discussion of the relevance of this work to research in accounting and management). Alvesson and Skoldberg (2000, p. 13) È point out the very close connection between qualitative research and grounded theory and present Turner 's (1988, p. 112) argument that, in qualitative work, there is no real alternative to the grounded theory approach. This does not imply that all qualitative researchers follow the strict canons of Strauss and Corbin (1990) or, indeed, utilise the methodological concepts of Glaser and Strauss (1967) but it does press the recognition of the theoretically-informative nature of ``data ' '. This analysis of levels of theory points to the greater likelihood of empirical research generating theory at levels one, two and three, than at four or five. The types of techniques used in qualitative research (interviews, focus groups, participant observation and archival work) tap into issues of agency and actors ' understanding rather than structure and systems. Hence ``grounded theory ' ' may generate new organizational metaphors, typologies, categorizations and concepts; it not as likely to be fruitful with respect to theory that is concerned with large scale relationships and regularities. The primary reason for this is that the structures, institutional and social issues at stake at level four and five may be opaque to organizational actors. People can be subject to social structures without being subjectively aware of them. Cause and effect relationships may operate outside of individual cognition. Actors may not feel themselves to be, for example, imprisoned within a bureaucratic hierarchical system that maintains power relations through surveillance or working in an organization that seeks external legitimacy through conforming to environmental reporting requirements. But that organizational members do not conceive of their own experience in this way does not necessarily invalidate such theories. Nor, indeed, do individuals have to be aware of the structural social forms that are analysed at level four to be engaged in reproducing them. Actors may participate in performance appraisals, for example, and be theorized as reproducing systems of surveillance in organizations while they see themselves (or theorize themselves) as engaged in, for example, furthering their own careers. A multiplicity of legitimate theoretical explanations can be offered for social phenomena and these explanations may hold at different levels of analysis (individual, group, organizational or societal). Moreover, monolithic theories, with their roots in the

nineteenth century, no longer seem to capture the complexity of life in the twenty-first; Czarniawska (1997, p. 2) comments as follows:
Companies meddle in politics while public administration enters the market, the global seeks to control the local while the local constructs and subverts the global, and people go networking in virtual meeting rooms. Citizens become customers, employees become economic citizens, public becomes private, and private goes public as identities multiply. All this seems to provoke a variety of reactions from indignation to fear to celebration.

What counts as ``theory ' '?

The world is now characterized by the inversion of what previously seemed to be ongoing, the intermingling of what previously seemed to be separate and the expansion of what previously seemed to be limited. Multiple layers and overlaps emerge ± necessitating many different conceptual frames both to give insight into this intricacy and to guide how people should react. Concluding comments There have been considerable changes in the type of empirical research carried out in the management and accounting disciplines over the last 20 years. Although quantitative work informed by positivism still dominates, qualitative research using interpretive methodologies has become increasingly influential. In consequence, there have been changes to research design (as, broadly speaking, case studies have replaced surveys) and shifts in methods (as interviews, focus groups and participant observation have been substituted for structured questionnaires). The way that research accounts are constructed has also changed as narratives displace analyses. But through all of these methodological (and method) developments the view of ``what theory is ' ' has not changed. Qualitative researchers in the management and accounting disciplines may have moved away from hypothesis testing, may be sceptical about ``operationalizing variables ' ' and may not expect theory to be predictive. But the mainstream view of ``theory ' ' among qualitative researchers is still that it represents an observed reality and explains regularities and relationships between empirical phenomena. This idea of theory is misguided in so far as it ignores the interdependence between theory and data in the social sciences. It is also too narrow to be adequate for all aspects of qualitative empirical work. Across the research included in the review there were no psycho biographical studies nor were there any papers that tracked the impact of specific individuals or groups of individuals on organizational processes or structures. Also, with the exception of the papers mentioned above, there was little work on emergence, on phenomena that were ending, or on unique events. This is surely connected to the lack of adequate theorising on agency and change that, in turn, follows from the view of ``theory ' ' as explaining regularities in social life. In the terms outlined in this paper, ``what counts as theory ' ' has been restricted to levels four and five. The lower levels (one, two and three) have not ``counted ' ' as theories and, yet, these types of conceptual framing are essential to understanding and explaining action, agency, emergence and change. Moreover, the lack of theorizing at levels one, two and three has led sometimes to an inappropriate

