Discourse Studies, 1/3, 1999, 259-295.
Discourse studies and ideology:
On ‘liberalism’ and ‘liberalisation’ in three large corpora of English*
Robert de Beaugrande
The word is the fundamental object of the study of ideologies.
— Vološinov (1973 [orig. 1929]: 15)
Look at the very connotations of the word ‘liberal’, which outside the political context is taken to mean gentle, generous, nice. But put the political wrapping around it…
— data from the Corpus of South African English
particularly in the West, we often use the word ‘liberalization’. That is not a bad word.
— data from the Bank of English
A controversial question in critical discourse analysis has been whether and how discourse may manifest or at least implicate the ideologies of the discourse participants. This question should be seen in the context of the long history of uneasiness concerning whether ideology can be an object of inquiry for science, whose stance of authority and objectivity implies a claim to be freed of all ideology. This claim has been quite emphatic in formalist linguistics, which has even proposed to investigate human language in isolation from human society. By restoring the focus upon discourse in society, critical discourse analysis offers an occasion to subject ideology to new methods of investigation, and to formulate an explicit ideology for the field itself. As a new source of evidence, large corpus data can be collated for expressions which are presumed to be undergo ideological contestation, such as 'liberal' and 'liberalism', which are examined here in data from the UK, the US, and South Africa.
A. Science and ideology
1. ‘Ideology’ is a deeply problematic term. In many registers of discourse, it routinely connotes some fixed, unreasoning dogma that foments conflicts, as when Shils (1958) cited Fascism, Nazism, Bolshevism, and McCarthyism as examples. Geertz (1973: 197f) commented upon this conception:
Like the politics it supports, it is dualistic, opposing the pure ‘we’ to the evil ‘they’, proclaiming that he who is not with me is against me. It is doctrinaire in that it claims complete and exclusive possession of political truth and abhors compromise. It is totalistic in that it aims to order the whole of social and cultural life in the image of its ideals, futuristic in that it works toward a utopian culmination of history in which such an ordering will be realised.
The favoured candidate to defend against ‘ideology’ in this darkly pejorative sense is held to be ‘science’ (cf. Geertz 1973; Zima 1981; Pêcheux 1982):
the essential criteria of an ideology [are its] deviations from scientific objectivity […] The problem of ideology arises where there is a discrepancy between what is believed and what can be [established as] scientifically correct (Talcott Parsons):
Such a pronouncement complacently implies that science itself can and should be free of all ideology, and indeed heralds the ‘end of ideology’ (Geertz 1973: 199) in the final triumph of ‘objectivity’ and ‘correctness’.
2. Yet dictionary definitions suggest that ‘ideology’ is a more neutral, normal, and even necessary framework, viz.: ‘a systematic body of concepts esp. about human life or culture’ (Webster’s Seventh, p. 413); ‘a body of doctrine or thought that guides an individual, social movement, institution, or group’ (Random House Webster’s, p. 668); or ‘a belief or set of beliefs, especially the political beliefs on which people, parties, or countries base their actions’ (Collins COBUILD, p. 718). Surely these broad definitions might hold for science too.
3. Perhaps the ideology of science is so hard to recognise because there we encounter not so much a particular ‘body of concepts’ (or ‘doctrines’ or ‘thoughts’) as a set of general methods for arriving at ‘concepts’ — not so much a ‘theory’ as a modality for producing and testing theories. Such an ideology effectively functions as a meta-ideology — an ideology of methods as distinct from an ideology of content — and cannot be challenged and discredited in the direct and definitive ways that radical political ideologies like Nazism or McCarthyism can. Refuting a scientific theory like the ‘steady-state theory’ of the universe did not refute the science of astronomy or its methods, but rather vindicated its methods (in this case, observing red shift in the spectra of light dispersion).
4. But we do find a popularised ideology which can be called scientism, holding that only scientific knowledge is true and valid, and that the progress of science will eventually explain the entire universe and solve all of humanity’s problems. This fits Geertz’s pejorative sense quoted above (§ 1) by being ‘totalistic’ in that it aims to ‘order the whole’ of the universe, and ‘futuristic in that it works toward a utopian culmination’ of human knowledge. And the scientists may tend to be ‘doctrinaire’ in ‘claiming complete and exclusive possession of the truth and abhorring compromise’.
5. The ‘doctrinaire’ stance can entrain the scientist in a lonely and vertiginous paradox: the history of science demonstrates that all scientific theories so far have been proven wrong, yet the ideology of scientism encourages you to claim your current theory to be, at long last, the sole true one, whereas the scientists who support rival theories must be mistaken or misled by personal or institutional biases (cf. Kuhn 1970; Gilbert and Mulkay 1984). The paradox can erode the essence of science as an institution that should freely grant and defend the legitimacy of alternative explanations, and should acknowledge that these may contribute the most when we integrate them.
6. Scientism has also helped to inhibit science from accrediting ideology as a major object of scientific inquiry. Karl Mannheim’s (1936) attempts to incorporate a ‘non-evaluative conception of ideology’ into his ‘sociology of knowledge’ were stymied by the prospect that ‘nowhere is resistance to claims of objectivity greater than in the study of ideology’, as remarked by Geertz (1973: 195), who added: ‘men do not care to have beliefs to which they attach great moral significance examined dispassionately, no matter for how pure a purpose’; ‘they may find it simply impossible to believe that a disinterested approach to critical matters of social and political conviction can be other than a scholastic sham’. Yet Geertz also saw a major ‘problem’ in the way ‘social science’ tries to ‘handle ideology as an entity in itself — as an ordered system of cultural symbols rather than in the discrimination of its social and psychological contexts’ (1973: 195f).
7. Scientism thus subverts the authentic interests of science when the exaggerations and mystifications of the authority and power of science render the individual scientist unproductively possessive and defensive about his or her own ‘objective truth’. Moreover, science is prevented from effectively confronting and discrediting anti-scientific ideologies such as racism and sexism, and from blocking their attempts to expropriate and exploit scientism and raid sociology, psychology, or genetics for ‘proof’ of the ‘natural inferiority’ of minorities and women.
8. If, as philosophers of science like Kuhn (1970) have emphasised, a scientific theory can be discredited only by another theory, then we may reasonably assume that an ideology can be effectively discredited only by another ideology — and not by some ivory-tower standpoint purporting to be free of all ideology (§ 22, 31). Science can profitably investigate ideology, and, at the same time, explicitly develop creditable alternative ideologies for itself and for the institutions it sustains. Building upon Ulric Neisser’s (1976: 2) concept of ‘ecological validity’, i.e., whether ‘a theory has something to say about what people do in real, culturally significant situations’ and says it in ways that ‘make sense to the participants’, we could promote the ideology of ecologism, wherein the theory and practice of science are dialectically reconciled in a transdisciplinary enterprise of sustaining humane and democratic practices of action, interaction and discourse (Beaugrande 1997). Some trends in this direction have already been emerging, witness the impact of the Gaia Atlas of Planet Management (Myers et al. 1993) and of Gell-Mann’s (1994) landmark exposé of The Quark and the Jaguar. I would also see a similar drift in ‘critical linguistics’ and ‘critical discourse analysis’, even if these do not seem to provide for humane counter-ideologies (section D).
B. The ideology of modern linguistics
9. One science that has been deeply reluctant to study the nature and functions of ideology is modern linguistics. In the major treatises I have surveyed in fine detail, dating from the early 20th century up into the 1970s (Beaugrande 1991), the term ‘ideology’ simply doesn’t appear. Adapting Geertz’s terms from sociology (§ 6), its absence might reflect the mainstream programme of linguistics, believing that the properly scientific method would be to ‘handle language as an entity in itself — as an ordered system of symbols rather than in the discrimination of its social and psychological contexts’.
