3.1 What is Discourse?

The concept of discourse is defined in highly diverse manners. These differences arise from its heterogeneous background, which encompasses disciplines such as linguistics, philosophy, sociology and cognitive anthropology with their specific understandings and research traditions.37 It is therefore not surprising that ‘discourse’ is not consistently defined. In fact, the term has been broadened and differentiated during the last decades, and is used differently and filled with partly overlapping, partly contradictory concepts (Mills 2004; Keller 2007: ch. 2).38 The quest for a definition begins with the question of whether discourse is analyzed in a narrow linguistic sense – which is a rather descriptive enterprise – or in the broader sense as a social practice. Torfing (2005: 5–9), for example, distinguishes three generations of discourse theory, whereby the definition of discourse expands with each generation. According to this categorization, the first generation very much sticks to a linguistic perspective: it is mainly interested in semantics and defines discourse rather straightforwardly as spoken and/or written text that (mis-)represents reality. The second generation extends the definition of discourse and views it as both language use and social practice/interaction. Discourse is considered constitutive of a broad range of practices and part of the whole. A prominent representative of the second generation is the linguist Norman Fairclough (1992, 2013) who developed Critical Discourse Analysis (CDA).39 This approach analyzes semiotic systems and elements, including images and gestures, which are used to produce and interpret meaning. CDA mainly focuses on politics in terms of “semiotic dimensions of power, injustice, abuse, and political-economic or cultural change in society” (Fairclough et al. 2011: 357). The third generation broadens the concept of discourse even further. Whereas CDA views social questions as “in part questions about discourse” (Chouliaraki/Fairclough 2007: vii, my emphasis), the third generation subordinates them totally to discourse. From this post-structuralist position, as held by Jacques Derrida (1976) as well as Ernesto Laclau and Chantal Mouffe (2001 [1985], 1987), everything is discourse.40 Post-structuralists do not believe in the existence of any kind of extra-discursive ‘given’ essence, be it nature, reason or humanity that assists in structuring our being or social life (Torfing 2005: 13; cf. Derrida 1978: 279–281). The assumption that everything is discursive and permanently modified through actual practice leads to the postulate of ultimate contingency. This, however, leads post-structuralists to be criticized for denying “the possibility of any foundations for knowledge altogether” (Donnelly 2013: 19; cf. Onuf 1989: 40, 2001: 246; Chouliaraki/Fairclough 1999: 125; Jørgensen/Phillips 2002: 54).

Whereas the second (constructivist), and the third (post-structuralist) generation are divided on the question as to what extent reality and truth are internal to discourse, these generations also share commonalities. It is (largely) through language that the identities assigned to subjects and objects are constructed and bear meaning. Moreover, what becomes accepted as normal or legitimate is constituted in and through discourse and in turn this conditions our behavior and political action. Analyzing discourse therefore means exploring the constitutive link between language use and social practice, that is, between textual characteristics and political action, the (re-)production or transformation of knowledge and the perception of reality. Discourse analysis thus links linguistics, hermeneutics and social theory by relating the internal structures of discourses to larger social structures (van Dijk 1985c, 1997a; Torfing 2005). However, in a similar vein to what has been said about constructivism, discourse analysis does not aim to provide general theory for particular fields or topics. Rather, it addresses various social puzzles to help understand how the situation under investigation came about, and it often does so in combination with other theories that are relevant to the particular research question (Torfing 2005: 22). Furthermore, the second and third generations of discourse theory also share a critical approach. They view discourses as a means of (re-)producing unequal power relations between social/societal groups (Jørgensen/Phillips 2002: 63; cf. Fairclough/Wodak 1997: 258).41 In this case, ‘critical’ indicates that these power relations are analyzed from a normative perspective that provides for critique and aims to produce positive social change. Moreover, while Torfing attests the three generations of discourse theory a “gradual development towards a more inclusive and quasi-transcendental notion of discourse” (Torfing 2005: 9), he also holds that in practice the differences between the second and the third generation are small when actually carrying out analytical work.

