7   Discourses
Text analysis issues
Discourses at different levels of abstraction
'Interdiscursive' analysis of articulation of discourses in texts
Equivalences and differences
Semantic relations between words (synonymy, hyponymy, antonymy)
Collocations
Social research issues
The 'new spirit of capitalism' Classification
The identification and analysis of discourses is now a preoccupation across the humanities and social sciences. Foucault (1972, 1984) has been a decisive influence. Commenting on his own use of the word 'discourse', he writes:
I believe I have in fact added to its meanings: treating it sometimes as the general domain of all statements, sometimes as an individualizable group of statements, and sometimes as a regulated practice that accounts for a number of statements.
(Foucault 1984)
The analysis of discourse for Foucault is the analysis of the domain of 'statements' — that is, of texts, and of utterances as constituent elements of texts. But that does not mean a concern with detailed analysis of texts — the concern is more a matter of discerning the rules wliich 'govern' bodies of texts and utterances. The term
124   Discourses and representations
'discourse' is used abstractly (as an abstract noun) for 'the domain of statements', and concretely as a 'count' noun ('a discourse', 'several discourses') for groups of statements or for the 'regulated practice' (the rules) which govern such a group of statements. Foucault's work has been taken up in many different theories and disciplines, producing a rather bewildering range of overlapping and contrasting theorizations and analyses of 'discourses' (Dant 1991, Macdonell 1986, Mills 1997). I see discourses as ways of representing aspects of the world — the processes, relations and structures of the material world, the 'mental world' of thoughts, feelings, beliefs and so forth, and die social world. Particular aspects of the world may be represented differently, so we are generally in the position of having to consider the relationship between different discourses. Different discourses are different perspectives on the world, and they are associated with the different relations people have to the world, which in turn depends on their positions in the world, their social and personal identities, and the social relationships in which they stand to other people. Discourses not only represent the world as it is (or rather is seen to be), they are also projective, imaginaries, representing possible worlds which are different from the actual world, and tied in to projects to change the world in particular directions. The relationships between different discourses are one element of die relationships between different people — tiiey may complement one another, compete with one another, one can dominate others, and so forth. Discourses constitute part of the resources which people deploy in relating to one another — keeping separate from one another, cooperating, competing, dominating — and in seeking to change the ways in which they relate to one another.
Levels of abstraction
In talking about discourses as different ways of representing, we are implying a degree of repetition, commonality in the sense that tiiey are shared by groups of people, and stability over time. In any text we are likely to find many different representations of aspects of the world, but we would not call each separate representation a separate discourse. Discourses transcend such concrete and local representations in the ways I have just suggested, and also because a particular discourse can, so to speak, generate many specific representations.
But discourses differ in their degree of repetition, commonality, stability over time, and in what we might call dieir 'scale', i.e. in how much of the world they include, and therefore in die range of representations tiiey can generate. As in the case of genres (see chapter 4), it makes sense to distinguish different levels of abstraction or generality in talking about discourses. For instance, there is a way of representing people as primarily rational, separate and unitary individuals, whose identity as social beings is secondary in that social relations are seen as entered into by pre-existing individuals. There are various names we might give to this discourse
Discourses    125
— for instance, the individualist discourse of the self, or the Cartesian discourse of the subject. It has a long history, it has at times been 'common sense' for most people, it is the basis of theories and philosophies and can be traced through text and talk in many domains of social life, and its 'scale' is considerable — it generates a vast range of representations. On a rather less general, but still very general, level, we might identify in the domain of politics a discourse of liberalism, and within the economic domain a 'Taylorist' discourse of management. By contrast, in Fairclough (2000b) I discussed the political discourse of the 'third way', i.e. the discourse of 'New Labour', which is a discourse attached to a particular position within the political field at a particular point in time (the discourse is certainly less than a decade old). Example 9 is taken from a management 'guru' book which is the focus of Chiapello and Fairclough (2002). The background to that paper is the analysis by Boltanski and Chiapello (1999) of what they call the 'new spirit of capitalism' — or die ideology of what I have been calling new capitalism. Their analysis is based upon management texts rather like Example 9, and the objective of my paper with Chiapello was to see how their 'new sociology of capitalism' could be enhanced by using critical discourse analysis, allowing a more detailed account of how the 'new spirit of capitalism' is textured in management texts. We might see the 'new spirit of capitalism' as a new discourse which has emerged from combining existing discourses. Here is a brief illustration (not included in Example 9) of how such combinations are textured:
Seven classic skills are involved in innovation and change: tuning in to the environment, kaleidoscopic thinking, an inspiring vision, coalition building, nurturing a working team, persisting through difficulties, and spreading credit and recognition. These are more than discrete skills; they reflect a perspective, a style, that is basic to e-culture.
