Historical perspectives and key definitions: defining discourse and the discursive

Many dictionary definitions will identify a pre-scientific meaning of the term discourse as “a long and serious treatment or discussion of a subject in speech or writing” (Oxford Advanced Learner’s Dictionary of English, 2005, p. 434), a definition sometimes echoed with more emphasis on public oral performance, as in the French Larousse, as a “développement oratoire sur un sujet déterminé, prononcé en public” (LPLI, 2003, p. 328). ¹ The two dictionary entries I refer to here include a second meaning connected to the specific ‘discipline’ of discourse analysis, one which highlights the dimension of language-in-use, while identifying the naturally occurring utterance and meaningful language use as typical units and foci of analysis: “the use of language in speech and writing in order to produce meaning” (OALDCE, 2005, p. 434). The equivalent definition from Larousse is: “Réalisation concrète, écrite ou orale, de la langue considerée comme un système abstrait” (LPLI, 2003, p. 328). ² By positing a contrast with language as an abstract system, de Saussure’s concept of parole is evoked as an alternative but now displaced term, as history has caught up with it. Whereas the early 20th-century Saussurean legacy identified parole in order next to suppress its study as ‘uninteresting’ to the scholar of language because language use was deemed ‘idiosyncratic’ and really the terrain of ‘user psychology’, the term discourse emerged in the second half of the century with the promise that meaningful language use in a real situation is an area worthy of detailed linguistic enquiry and that the concept of discourse opens a window on a range of relevant phenomena and considerations which linguists until then had not explored systematically in great detail.

One way to throw light on the term and concept is to examine how its use emerged, how the concept and tradition have been shaped historically, while recognising that the appeal of the discursive realm has mostly resided in the development of a particular agenda for language inquiry, at times complementary to developments in linguistics, and at other times in opposition with received practice at the time. Three defining, interrelated areas can broadly be identified in this regard.

• Viewed from within a linguistic project, ‘discourse’ emerged in the 1960s and 1970s to refer to specific language phenomena which are characteristic of running text and ongoing interaction, in short, with a stress on authentic language data and on pushing formal and functional enquiry both beyond the bounds of the isolated grammatical sentence and the self-constructed language datum.
• From within a sociolinguistic and anthropological project, ‘discourse’ has been instrumental in developing a qualitative research agenda on the role of language use in social life. In the background is an interactional and a performative view on language use in its complex manifestations. Language use is viewed as social action, and studying this is key to an understanding of its role in a larger project of understanding societies and cultures.
• Finally, as the term ‘discourse’ also surfaced in social theoretical work, it has become a metaphor for capturing processes and practices of socio-cultural representation. In this view, representations are seen as imbued by world view, permeated by ideology, central to the construction of identities, and they often count as rationalisations of power relationships. This tradition descends from post-structuralist thought and deconstructivist traditions in critical enquiry.

Attending to the first area of definition, the development of a scientific object referred to as ‘discourse’ resulted in a major breakthrough in linguistics’ efforts to unravel the nature of naturally occurring language use and to do so with specific reference to formal-functional properties characteristic of running text and ongoing interaction, as well as locating areas of meaning relative to situation, purpose and user. It is worth reminding readers here that some instances of early discourse research prioritised the conversational domain of spoken exchanges, while other early developments focussed more on the properties of written texts. Anglo-American work, in particular, interacted directly with speech act theory and conversation analysis and was quick to integrate key concepts from these fields – e.g. turn, sequence, speech act, speaker intention, etc. At the same time, examples in this tradition such as Sinclair et al. (1972) on classroom interaction, Labov and Fanschel (1977) on psychotherapeutic encounters and Brown and Yule’s (1983) discourse analysis textbook were still very much bracketed by fairly traditional linguistic concerns such as the detection of a hierarchical structure in speech events in ways akin to what had been described earlier for the constituents of the sentence. Corresponding continental European work, which often identified itself as ‘text linguistics’, focussed on descriptions of the functional components of structure beyond the sentence level and the ways in which textual make-up is tied to situation of use (e.g. anaphoric, cataphoric and situational reference; constituents of argumentative and rhetorical structure; devices to accomplish textual cohesion and establishing textual coherence, etc.), as well as on the cognitive processing of textual units. In each of these cases, the impetus was to push the linguistic agenda beyond the confines of the isolated sentence, locating the text in its contexts-of-use. Well-known examples include Harweg (1968) on reference, Werlich’s (1976) text grammar of English, De Beaugrande and Dressler’s (1981) identification of seven standards of textuality and Van Dijk and Kintsch’s (1983) work on the processing of textual information.