3.2 CDA: Reality is More than Discourse

Similar to the discussion among constructivists about the role of language, there is dispute among discourse theorists about ontological and epistemological positions and hence the significance of discourse as the kernel of scientific research. Is ultimately everything determined through discourse or are there other parameters to reality? Many scholars do not agree with Derrida’s famous position that “there is nothing outside of the text” (Derrida 1976: 158). In their view, the post-structuralist perspective overstates the constitutive power of discourse. A major aspect of this controversy is represented by the question of whether the non-discursive exists.42

As mentioned above, for representatives of the third generation, there is indeed nothing outside discourse. All non-discursive actions and perceived phenomena are based one way or another on sedimented discourses – and are thus still discursive. In this understanding, discourse constitutes the social completely: discourse is the social (Laclau 1981: 176; Foucault 1984 [1971]: 127). In line with this assumption, individuals and identity are in their entirety the effect of their continued constitution (Butler 1993). Critics reject this stance since it indicates that the subject is ‘dead’. Norman Fairclough and his CDA disclaim the foregrounding of discursive performativity to the level where agents are merely interpreted as products of discursive structures. While Fairclough does acknowledge the constitutive power of language, he still emphasizes actors’ agency and intentionality. Actors “position themselves” inside discourses and deploy rhetoric self-consciously up to the point of normalization where people come “to ‘own’ discourses” (Fairclough 2003: 208). In this respect, CDA claims a truly mutual constitution of structure and agent and thus views individuals as both discursive products and producers of discourses (Fairclough 1992: 45).43 In general, Fairclough sees social life as produced by more ‘mechanisms’ (generative powers) than just discourse.44 In his understanding, to state that language is not a reducible component of social life is “not a matter of reducing social life to language, saying that everything is discourse – it isn’t” (Fairclough 2003: 2, cf. 1992: 66).

Researchers who use CDA define their approach to be both interpretive and explanatory at the same time (Fairclough/Wodak 1997: 278f). As mentioned earlier, according to Fairclough, texts have causal effects and can therefore explain changes in the social and material world. The kind of causality involved here, however, is not a mechanical one. There is nothing inevitable about its effects. It is not that particular “texts automatically bring about particular changes in people’s knowledge or behaviour” (Fairclough 2003: 8). In the end, the effect a text will have is determined by the interplay of the various factors that make up the context; the impact of a discourse “depends upon how it interacts with the preconstituted reality” (Fairclough 1992: 60, cf. 2003: 8f; see also section 3.3 below). Here, Fairclough assumes that a form of inertia exists within meanings that have already been established. According to him, constructivism becomes problematic “where it disregards the relative solidity and permanence of social entities, and their resistance to change” (Fairclough 2003: 209). In sum, CDA shows intersections with the ‘moderate’ positions of Alexander Wendt, yet strongly accepts the constitutive power of discourse and the possibility of its enactment and materialization. What is important for the study at hand is CDA’s acknowledgement of the mutual constitution of actors and discourses. In this sense, CDA is truly constructivist and thus apt for the discourse analysis conducted in this book.

3.3 Intertextuality, Context, and Change

As discussed in Chapter 2, constructivists point to the possibility of change. Whereas ‘moderate’ constructivists believe in more inertia than their ‘radical’ colleagues: ideas and norms, identities and interests are considered (at least in part) as constituted and thus subject to change. In the view of ‘radical’ constructivists, these transformations largely take place through discourse: “changes in discourse are a means by which the social world is changed” (Jørgensen/Phillips 2002: 9, emphasis in original; cf. Foucault 1972 [1969]: ch. IV/5: 166–177; Fairclough 1992: 5, 97, ch. 3, 2003: 8; Wodak 1996: 15; Fairclough/Wodak 1997: 264f). With regard to how these changes come about, they draw on what Julia Kristeva (1986: 37) terms ‘intertextuality’. This refers to the interdependency of texts and discourses that are defined by their relation to other texts and discourses and “can only be understood on the basis of others” (Wodak 1996: 14). According to this view, texts always draw on other texts, reproduce and transform them. After all, as Bakhtin (1986: 69) nicely puts it, no matter who is speaking, the speaker is not “the one who disturbs the eternal silence of the universe.” Rather, we build on former utterances, reactualize them and adapt their inherent evaluation (Bakhtin 1986: 69, 89; cf. Foucault 1972 [1969]: 98). This is the realm where change takes place. Intertextuality, as Fairclough (1992: 96) notes, “builds creativity in as an option.”45 Discourses provide the space for linguistic innovation, which enables new things to be stated by combining discursive elements in a new way or by introducing them into new situations and contexts. In short, rearticulation and recontextualization provides for shifts in meaning. Whereas changes like these might first occur in single discursive events, they can accumulate to finally change agents and structures on all levels from the individual to the (inter-)national (Fairclough 1992; Jørgensen/Phillips 2002: 7, 17).