The 'style' which is 'reflected' in this list is how the 'new spirit of capitalism' represents the 'leader' in business enterprises. The list works together into a relation of equivalence expressions which emanate from and evoke different discourses — listing is a texturing device for effecting the combination of discourses which constitute the new discourse. But one can see tiiis process of combination as layered. Boltanksi and Chiapello (1999) suggest that the 'new spirit of capitalism' centrally articulates together 'inspirational' and 'connexionisť discourses (or what they actually call 'cites', or 'justificatory regimes') — leaders are people who combine vision with good networldng, to put it rather crudely. The first three listed elements ('tuning in to the environment, kaleidoscopic thinking, an inspiring vision') emanate from the 'inspirational' discourse, whereas the fourth ('coalition building') emanates
126   Discourses and re-presentations
from the connexionist discourse. Yet the first three listed elements can themselves be seen to emanate from different discourses — 'tuning in' is a metaphorical use of an expression in technical discourse which evokes a discourse of personal relationships, perhaps a counselling discourse, where the quality of how one listens to others is in focus; 'kaleidoscopic thinking' evokes perhaps popular psychology texts on creative thinking; whereas 'inspiring vision' would seem to emanate from a discourse of art criticism). So the 'inspirational' discourse can itself be seen as an articulation of discourses.
Example 9 itself shows a similar texturing together of discourses, though in this case it is a matter of both equivalences within the 'new spirit of capitalism', and differences between it and the 'old' discourse (see the discussion of equivalences and differences in chapter 5). The texturing of the relationship of difference is effected through a range of contrastive or antithetical relational structures and expressions: X instead qfY, X not just Y, X but also Y, X is different from Y, more like X than Y. The clearest case is in the list in the centre of the extract, where what we might call the 'protagonist' discourse (the 'new spirit of capitalism') represented before the brackets is set off against the 'antagonist' discourse within the brackets. At the same time, elements in the list before the brackets are textured as equivalent, so are elements in the list within the brackets, and the different discourses from which these elements emanate are thereby articulated together.
The point of referring to different levels of 'abstraction' becomes clear as soon as we look in any detail at any of the discourses I have mentioned. They are all internally variable. Practically any treatment of Liberalism, for instance, is likely to on the one hand identify certain commonalities in Liberal representations of political life, but then go on to differentiate varieties of Liberalism. Even the discourse of the 'Third Way' is not homogeneous — one theme in the analysis I referred to is precisely how that discourse has varied and shifted in a rather short period of time. Why then talk about these heterogenous entities as 'discourses' at all? The answer cannot simply be based upon there being a certain commonality and continuity in the way the world is represented, as well as variation. It is also based upon the dialectical relationship between discourse and other elements of social life — that one distinguishes 'discourses' when particular ways (partly stable, partly variable) of representing the world are of social significance, perhaps in terms of the effectivity of discourse, its 'translation' into non-discoursal aspects of social life. Discourses can therefore be seen as not just ways of representing with a degree of commonality and stability, but such ways of representing where they constitute nodal points in the dialectical relationship between language and other elements of the social.
A further complexity is that discourses, except at the lowest level of generality, the level of the most specific and localized discourses, can themselves be seen as combinations of other discourses articulated together in particular ways. This is
Discourses   127
how new discourses emerge — through combining existing discourses together in particular ways. So, for instance, my analysis of the political discourse of the 'Third Way' saw it as a specific articulation of other discourses including social democratic and 'New Right' (Thatcherite) political discourses. The new is made out of a novel articulation of the old.
