The linguistic turn of the later twentieth century has led to a widespread and growing interest in discourse, both in organization studies and in the social sciences more generally. Since the late 1970s, organization scholars have began to move beyond a conception of language as a functional, instrumental conduit of information, and drew attention to its symbolic and metaphorical aspects as constructive of social and organizational reality (Dandridge, Mitroff and Joyce, 1980: Manning, 1979), constitutive of theory (Morgan, 1980, 1983), and enabling of shared meanings, co-ordinated action, and even organization itself (Daft and Wiginton, 1979; Louis, 1983; Pondy and Mitroff, 1979; Smircich, 1983). Subsequent scholars have adopted a wide range of approaches to the analysis of organizational discourses and have conceptualized discourse itself, and its relevance to organizational interpretations, actions and subjectivity, in a variety of ways (Grant, Keenoy and Oswick, 1998; Heracleous and Hendry, 2000; Mumby and Stohl, 1991; Phillips and Hardy, 2002).
Discourse analysis, in the broad sense of utilizing textual data in order to gain insights to particular phenomena, has had a rich and varied heritage in the social sciences, spanning the fields of sociology, anthropology, psychology, political science and history (OConnor, 1995), and this same richness and diversity is evident in the organizational sciences. Approaches include hermeneutics (Kets de Vries and Miller, 1987; Phillips and Brown, 1993; Thachankary, 1992), ethnomethodology (Atkinson, 1988), rhetorical analysis (Alvesson, 1993; Keenoy, 1990; Watson, 1995), deconstruction (Kilduff, 1993; Noorderhaven, 1995), metaphorical analysis (Jacobs and Heracleous, 2006; Lakoff and Johnson, 1980; Ortony, 1979), critical discourse analysis (van Dijk, 1993; Garnsey and Rees, 1996; du Gay and Salaman, 1992; Knights and Morgan, 1991, 1995), narrative analysis (Barry and Elmes, 1997; Manning and Cullum-Swan, 1994), and semiotic analysis (Barley, 1983; Fiol, 1989).
Both discourse and related terms, such as language, text or narrative, have been conceptualized and categorized in diverse ways in organization theory (van Dijk, 1997; Grant, Keenoy and Oswick, 1998; Grant et al., 2004). In my work, I have employed the term “discourse” to mean collections of texts, whether oral or written, located within social and organizational contexts that are patterned by certain structural, inter textual features and have both functional and constructive effects on their contexts. In this sense, language can be seen as the raw material of discourse, and individual texts are both manifestations, and constitutive, of broader discourses (Heracleous, 2004 and Hendry, 2000).
In spite of the variety of conceptualizations and operationalizations, three dominant approaches to the study of organizational discourse can be discerned interpretive, functional, and critical (Heracleous, 2006; Heracleous and Barrett, 2001; Heracleous and Hendry, 2000). These approaches are not mutually exclusive, but they can be seen as analytically distinct. A key distinction has been made between interpretive and critical approaches to discourse (Mumby and Clair, 1997) that parallels the related distinction of research focusing on meaning construction processes or on issues of power (Oswick, Keenoy and Grant, 1997), as well as the distinction between monological accounts presenting the perspective of a dominant group and dialogical accounts presenting a multiplicity of conflicting perspectives and multiple realities (Boje, 1991; Grant, Keenoy and Oswick, 1998; Keenoy, Oswick and Grant, 1997).
Interpretive approaches conceptualize discourse as communicative action that is constructive of social and organizational realities. Functional approaches view discourse as a tool at actors' disposal, to be employed for facilitating managerially relevant processes and outcomes such as effective leadership, employee motivation, and organizational change. Critical approaches conceptualize discourse as power knowledge relationships, constitutive of subjects' identities and of organizational and societal structures of domination. The emerging structurational approach finally views discourse as a duality of communicative actions and deep structures, interrelated through the modality of interpretive schemes (Heracleous, 2006; Heracleous and Barrett, 2001; Heracleous and Hendry, 2000).