Researchers agree that change through rearticulation and recontextualization happens to some extent unconsciously. In CDA and other ‘second generation’ approaches to discourse theory, however, this is also a power that can be seized by actors (Viehöver 2001: 178). Agency lies in the ability to deliberately shift meanings in order to bring about change in social structure. There is no automatism to this process, however, and the chances for change are limited. Just as Austin’s speech act theory demonstrated that single utterances have to meet certain conditions in certain contexts in order to have an effect, in a broader sense, this is also true for discourses. The ability to successfully contest or alter certain issues in discourse is a question of the larger context:

People are active – agents – to the extent that they are capable of pursuing collective or individual strategies in their discourse. But whether articulatory shifts in discourse constitute substantive shifts in identity […] depends […] on how the moment of discourse is inserted within the social process overall. (Chouliaraki/Fairclough 2007: 14)

In this light, the possibility for change is a matter of the particular situation with its sum of particular features and influences. The overall political or socio-cultural environment plays a role, as well as the meta-narrative in which the discourse is embedded. Agents might give the impetus for change, but the context establishes “the rationality of the situation” and provides “the framework for processes of deliberation, interpretation, and argumentation” (Zehfuss 2001: 68). Part and parcel of the context is the characteristics of structures, the degree of inertia underlying the established meanings and identities, as well as the discourses’ relation to society (van Dijk 1997a, 1997b). Moreover, as Fairclough (1992: 102f) points out, the possibility for change can also be constrained by social power relations (see section 3.4).

In general, text, including talk, is always “text in context” (van Dijk 1997a: 3) and thus historically and culturally embedded. By drawing on former texts and discourses, the new ones are nourished by this storied background while they transform past conventions into those of the present (Fairclough 1992: 85). This conditioning implies “the insertion of history (society) into a text and of this text into history” (Kristeva 1986: 39). Intertextuality thus includes historicity and suggests interpreting something in terms of a larger historical whole. Besides that, it points to the fact that discourses are constituted as part of a continuous process; consequently, there is no clear beginning and no clear end to it. Discourses have histories but they are also components of the present and the future.

3.4 The Power of Discourse

So far, it has been stated that language and thus discourse shapes our cognition and perception, identity and interest. Language thus has constitutive power with regard to who we are and how we see the world. However, there is more to say about the power of discourse, as the French philosopher Michel Foucault found out. In fact, his insights proved to be highly influential for the development of discourse theory.46 Most contemporary approaches to critical discourse analysis build one way or another on his work.

Foucault’s reasoning emanates from the assumption that discourse as social practice is integral to societal power relations and is therefore not reducible to speech (Foucault 1972 [1969]: 49). In contrast to Austin and Searle, who focused mainly on the force and performative power of individual utterances, Foucault investigated the collectivity of conditions that are inherent to discourses.47 He focused on the internal rules of discourse that govern the production of statements and practices (Torfing 2005: 7). He was interested in “the law of existence of statements, that which rendered them possible” (Foucault 1991a [1968]: 59, emphasis in original). These systems of rules govern the production of speech acts in two ways: first, they constrain what can be said; and second, they constrain by whom it can be said. Discourses make it possible or impossible for certain statements to occur in particular contexts. In this sense, discourses are “sets of sanctioned statements” (Mills 2004: 55; cf. Fairclough 1992: 40). Due to this characteristic, they narrow one’s field of vision as they “exclude a wide range of phenomena from being considered as real or as worthy of attention, or as even existing” (Mills 2004: 46). Certain perspectives towards a topic – including alternative interpretations that might have set other guidelines for social action – remain unexpressed. In a similar way, discourses set boundaries on who can participate in a particular discourse, that is, who is regarded as authoritative enough to contribute to it. Not all discourses are accessible to everyone: “None shall enter the order of discourse if he does not satisfy certain requirements or if he is not, from the outset, qualified to do so” (Foucault 1984 [1971]: 120; cf. Milliken 1999: 229).48 So, in terms of authority and legitimacy, discourses direct who has the right to speak on a particular occasion. As discourses constrain the range of possible statements as well as the people who can voice them, they are constructions of exclusion. This effect can be amplified by either the institutionalization of discourses – the more discourses are institutionally embedded the more they succumb to the specific rules and norms the institution demands – and/or by the sheer dominance of the discourse by which alternative discourses are marginalized or suppressed. As Fairclough points out for the latter case, “the seemingly limitless possibilities of creativity in discursive practice […] are in practice limited and constrained by the state of hegemonic relations and hegemonic struggle” (Fairclough 1993: 137, cf. 2013: 95).