The second area in delineating the discursive has been tied up with the emergence of an integrated sociolinguistic project in the 1960s and 1970s, especially the more qualitative, interactional sociolinguistic traditions, which developed out of the work of, among others, John Gumperz and Dell Hymes. Not surprisingly, their insights on the nature of social interaction and situated language use drew substantially on theoretical conversations with speech act theory, conversation analysis, ethnomethodology and Goffman’s analysis of the interaction order (see also the integrative project which is suggested by the list of contributing authors in Gumperz and Hymes’s edited volume, titled: Directions in Sociolinguistics. The Ethnography of Communication (1972), and see Rampton, this volume). While for some authors (Fairclough, 1989, p. 9) speech act theory’s performative view on the language utterance, as outlined in Austin (1962), counted as a primary point of departure for a generalised social-actional view on language use, other authors invoked the groundbreaking contributions in ethnomethodology and conversation analysis as primary loci for connecting the use of language with interaction and detailing the in-course ‘linguistic’ production of a situational and social context (Duranti & Goodwin, 1992, p. 22). It is to these developments that we owe the axiom that language use and interaction count as mutually constitutive (see esp. Fairclough (1989) and Duranti and Goodwin (1992) for parallel claims in this respect on separate sides of the Atlantic). Various key tenets of interactional sociolinguistics can be noted: interaction is seen as unfolding, it does so sequentially and in real time; interactional behaviour is a major site for uncovering participants’ meaning perspectives on a situation; interaction amounts to talking social categories and situations into being; interaction is a major site for studying where/how the social is being created and sustained; interaction comes with real-life consequences and outcomes. Whether language use in a performative perspective is seen as an instrumental matter of ‘getting things done’ or of ‘relating to others in a particular way’, the underlying view is one of the social as originating and reflected in interactants’ engagements with the real world. Hymes’s formulation of the SPEAKING-project (1972, 1974) is nowadays generally accepted as the formulation of a discourse perspective within sociolinguistics which is rooted in ethnographic enquiry and takes the ‘speech event’ and ‘ways of speaking’ as primary units of analysis (cf. Johnstone & Marcellino, 2011). As noted by Duranti (2001), this view played a central role in reformulating linguistic anthropology’s share in a larger project of cultural anthropology: it marked the transition from ‘anthropological linguistics’ (description of languages) to linguistic anthropology’ (understanding speech events).

Third and finally, as it also surfaced in a social theoretical context, ‘discourse’ has since the late 1960s and early 1970s become a metaphor for understanding socio-cultural representation. This area of definition signals how the concept of discourse has been implicated in some of the theoretical and epistemological challenges posed to the human and social sciences by post-structuralist theory. In Michel Foucault’s version of this (Foucault, 1972), discourse covers what is socio-historically “sayable” about a particular social domain, as well as the rules which prescribe the ways of talking about these topics (including who gets to speak about this in the first place). Discourse is seen as centrally connected to the production of truth and it is located in a field of power relationships which enable the production of social life in its historically shifting and evolving forms and manifestations. It is particularly in this area that a discourse analytic perspective has spilled over into various other disciplines (law, social work, history, etc.), where it has given rise to ‘a linguistic turn’ which stresses how the production of truth is constituted by discursive articulations, while highlighting the social, institutional and organisational conditions that enable its expression (see also Cheek, 2008). Examples include Fook (2016) for social work, Holdgaard (2008) and Golder (2013) for law, Shaw (2012) for teaching, etc. The list of possible references is a very long one. Other examples which do not necessarily entail a Foucauldian perspective include Dumolyn (2008) for medieval institutional history, Doty (1996) for North/South relations and Campbell and Dillon (1993) for international relations. Note that these studies are often culminations of lines of theoretical and empirical engagement with situated language use in ways specific to these disciplines. For discourse theorists such as Laclau and Mouffe (1985), Torfing (1999) and Howarth (2000), discourse study has become an epistemology for reading the societal state of hegemonic relationships vis-à-vis particular ideological formations. Discourse theory is a form of political discourse analysis, which, however, rarely comes with specific text or language-oriented empirical imperatives, as projects are more oriented to the macro-levels of power/knowledge relationships and the articulation of collective identities (Keller, 2005). For instance, for Cheek (2008, p. 356), Foucauldian discourse analysis specifies more a way of doing discourse analysis than an empirical methodology. The same can be said about discourse theory.