Texts and discourses
Different texts within the same chain of events or which are located in relation to the same (network of) social practices, and which represent broadly the same aspects of the world, differ in the discourses upon which they draw. For instance, Example 13 is an extract from a book written by two long-standing left-wing members of the British Labour Party on 'New Labour's' view of what it calls 'the global economy' and what they call 'capitalist globalization'. One difference between the representation of global economic change in this left-wing political discourse and the New Labour political discourse of the 'Third Way' is that 'transnational companies' are referred to as the agents dominating economic change — who 'divide and conquer'. In New Labour representations of global economic change, by contrast, these companies are not represented at all, and economic change ('globalization' and so forth) is represented as a process without social agents — as something which is just happening rather than something that people or companies or governments are doing (see Fairclough 2000b for comparative analysis of New Labour texts). Another significant feature of the left-wing discourse drawn upon in this extract is the semantic relations which obtain within it. Notice the different expressions used to represent transnational corporations — 'transnational companies', 'transnational capital', 'international capital'. Through rewording, a relation of equivalence, or synonymy, is textured between 'companies' and 'capital', between concrete and abstract. This sort of mapping of concrete, phenomenal forms of appearance ('companies') onto abstract, structural entities ('capital') is characteristic of a Marxist element which is evident in leftwing Labour discourse and which differentiates it from right-wing (and New Labour) discourse. Moreover, national governments (and the European Union) are represented as in a potentially antagonistic relationship to transnational companies/capital ('employing powers against' them, and acting in 'response' to them). This again is a characteristic of left political discourse — 'capital' is to be contested, fought against. And national governments are represented as potentially acting in alliance with trade union organizations (as well as non-governmental organizations more generally) on an international basis in accordance with 'internationalist' traditions — 'internationalism' here maintains its sense of the solidarity of labour, whereas in the discourse of New Labour is has come to refer to 'cooperation' between nation-states in the 'international community' (e.g. in bombing Yugoslavia). Notice also the concept of 'clientism', set up against
128    Discourses and representations
'employing powers against' or 'bargaining' with capital, which has no part in the political discourse of New Labour.
Texts also set up dialogical or polemical relations between their 'own' discourses and the discourses of others. In this case there is a critique of what New Labour says about 'partnership' and 'cooperation'. This is partly contesting the meanings given to these words within the discourse of New Labour, setting up a different discourse in which 'partnership' and 'cooperation' are articulated with 'trust', 'openness', 'respect'. And it is partly claiming (in an apparent allusion to New Labour's favoured 'not only but also' relations, e.g. 'cooperation as well as competition') that there is a covert hierarchy in New Labour discourse - 'enterprise' and 'competition' always comes before 'partnership' and 'cooperation'.
This dialogical/polemical relationship is one way in which texts mix different discourses, but their 'own' discourses are also often mixed or hybrid. An inter-discursive analysis of texts (see chapter 2) is partly concerned with identifying which discourses are drawn upon, and how they are articulated together. (We can see a text as drawing upon a discourse even if the realization ofthat discourse in the text is minimal — perhaps no more than a single word.) Let's look at Example 4 in this respect. Wodak, in the article from which the example is taken, traces the transformation of this text through successive versions in the course of meetings of the EU Competitiveness Advisory Group. Sentences 5-7 were added in later versions 'as a concession to the unions'. We can see this addition as a hybridization of discourses. The two main discourses here are, first, the neo-liberal discourse of economic change which represents 'globalization' as a fact which demands 'adjustments' and 'reforms' to enhance 'efficiency and adaptability' in order to compete; and, second, a political discourse which represents societies in terms of the goal of 'social cohesion' and threats to 'social cohesion'. These different discourses entail different policy priorities - policies to enhance competitiveness on the one hand, and social cohesion on the other. The discourse of social cohesion represents people in ways which are foreign to neo-liberal discourse - in terms of their feelings ('sense of unease, inequality and polarization'), and their 'hopes' and 'aspirations'. But sentence 7 is particularly significant in the way in which it articulates these discourses together by using vocabulary which works key categories within the two discourses into semantic relations — 'social cohesion' is reconstrued in economic terms as 'human quality' and 'the ability to work as a team' and as a 'source' of 'efficiency and adaptability'. Whereas the discourse of 'social cohesion' is a fundamentally moral and humane discourse which is oriented to people who have a 'sense' of belonging to a community, 'human quality' in particular reduces people to forces of production which rank along with others, such as information technology. Yet although these discourses can be seen as fundamentally incompatible in how they represent and imagine people, what we have here is a strategy of legitimizing the discourse of social cohesion in terms of the neo-liberal discourse.