10. This austere programme encouraged mainstream linguistics to develop an ideology of idealisation, holding that language is based upon an ideal mode of order that is not readily evident in the ordinary practices of real discourse within society. Such was the source of the many static dichotomies that in effect disconnect ideal language from real language, such as ‘langue versus parole’ or ‘competence versus performance’ (Beaugrande 1998a) (§ 19). One label for this ideology, which is also being heavily camouflaged as a standpoint freed of all ideology (§ 1), might be the neologistic term linguisticism. ‘Linguisticism’ sustains the doctrine that ‘language’ is an abstract, uniform, and stable system whose nature and properties can be determined only by the ratiocinations of ‘linguistic theory’ and not from observing and recording discursive practices. Saussure (1966 [orig. 1916]: 8) hinted as much when he speculated that ‘other sciences work with objects that are given in advance’, whereas in ‘linguistics’, ‘it is the viewpoint that creates the object’. Hjelmslev (1969 [orig. 1943]: 18), who aspired to be a devout successor to Saussure, was more dramatic: ‘linguistic theory cannot be verified (confirmed or invalidated) by reference to any existing texts and languages’.
11. The same ideology was signalled, albeit less patently, by the declaration that ‘language is a well-defined object in the heterogeneous mass of speech facts’, whereas ‘speech cannot be studied’, nor indeed can it be ‘put in any category of human facts, for we cannot discover its unity’(Saussure 1966 [orig. 1916]: 14, 9, 11). The trend was reinforced half a century later, when Chomsky (1965: 3f, 201) announced that ‘linguistic theory is primarily concerned with an ideal speaker-hearer in a completely homogeneous speech-community, who knows its language perfectly’, whereas the ‘observed use of language’ ‘surely cannot constitute the subject-matter of linguistics, if this is to be a serious discipline’; indeed, ‘from the point of view of the theory’, ‘much of the actual speech observed consists of fragments and deviant expressions of a variety of sorts’.
12. The key question to challenge linguisticism might be: what shall be the sources of evidence for this ‘primary concern’, if, as Chomsky (1977: 192) cheerily grants, the ‘ideal speaker does not exist in the real world’? The de facto answer would be: the ‘linguistic intuition’ of the linguist ‘himself’ as a ‘native speaker’ (Chomsky 1965: 20). Yet Chomsky (1965: 8) has expressly denied that the ‘speaker of a language’ ‘is aware of the rules of the grammar or even’ ‘can become aware of them’. So either linguists acting as native speakers are also unable to ‘report’ ‘the rules of the grammar’; or else they hold superhuman powers for ‘becoming aware of and reporting’ what other speakers cannot. Either linguists do represent the community and cannot ‘report’; or they can report and do not represent the community.
13. Viewed in this light, the ideology of linguisticism legitimises a remarkable arrogation of power by the theoretical linguists to be the sole authorised representatives of a purely hypothetical ‘speech-community’ and therefore immune to contradiction or counter-evidence from real speakers in a real community. This arrogation sharply illuminates the consequences of linguisticism declining to study ‘actual speech’, and accounts for the tendency, typical of ideology, to foment conflicts within the discipline (cf. § 1). Once linguists have cut themselves off from the rational basis for developing and validating empirically sound theories of language, the fashion arises of fabricating ‘linguistic theories’ from the top down and applying them to odd handfuls of trivial invented sentences; and the contentious fragmentation of ‘theoretical linguistics’ we have actually seen is a natural consequence (Beaugrande 1998a).
14. The ideology of linguisticism has evidently foundered upon its own implicit dualism: equating language (‘langue’, ‘system’, ‘competence’ etc.) with perfect order, whilst equating discourse (‘parole’, ‘speech’, ‘performance’ etc.) with massive disorder. The direct corollary would be that using a language to produce discourse triggers an abrupt and catastrophic transition from stable and integrative order over to unstable and disintegrative disorder. Since this corollary is patently absurd, we must conclude that the linguisticism has make a capital mistake by attributing to ‘language’ an idealised mode of order which is fully determined and finalised within the abstract system. The ideology of ‘ecologism’ proposes instead a dialectic whereby the real order of language elaborately supports the order of discourse without fully determining or finalising it; discovering how that support actually operates is now a stimulating challenge for a transdisciplinary science of text and discourse (Beaugrande 1997).
C. The ideology of corpus linguistics
15. The early stages of corpus linguistics at present might be favourable for deliberating upon what ideology it could develop out of its strong potential to discredit the mainstream ideology of linguisticism. Corpus data immediately deconstruct the vision of ‘observed use of language’ constituting a mass of disorder and ‘deviance’ (§ 11, 14). But the order of discourse, not surprisingly, is not the mode of static and abstract order envisioned by linguisticism. We can easily recognise how the standing constraints that persist on the plane of the system (e.g., the English article going before the noun, not after it) continually interact with emergent constraints that are only decided on the plane of the discourse (e.g., the lexical choices appropriate to a political debate). And the local constraints among sets of selections and combinations interact with the global constraints of register, discourse domain, topic, and so forth. Linguisticism naturally projects disorder onto discourse after taking into account only a narrow and arbitrarily defined subset of local standing constraints which get reconstructed in a ‘grammar’ and hugely overburdened with the task of sustaining a fully determined and finalised system which is not (and cannot be) ‘reflected’ in discourse. As Sinclair (1991: 496) has remarked, ‘much of the apparent disorder is created by the perspective that is initially adopted’.
16. Corpus research also suggests describing the order of real language in terms of colligability, subsuming the ‘preferences’ of some grammatical options for appearing with certain others and holding the ‘grammar’ together; and of collocability, subsuming the ‘preferences’ of some lexical options for appearing with certain others and holding the ‘lexicon’ together. In a dialectical parallel, the order of discourse partly realises and is partly realised by the order of language in the actually occurring grammatical colligations and lexical collocations.1
17. The concept of the integrated lexicogrammar in systemic functional linguistics (e.g. Halliday 1994) further suggests that the grammar and lexicon hold each other together. This factor could explain why so many ‘grammars’ sponsored by linguisticism and designed to be independent of the lexicon have remained so fragmentary and so remote from authentic data. The lexicon, in its turn being regarded as another heterogeneous mass of disorder, has received scant attention in mainstream linguistics (cf. Bolinger 1970).
18. A major principle that corpus linguistics could now field against linguisticism would be that the order of language is dynamic and transitory, and in principle not describable by any static or ‘synchronic’ theory. To a significant degree, a language is always in the process of being created and negotiated whilst discourse is in progress; and a different generation of linguistic theories will be needed to explain how. The ideology sustained by corpus linguistics would accordingly be a version of dynamism wherein the specification of a theory will be far more actively data-driven and ‘bottom-up’ than the theories sponsored by linguisticism. Theories will no longer originate ‘from the top down’ every time some ambitious linguist chooses to fabricate his or her personal idealisation and illustrate it with a handful of fictional sentences (§ 13). Instead, data-based theories will evolve by being adjusted and tuned through continuing research on steadily larger corpora of authentic discourse (Sinclair 1997).
19. This evolution will undoubtedly affect the familiar dichotomies and divisions whereby those older theories aspired to freeze language into a Saussurian ‘well-defined object’ (§ 11). If these are to survive the test of authentic data, they need to be deconstructed and re-theorised as dialectical interactions: langue - parole, competence - performance, synchronic - diachronic, syntagmatic - paradigmatic, grammar - lexicon, language acquisition - language learning, and so forth.