This conceptual diversity is symptomatic of a similar diversity of approaches to discourse in the social sciences more generally, and reflects long standing divisions between agent-centered and structuralist-oriented theories in sociology (Burrell and Morgan, 1979, Thompson, 1989). The interpretive and functional approaches to organizational discourse tend to privilege the action level, giving primacy to human agency, the hermeneutic nature of discourse at the individual and organizational levels, and how agents can employ discourse to shape their own or others' understandings of situations. The critical approach, on the other hand, tends to privilege the structural level, giving primacy to how human agency, identity and subjectivity are constituted, shaped, and may even be lost in the webs of discursive structures and the patterns of social domination that these structures surreptitiously help to legitimize and sustain. The structurational approach, in line with Giddens's efforts to transcend the structure/agency dualism, aims to address both communicative actions and discursive deep structures as inherently interlinked and mutually constituted levels via actors' interpretive schemes, in which communicative actions are both a manifestation and instantiation of deep structures.
As noted earlier, organizational discourses can be seen as collections of texts, both spoken and written. The term “text” has been interpreted in a variety of ways, with texts viewed broadly as “all types of data that contain messages and themes that can be systematized” (Kets de Vries and Miller 1987: 235; Phillips and Brown, 1993), for example structured patterns of actions and interpretations, or even organizations (Putnam, Phillips and Chapman, 1996: 391; Thachankary, 1992); as well as in a more literal way as primarily language-based artifacts (Gephart, 1993; Giddens, 1979).
An understanding of context is crucial to the interpretive validity and potential insight afforded by discourse analyses. According to Cicourel, “the study of discourse and the larger context of social interaction requires explicit reference to a broader organizational setting and aspects of cultural beliefs often ignored by students of discourse” (1981: 102). Unfortunately, some approaches that began with interpretive or hermeneutic inspirations like ethnomethodology, stressing features of language such as indexicality (the notion that language use and interpretation depends on contextual features) and the temporality of social activity (where social action is understood and analyzed with regard to its temporal location), have gradually proceeded to restrict themselves to behaviorist straitjackets which can hinder them from grasping the richness of social life, as in the form most ethnomethodological conversation analysis has taken (Atkinson, 1988).
Fairclough has observed that in practice “analysis of text is perceived as frequently proceeding with scant attention to context discourse analysis needs a developed sense of and systematic approach to both context and text” (1992: 212–213). Fortunately, several useful approaches for integrating context in organizational discourse analysis have been developed. These include critical discourse analysis (van Dijk, 1993; Fairclough and Wodak, 1997) social semiotics (Hodge and Kress, 1988; Kress, Leite-Garcia and van Leeuwen, 1997); rhetorical analysis (Aristotle, 1991; Gill and Whedbee, 1997); or ethnography of communication (Hymes 1964, 1972; Gumperz and Levinson, 1991).
From a sociological perspective, Giddens has suggested that the influence of structuralist and post-structuralist thought has encouraged the neglect of context and temporality in discourse analysis, indicating that although structuralism and post-structuralism have brought to the fore of social theory important issues such as the importance of temporality as reversible time, the properties of signification systems as existing outside time-space, and the relevance of decentring the subject, they are fraught with theoretical difficulties that make them unsuitable theoretical traditions through which the themes they have highlighted can be pursued (Giddens, 1979; 1987).
Saussures (1983) basic distinction between langue and parole, for example, and the emphasis on langue, is deemed as inadequate because it isolates language from its social environments of use and therefore does not promote the need for a theory of the competent speaker or language-user (Giddens, 1979). As a result, a conception of human subjects as agents has not been reached in structuralism, and the theoretically decentered elements (such as the author) are not satisfactorily recombined in the analysis (Giddens, 1987).