Discourses    129
Identifying and characterizing discourses
How do we go about identifying different discourses within a text? We can think of a discourse as (a) representing some particular part of the world, and (b) representing it from a particular perspective. Correspondingly, in textual analysis one can:
(1)    Identify the main parts of the world (including areas of social life) which are represented - the main 'themes'.
(2)    Identify the particular perspective or angle or point of view from which they are represented.
For instance, the themes of Example 7 (Appendix, pages 239—41) include: economic processes and change, processes of (global and national) governing, political protest (the misnamed 'anti-globalization' protests), and views of globalization (in 'the Soudi'). Each of these themes is open in principle to a range of different perspectives, different representations, different discourses. In this case, economic processes and change (for instance the penultimate paragraph, on Ghana) are represented in the terms of the 'neo-classical', market-liberalization discourse of the 'Washington consensus' — in contrast for instance to Keynsian economic discourse. Governing is represented as 'governance', a term which is itself very much part of a neo-liberal discourse of governing which on the one hand represents governing as not only the business of governments but also 'the framework of global governance' (international agencies such as the World Trade Organization and the International Monetary Fund, which has been central in imposing the ' Washingtom consensus'), and on the other hand prescribes changes in governing in terms of 'transparency', 'accountability', and so forth. One can contrast this with more traditional state-centred discourses of governing.
I have suggested that discourses are distinguished both by their ways of representing, and by their relationship to other social elements. Focusing on the former we can specify ways ofrepresenting in terms of a range of linguistic features which can be seen as realizing a discourse.
The most obvious distinguishing features of a discourse are likely to be features of vocabulary — discourses 'word' or 'lexicalize' the world in particular ways. But rather than just focusing atomistically on different ways of wording the same aspects of the world, it is more productive to focus on how different discourses structure the world differently, and therefore on semantic relationships between words. One example is the relationship between 'transnational companies' and 'transnational capital' in Example 13, discussed above. The former is reworded as the latter in the text. One might see this as a local texturing of semantic relations — new semantic
130   Discourses and representations
relations are indeed set up in texts, and that is part of the work of social agent (and the causal effects of agency, see chapter 2) in making meaning. But in this case a comparison of texts within this political tradition, including other texts of these authors, would suggest that the rewording draws upon and evokes the way of structuring the world associated with this discourse, rather than setting up a new relation. One might say that the text takes as given, presupposes, what one will find explicitly asserted and argued for elsewhere in texts which draw upon this discourse: that (transnational) companies are a phenomenal form of appearance of (transnational) capital. Semantically, we can say that 'companies' is a hyponym of 'capital', along with other 'co-hyponyms' such as 'trusts' and 'financial markets'. This presupposed structuring of the world, and this presupposed semantic relation, is both what allows the writers to reword 'companies' as 'capital' without having to make the relation explicit, and what allows readers to make sense of the text.