20. Alternatively, less familiar terms or concepts can be introduced wherever the data seem to justify them, such as the parallel dialectics between ‘colligabilities’ and ‘colligations’, and between ‘collocabilities’ and ‘collocations’ (§ 16). In the present paper, I shall explore the prospect that ‘ideology’ may also find a new home as one of the sources of global constraints upon the orders of language and discourse, e.g., in the broad sense of ‘a systematic body of concepts about human life or culture’ (§ 2) (compare Christie and Martin eds. 1997; van Leeuwen and Wodak 1998). These concepts would be applied during discourse below conscious awareness and thus seem to be fully identical with life or culture; and those who hold contrasting ideologies seem to be distorting the ‘real world’, whence the well-known tendency of ideologies to foment conflicts (§ 1). To recognise the value of a contrasting ideology, you might first need to register the contingent and partial quality of your own; but doing so would require adopting, at least temporarily, another ideology from whose standpoint your own could be contemplated and contrasted. A vicious cycle impends: how to step outside your own ideology when you are not even able to see it as anything but the reality of life.
21. For similar reasons, criticising or attempting to change another person’s ideology may be perceived as acutely threatening, perhaps like trying to make them switch from reasonable over to unreasonable (§ 31). People may well defend and cling to their ideology even when they perceive symptoms of disorientation, fearing that they might lose whatever orientation they still have. The accumulating symptoms would foment increasing alienation, which could readily be expressed (though not alleviated) by hostility and aggression against people who hold a contrasting ideology.
22. How then can science investigate ideology if the latter is so deeply anchored at the base of human awareness? The principle of making an ideology explicit from the standpoint of a contrasting ideology might incur the usual risk of construing differences as distortions, especially where the ideology under investigation is defiantly positioned against science itself, as is religious fundamentalism. Also, even if we can design non-threatening techniques for bringing ideology to the conscious attention of the people who subscribe to it, we may risk transforming its functions we want to describe, perhaps in the way that literary techniques deliberately use framing or irony to undermine the illusion of reality in a narrative, and thereby disrupt the ‘willing suspension of disbelief’ (Coleridge) that literature solicits.
23. An alternative method would be to investigate large corpora of authentic data in terms of how ideologies leave traces within the order of discourse, even (or especially) when the discourse reflects or works out the disorders within the society (Wodak 1996). The traces should be conspicuous wherever the discursive order sponsored by one ideology is experienced as discursive disorder from the standpoint of another ideology. Conversely, discursive order may be merely superficial when our data show the same conception being appropriated by contrasting ideologies in order to project a deceptive consensus, e.g., onto the meaning of ‘democracy’ (Beaugrande and Williams in press).
24. However, research with large corpora may well discover that far less conspicuous selections and combinations of discourse options are also sensitive to ideological groundings (§ 88). Such was the discovery when feminism, though using more limited data, moved beyond the usual issues of ‘sexist language’ (e.g. male pronouns for everyone) to investigate the traces of the ideology of patriarchy inherent in the much deeper organisation of domains like ‘grammar’ (Beaugrande 1988; Cameron 1992; Wodak ed. 1997; Kotthoff and Wodak eds. 1998). Certainly, the degrees of detail and delicacy within language that have been exposed by sorting and searching large corpora of data extend far beyond anything detected before (Sinclair 1996). Might not a similar exposure be achieved for the traces of ideology?
25. What then of the ideology of research for the corpus linguists themselves? By the arguments advanced here, might they not overlook the traces of their own ideology and focus unduly on contrasting ideologies? One answer might be that research could recruit representatives of diverse ideologies, whereby the potential blind spots of any one would be compensated by the vigilance of the others. To be sure, recruiting scientists by explicit reference to their personal ideologies would be a sensational tactic after scientism has cultivated for so long the illusion of science standing free of ideology. And the problem of even identifying the suitable range of ideologies would be far from trivial, especially during early stages of the research.
26. An alternative answer could be that corpus research has a general effect of sensitising researchers toward the normally naturalised constraints upon discourse, including ideological ones. The patterns that emerge from the multiple means to query a data base are often unpredictable and surprising enough that we behold an image of own intuitions and habits of speaking which had never entered our conscious awareness before. As we repeatedly discover just which among the staggering range of potential selections and combinations are typically made, we are prompted to examine the order of discourses in unprecedented detail, a bit like the biologists who examined cell tissue under powerful microscope for the first time. At the same time, we can escape the quandary of linguisticism which replaced real language with ideal language and then claimed superhuman access to the ‘perfect knowledge’ of the ‘ideal speaker’ by virtue of ‘intuition’ as distinct from data (§ 12). Working with large corpus data emphatically refers the linguist back into the community of real speakers who produced and received the data and among whom we ourselves informally belong. How typical and representative we might be and how useful our intuitions might be are questions to be decided during the research and not by the glib arrogations of linguisticism (§ 13) (cf. Francis and Sinclair 1994).
D. Ideology in critical discourse analysis
27. The ancestors and precursors of what is presently called ‘discourse analysis’ include the data-driven approaches in linguistics and neighbouring disciplines (e.g. anthropology, ethnography, sociology) that were arrayed against the linguisticism of idealisation, and were for decades targeted with dismissive polemics promulgating the doctrine that ‘observed use of language’ ‘surely cannot constitute the subject-matter of linguistics’ as a ‘serious discipline’ (§ 11).
28. When ‘discourse analysis’ finally emerged as a field with that name in the 1970s, the early concerted efforts of Western ‘critical linguistics’ were also getting under way (e.g. Mey. 1979 [orig. 1974]; Fowler, Hodge, Kress, & Trew 1979). The ‘critical’ work was clearly distinguished by its resolve to accept ‘ideology’ (along with ‘power’, ‘domination’ etc.) as a legitimate object of investigation, finally in line with the neglected proposals of Vološinov (1973 [orig. 1929]: 9), who envisioned a ‘Marxist theory of ideologies’ as ‘the bases for studies of scientific knowledge’ and was targeted for Stalin’s death camps.
29. Perhaps to eschew the idealising and formalist tendencies of conventional ‘linguistics’, this work adopted the broader heading critical discourse analysis, devoted to ‘the analysis of linguistic and semiotic aspects of social processes and problems’ (Wodak 1996: 15). A key question for the new discipline is naturally how ‘ideology’ as such should be defined. In most of this research, the old pejorative connotations of ‘ideology’ have persisted but in a more precisely defined function, viz.: ‘ideologies are particular ways of representing and constructing society which reproduce unequal relations of power, relations of domination and exploitation’ (Wodak 1996: 18; compare now Wodak 1997; Wodak et al. 1998). Fairclough (1992: 67) offers a similar definition: ‘ideology is significations generated within power relations as a dimension of the exercise of power and struggle over power’. Lemke’s (1995: 12f, his italics) definition seems the most darkly pejorative: ‘ideology supports violence and is critically shaped by and in a context of violence’ and by ‘physical pain and social dehumanisation’.
30. No doubt such definitions reflect the justified urgency to critically analyse and deconstruct those ideological discourses which most actively legitimise or mystify power, inequality, domination, exploitation and violence, such as racism and sexism. Moreover, the traces of such ideologies in discourse should be the most accessible to practical analysis (cf. § 23). But in terms of the present discussion, building these pejorative effects into our basic definition of ‘ideology’ incurs at least three serious drawbacks. The first drawback concerns the ideological commitment of critical discourse analysis itself, especially when it repudiates the idealising ideology I have called ‘linguisticism’. By the definitions I have just quoted, the ideology of critical discourse analysis might be charged with sustaining one more ‘exercise of power’ and ‘domination’ — a prospect Fairclough (1996) himself has recently aired. How could we prevent our critical engagements with ideologies from being hit by the fall-out of a rigorously pejorative concept of ideology (§ 92)?