As indicated earlier, a social constructivist perspective in which language use, often in combination with other practices, is seen as constitutive of social reality is intrinsic to many traditions of discourse analysis. Critical discourse analysis (Fairclough, 1992; Wodak, 1996; Jäger, 1999; Locke, 2004) must be accredited for programmatically articulating a comprehensive multi-level model for discourse analysis, which combines close textual analysis, the study of discourse processing and macro socio-cultural analysis and, in doing so, linking micro-analysis of text and talk with the explanatory ambitions of social theory and emancipatory goals of social critique. CDA has been agenda-setting for a scale-sensitive discussion of the connections between situated language use, power and ideology and has acted as a broker for much social theoretical work within discourse studies. The publication of Fairclough (1989) certainly counted as a watershed publication in British applied linguistic and sociolinguistic circles, more generally. In fairness, one must note that this development occurred alongside, and throughout the 1990s has in fact been in growing interaction with, comparable programmatic aspirations which originated in other traditions (Briggs, 1996, in linguistic anthropology).

Since the mid-1980s and into the present, one can identify in the history of discourse analysis an accumulative dialogic engagement with social theoretical work on questions of language, representation, ideology, power, equity and identity. The oeuvres of Michel Foucault and Pierre Bourdieu have undoubtedly been central, although the list has grown considerably since. The conversations with social theoretical work, alongside broader analyses of contemporary socio-cultural and economic developments, have enabled discourse analysis to keep close to work that is on the pulse of the contemporary social era and its specific currents. One can refer here to the impact of Giddens’ (1991) thematisation of the reflexive individual as an aspect of the late modern condition; Harvey’s writings on globalisation and neo-liberalism (Harvey (1996); Wernick (1991) on promotional culture; Anderson (1991) on nationalism and narrating the nation; Bhabha on cultural hybridity within a postcolonial perspective (Bhabha, 2004)). These conversations, in their turn, have played an important role in shaping the research agendas of discourse analysts.

When the term ‘discourse’ is used in its countable form with the particular qualification of its reality-creating and sustaining capacities (e.g. capitalist discourses, sexist discourses, medical discourses and discourses of education), this sometimes counts as a reference to typical patterns of interaction and language use, and sometimes as a reference to ‘bigger’ ways of reading, coding and interpreting the world which are associated with a particular locus or domain of social activity. Here it is imperative that we refer to Gee’s (1999) distinction between discourses (language-in-use; noted regularities of occurrence) and Discourses (which Gee defines as: language plus “other stuff”, i.e. practices which render language- in-use socially significant and consequential). A somewhat comparable attempt to integrate the two perspectives in a single model can be found in Wetherell’s (1998) suggested dual focus on “interactional sequence” and “interpretative repertoire”. Both instances can be seen as examples of how in the combination of complementary perspectives, micro-analysis of text and interactional sequence may be combined with the kind of macro perspective that is invited by much discourse research today.