Another example of such a covert semantic relation is in the relationship between 'globalization' and 'economic progress' in sentences 1 and 2 of Example 4 — the coherence of the text depends upon a relationship of hyponymy between them, that 'globalization' is a hyponym of 'economic progress'. In Example 1, the employees of the company are classified into three groups — 'senior management', 'the bottom end' and 'us', where 'us' is middle management. These can be seen as co-hyponyms, and constitute a taxonomy, though it is not clear what the superordinate term is (i.e. what they are hyponyms of): perhaps 'workforce' is used in this way when the manager explains 'bottom end' in response to the interviewer's question: 'Of the workforce'. But 'workforce' is in contrast with 'managers' when the manager says 'take the power from the unions and give it back to the managers and give it back to the workforce', and perhaps a synonym of 'the bottom end'. These are not meaning relations one is likely to find in a dictionary, because they are specific to particular discourses. In addition to hyponymy ('meaning inclusion'), and synonymy ('meaning identity'), they include antonymy ('meaning exclusion'). For instance, in the discourse of social cohesion drawn upon in Example 4, the antonyms of 'social cohesion' include 'polarization' (in the text) as well as 'social exclusion' (not in the text).
What is at issue here is classification, preconstructed classificatory schemes or systems of classification, 'naturalized preconstructions . . . that are ignored as such and which can function as unconscious instruments of construction' (Bourdieu and Wacquant 1992), preconstructed and taken for granted 'di-visions' through which people continuously generate 'visions' of the world. When different discourses come into conflict and particular discourses are contested, what is centrally contested is the power of these preconstructed semantic systems to generate particular visions of the world which may have the perfomative power to sustain or remake the world in their image, so to speak.
The vocabularies associated with different discourses in a particular domain of social life may be partly different but are likely to substantially overlap. Different
Discourses    131
discourses may use the same words (for instance, both neo-liberal and 'anti-globalization' discourses use 'globalization'), but they may use them differently, and again it is only through focusing upon semantic relations that one can identify these differences. One way of getting at this relational difference is through looking at collocations, patterns of co-occurrence of words in texts, simply looking at which other words most frequently precede and follow any word which is in focus, either immediately or two, three and so on words away. Sometimes one is struck by collocations in particular texts. For instance, the word 'globalization' occurs in Example 7 in collocation with 'overpowering' ('fear that overpowering globalization will force the extinction of national cultures and traditions'). This is a text produced by an organization which has been strongly supportive of neo-liberalism, but which is giving voice to concerns about the negative effects of 'globalization', and drawing in discourses, as this collocation indicates, which one is unlikely to find in more conventional neo-liberal texts. But the most effective way of exploring collocational patterns is through computer-assisted corpus analysis of large bodies of text (McEnery and Wilson 2001, Stubbs 1996). For instance, in a corpus analysis of texts of New Labour and 'old' Labour (i.e. texts from earlier stages of Labour Party history), it emerged clearly that although the word 'work' was, rather obviously, rather common in both, its collocative patterns were different. 'Back to work', 'into work', 'desire to work', 'opportunities to work', 'Welfare-to-work'reflect common collocations in the New Labour corpus, whereas 'out of work', 'right to work', 'democracy at work', 'health and safety at work' reflect common patterns in the 'old' Labour corpus. Generalizing over the results, the focus in New labour is on getting people off welfare and into work, the focus in 'old' Labour is on improving conditions and relations in work, on unemployment as an infringement of the 'right to work' and a responsibility for Government (Fairclough 2000b).
Discourses are also differentiated by metaphor, both in its usual sense of 'lexical metaphor', words which generally represent one part of the world being extended to another, and what I shall call in the next chapter grammatical metaphor (e.g. processes being represented as 'things', entities, through 'nominalization'). Let me make some comments on lexical metaphor (see Goatly 1997). In Example 9, competition between companies is represented metaphorically as a race. The 'best' companies are 'pacesetters', like the runner who takes the lead and sets the pace in a race. The 'worst' companies are 'laggards', those who trail behind. Unlike 'pacesetter', 'laggard' is not specifically part of the vocabulary of racing, it broadens the metaphorical representation of companies as being like people to include other activities in which people are evaluated and graded in terms of performance (e.g. there are 'laggards' classrooms). Example 9 also explicitly elaborates a metaphorical representation of companies as 'communities' with 'members' (rather than just 'employees') who have 'shared understandings' and a 'feeling of connection', and so forth. Such metaphors differ between discourses — metaphor is one resource
132   Discourses and representations
available for producing distinct representations of the world. But'it is perhaps the particular combination of different metaphors which differentiates discourses: the two metaphors I have identified here are common ways of representing companies which turn up in various discourses, and it is perhaps the combination of these and other metaphors which helps to differentiate this particular managerial discourse. The influential work of Lakoff and Johnson (1980) on metaphors which are deeply embedded within cultures (e.g. the metaphorical representation of arguing as fighting) is also relevant here.