31. The second and closely related drawback follows from the plausible prospect that an ideology can be effectively opposed or deconstructed only from the standpoint of another ideology (§ 8). A pejorative definition forecloses our efforts to develop ameliorative counter-ideologies, such as ‘ecologism’, which expressly promote equality and solidarity (§ 8), e.g., in alliances with feminism and multiculturalism; and for defining such a counter-ideology to guide the projects of discourse analysis itself. Fairclough (1996, 1998) has expressed concern that if ‘ideology’ as ‘understood in the critical tradition’ gets redefined as I propose, the concept might get ‘appropriated’, ‘disarmed’, or ‘compromised’, leaving us with the ‘relativist conclusion that values are merely different, that all values are as good as all others’ (cf. § 43). Such relativist tendencies can undeniably be diagnosed in several trends in ‘post-structuralism’ and ‘post-modernism’ but by no means in ‘ecologism’, which insists that we cannot evade the choices between promoting either power or solidarity, either inequality or equality, and so on, by retreating either into relativism or into objectivity (§ 95). Fairclough would probably concur with ecologism that the hallowed ‘scientistic’ stance of objectivity in pursuit of ‘pure truth’ can imply an acquiescence to the existing structures of power and domination, as when a linguisticism dedicated to the ‘ideal speaker-hearer in a completely homogeneous speech-community’ (§ 12) elided the significant social consequences of language variations in a real speech-community (Beaugrande 1998a).
32. My point is rather that critical discourse analysis should not run the risk, inherent in many leftist and Marxist critiques, of appearing to offer only opposition and negation without a coherent alternative programme. At this stage of post-modern society, we behold a one-sided array of entrenched but ecologically unsound ideologies, such as consumerism, individualism, capitalism, free-marketism, and so on, whilst very few effective counter-ideologies on the other side even have established names, let alone organised groups of adherents (§ 85). Persuading people to exchange one ideology for another is hard enough; persuading them to throw out their ideology in favour of none at all seems frankly unrealistic (cf. § 21).
33. The third and again closely related drawback concerns the question of whether all discourses are ideological (e.g. Hodge and Kress 1988, Gee 1990; Lemke 1995) or only some discourses are ideological and others are not (e.g. Fairclough 1995; Wodak 1996; van Dijk 1998a). If (as if often the case) both positions favour a pejorative definition, the first is plainly the more pessimistic than the second in implying that power and domination will always control discursive practices; such an implication would certainly be rejected by emancipatory ideologies such as feminism. The second position would imply an enterprise of ‘de-ideologising’ discursive practices: of purging ideologies rather than changing or replacing them. How this might be achieved without fostering some disorienting ideological vacuum remains to be seen.
34. Some of Fairclough’s (1992: 91, my italics) deliberations might signify differences in degree rather than in kind:
Is all discourse ideological? I have suggested that discursive practices are ideologically invested if they incorporate significations which contribute to sustaining or restructuring power relations.[…] But all discourse is not thereby irredeemably ideological. […] the fact that all types of discourse are open in principle, and no doubt to some extent in fact, in our society to ideological investment does not mean that all types of discourse are ideologically invested to the same degree.
In a similar vein, Wodak’s (1996: 19, my italics) conception of critical discourse analysis ‘does not claim that all discourse is ideological: “it does not follow that because all practices are in ideology or inscribed by ideology, all practices are nothing but ideology” (Hall 1985: 103)’. And Lemke (1998) counsels that ‘a discourse formation is not necessarily inherently ideological, but is so only by virtue of its uses; it is the social function of the discourse which is ideological, and not the discourse itself’.
35. The question bears on both theory and practice. If we favour the data-driven theories advocated in section C (§ 18), then we should not pre-empt our conception of ‘ideology’, which has only recently begun to be theorised in earnest. The broad conception I have advocated of ‘ideology’ being a source of global constraints upon discourse in general (§ 20) would be most conducive to putting our prospective theories of ideology onto the broadest empirical basis and would be more perceptive toward the covert ideological groundings which leave less conspicuous traces among the selections and combinations of discourse. A pejorative conception, in contrast, would tend to attract our vision to the more conspicuous traces we can already link to ideologies of power and domination in intuitive and pre-theoretical ways.
36. In terms of practice, fundamental problems impend when we try to distinguish the set of discourses (or the portions within a single discourse) which count as ‘ideological’ from the set which do not. Moreover, the pejorative definition implies that once the ‘ideological’ ones have been discovered, the ‘non-ideological’ ones can be left out of our analyses, at least for the time being. But critical analysts, including Fairclough and Wodak, have often noted the significant tendency of ideological discourses of power to become less overt in order to encourage illusions of openness and democracy. If so, the demarcation between the two sets would be transient and unreliable, and we would tend to exclude some discourses whose ideological groundings could be discerned only after detailed critical analysis. This factor could be especially acute if the findings of critical discourse analysis itself become popularised in the news media, and the institutions of power and domination respond by adopting more devious and covert discursive strategies to escape detection.
37. Some of these problems might be attenuated by Fairclough’s proposal to recognise differences in degree (§ 34). But the problem may not prove much simpler of distinguishing degrees of ‘ideological investment’ (how ardently you support the ideology) from degrees of discursive overtness (how forcefully your express your support). A devious institution could exploit strategic confusion here too, such as covertly accepting one ideology (e.g. racism) while overtly promoting another (e.g. ‘equal opportunity’ achieved by phasing out affirmative action).
38. Fairclough (1992: 88f, my italics) was evidently aware of such problems when he critiqued the ‘textual view of the location of ideology, which one finds in Critical linguistics — ideologies reside in texts’:
While it is true that the forms and content of texts do bear the imprint of (are traces of) ideological processes and structures, it is not possible to ‘read off’ ideologies from texts, […] because meanings are produced through interpretations of texts, and texts are open to diverse interpretations which may differ in their ideological import. […] Claims to discover ideological processes solely through text analysis run into the problem now familiar in media sociology that text ‘consumers’ (readers, viewers) appear sometimes to be quite immune to the effects of ideologies which are supposedly ‘in’ the texts.
In my view, these arguments might lead to a different conclusion. Rather than saying that ‘to read off’ ideologies from texts’ is just ‘not possible’, we might say that different interpreters can and often will ‘read off’ different ideologies from the same texts. Even if we leave aside the old disputes over whether the ‘sameness of the text’ is a meaningful concept — an issue which has hardly been clarified by the ‘post-structuralism’ with its concepts like ‘dissemination’, ‘bricolage’, ‘jouissance’, ‘free-play of the signifier’, and so on (Beaugrande 1988a) — we may incur a replay of the chicken-and-egg problem: you ‘read off the ideology’ of a text under the controls of the ideology you already hold. But, as I have indicated (§ 20), you are likely to focus on a contrasting ideology and interpret it as a distortion; conversely, you are not well positioned to ‘read off’ your own ideology, which would appear natural and transparent and so largely invisible.
39. Moreover, ‘immunity’ confirms rather then refutes my view: the control unconsciously exerted by your own ideology naturally generates resistance to the ‘effects’ of another sustained by a ‘text’. Actual conversions would be limited chiefly to cases where you are uncommitted regarding the relevant ideological parameters, or where you become aware of some newer inclinations or sympathies, e.g., when your social position has switched from ‘have-not’ over to ‘have’ and your populism changes into elitism.
40. At all events, whether ideologies can be ‘read off’ from texts and how ought to be empirical questions to be resolved through practical tests and not just theoretical pronouncements. At least, the prospects do not seem unfavourable for gathering and comparing a representative spread of ‘read-offs’ from a team whose ideologies mutually contrast (§ 25). Also, extensive experience with corpus work might render critical analysts keenly attuned to the discursive traces of ideologies and hence far better at ‘reading off’ than we might have expected (§ 26).