Current contributions: the contemporary range of discourse analytical engagements

While one can historically detect a gradual progression from an early, more narrow linguistic interest in the formal and functional mapping of various aspects of the situated utterance to a later, more process-oriented research perspective which lends priority to social questions such as the connections between language and identity, it is probably more accurate to state that discourse analysis has crystallised within language studies in two directions. One can note, on the one hand, a continuation of more linguistic uses of the term, in which discourse is viewed as the layer of meaning which is tied more directly to situations of language use. The focus here is often on large collections of verbal material of a particular situation or activity type and the use of quantitative methods and techniques of corpus linguistics (Biber et al., 2007). Specific discourse-related themes are selected for closer attention (e.g. Warren, 2006 on aspects of naturalness in informal conversational language use). On the other hand, recent decades have witnessed the formulation of a broad project of discourse studies which more holistically views language use, often in combination with other forms of semiotic behaviour, from the angle of ‘social practice’. Much discourse research thus simultaneously attends to aspects of text and talk, processes of interpretation and cognition, and social- actional dimensions of communicative behaviour as well as its functioning at the level of ideological reproduction and socio-cultural transformation.

Especially in the latter tradition, discourse analysis has often (if not mostly) stood in an applied relationship to the social world, with discourse research oriented to the identification of recommendations for practice (here echoing Brumfit’s definition of applied linguistics as engaged with real-world problems and issues in which language plays a central role (Brumfit, 2001)). A quite arbitrary list which attempts an impression of the range of possible themes could include: contextual complexity in the processing of cartoons by patients with unilateral lesions (Dagge & Hartje, 1985), the effects of story sequencing on affective reactions to news broadcasts (Mundorf & Zillman, 1991), self-commodification in dating advertisements (Coupland, 2006), pedagogical focus in foreign language classrooms and the use of repair strategies (Kasper, 1986), media coverage of the genetically modified food debate (Cook et al., 2006), the communication of rights in contexts of police arrests and detention (Rock, 2007), and so on and so forth. Themes under the heading of registering discourse change in response to shifts in socio-cultural values are also many and varied. One such theme has been the interest in ‘technologies of discourse’ (Fairclough, 1992; Cameron, 2000).

Specific fields of application for discourse research have given rise to specialist off-shoots, such as professional discourse studies (Gunnarsson et al., 1997; Sarangi & Roberts, 1999; Candlin & Crichton, 2011) with further sub-divisions for medicine (Gotti & Salager-Meyer, 2006; Iedema, 2007), law and forensic science (Philips, 1998; Coulthard & Johnson, 2007; Rock, 2007), social work (Hall et al., 2006, 2014), etc. Discourse perspectives have been articulated for specific language-related interests. For instance, Hatim and Mason (1990) have done this for translation studies; Wadensjö (1998) and Roy (2000) for interpreting studies; Carter and Simpson (1989) for stylistics. While Barton (2007), Street (2003) and Collins and Blot (2003) have formulated a (critical) discourse analytical programme for literacy studies (see Gillen and Ho, this volume), Larsen-Freeman (1980) brokered comparable territory for second language acquisition research. Education has, of course, also been a major focus (Roberts, 2004; Wortham, 2006; Mosckovich, 2007), and media discourse (Scannell, 1991; Bell & Garrett, 1998; Jacobs, 1999; Tolson, 2001; Montgomery, 2007). Discourse analysis can thus be summed up as entailing a multi-perspectival take on language use and social life. The themes of identities-in- discourse and identities-as-outcomes-of-discourse are undeniably among the most commonly addressed in research across social domains and fields of application. Instances of discourse analysis will in many cases also draw seminally on various traditions in the study of language use or semiotics. For instance, discursive psychology (Billig, 1987; Potter & Wetherell, 1987; Antaki, 1988; Edwards, 1997) has developed themes from cognitive psychology such as the nature of everyday explanations, the functioning of memory and attitude by bringing together a conversation analytic perspective with social psychological constructivism.

Multimodal discourse analysis (Kress & van Leeuwen, 2001; Scollon & Wong-Scollon, 2003; O’Halloran, 2004; Norris, 2004; Jewitt, 2005) has drawn substantially on a systemic- functional perspective on meaning making for the development of a discourse analysis which is not restricted by an interest in verbal modes of communication, developing instead an interest in how the use of language interacts with other modes and modalities. Having said that, discourse analysis today often continues to lack on the side of a fuller engagement with the terrain of the body-actional behaviour and interactants’ engagement with objects within visual range, despite pioneering work by authors such as Charles Goodwin, Christian Heath and Adam Kendon (e.g. Goodwin (1997) on colour categories, Goodwin (2000) on moves in a game of hopscotch and Goodwin (2014) on gesture in a framework of co-operative action; Heath (1986) on body movement in medical interaction, Heath and Luff (1992) on multimedia crisis management in London Underground line control rooms and Heath (2002) on the gestural embodiment of symptoms; Kendon (2004) is a landmark publication for the study of gesture).