I referred above to semantic relations being presupposed. In fact, presuppositions and assumptions can more generally be seen as discourse-relative — the categories of assumption I distinguished in chapter 3 (existential assumptions, propositional assumptions, value assumptions) can all be seen as potentially tied to particular discourses, and as variable between discourses. Potentially, because there many assumptions which are more or less pervasively held throughout societies or social domains or organizations. I made the point in chapter 3, for example in discussing Example 4, that assumptions may be discourse-relative, so I shall not repeat the argument in detail here. I also suggested in Chapter 4, in discussing Argument as a genre, that arguments often rest upon assumptions which are discourse-specific and discourse-relative (see Gieve 2000).
I referred earlier to the two main discourses in example 4, the neo-liberal discourse, and the discourse of social cohesion. Despite the contrast between them, there is one thing they have in common: they represent real social processes and events in a highly abstract way. Although one can say that they are ultimately referencing concrete and particular events, if highly complex sets and series of such events, they represent the world in a way which abstracts away from anything remotely concrete. One corollary of this is that many of the elements of concrete events are excluded. Processes ('globalization', 'progress') and relations ('social cohesion') and even feelings ('hopes', 'aspirations') — I shall use 'processes' in a general sense to include all these — are represented, but the people involved are for the most part excluded ('people' in sentence 6 is an exception, but the representation here is again very general — in fact 'generic', see chapter 8), as are other elements of social events, such as objects, means, times, places. Processes are in fact 'nominalizeď, not worded with verbs as they most commonly are, but with noun-like entities called 'nominalizations' ('globalization', 'cohesion'), or what one might call 'process nouns', nouns with the verb-like quality of representing processes and relations and so forth ('progress', 'hope'). Syntactically, these process-expressions operate like nouns — so, for example, 'social cohesion' in (5) is the subject of a (passive) sentence. When processes are nominalized or worded as process nouns, their own subjects, objects and so forth tend to be excluded. Contrast Example 12 (Appendix, pages 248—9) with Example 4 (page 236). The sort of ethnographically oriented sociological discourse of the former represents
Discourses    133
events more concretely, and includes more elements of events (including the people involved in them) in its representations, than either the neo-liberal discourse or the discourse of social cohesion, both of which are oriented to abstraction from and generalization over events in mainly policy-formation contexts.
What these comments point to is that discourses are characterized and differentiated not only by features of vocabulary and semantic relations, and assumptions, but also by grammatical features. Discourses differ in how elements of social events (processes, people, objects, means, times, places) are represented, and these differences can be grammatical as well as lexical (vocabulary). The difference between a nominalization and a verb is a grammatical difference, so also is the difference between transitive and intransitive verbs, the difference between generic and specific noun phrases (e.g. generic, general and inclusive, reference to 'the police', as opposed to specific reference to 'this policeman'), and so forth. These are some of the ways in which discourses differ in die representation of social events (see chapter 8 for more detailed discussion).
Summary
We have seen that discourses are ways of representing the world which can be identified and differentiated at different levels of abstraction - so that for instance what Boltanski and Chiapello (1999) identify as the 'new spirit of capitalism' can be seen as a discourse at a high level of abstraction which develops as an articulation of discourses. Texts differ in the discourses they draw upon to represent particular aspects of the world, and they articulate different discourses together (hybridize or mix discourses) in various ways. Discourses can be differentiated in terms of semantic relations (synonymy, hyponymy, antonymy) between words - how they classify parts of the world - as well collocations, assumptions, and various grammatical features.