41. Still, Fairclough is indisputably justified in contending that analyses of the texts by themselves are far from sufficient. In his view,
ideology is located both in the structures (i.e. orders of discourse) which constitute the outcome of past events and the conditions for current events, and in events themselves as they reproduce and transform the conditioning structures. It is an accumulated and naturalised orientation which is built into norms and conventions, as well as an ongoing work to naturalise and denaturalise such orientations in discursive events (1992: 89)
The analyst would seemingly require extensive background data about ‘past events’ and ‘current events’, as well as about the processes of ‘accumulating and naturalising orientations’ and ‘conditioning structures’ which have perhaps been ‘transformed’ anyway. Surely some ‘reading off’ from texts would be an allowable heuristic strategy for telling us where to look?
42. The insufficiency of texts in isolation has been made one point of contention in a tireless series of attacks by H.G. Widdowson against critical discourse analysis. He diagnoses ‘confusion’ ‘about the nature of discourse (as distinct from text) and about analysis (as distinct from interpretation)’, and has ‘suggested that this confusion is bred of commitment’ (Widdowson: 1995). Fairclough (1996) infers that critical discourse analysis (CDA) is being accused of ‘ideological commitment’ and ‘prejudice’, and sees in Widdowson’s attacks ‘a version of the classical liberal distinction between’ ‘science and impartiality’ versus ‘ideology, commitment, prejudice and partiality’. But matters can hardly be so simple after Widdowson’s himself has affirmed:
all the discourses of theory, including those of linguistics, are ideologically loaded, cultural constructs designed to establish control and a sense of security. This is not in the least surprising of course, since theories are made out of language (Widdowson 1991a: 39)
Notice the implication here that all discourses of ideological! Nor would Widdowson accept the attitude of mainstream linguistics to achieve an ideology-free scientific status by idealising language (section B), after he has roundly vowed that ‘there can be no idealisation without ideology’ (1991a: 39). His advocacy is substantially more general: he would
not want to suggest that we should avoid the cultural partiality of disciplinary discourses and strive instead to be neutrally objective, [but rather that] we should guard against being too readily persuaded into believing in the validity of relevance of any particular discourse, no matter what apparent authority it might have (1991a: 40)
So Widdowson is rebuking Fairclough and others on the grounds that
a good deal of critical discourse analysis talks about the linguistic features of texts as if they inevitably expressed meaning, particularly ideological meaning, whether the writer intended them to or not. […] Such analysis […] states what the text means to the particular reader who happens to be assuming the role of analyst, and then claims that this is what the text itself means. (1991b: 5)
There is rarely a suggestion that alternative interpretations are possible. There is usually the implication that the single interpretation offered is uniquely validated by the textual facts.
What is being accurately described here is a favoured tactic of both traditional and formalist literary criticism, though neither of the two would wan to appeal to ‘ideological meanings’. Critical discourse analysis, in contrast, quite plainly asserts that
interpreters are more than discourse subjects in particular discourse processes; they are also social subjects with particular accumulated social experiences, and with resources variously oriented to the multiple dimensions of social life, and these variables affect the ways they go about interpreting particular texts. […] it is important to take account of the ways in which interpreters interpret texts if one is properly to assess their political and ideological effectiveness. (Fairclough 1992: 136)
The ‘overwhelming emphasis’ in Fairclough’s ‘recent work’ has accordingly been consigned to ‘showing how shifting discursive practices, manifested in texts which are heterogeneous in forms and meaning, can be analysed as facets of wider processes of social and cultural change’ (1996). So Widdowson’s rebukes are quite simply misplaced and vacuous
43. The key difficulty I would see — one not raised by Widdowson — is that Fairclough’s method consigns ‘ideological effectiveness’ only to the pursuit of domination, and not to resistance or solidarity. The field is thus not well fortified against relativism, as we encounter it when Widdowson elsewhere erases the profound ideological differences among three approaches to language. Labov’s (1970: 192) contention that ‘working-class speakers are more effective narrators, reasoners, and debaters than many middle-class speakers who temporise, qualify, and lose their argument in a mass of irrelevant detail’ is commented in these terms:
we may wish to acknowledge that Labov’s argument is devoted to a good cause and wish to be associated with the ideology which informs it. But […] its promotion of a particular mode of use by referring to some absolute criterion of precision is, like the campaign of the Plain Language Movement, not essentially different from the efforts of […] custodians to promote the mode of use of their preference by referring to some absolute criterion of correctness (Widdowson 1988: 346)
Here, relativism scales stunning heights by equating egalitarian projects to restore the dignity of stigmatised language varieties and to defend ordinary citizens against manipulation by powerful corporations and bureaucrats, with elitist projects to perpetuate and reinforce that very stigmatisation.
44. Moreover, only the ‘criteria’ of ‘correctness’ are absolute because they cannot be rationally measured or justified. The criteria of ‘plainness’ and ‘effectiveness’ are not absolute because they can be rationally measured and justified in terms of mutual respect and openness among socially and linguistically diverse groups. The sociolinguistics of Labov and the ‘the campaign of the Plain Language Movement’ are sustained by counter-ideologies against a linguistic elitism which modern linguistics, in its aspirations to be an ideology-free science, could not effectively combat as long as it merely replaced a simplistic, crude idealisation with complex, sophisticated one.
45. Fairclough’s writings imply that a counter-ideology would be ‘partial’ and ‘sustain relations of domination’:
CDA would argue that we are all […] writing from within particular discursive practices, entailing particular interests, commitments, inclusions, exclusions, and so forth; […] Aspects of these discursive practices may serve to sustain relations of domination and may hence be ideological — no theory or science is immune from that possibility. (1996)
But some optimism persists. His aspirations would thus be that ‘the values of CDA do not (unlike others) sustain relations of domination — they do not actually work ideologically’ (1996). Moreover, CDA is theoretically better-placed to recognise its own “partiality” than most theories’.
46. In the present paper, I shall pursue a different tactic than has been advocated in these discussions. Instead of seeking to ‘read off’ the ideology of a given text (pace Fairclough), and also instead ‘offering a single interpretation as uniquely validated by the textual facts’ (pace Widdowson), I shall examine a large set of contexts for the terms derived from one central term which I think we would all peacefully grant to harbour ideological resonances, although (as we shall see), these are unstable and contested. My analysis is not intended to be authoritative, much less ‘validated’ but rather heuristic in suggesting some topics and directions for the deeper analysis recommended by Fairclough. Just as corpus research continually points to questions we can pursue with a still larger or more specialised corpus, so might discourse analysis working with corpora find unexpected leads toward relations between discourse and society which only become well-defined whilst we examine large sets of authentic data.
E. On the meanings of ‘liberal’ in three corpora
47. I chose the term ‘liberal’ and its derivatives, such as ‘liberalism’ and ‘liberalisation’, because they have been much used of late by groups with disparate interests in various meanings, e.g. for freedom from government regulations and for solidarity of white people with black people. The same tendency may well emerge from data-studies of other key terms in the discourses of post-modern society. I shall compare data drawn from three corpora representing different regions, the United Kingdom, the United States, and South Africa.2 Before you read on, you might want test your own intuitions about what the terms would mean if you used them yourself. Who are the ‘liberals’, and would you want to be one of them?
E.1 United Kingdom data
48. The Bank of English was developed at Birmingham University by John Sinclair and his team, and as of June 1996 had reached the unprecedented size of some 323 million words of running text from contemporary spoken and written sources (Sinclair 1996). The UK sources in this corpus returned exactly 1,000 lines centring on occurrences of the keyword ‘liberal’ by itself or in its derivatives. I shall disregard the 288 lines for ‘Liberal Democrat(s)’ (and 34 for ‘Liberal Party’) on the grounds that the name of a political party can be notoriously arbitrary and opportunistic, witness the vastly different parties bearing this same name in the UK, Russia, and Japan.