Critical issues and debates: discourse analysis and linguistic ethnography

Discourse analysis has always been a seminal and hybrid exercise. As a result, it may well have been insufficiently theoretically cumulative and methodologically systematic across different studies for it to constitute a bounded discipline in its own right. Lack of standardised terminology and a unifying conceptual framework may put it in a disadvantaged position in a cross-disciplinary perspective. This raises the question of its suitability to fill in the ‘linguistic’ component within a programme of linguistic ethnography. Rampton (2007, p. 3) expresses the view that linguistic ethnography amounts to an umbrella term for different research traditions. His list includes “interactional sociolinguistics and new literacy studies, as well as certain types of critical discourse analysis, neo-Vygotskyan research on language and cognitive development, and interpretive applied linguistics for language teaching”. The point about linguistic ethnography then is not the range of sub-traditions and the differences between these, but instead a common subscription to a complementarity between, on the one hand, an ethnographic take on the empirical and detailed investigation of contexts of communication, and on the other hand, detailed analysis of the internal organisation of verbal, linguistic and other kinds of semiotic data.

In the words of Tusting and Maybin (2007, p. 578),

UK linguistic ethnography includes a cluster of research which studies relationships between the micro-level of language practices and the broader context and social order, drawing on linguistics, social theory, and an ethnographic methodology which places the researcher at the heart of the research. This approach highlights particular issues for research into language practices.

In this respect, it is interesting to note that, while discourse analysis most surely presents itself as one of the ways of doing the linguistic in linguistic ethnography, the relationship is by means exclusive of other traditions that may fill in what can be meant by ‘detailed language analysis’, although Wetherell (2007, p. 661), in a comment on linguistic ethnographic publications, observed how it produced “some hugely important pieces of research and some of the most inspiring and exciting discourse work to be found anywhere”.

The question can also be phrased in different terms. Is linguistic ethnography likely to involve another linguistic component than one which is oriented to understanding the creation of meaning in contexts of real and naturally occurring language use? The answer may well be ‘no’, but diverse traditions of analytical engagement with naturally occurring language use of course do exist and turn out to be relevant. So, while discourse analysis (following Rampton) is not necessarily a preferred partner in a linguistic ethnographic research set-up, a discourse analytical perspective often does come in, and it may come in sideways. This is, for instance, the case where linguistic ethnography involves the adoption of, say, a new literacy studies perspective, which is discourse analytical in its orientations but clearly involves more than just the analysis of discourse.

In addition, one must note that the relationship between linguistic ethnography and traditions of discourse analysis has at times been a critical one. For instance, Tusting and Maybin (2007), wary of the risk of foregone interpretative closure and more tentative in the expression of critical engagements, insist on a cautious relationship with critical discourse analysis. More strongly, such caution may well have provided one of the reasons for the emergence of linguistic ethnography as a project of language study. Slembrouck (2005) develops a similar argumentation by locating the critical efforts of discourse analysts in the processual and dialogic terms of ongoing ethnographic contact. So, is this a matter of expected convergence of perspectives? Arguably, discourse analysis requires an ethnographic turn in its own right, irrespective of its teaming up with work that self-identifies as ‘linguistic ethnographic’. Ethnography is then a matter of how discourse analysis is informed theoretically, how it does not take context for granted, insists on separate analyses of context and an engagement in the terms which speak from it (terms which will often be language-use-related but may be quite different). The complementary of discourse analysis and ethnography within linguistic ethnography may well actually be about the necessary intertwining of accounts of interaction and their contexts, on the one hand, and analysis and interpretation of interactional or textual detail, on the other hand.

It is worth continuing to quote Tusting and Maybin (2007, pp. 578–579) here:

Broadly speaking, these [issues] relate to the nature of the truth claims which can be made on the basis of such work, in regard to the position of the researcher, tensions between participants and analysts, the impact of researchers’ political commitments, the relationship between different levels of social structure, and debates between realist and constructionist views.