699

AAAJ 16,4

700

use of levels four and five theories to understand and explain action, agency, emergence and change. The range of empirical phenomena that can be usefully explored in the management and accounting disciplines will remain restricted until ideas on ``what counts as theory ' ' are expanded. Through introducing these five levels of theory, this paper has sought to make a contribution to a fuller, more transparent and more self-conscious understanding of ``what theory is ' '.
Notes 1. This paper considers the management and accounting disciplines together in terms of theorization for empirical research. Parker (1999) has argued that the two areas have much in common (a focus on planning, control, budgeting and decision making) and has provided a convincing rationale for assuming a joint research perspective. 2. ``Contextualization ' ' is understood here as a hermeneutic endeavour that relates parts to wholes, whereas explanation locates a phenomenon in a trajectory of events. 3. For an extended discussion of some of the issues raised in this section see Sayer (1992, pp. 45-84). 4. See also Alvesson and Skoldberg (2000, pp. 58-9), for a discussion of meaning and È significance in hermeneutics. 5. The concept-dependence of social phenomena should not be taken to imply that concepts have no material aspects. The meaning of the concept of accountability may be socially constructed but structures of accountability in organizations (e.g. hierarchy, roles, assignments and reports) have a materiality external to the thought processes of organizational members. 6. For a notable exception to this neglect, see, Argyris and Schon (1974) who analyse ``theory È in practice ' ' and contrast agents ' espoused theories with their ``theories-in-use ' '. 7. The metaphor of ``levels ' ' is not unproblematic. As discussed in the section on how different ``levels ' ' of theory relate to different empirical issues, the metaphor of the circle and the spiral also capture aspects of what is being conveyed here ± as is usual in social science many different (and often contrasting) metaphors are required to explicate a new idea or phenomenon. 8. See Giddens (1987, p. 60) for further discussion on dualisms and dualities. 9. Others have commented on the difficulty of categorizing Foucault 's work ± see for example Silverman (1985, p. 71). 10. This review is limited in several ways. First, it does not claim to be completely exhaustive in its coverage of articles employing a qualitative research methodology published in AAAJ in the last five years. Second, it does not imply that all the papers discussed can be unambiguously classified in terms of the ``level ' ' of theory used. Third, no attempt is made to determine the numerical proportion of work falling into each of the ``levels ' '. 11. Reiter (1997) is not an article based on empirical research but is included here as it draws on Lakoff and Johnson (1980), a book quite extensively discussed in this paper. 12. Page and Spira (1999) is also not a article based on empirical research but is included in this review as its insights into metaphor are very relevant to the themes of this paper. 13. Explanation in the social sciences locates phenomena in a human project but this should not be taken to imply that all aspects of this project are intentional. Many of the consequences of the project will be unintended because of interactions with other ``projects ' '. The ``active ' ' phrase of the project may be over, hence what remains are the `` . . . remnants of human efforts ' ' (Ahrne, 1990, p. 136).