49. Predictably, a fair number of occurrences, especially for the Process-Verb3 ‘liberalise’ and its Noun-Form ‘liberalisation’ concerned economic policy. Prominent collocates included ‘economic’ and ‘economy’ themselves as well as ‘technological progress’, ‘capitalism’, ‘financial policies’, ‘free market’, ‘trade’, ‘protection’, and ‘privatisation’, with the ‘IMF’ near at hand [1-4]. I underline the contextual clues I find relevant to determining the meanings of the key-words.
 Driven by technological progress and economic liberalisation, global capitalism is changing fundamentally
 financial polices that are consistent over time; liberalisation and privatisation to promote free markets
 about the economy. The IMF wants a new government to liberalise the economy and privatise public companies.
 even essential. One is trade; to persuade voters that liberal trade is in their interests, less protection at home
In such contexts, the core meaning would plausibly be: activities and policies designed to remove restraints on the ‘free’ manipulation of the ‘market’ of goods, services, and labour. In view of the adverse effects upon ordinary wage-earners and consumers, a deal of mystification may be needed to ‘persuade voters that liberal trade is in their interests’  (§ 88).
50. Some corpus data gave more precise indications of what a ‘liberal’ system offers to ‘investors’, such as ‘cheap labour’ and ‘few environmental restraints’ . The collocation ‘liberalise + prices’ (86 occurrences in corpus) evidently offers a handy euphemism for raising prices with no controls, and consumers can be thankful if a ‘small basket of foods’ is exempted, though these may still be unaffordable to the ‘unemployed’ whose ‘benefits’ have been ‘cut’ .
 in Taiwan finds cheap property, labour, and raw materials, liberal investment laws and few environmental restraints.
 the prices of everything but a small basket of foods would be liberalised and generous unemployment benefits would be cut.
51. Such data may suggest why public discourse might be eager to maintain that that ‘capitalism’ and ‘liberal democracy’ are ‘inexorably’ allied, e.g. by asserting that ‘American micro-electronics accelerated the fall of communism’ . Yet that optimism is belied by data on societies like Chile, where ‘economic liberalisation’ and ‘market reforms’ got so easily combined’ with ‘political repression’ , and by data explicitly linking ‘liberal’ with ‘rapaciousness’  and with opposition to ‘populism’ .
 a movement towards ever higher technological capability. This movement has led, he argues, inexorably towards capitalism and liberal democracy. It was, for example, American micro-electronics that threatened to render obsolete the entire Soviet arsenal and thereby accelerated the fall of communism.
 Democracy is well established after the traumatic Pinochet interlude, which combined political repression with economic liberalisation. Chile had a ten-year start over other Latin American countries in market reforms, and began bringing down its own trade barriers without waiting for others to reciprocate.
 A tried-and-tested alternative to rapacious western liberalism was at hand. India’s colonial past still shapes
 He will not turn populist, but instead push ahead with liberal policies and go faster with privatisation,
52. Particularly intriguing were the attestations reminding us that favouring big capital over labour and consumers is also a cherished principle of the traditional opposite of ‘liberal’, namely ‘conservative’ ideology [11-15]. Perhaps a semantic merger is under way to give us not just occasional alliances on specific issues like opposing ‘monopoly’  — a stance many current ‘liberals’ have evidently abandoned anyway — but a political species of ‘liberal conservatives’ [14-15], or at least ‘conservative governments’ that find it expedient to call themselves ‘liberal’ .
 viewing city development conservative and liberal plans, plans to support the private market
 its American democratic inheritance. That conservative-liberal philosophy remained, and still remains, powerful
 In theory, it [the argument against monopoly] could be shared by liberals eager to break the power of big business and conservatives intent on restoring some semblance of perfect competition
 trace their ancestors to the liberal, parliamentary, elitist and moderate conservatives served any regime which served them.
 many Conservatives, those who can be called liberal Conservatives, were thoroughly opposed to the increasing trend towards economic collectivism
 voting system, it was not immediately known whether the Liberal (conservative) government would continue to run
The irony is all the richer in view of the habit of the British Conservative Party to
 blame everything that has gone wrong in British society on the liberal 1960s rather than on their party which has been in power since 1979
53. Toward the opposite pole of the political spectrum, a few data alluded to the prospect of being a ‘liberal communist’, but specifically in Eastern Europe, where ‘new political movements’  or ‘expulsion from the Party’ would be imminent consequences .
 But it would also encourage millions of liberal communists to form new political movements
 largely communists expelled from the Party for their liberal views. The second was the so-called Left Alternative,
The core meaning of ‘liberal’ could still be: policies designed to remove restraints on the ‘free’ manipulation of the ‘market’, but the implications would be starkly different within a strictly state-controlled economic system. In view of the later deterioration of the Russian economy, the Western economists who congratulate ‘capitalism’ on its victory over ‘communism’ (e.g. in sample ) might ask themselves (or the Russians) if the horrendous effects of naked profiteering and corruption were perhaps just results of the most thorough and relentless application of ‘capitalism’ to a society which had no understanding of how to control and modulate it.
54. I was surprised to find ‘bourgeois liberals’ being opposed to ‘affluent proletarians’ until I noticed the context concerning how ‘jokes are becoming less subtle’ ; and another data entry ominously indicated how .
 Unfortunately, as our humane, bourgeois liberal culture gives way to the affluent proletarian mass culture of Essex and Hollywood, our jokes become less subtle and a new element begins to emerge.
 So why is he so popular # Because thousands love his sexist, racist and deformed-people jokes and he has brilliant comic timing. Believes that wishy-washy white liberal do-gooders are undermining the soul of the British comedy
Here, being ‘liberal’ would mean defending racial minorities, women, and the handicapped against the public ridicule which, we are briskly told, ‘thousands love’ and which constitutes ‘the soul of the British comedy’. I have always wanted to ask the users of the dismissive label ‘do-gooders’ if they admire ‘do-badders’ and aspire to be among them.
55. ‘Liberals’ who do defend racial minorities and women might be targeted by the New Right conservative backlash as ‘radicals’, but I found only one instance in UK data , plus one in which a concern for ‘minority cultural rights’ was contrasted with the ‘language of tradition’ and was ‘despised’ . The routine attack upon liberalism in the UK data was closer to the opposite: being mindlessly uncommitted and vague in one’s good intentions [24-26], rather than, say, being ‘tough on crime’ .
 And that fits into radical sort of liberal adult education awareness for some encouraging
 language of tradition but instead deploy the despised liberal rhetoric of minority cultural rights. In defending the
 How should one describe Bresson’s cinema? A liberal humanist approach flounders in generalities
 men like Raplin exist shows what you get with a liberal agenda # You pump people full of drugs and pipe dreams
 Women can’t be sexist as they are the oppressed? Bollocks. That is a lily-livered liberal wank view and holds no water with me. I can spot massive violent prejudice when I see it
 his ‘tough on crime’ policies unimpeded by Tumim’s liberal interference. For how long, though, is a moot point.
The vulnerability of women emerged again when they, rather than men, were bluntly called ‘sexists’ who harbour ‘massive violent prejudice’ , much as the victims of racism and their defenders are called ‘racists’ by the New Right (cf. van Dijk 1993, 1998b). A vicious twist with a long tradition was to cast aspersions (‘lily-livered’, ‘wank’) on the manhood of ‘liberals’ who champion women’s rights.