This view also led Foucault to postulate the power-knowledge nexus. In other words, he analyzed discourses on the basis of power-knowledge relationships.49 As indicated above, discourses constitute our knowledge about the world and transport it. Consequently, knowledge is neither a neutral reflection of reality nor a simple expression of ideas. It is a discursively constituted awareness of the real. This form of knowledge production, however, is shaped by the power relations pervading the social world as they influence who can be heard, and who can dominate and control a discourse. In turn, the kind of knowledge that is produced and the way it is institutionalized constrains the range of actors who can participate in a discourse and what can be voiced within it. This contributes to the structure and practices of power relations in society (Jäger/Jäger 2007: 20; Dunn 2008: 80). Put simply, if the powerful express their belief systems through discourses, the knowledge generated will correspond to their world view. This in turn stabilizes their own power position while excluding contradicting or alternative meanings from discourse. Therefore, Foucault declares knowledge and power to be mutually constitutive, with knowledge emanating from power struggles. As a consequence, Foucault views truth – or, more correctly, the perception of it – as shaped by power relations whereas ongoing battles are not “in favour” of truth but “about the status of truth” and its political role (Foucault 2006 [1979]: 46f). It is the dominant, socially accepted discourses that shape our understanding and knowledge of the world. If statements belong to the sayable and are socially accepted or even dominant within a certain context, they are amenable to consistent repetition. Certain ‘truths’ or perspectives, thereby, become naturalized, which further strengthens their durability. This stabilizes and legitimates systems and practices which then become accepted as social objectivity.

Thus, for Foucault, discourses are instruments of power, which also means that there is an ongoing struggle to dominate them. There are power struggles within discourse and power struggles over discourse, or, as Foucault puts it: “Discourse […] is the thing for which and by which there is struggle, discourse is the power which is to be seized” (Foucault 1984 [1971]: 110).50 In sum, discourse is both a means of power and an effect of it (Foucault 1998 [1976]: 101). This very characteristic embodies the political nature of discourse. Critical discourse analysis often tries to unfold and examine the power relations inherent in discourse, which means the constraints on the acceptable, what can be said and what is silenced. It is an investigation of how meanings are established as truths – with all the consequences that follow from them.

3.5 Summary

This chapter briefly introduced different approaches to discourse theory. Following Torfing (2005: 5–9), the field can be structured into three ‘generations’, which can be classified as linguistic, constructivist, and post-structuralist. The second and third generations define discourse as constituting social reality. The discussion outlined that Norman Fairclough positions CDA in the field with a clear constructivist stance. While this approach was the main interest, the chapter also shed light on important insights into the way discourse is defined in third generation discourse theory.

The discussion also focused on in how far discourses are sources of change. Discourses are continuous processes that build on one another and thus link past, present, and future. Due to their intertextuality, they are platforms for rearticulation and recontextualization, and this provides for shifts in meaning and – as understood here – engaging in this realm can involve deliberate agency. Nevertheless, the rewording of things does not automatically lead to transformations in meanings and actions since this is also a matter of the larger social context. Moreover, the discussion demonstrated that discourses reflect power relations and the knowledge (re-)produced by and in them substantiate these power relations. Discourses confine the range of contributors and only permit certain positions to be voiced, while others are ignored or rejected as illegitimate. In this sense, discourses are exclusive constructions of power.

Having said that, analyzing discourse involves reconstructing the (historical) conditions under which particular discourses developed, that is, how ideas and identities, power and policies have been constituted and legitimized. The focus is on texts and the actions they enable – on what is said by whom and what is done within the political latitude that these discourses created. This also includes investigating the kind of effect that this has on society or the international level.

The aim of this book is to lay open the way in which conditions of possibility are constructed; this means demonstrating how discourse generates room for maneuver and makes particular policies reasonable or apparently inevitable – including those, that otherwise would have been discarded. This is discussed in the following chapter.