14. A complete review of all qualitative methodologies lies outside of the scope of this paper, but this section sketches some associations between levels of theory and methodologies. 15. The popularity of the term ``thick description ' ' in relation to ethnography can be taken to imply that ethnographic studies are a-theoretical (i.e. that they are devoid of theory; their value lying in the richness and subtlety of the observations described). 16. Where the ``contemporary phenomenon ' ' of interest is an individual or a social group theorization will be at levels one or two. 17. ``Texts ' ' are understood here to encompass the written word, the spoken word and also readings of social practice. 18. See Alvesson and Skoldberg (2000, pp. 69-75) for details of four canons of source criticism: È remnants/narrating sources; bias; distance/dependence; and empathy. References Abbott, A. (1988), The System of Professions, The University of Chicago Press, Chicago, IL. Ahrne, G. (1990), Agency and Organization, Sage, London. Alvesson, M. and Skoldberg, K. (2000), Reflexive Methodology: New Vistas for Qualitative È Research, Sage, London. Amarglio, J., Resnick, S. and Wolff, R.D. (1993), ``Division and difference in the discipline of economics ' ', in Messer-Davidow, E., Shumway, D.R. and Sylvan, D.J. (Eds), Knowledges: Historical and Critical Studies in Disciplinarity, University of Virginia, Charlottesville, VA. Archer, M. (1996), Culture and Agency, rev. ed., Cambridge University Press, Cambridge. Argyris, C. and Schon, D.A. (1974), Theory in Practice: Increasing Professional Effectiveness, È Jossey-Bass, London. Argyris, C., Putnam, R. and McLain Smith, D. (1985), Action Science, Jossey Bass, San Francisco, CA. Beeson, I. and Davis, C. (2000), ``Emergence and accomplishment in organizational change ' ', Journal of Organizational Change Management, Vol. 13 No. 2, pp. 178-89. Berger, P.L. and Luckmann, T. (1966), The Social Construction of Reality: A Treatise in the Sociology of Knowledge, Doubleday, Garden City, NY. Bhaskar, R. (1979), The Possibility of Naturalism, Harvester Press, Hassocks. Boden, R., Gummett, P., Cox, D. and Barker, K. (1998), ``Men in white coats . . . men in grey suits: new public management and the funding of science and technology services to the UK government ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 3, pp. 267-91. Bourdieu, P. (1977), Outline of the Theory of Practice, Cambridge University Press, Cambridge. Bourdieu, P. (1988), ```Vive la crise '!: for heterodoxy in the social sciences ' ', Theory and Society, Vol. 17, pp. 773-87. Braverman, H. (1974), Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century, Monthly Review Press, New York, NY. Broadbent, J. and Laughlin, R. (1998), ``Resisting the `new public management ': absorption and absorbing groups in schools and GP practices in the UK ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 4, pp. 403-35. Broadbent, J., Jacobs, K. and Laughlin, R. (2001), ``Organisational resistance to unwanted accounting and finance changes: the case of general medical practice in the UK ' ', Accounting, Auditing & Accountability Journal, Vol. 14 No. 5, pp. 565-86. Brunsson, N. (1994), ```Politicization ' and `company-ization ' ± on institutional affiliation and confusion in the organizational world ' ', Management Accounting Research, Vol. 5, pp. 323-35.

What counts as ``theory ' '?

701

AAAJ 16,4

702

Brunsson, N. and Olsen, J.P. (1998), Organizing Organizations, Forbokforlaget, BergenSandviken. Burns, J. (2000), ``The dynamics of accounting change: inter-play between new practices, routines, institutions, power and politics ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 5, pp. 566-96. Burrell, G. and Morgan, G. (1979), Sociological Paradigms and Organizational Analysis, Heinemann, London. Burritt, R.L. and Welch, S. (1997), ``Accountability for environmental performance of the Australian commonwealth public sector ' ', Accounting, Auditing & Accountability Journal, Vol. 10 No. 4, pp. 532-61. Bryman, A. (1989), Research Methods and Organization Studies, Unwin Hyman, London. Carson, P.P. and Carson, K.D. (1998), ``Theoretically grounding management history as a relevant and valuable form of knowledge ' ', Journal of Management History, Vol. 4 No. 1, pp. 29-42. Chapman, C. (1997), ``Reflections on a contingent view of accounting ' ', Accounting, Organizations and Society, Vol. 22 No. 2, pp. 189-206. Child, J. (1972), ``Organizational structure, environment and performance: the role of strategic choice ' ', Sociology, Vol. 6 No. 1, pp. 1-22. Child, J. (1997), ``Strategic choice in the analysis of action, structure, organizations and environments: retrospect and prospect ' ', Organization Studies, Vol. 18 No. 1, pp. 43-76. Coy, D. and Pratt, M. (1998), ``An insight into accountability and politics in universities: a case study ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 5, pp. 540-61. Czarniawska, B. (1997), Narrating the Organization, The University of Chicago Press, Chicago, IL. Davie, S.S.K. (2000), ``Accounting for imperialism: a case of British-imposed indigenous collaboration ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 3, pp. 330-59. Deegan, C. (2002a), ``Social and environmental reporting and its role in maintaining or creating organizational legitimacy ' ', Special ed., Accounting, Auditing & Accountability Journal, Vol. 15 No. 3, pp. 277-436. Deegan, C. (2002b), ``Introduction: the legitimising effects of social and environmental disclosures ± a theoretical foundation ' ', Accounting, Auditing & Accountability Journal, Vol. 15 No. 3, pp. 282-311. Delanty, G. (2003), ``Michel Foucault ' ', in Elliott, A. and Ray, L. (Eds), Key Contemporary Social Theorists, Blackwell, Oxford, pp. 123-8. Dickens, L. and Watkins, K. (1999), ``Action research rethinking Lewin ' ', Management Learning, Vol. 30 No. 2, pp. 127-40. Donaldson, L. (1984), In Defence of Organisation Theory: A Reply to the Critics, Cambridge University Press, Cambridge. Douglas, M. (1986), How Institutions Think, Syracuse University Press, New York, NY. Dyball, M.C. and Valcarel, L.J. (1999), ``The `rational ' and `traditional ': the regulation of accounting in the Philippines ' ', Accounting, Auditing & Accountability Journal, Vol. 12 No. 3, pp. 303-27. Fairclough, N. (1995), Critical Discourse Analysis, Longman, London. Featherstone, M. (1997), Undoing Culture, 2nd ed., Sage, London. Ferreira, L.D. and Merchant, K.A. (1992), ``Field research in management accounting and control: a review and evaluation ' ', Accounting, Auditing & Accountability Journal, Vol. 5 No. 4, pp. 3-34.