56. In sum, the UK data indicated that the meaning of ‘liberal’ was most precisely determined in contexts of economic activity and policy, as a designation for ‘privatising’ everything [2-3] and allowing the free market to run its course, unrestricted by such annoyances as ‘environmental restraints’  and ‘unemployment benefits’ . This stance happens to be dear to the traditional opposite, ‘conservative’ ideology [11-15]. The economic meaning was usually contextualised amelioratively to uphold the credo that
 the virtues of the economic system that underpins liberal democracy significantly outweigh its vices
along with congratulatory invocations of benefits like ‘higher technological capability’ . For foreign investors, the ‘cheap labour’ and the lack of ‘environmental restraints’ in Taiwan  simply constitute attractions for shrewd businessmen, rather than short-term and long-term degradation’s in the lives of the local inhabitants. Even ‘political repression’ can get contextualised into a laudable impulse for a ‘head start’ in ‘market reforms’ to ‘bring down the trade barriers’ , whereby Pinochet thanked the US and its CIA for overthrowing the democratically elected socialist government and installing him as dictator.
57. Intriguingly, the further away the contexts moved from the economic meaning, the more pejorative they became. Only a small part seemed to be motivated by threadbare New Right tactic of ‘blaming everything that has gone wrong in British society on the liberal 1960s’ . A more decisive motive seemed to be the trenchant hostility in public discourse of spokespersons and groups against a different type of restrictions, namely those protecting of the rights of women, minorities, and accused offenders of the law. Such protections were defamed as the concern of ‘wishy-washy do-gooders’ and ‘lily-livered wankers’ , whilst the British public is being re-educated by a ‘comedy’ of ‘sexist, racist and deformed-people jokes’ . So the ameliorative and pejorative contextualisations did not contradict each other after all, however much they may contradict the spirit of genuine democracy.
E.2 United States data
59. The US data came from the US-based sources in the Bank of English and totalled 287 lines. Since ‘the proportion of English from identifiable US sources in the Bank of English is maintained at around 25%’ (Sinclair, personal communication in March 1998), the overall frequency of occurrences was not significantly lower than for the UK data with 1000 occurrences for a volume three times at large, if we set aside the party name of ‘Liberal Democrat(s)’ with just 7 occurrences in the US data versus 288 in the UK data. Perhaps the Americans, with their fossilised two-party system, take no keen interest in discussing the political parties of other nations.
59. Some data concerned the same economic activities and policies we saw in the UK data, again with ‘liberalizing prices’ as a handy euphemism for freely jacking them up [29-30], a move sometimes foisted upon a nation from the outside . The notion that maintaining a ‘liberal free-market’ makes you a ‘democrat’ scaled the peaks of irony for the ‘changed Khmer Rouge’ , but then who can predict just how many ‘millions’ may ‘die’ in the ‘killing fields’ of the ‘free market’ as its rationalises people out of their jobs and abolishes welfare and unemployment compensation (§ 85)?
 He wants a market economy instituted rapidly, with liberalized prices. He wants the state bureaucracies to drop the subsidies
 price reform. Jan Vanous: If Russia liberalizes prices, Ukraine has no choice but to follow
 some one million Cambodians died in the infamous Khmer Rouge killing fields. <p> Until this month, the Khmer Rouge appeared intent on proving to the world that they’ve changed and are now liberal, free-market democrats.
60. Yet unlike the UK data, the US data also reflected the contradictory view that ‘liberals’ are opposed to current economic ‘realities’ or cannot apply their principles to these , and are out of touch with the broad population anyway .
 is a little like Bill Clinton’s: how to reconcile liberal principles with certain harsh realities
 The vast majority of Americans are not liberals <p> and they should simply listen to the real speeches
Alternately, ‘liberalism’ was identified as the ideology specific to the ‘bourgeoisie’  and the ‘middle class and the rich’, where the term ‘money liberals’ did seem piquantly apt .
 The bourgeoisie is assumed to be necessarily liberal in politics and if a liberal order is not created then
 This is the stuff of social equality. But virtually none of these virtues are evident when all the government does is distribute cash benefits — even if, as money liberals typically recommend, benefits go to the middle class and rich as well as the poor.
61. Another contrast to the UK data emerged where ‘liberals’ were said to actually favour social welfare programs [36-37]. In both cases, the contexts connoted disapproval, e.g., in suggesting that ‘liberals’ were somehow acting unfairly by ‘simply imposing taxes on the rich minority’ in order to the ‘fund the public sphere’.
 thought that congressional Democrats were tax-and-spend liberals, wedded to big government.4
 large tax increases needed to fund the public sphere — liberals can simply impose the increases on the rich minority
The diversity of social groups among or close to the ‘liberals’ was conspicuous in the U.S. data, e.g., ‘human rights’ advocates and environmentalists , feminists  (whose cause disqualifies male supporters as ‘lily-livered wankers’ ?), blacks , or white spokespersons for blacks  (the core meaning in South Africa, as we shall see in section E.3) (§ 68ff).
 to recoup with a coalition of human rights-oriented liberals and antiwaste moderates
 Socialist feminism in the United States reemerged in the 1960s, when liberal feminists became frustrated with the pace of social reform and began to seek more fundamental sources of women's oppression. By focusing on women’s economic roles
 says, ‘He’s got the wrong vision. He’s not liberal and most blacks are liberal’.
 It just is another example of benevolent white liberals dissecting and analyzing the attitudes of black people
62. Discursive moves to discredit ‘liberal’ ideology were more imaginative than in the UK data (back in [22-26]). The US data variously portrayed it not just as wishy-washy [42-43], but also subservient to ethnic groups like Arabs in ‘caftans’ and African Americans in ‘dashikis’ , ‘not progressive’ , deficient in ‘understanding the ‘stubborn facts of human experience’ , incapable of integrating ‘affective life’ with ‘intellectual interests’ , ‘sanctimonious’ , ‘fearful’ of the ‘plebs’ , or compromised if not indeed criminalised by the support of ‘drug users’ who hold a ‘permissive and anti-conventional outlook’ [50-51].
 all we ever get to see is the conservative against the liberal, and the liberal has no politics at all, anyhow.
 the differences among individuals. This is because, to liberals, there is no absolute truth, everything is relative
 of a caftan or whatever those things are the Yankee liberals are all running around in now’. ‘Dashikis’
 an idea of an agency to put even these into effect. The liberals of today are not a part of a progressive movement
 Lacking any true insight into these stubborn facts of human experience — corruption, evil, irrational desire — liberals also fail to understand that evil often lies beyond purely rational treatment
 The body, the unconscious, the pre-rational are all important to sound thought. But because the liberal has sought no positive discipline for emotion and feeling, there is an open breach between his affective life and his intellectual interests.
 GOP ticket in 1952. And the privileges of sanctimonious liberals chafed against his [Nixon’s] prepolitical past: a childhood
 long enough to explore. In any event, among most French liberals, fear of the plebs outran their hopes for it.
 These veterans of the ‘60s era of sex and drugs and rock’n’roll support a liberal social agenda. And they're more attuned to the special-interest groups linked to the Democratic Party than to the big-business associations of the GOP.
 among those who have gone to college, drug users are more likely to have majored in the social sciences, fine arts, and humanities than in the natural sciences. They are also more likely to favor liberal politics, to be estranged from religion, and to have a generally permissive and anti-conventional outlook
The oddest linkage of all was made among groups who are not convinced that ‘the principle of free speech’ should protect the hate speech of ‘racists’: ‘bigots’ got juxtaposed with ‘liberals’, and ‘feminists’ with ‘sexists’ (compare sample  again).
 not when they engage in violence or vandalism. But when they speak or write, racist assholes fall right into this Oliver Wendell Holmes definition — highly unpopular among bigots, liberals, radicals, feminists, sexists, and college administrators: ‘If there is any principle of the Constitution that more imperatively calls for attachment than any other, it is the principle of free speech’
Applying ‘the principle of free speech’ to ‘hate speech’, where the ‘negative presentation of the outgroup’ (van Dijk) casts aside all restraint, is an alarming demonstration of how the principles of democracy can be used to subvert it. This notion of ‘freedom’ is difficult to oppose from an ‘ideology’ of ‘liberalism’ that purports to ‘outflank or transcend ideology’ and puts its ‘faith’ in ‘reason as a faculty that operates independently of any particular world view’ (Fish 1994: 134f).