Furthermore, because of the stress on form rather than substance, and because of the thesis of the arbitrary character of the sign (Saussure, 1983), structuralism and poststructuralism have promoted a “retreat into the code,” where the aim was “to determine the forces operating permanently and universally in all languages, and to formulate general laws which account for all particular linguistic phenomena historically attested” (Saussure, 1983: 6). This “retreat into the code” means that structuralism and post-structuralism have been unable to provide satisfactory accounts of reference, or of meaning. Meaning, for example, is said to derive from the intra- or inter textual play of differences of the signifiers, ignoring the relationship of such signifiers with their contexts of use (Giddens, 1987). The focus on the signifier/signified distinction as arbitrary has led to an elision between the “signified” and the “object signified,” the reality to which the sign is related (Giddens, 1979).
Further, Saussures theoretical distinction between synchrony and diachrony has been utilized by structuralism as a methodological division, which is deemed unjustifiable because one can often gain a deeper understanding of linguistic and social systems in longitudinal rather than cross-sectional study (Lewin, 1952). The general “repression of time” in social theory has been attributed to the maintenance of this distinction between synchrony and diachrony, or statics and dynamics (Giddens, 1979).
While structuralism isolates texts from their contexts, a tradition such as hermeneutics stress their essential contextuality and the role of context in valid textual interpretations (Giddens, 1979). Ricoeur has defined hermeneutics as “the theory of the operations of understanding in relation to the interpretation of texts,” and posed as a key idea the transformation of spoken discourse in written text (1991: 53). Spoken discourse is seen as an event in that (1) it is realized temporally and in the present; (2) the “instance of discourse” is self-referential because it refers back to its speaker; (3) discourse is always about something: it refers to a world that it attempts to describe, express or represent; and (4) discourse is in practice addressed to an other (Ricoeur, 1991: 77–78).
Ricoeur argues, however, that as soon as discourse is “fixed” in writing as text, several hermeneutic issues emerge. First, whereas discourse is realized temporally as a speech event, the written text fixes, in decreasing order of susceptibility to such fixing, the locutionary, illocutionary and perlocutionary acts of spoken discourse and divorces them from their temporal and social contexts. Second, whereas spoken discourse is self-referential in that it refers back to its speaker, the intended meanings of the author and the semantic meanings of the text do not necessarily coincide when spoken discourse is fixed as a text, because the text is open to a potentially unlimited series of interpretations. Third, whereas spoken discourse displays ostensive references deriving from the common situation and context within which the interlocutors find themselves, texts, divorced from such conditions, display non ostensive references, ideally projecting new possibilities of being-in-the-world – a concept that is for Ricoeur the ultimate referent of all texts. Finally, whereas spoken discourse is addressed at a specific interlocutor, texts are in principle available to anybody who can read (Ricoeur, 1991: 146–150).
On the basis of Ricoeur's distinction between spoken discourse and written text, I would suggest that organizational texts (not only oral communicative actions but also those fixed in writing) can be seen as implicated in particular conditions and imperatives which necessitate that they are understood and analyzed as being ontologically closer to spoken discourse than written text. This proposal can be clarified through a comparison of Roland Barthes' (1972, 1977, 1994) early structuralist and later post-structuralist conceptions of text, with the particular conditions that organizational texts tend to be implicated in.
Organizational texts are often bound up with and shaped by, imperatives such as rules of communicative appropriateness in particular organizations, and overarching purposes as espoused by dominant coalitions. Further, due to the need for co ordination and collective action, organizational texts most often aim to display unambiguous references that suppress the plurality of meanings that, according to Barthes' (1977: 155–164) suggestions, should characterize texts. The organizational imperative of effective cross-functional coordination fosters demands for such organizational texts to have a relatively unambiguous, representational (or informational) aspect, and to suppress the plurality of possible meanings.