Fogarty, T.J. and Radcliffe, V.S. (1999), ``Extending practice: accountants ' constructions of the industrial relations arena in the USA ' ', Accounting, Auditing & Accountability Journal, Vol. 12 No. 5, pp. 525-60. Foucault, M. (1980), Power/Knowledge: Selected Interviews and Other Writings 1972-1977, Gordon, C. (Ed.), Harvester Press, Brighton. Foucault, M. (1991), ``Governmentality ' ', in Burchell, G., Gordon, C. and Miller, P. (Eds), The Foucault Effect, The University of Chicago Press, Chicago, IL, pp. 87-104. Funnell, W. (1996), ``Preserving history in accounting: seeking common ground between `new ' and `old ' accounting history ' ', Accounting, Auditing & Accountability Journal, Vol. 9 No. 4, pp. 38-64. Funnell, W. (1998), ``The narrative and its place in the new accounting history: the rise of the counternarrative ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 2, pp. 142-62. Gadamar, H.G. (1975), Truth and Method, 2nd ed., Sheed and Ward, London. Gasch_ , R. (1986), The Tain of the Mirror, Harvard University Press, Cambridge, MA. e Geertz, C. (1973), The Interpretation of Cultures, Basic Books, New York, NY. Gellner, E. (1970), ``Concepts and rationality ' ', in Wilson, B.R. (Ed.), Rationality, Oxford University Press, Oxford. Gentner, D. (1989), ``The mechanisms of analogical reasoning ' ', in Vosniadou, S. and Ortony, A. (Eds), Similarity and Analogical Reasoning, Cambridge University Press, New York, NY, pp. 199-241. Glaser, B. and Strauss, A. (1967), The Discovery of Grounded Theory: Strategies for Qualitative Research, Aldine, New York, NY. Giddens, A. (1979), Central Problems in Social Theory, Macmillan, London. Giddens, A. (1984), The Constitution of Society, Polity Press, Cambridge. Giddens, A. (1987), Social Theory and Modern Sociology, Polity Press, Great Britain. Goffman, E. (1959), The Presentation of Self in Everyday Life, Doubleday, Garden City, NY. Goffman, E. (1961), Asylums, Anchor, New York, NY. Goffman, E. (1983), ``The interaction order ' ', American Sociological Review, Vol. 48, pp. 1-17. Greer, S. and Patel, C. (2000), ``The issue of Australian indigenous world-views and accounting ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 3, pp. 307-29. Grendron, Y., Cooper, D.J. and Townley, B. (2001), ``In the name of accountability: state auditing, independence and new public management ' ', Accounting, Auditing & Accountability Journal, Vol. 14 No. 3, pp. 278-310. Gummesson, E. (1988), Qualitative Methods in Management Research, Chartwell Bratt, Sweden. Guthrie, J. and Parker, L.D. (1999), ``A quarter of a century of performance auditing in the Australian public sector: a malleable masque ' ', Abacus, Vol. 35 No. 3, pp. 302-32. Habermas, J. (1984), The Theory of Communicative Action Volume 1: Reason and the Rationalization of Society, Polity Press, London. Habermas, J. (1987), The Theory of Communicative Action Volume 2: The Critique of Functionalist Reason, Polity Press, London. Hacking, I. (1999), The Social Construction of What? Cambridge, Harvard University Press, Cambridge, MA. Hanson, N.R. (1958), Patterns of Discovery, Cambridge University Press, Cambridge. Hassard, J. (1988), ``Overcoming hermeticism in organization theory: an alternative to paradigm incommensurability ' ', Human Relations, Vol. 14 No. 3, pp. 247-59.