What is perhaps centrally at stake, then, is an openness which is required – and more than that, a necessary epistemological reflexivity which extends not only in the direction of the specific dialogic engagements with society and its actors-participants (see Patiño-Santos, this volume), but just as much in the direction of how one’s own field of scientific enquiry and activity is being shaped. Tusting and Maybin (2007, p. 580) develop the case of linguistic ethnography’s ancestry in a liberal, humanist commitment:

[this] may be a strength of linguistic ethnography, in that it leaves the range of questions that can be addressed relatively wide, and reduces a risk of intrinsic bias for which approaches like critical discourse analysis have been criticised (e.g. Widdowson, 1998). However, the lack of explicit articulation of a political position for linguistic ethnography may also be a weakness. Where most work in a particular field shares broadly similar underlying political perspectives, the answers to some important questions about the social structures within which action takes place may be assumed rather than examined. These hidden assumptions can shape the development of the field quite considerably, by framing what questions can be asked and the sorts of answers which are acceptable.

For discourse analysis specifically, the main conclusion then is that it serves to benefit considerably from adopting an ethnographic perspective, while embracing the role which it can play within specifically linguistic ethnographic research designs and epistemologies.

Research methods and future directions: recording, transcription and the relationship with fieldwork

It is hard to think of discourse analysis and not to include one or two paragraphs on the many advantages which audio, audio-visual and later digital recording have introduced, when it comes to producing reliable and replicable data and the potential of this for analysing the social-in-the-interactional. Recording technology has over the past century been improved and refined and its current scope has been expanded tremendously by the advent of digital recording and storage (think, for instance, of the use of radio-microphones, which register the sound within a person’s aural range). The current methodological dictate is still that recordings are to be transcribed before analysis. A transcription then counts as a written re-entextualisation of a stretch of recorded speech, often using a combination of conventions specific to writing together with other graphic devices which highlight the relevance of particular aspects of speech, other relevant behaviour and dimensions of activity and situation. The purpose of a transcription is to ‘freeze’ the interaction so that it becomes amenable to repeated detailed scrutiny for purposes of analysis. Transcriptions come with varying degrees of sophistication and detail, and transcription practice has been detailed as posing particular challenges of authenticity/credibility, accuracy, accessibility/readability, translation and interpretative relevance (Bucholtz, 2008), while being ‘coloured’ by specific histories and traditions in the representation of orality in print (Jaffe, 2000). While transcription practice in discourse analysis originally tended to restrict itself to the representation of features of talk (including some paralinguistic features; cf. the Jefferson model, Jefferson, 2004), nowadays, with the use of computer software, some of the burdens put traditionally on the written transcription have been lifted. On screen, transcriptions and video recordings are now graphically aligned enabling simultaneous access to a transcribed section and its corresponding video fragment. This is made possible by software such as NVivo transcription.

At the same time, recording practices and the practice of recording have also been debated, highlighting, among other things, the importance of preparatory conditions of sustained observation before recording takes place, the issue of researcher presence during and note-taking during/after the recorded events, the effects of recording itself on the interaction which is being captured, as well as attention to how recordings are selectively invested with viewpoint (Slembrouck, 2005, p. 634). Very recent discussions have swung the pendulum back in the direction of sustained direct observation and researcher field notes (see Papen, this volume), as enabling a more comprehensive take on ‘data events’. Viewed from this angle, the data techniques of participant observation and interviewing as part of doing ethnography, and those of recording interaction in preparation of detailed analysis of language data present themselves as a complementarity which comes with different kinds of replicability.

My final point is that recent developments in software have actually contributed to this complementarity, as the digital condition also affords new possibilities in the area of integrative databases in which it is not only possible to read the transcription while the recording is being viewed or listened to, but other forms of data can be accessed alongside, with potentials for cross-coding and cross-medial analysis. Not only has transcription practice been relieved of some of the pressures of exhaustiveness, detailedness and authenticity; at the same time, the terms of what can be included in a digital data corpus have shifted, enabling data holdings of quite diverse ancestry and nature.