The possibility of varying interpretations of organizational texts (or, in Barthian terms, a plurality of meanings) of course cannot be fully suppressed. But the imperatives of competitiveness and effective organizational processes tend to limit the signified. Time-starved, goal-oriented readers of organizational texts are usually not disposed in this context to write the text anew or metaphorically participate in textual production through active reading. The instance of the Barthian text is the signifier, but that of organizational texts is the signified. In this sense, organizational texts cannot practice “the infinite deferment of the signified” (Barthes, 1977: l58). They are thus potentially reducible, as opposed to Barthian text, which is not only plural but also irreducible (1977: l59).
The content of organizational texts, moreover, tends to be of a different, more intentional and indexical nature from that of other types of texts. Barthes (1994) does not make explicit to what types of narrative his structural analysis might apply, but the fleeting references to “the story,” the previous research on which he draws (Propp, Bremond, Todorov, Greimas, Levi-Strauss), many of the examples he uses (e.g. from James Bond movies), and the important part played by the actional level in his mode of analysis make it clear that the structural analysis of narrative, as developed by him, would be more suited to stories (at the social system level), novels, or myths. Organizational texts may not exhibit similar discourse-level structures to those discovered for stories, myths, or novels, and thus a homology of textual ontology and the analytical process between these texts cannot be assumed.
From the perspective of Barthian structuralist analysis, textual content would be of interest merely as a manifestation of deeper structures due to the assumed supremacy of form over substance (an idea originating from Saussure, 1983). In interpretive-oriented studies of the constructive role of discourse in organizations, this structural-level perspective needs to be complemented with consideration of textual content in its own right, in the light of the particular context, since the meaning of texts does not reside solely in intra- or inter textual relations but also in the dynamic interaction of these domains with the social context within which agents act (Giddens, 1987: 91).
With regard to textual functions, Barthes (1994) does not make clear in his structural analyses what functions stories or novels might have in their wider social context. “Functions” in Barthes structuralism relate solely to signifying units within the text, and do not refer to the interrelation of text with its social context. Barthes draws an analogy between narrative and linguistics, viewing narrative as a “great sentence” (1994: 99–100), and between narrative analysis and linguistic analysis holding that “just as linguistics halts at the sentence, the analysis of narrative halts at discourse” (1994: 127). In Mythologies (1972) his narrative analyses revealed critical concerns, relating to the unmasking of ideological processes working in the interests of the bourgeoisie, while in his “post-structuralist” period the consumption of the text was bound to “a pleasure without separation” (1977: 164).
Organizational texts on the other hand, as argued earlier, in addition to their constructive potential, tend to be imbued by a functional, representational nature that suppresses an infinite plurality of meanings due to the imperatives of systemic co ordination, collective action, and organizational competitiveness. Organizational texts have particular functions in their social and organizational contexts; they are normally not concerned with the critical aims of unmasking social domination, and any pleasure they bring to the reader is incidental. The latitude of interpretation of organizational texts varies according to the type of text, but all texts have an underlyings purposive construction by agents who have specific intentions in producing them for particular audiences, and intentionally wish to limit the potential plurality of textual meanings (except in special cases in which, for example, metaphorical discourse can aid organizational change processes because of its wide latitude of interpretation).
With regard to textual authorship, the structuralist tendency to “equate the production of texts with their inner ‘productivity’,” the decentering of the author, ultimately derives from the preoccupation with signifiers rather than signifieds (or the emphasis on form over substance), and often leads to an impression that texts wrote themselves (Giddens, 1987: 94–95). Organizational texts, in line with Barthes concept of “work,” however, are “caught up in a process of filiation” (1977: 160). Their authors are not “paper authors” (1977: 161) but flesh- and -blood individuals whom the audience knows and has opinions and thoughts about. Individuals referred to in organizational texts are not “paper beings” (1994: 123) but people bound up with the textual context. Various characteristics of the author are highly relevant for the interpretation and persuasive potency of a text (Burgoon, Hunsaker and Dawson, 1994; Petty and Cacioppo, 1986). This would not be the case for the kinds of stories, myths or novels that Barthian structuralist analysis was concerned with, however, where their interpretation (at least by readers if not by literary critics) does not depend on who the author is, and there is usually no immediate, context-dependent persuasive intention attached to them.