What counts as ``theory ' '?

703

AAAJ 16,4

704

Hines, T., McBride, K., Fearnley, S. and Brandt, R. (2001), ``We 're off to see the wizard: an evaluation of directors ' and auditors ' experiences with the financial reporting review panel accounting ' ', Auditing & Accountability Journal, Vol. 14 No. 1, pp. 53-84. Hopwood, A. (1983), ``On trying to study accounting in the contexts in which it operates ' ', Accounting, Organizations and Society, Vol. 8 No. 2/3, pp. 287-305. Hussain, M.M. and Hoque, Z. (2002), ``Understanding non-financial performance measurement practices in Japanese banks: a new institutional sociology perspective ' ', Accounting, Auditing & Accountability Journal, Vol. 15 No. 2, pp. 162-83. Jacobs, K. (2000), ``Evaluating accountability: finding a place for the Treaty of Waitangi in the New Zealand public sector, Accounting, Auditing & Accountability Journal, Vol. 13 No. 3, pp. 360-80. Jacobs, K. and Kemp, J. (2002), ``Exploring accounting presence and absence: case studies from Bangladesh ' ', Accounting, Auditing & Accountability Journal, Vol. 15 No. 2, pp. 143-61. Johnson, H.T. and Kaplan, R.S. (1991), Relevance Lost: The Rise and Fall of Management Accounting, 2nd ed., Harvard Business Press, Boston, MA. Jones, C.S. (1999), ``Hierarchies, networks and management accounting in NHS hospitals ' ', Accounting, Auditing & Accountability Journal, Vol. 12 No. 2, pp. 164-87. Klumpes, P. (2001), ``Generational accountability of public sector management: a case study of the state authorities superannuation board of New South Wales ' ', Accounting, Auditing & Accountability Journal, Vol. 14 No. 2, pp. 166-89. Lakoff, G. and Johnson, M. (1980), Metaphors We Live by, The University of Chicago Press, Chicago, IL. Lakoff, G. and Turner, M. (1989), More than Cool Reason, The University of Chicago Press, Chicago, IL and London. Latour, B. (1987), Science in Action: How to Follow Scientists and Engineers through Society, Open University Press, Buckingham. Latour, B. (1993a), We Have Never Been Modern, Harvester Wheatsheaf, Brighton. Latour, B. (1993b), ``Pasteur on lactic acid yeast: a partial semiotic analysis, Configurations, Vol. 1 No. 1, pp. 129-46. Latour, B. (1996), Aramis, or the Love of Technology, Harvard Press, Cambridge, MA. Latour, B. and Woolgar, S. (1979), Laboratory Life: The Construction of Scientific Facts, Princeton University Press, Princeton, NJ. Laughlin, R. (1995), ``Empirical research in accounting: alternative approaches and a case for middle range thinking ' ', Accounting, Auditing & Accountability Journal, Vol. 8 No. 1, pp. 63-87. Laughlin, R. (1997), ``Developing empirical research: an example informed by a Habermasian approach ' ', Accounting, Auditing & Accountability Journal, Vol. 10 No. 5, pp. 622-48. Layder, D. (1993), New Strategies in Social Research, Polity Press, Cambridge. Layder, D. (1994), Understanding Social Theory, Sage, London. Lewin, K. (1946), ``Action research and minority problems ' ', Journal of Social Issues, Vol. 2 No. 4, pp. 34-46. Lightbody, M. (2000), ``Storing and shielding: financial management behaviour in a church organization ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 2, pp. 156-74. Lindkvist, L. and Llewellyn, S. (in press), ``Accountability, responsibility and organization ' ', The Scandinavian Journal of Management. Llewellyn, S. (1992), ``The role of case study methods in management accounting: a comment ' ', British Accounting Review, Vol. 24, pp. 17-31.

Llewellyn, S. (1994), ``Managing the boundary: how accounting is implicated in maintaining the organisation ' ', Accounting, Auditing & Accountability Journal, Vol. 7 No. 4, 1994, pp. 4-23. Llewellyn, S. (1997), ``Purchasing power and polarized professionalism in British medicine accounting ' ', Accounting, Auditing & Accountability Journal, Vol. 10 No. 1, pp. 31-59. Llewellyn, S. (1998a), ``Boundary work: costing and caring in the social services ' ', Accounting, Organizations and Society, Vol. 23 No. 1, pp. 23-48. Llewellyn, S. (1998b), ``Pushing budgets down the line: ascribing financial responsibility in the UK social services ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 3, pp. 292-308. Llewellyn, S. (2001), ``` Two-way windows ': clinicians as medical managers ' ', Organization Studies, Vol. 22 No. 4, pp. 593-624. Llewellyn, S. and Walker, S.P. (2000), ``Accounting at home ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 4, special issue, pp. 413-560. Lowe, A. (2000), ``The construction of a network at Health Waikato: the `towards clinical budgeting ' project ' ', Accounting, Auditing & Accountability Journal, Vol. 14 No. 1, pp. 84-114. Lowe, A. (2001), ``After ANT: an illustrative discussion of the implications for qualitative accounting case research ' ', Accounting, Auditing & Accountability Journal, Vol. 14 No. 3, pp. 327-51. Lyotard, J.-F. (1984), The Post-modern Condition: A Report on Knowledge, Bennington, G. and Massumi, B. (trans), Manchester University Press, Manchester. McGregor D. (1960), The Human Side of Enterprise, McGraw-Hill, New York, NY. Macintosh, N.B. and Baker, C.R. (2002), ``A literary theory perspective on accounting: towards heteroglossic accounting reports ' ', Accounting, Auditing & Accountability Journal, Vol. 15 No. 2, pp. 184-222. Mangham, I.L. (1996), ``Some consequences of taking Gareth Morgan seriously ' ', in Grant, D. and Oswick, C. (Eds), Metaphor and Organizations, Sage, London, pp. 21-36. Markovic, M. (1974), From Affluence to Praxis, Michigan University Press, Ann Arbor, MI. Maslow, A. (1943), ``A theory of human motivation ' ', Psychological Review, Vol. 50, pp. 370-96. Merton, R.K. (1968), Social Theory and Social Structure, Free Press, New York, NY. Miller, P. (2003), ``Management accounting practices and assemblages ' ', paper presented at the 8th Biennial Management Accounting Research Conference, Sydney, Australia, February. Miller, P., Hopper, T. and Laughlin, R. (1991), ``The new accounting history: an introduction ' ', Accounting, Organizations and Society, Vol. 16 No. 5/6, pp. 395-403. Mintzberg, H. (1989), Mintzberg on Management, Collier Macmillan Publishers, London. Morgan, G. (1983), ``More on metaphor: why we cannot control tropes in administrative science ' ', Administrative Science Quarterly, Vol. 28, pp. 601-7. Morgan, G. (1986), Images of Organization, Sage, Beverly Hills, CA. Mouritsen, J., Larsen, H.T. and Bokh, P.N. (2001), ``Valuing the future: intellectual capital supplements at Skandia ' ', Accounting, Auditing & Accountability Journal, Vol. 14 No. 4, pp. 399-422. Mouzelis, N. (1995), Sociological Theory: What Went Wrong?, Routledge, London and New York, NY. Nagel, T. (1986), The View from Nowhere, Oxford University Press, New York, NY. Napier, C.J. (2001), ``Accounting history and accounting progress ' ', Accounting History, Vol. 6, pp. 7-31. Neu, D. (2000), ``Accounting and accountability relations: colonization, genocide and Canada 's first nations ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 3, pp. 268-88.

What counts as ``theory ' '?