Temporality, in addition, is seen in structuralist approaches such as Barthes' (1994: 112) as “only a structural class of narrative,” divorced from the texts social context. In analyzing organizational discourse to gain ideographic insights to social settings, however, temporality must ideally be considered in terms of real-time, recursive, and historical events. Organizational texts, especially intra-organizational ones, while “fixed” in writing (and thus according to Ricoeur available to anyone who can read and potentially subject to an unlimited series of interpretations), they are read, if at all, a relatively short amount of time after they are written and are usually read only once. Their functional, intentional relevance tends to diminish the longer they remain unread, and after a certain period of time the only individuals likely to have an interest in them are not organizational actors themselves but organizational researchers and historians. Such researchers, ironically, may themselves in fact be trying to utilize texts as a source of information in order to reconstitute retrospectively actual events or situations that they would have ideally preferred access to in real time but could not, because of various constraints.
The above discussion suggests that, because of the particular contextual conditions in which organizational texts are implicated, irrespective of whether they are spoken or written, they should be understood and analyzed more as spoken discourse or language-events (temporal, self-referential, representational, occurring among identifiable agents), rather than texts in a Ricoeurian or Barthian sense. This perspective, of course, does not discount or discourage a focus on such aspects as inter textual patterns and their constructive effects, or effects on agents' subjectivity. What the above discussion suggests, however, is that attention to the various dimensions of organizational context is indispensable for higher validity in textual interpretations.
Discourse analysis approaches, at least as employed in organization theory, sociology and literary studies, are not methods in the positivist sense of precisely defined sequential steps in search of universally applicable laws, but rather approaches emphasizing hermeneutic, iterative journeys of discovery by (re)reading individual texts in the context of the whole and their social context and then (re)considering the whole as manifested in individual texts. Several authors have drawn attention to the unstructured, interpretive nature of discourse analysis (Fairclough, 1992; Potter and Wetherell, 1987). Narrative analysis, for example, is said to be “rather loosely formulated, almost intuitive, using terms defined by the analyst” (Manning and Cullum-Swan, 1994: 465) discourse analysis is “neither systematic nor detailed” (Fairclough, 1992: 196) and deconstruction is “not reducible to a set of techniques, . . . [and] cannot be summarized as a mechanical series of operations to be applied to any piece of language” (Kilduff, 1993: 16). Barthes has on repeated occasions consciously refused to refer to his analyses as exemplifying a “method” which he saw as having positivistic connotations (Barthes, 1994: 223, 248, 263). Contrary to his early structuralist statements that narrative was to be studied in a deductive fashion, he later denounced an inductive deductive science of texts as illusory (Barthes, 1977: 159–160).
As discussed in more detail in chapter 2, this situation does not necessarily imply insufficient or inadequate methodological rigor, or degeneration to totally subjective opinions as a basis of textual interpretation. Rigor in organizational discourse analysis however has a different meaning than in positivism; replicability, especially in ethnograpically oriented studies is not possible, and the search is for broad principles relating to the nature and functioning of social systems rather than mechanistic “universal laws” that would foster the same outcome if the technologies they imply are implemented in different settings. Discourse analysis aiming to identify such entities as genre repertoires (Orlikowski and Yates, 1994; Yates and Orlikowski, 1992), generative metaphors (Schön, 1979) or deep structures (Heracleous, 2006; Heracleous and Barrett, 2001) as opposed to more narrowly defined discursive aspects (e.g. turntaking in conversational analysis) is necessarily a loosely structured, interpretive exercise in which the researchers own competencies and judgment are critical (Fairclough, 1992: Potter and Wetherell, 1987).