705

AAAJ 16,4

706

Northcott, D. and Doolin, B. (2000), ``Home accountants: exploring their practices ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 4, pp. 475-501. O 'Donovan, G. (2002), ``Environmental disclosures in the annual report: extending the applicability and predictive power of legitimacy theory ' ', Accounting, Auditing & Accountability Journal, Vol. 15 No. 3, pp. 344-71. O 'Dwyer, B. (2002), ``Managerial perceptions of corporate social disclosure: an Irish story ' ', Accounting, Auditing & Accountability Journal, Vol. 15 No. 3, pp. 406-36. Otley, D.T. (1980), ``The contingency theory of management accounting ' ', Accounting, Organizations and Society, Vol. 5, pp. 413-28. Ouchi, W.G. (1980), ``Markets, bureaucracies and clans ' ', Administrative Science Quarterly, Vol. 25, pp. 129-41. Page, M. and Spira, L. (1999), ``The conceptual underwear of financial reporting ' ', Accounting, Auditing & Accountability Journal, Vol. 12 No. 4, pp. 489-501. Pahl, J. (2000), ``Couples and their money: patterns of accounting and accountability in the domestic economy ' ', Accounting, Auditing & Accountability Journal, Vol. 13 No. 4, pp. 502-17. Parker, L.D. (1997), ``Informing historical research in accounting and management: traditions, philosophies, and opportunities ' ', The Accounting Historians Journal, Vol. 24 No. 2, pp. 111-49. Parker, L.D. (1999), ``Historiography for the new millennium: adventures in accounting and management ' ', Accounting History, Vol. 4 No. 2, pp. 11-42. Parker, L.D. (2002), ```Presenting ' the past: perspectives on time for accounting and management history ' ', Plenary Address at the 9th Word Congress of Accounting Historians, Melbourne. Parker, L.D. and Roffey, B.H. (1997), ``Back to the drawing board: revisiting grounded theory and the everyday accountant 's and manager 's reality ' ', Accounting, Auditing & Accountability Journal, Vol. 10 No. 2, pp. 212-47. Parker, L.D., Guthrie, J. and Gray, R. (1998), ``Accounting and management research: passwords from the gatekeepers ' ', Accounting, Auditing & Accountability Journal, Vol. 15 No. 1, pp. 69-93. Pierce, C.S. (1955), Philosophical Writings of Pierce, Buchler, J. (Ed.), Dover, New York, NY, pp. 98-119. Polkinghorne, D. E. (1988), Narrative Knowing and the Human Sciences, State University of New York Press, New York, NY. Popper, K. (1972), Objective Knowledge, Clarendon Press, Oxford. Potter, B. (2002), ``Financial accounting reforms in the Australian public sector: an episode in institutional thinking ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 4, pp. 371-402. Potter, J. and Wetherell, M. (1987), Discourse and Social Psychology: Beyond Attitudes and Behaviour, Sage, London. Prichard, C. and Willmott, H. (1997), ``Just how managed is the McUniversity? ' ', Organization Studies, Vol. 18 No. 2, pp. 287-316. Rabinow, P. (1984), ``Introduction ' ', in Rabinow, P. (Ed.), The Foucault Reader, Penguin, Harmondsworth, pp. 3-29. Rahaman, A.S. and Lawrence, S. (2001), ``A negotiated order perspective on public sector accounting and financial control ' ', Accounting, Auditing & Accountability Journal, Vol. 14 No. 2, pp. 147-165 Rahman, S.F. (1998), ``International accounting regulation by the United Nations: a power perspective ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 5, pp. 593-623.

Reason, P. (1988), ``Introduction ' ', in Reason, P. (Ed.), Human Inquiry in Action, Sage, London. Reiter, S. (1997), ``The ethics of care and new paradigms for accounting practice ' ', Accounting, Auditing & Accountability Journal, Vol. 10 No. 3, pp. 299-324. Riessman, C.K. (1993), Narrative Analysis, Sage, London. Roberts, J. (1991), ``The possibilities of accountability ' ', Accounting, Organizations and Society, Vol. 16 No. 4, pp. 355-68. Rock, P. (1979), The Making of Symbolic Interactionism, Macmillan, London. Saussure, F. (1974), Course in General Linguistics, Fontana, London. Sayer, A. (1992), Method in Social Science: A Realist Approach, 2nd ed., Routledge, London and New York, NY. Sayer, A. (2000), Realism and Social Science, Sage, London. Scapens, R. (1990), ``Researching management accounting practice: the role of case study methods ' ', British Accounting Review, Vol. 20 No. 3, pp. 260-79. Seal, W. and Vincent-Jones, P. (1997), ``Accounting and trust in the enabling of long-term relations ' ', Accounting, Auditing & Accountability Journal, Vol. 10 No. 3, pp. 406-31. Sharpe, J. (1991), ``History from below ' ', in Burke, P. (Ed.), New Perspectives on Historical Writing, Polity Press, Cambridge, pp. 42-66. Silverman, D. (1970), The Theory of Organizations, Heinemann, London. Silverman, D. (1985), Qualitative Methodology and Sociology, Gower, Aldershot. Silverman, D. (2001), Interpreting Qualitative Data: Methods for Analysing Talk, Text and Interaction, 2nd ed., Sage, London. Strauss, A. and Corbin, J. (1990), Basics of Qualitative Research, Sage, Newbury Park, CA. Streatfield, P.J. (2001), The Paradox of Control in Organizations, Routledge, London. Strong, P. and Robinson, J. (1990), The NHS ± Under New Management, Open University Press, Buckingham. Susela, S.D. (1999), ```Interests ' and accounting standard setting in Malaysia ' ', Accounting, Auditing & Accountability Journal, Vol. 12 No. 3, pp. 358-87. Sutton, R.I. and Staw, B.M. (1995), ``What theory is not ' ', Administrative Science Quarterly, Vol. 40, pp. 371-84. Thompson, G. (1991), ``Comparison between models ' ', in Thompson, G., Frances, J., Levacic, R. and Mitchell, J. (Eds), Markets, Hierarchies and Networks, Sage, London. Thompson, G. (1999), ``Cultural capital and accounting ' ', Accounting, Auditing & Accountability Journal, Vol. 12 No. 4 pp. 394-412. Thompson, P. and McHugh, D. (2002), Work Organizations; A Critical Introduction, 3rd ed., Palgrave, Basingstoke. Tinker, T. (1999), ``The Hegelian logic of critical research: understanding Professor Yoshinori Shioawa ' ', Accounting, Auditing & Accountability Journal, Vol. 12 No. 1, pp. 39-46. Tinker, T. and Koutsoumadi, A. (1997), ``A mind is a wonderful thing to waste: `think like a commodity ', become a CPA ' ', Accounting, Auditing & Accountability Journal, Vol. 10 No. 3, pp. 454-67. Triandis, H.C. (2001), ``The study of cross cultural management and organization: the future ' ', International Journal of Cross Cultural Management, Vol. 1 No. 1, p. 17-20. Turner, J. (1988), A Theory of Social Interaction, Polity Press, Cambridge. Walker, S.P. and Shackleton, K. (1998), ``A ring fence for the profession: advancing the closure of British accountancy 1957-1970 ' ', Accounting, Auditing & Accountability Journal, Vol. 11 No. 1, pp. 34-71.

What counts as ``theory ' '?

707

AAAJ 16,4

Weick, K.E. (1989), ``Theory construction as disciplined imagination ' ', Academy of Management Review, Vol. 14 No. 4, pp. 516-31. Wodak, R. (2001), ``The discourse-historical approach ' ', in Wodak, R. and Meyer, M. (Eds), Methods of Critical Discourse Analysis, Sage, London, pp. 63-94. Yin, R.K. (1991), Case Study Research: Design and Methods, rev. ed., Sage, London. Further reading Habermas, J. (1971), Towards a Rational Society, Heinemann, London. Habermas, J. (1987), The Philosophical Discourses of Modernity, Polity, Cambridge. Lakoff, G. (1993), ``The contemporary theory of metaphor ' ', in Ortony, A. (Ed.), Metaphor and Thought, 2nd ed., Cambridge University Press, Cambridge, pp. 202-52. Llewellyn, S. (1993), ``Working in hermeneutic circles in management accounting research: some implications and applications ' ', Management Accounting Research, Vol. 4, pp. 231-49.

708

Citations: Mouzelis (1995, p. 9) argues that a pressing task for researchers is: Alvesson and Skoldberg (2000, foreword) set out a rationale for a project: È 666 Bhaskar 's (1979, p