63. At all events, ‘liberals’ in the US can expect a barrage of ‘attacks’ , and even their ‘contented moderation’ is anathema to the ‘red-hunters on the ‘Right’ .
 today and continued his attacks on what he calls the liberal establishment. His main target was New York, a city
 For the time being, the tones of contented moderation had driven out the discourse of anguished zeal. Great Liberal Fear <> But red hunters on the Right gave liberal intellectuals a terrible shock.
Not at all surprisingly, US politicians were far more wary than in the UK of being classed as ‘liberal’ [55-56], and ‘measures’ to ‘repeal liberal rights’ are being busily drawn up even (or especially?) for California .
 clear from Clinton’s reactions over the weekend to the liberal label, him fighting back, that this ticket is not liberal
 McGovern: Well, I think Bill Clinton is probably more liberal than his present image. I hope my saying that won’t harm his campaign
 the court overturned a portion of the initiative that would have stripped defendants of any rights that aren’t provided by the US Constitution. That part of the measure was designed to repeal more liberal rights provisions in the California constitution.
Scorning ‘liberals’ for a failure to be ‘tough on crime’ (sample ) is an ironic argument in age where the ‘liberal free-market’ is rapidly eroding opportunities for honest work (§ 85).
64. The old antipathy between liberalism and religious doctrine, which has furnished history with some bemusing sidelights  (but compare sample  below), is currently focused upon the inflammatory issue of ‘abortion’ [61-62], where being ‘liberal’ means allowing the women involved to decide for themselves without being threatened by criminal prosecution.
 When in 1889 a canon of Santo Domingo de Calzado who had preached against freedom of the press and the ‘damned error of liberalism’ and called it a sin for Catholics to vote for Liberals, was brought to trial for ‘having condemned political liberalism from the pulpit’ he was supported by many of the region's clergy.
 he argued for better social services and more liberal abortion laws. But the sharpest criticism today
 that voters have overwhelmingly rejected a proposal to liberalize the country’s strict abortion laws. Returns from
How this meaning might relate to economic ideologies is a deep question indeed. Is the fierce opposition to abortion (a) simply a gaudy political football of the New Right; (b) the centrepiece of an broad attack upon feminism and the rights or women; (c) a ploy to gain moral high grounds for the ideology of ‘Christian fundamentalism’ which can accuse the ‘liberal’ society of slaughtering the innocents; (d) a measure to encourage an oversupply of cheap labour in coming generations; or (e) some queasy mixture of all of these? Here is another issue for in-depth research
65. The US data signalled a curious mix of disparate visions regarding the history of ‘liberal’ ideology [61-63], also disagreeing on whether or not it is currently coming or going [62-68]:
 rests upon a quite different set of premises. Liberalism in this sense was symbolically a child of Voltaire
 Outside a small circle of theorists, liberalism connoted a worldview defined in the 1930s and reborn in the early 1960s that no longer inspired many activists or voters.
 the New Republic is profoundly distressed to discover that liberalism has undergone an eclipse since 1912 when the bulk of
 argues that the Western idea of economic and political liberalism has triumphed over alternatives such as fascism,
 own credit and did a little something to vindicate liberalism by opposing unjustifiable appeals to arms.
 they will have to pay more attention to the discrepancies in their record. They will have to either shift liberalism to new and stronger ground or repair the manifest breaches in their defenses. <p> liberals deserve the eclipse from which liberalism is suffering.
 the liberal remains unprepared to face the worst; and on the brink of what may turn out another Dark Ages, he continues to scan the horizon for signs of dawn. <p> The record of liberalism during the last decade has been shameful evasion and inept retreat. <p Liberalism has compromised with despotism because despotism promised economic benefits to the masses
 Pragmatic liberalism has flatly betrayed ideal liberalism. The values that belong to the latter have been compromised away, vitiated, ruthlessly cast overboard. The permanent heritage of liberalism has been bartered for the essentially ignoble notion of national security, in itself a gross illusion
66. Taken all together, the US data were substantially less focused than the UK data, though important similarities could readily be detected. Perhaps because there is no respected ‘Liberal Party’ in US politics, the term has not stabilised an ameliorative economic meaning, which has been heavily coloured by the ‘tax-and-spend’ anathema accompanying the long-term attack on social welfare — ironically, a policy elsewhere proudly called ‘liberal’ (§ 85). The non-economic meaning of ‘liberal’ is evidently being blurred by the highly vocal New Right offensive, enlisting the to cast scorn upon their long and diffuse list of enemies all across society, including not just minorities and women (especially feminists), but also, intellectuals, environmentalists, and advocates of civil rights or of curbs on the economic power of the private capital that bankrolls the New Right. The blurring is consummate when those who obviously fall under the traditional economic core meaning of the term, such as Bill Clinton (contrast  with ), repudiate the label.
E.3 South Africa data
67. The Corpus of South African English (CSAE) has recently been established at the University of Port Elizabeth under the supervision of Chris Jeffries and Linda Pearce Williams, and has grown from one million to some three million words. This corpus was originally inaugurated as part of the International Corpus of English (ICE) under the direction of the late Sidney Greenbaum at University College London (cf. Greenbaum ed. 1996). These corpora would provide, for the first time in history, a base of authentic data about local varieties of English that have hitherto led a contradictory existence of being used by large populations but not rarely recognised as valid alternative language systems.
68. Linda Pearce Williams has kindly supplied the total of 247 data lines from the CSAE. This total represents a vastly higher proportional frequency rate than for the UK and US data from the Bank of English, which is roughly 100 times larger. To appreciate why, we should keep in mind that in South Africa, the terms ‘liberal’ and ‘liberalism’ — but not ‘liberalise’ and ‘liberalisation’, these being more recent economic terms — have long carried the highly specific meaning of mostly white people and policies that promoted the human rights of black Africans.5 Such meaning aroused much discussion and controversy in the grim landscape of apartheid especially the founding of the Liberal Party in 1953. It was eventually outlawed in 1968 by a decree with the amazingly candid name, for a supposedly parliamentary state, of ‘Prohibition of Political Interference Act’. Thus, the only ‘Liberal Party’ in my data to be actually banned was also the only one whose ideological position was most sharply defined in terms of promoting human rights.
69. Being ‘liberal’ in South Africa as the thumbscrews of apartheid were ruthlessly tightened would expose you to drastic hostility of the government and its allies in institutions and the media, in the form of house arrests, ‘banning orders’ and so on, but — presumably to avoid galvanising white resistance — usually stopping short of the martyrdom by detention, torture, and killing, visited upon black activists like Steve Biko, the key advocate of ‘Black Consciousness’. Moreover, some black Africans mistrusted the ‘liberals’ of trying to preserve the system in a milder modality to keep from radically transforming it. Biko himself once described ‘the liberal establishment, including radical and leftist groups’, as ‘a curious bunch of non-conformists’ and ‘do-gooders’ (that slap again!) whose ‘artificially integrated circles are a soporific to the blacks while salving the conscience of the guilt-stricken whites’ (1978 [orig. 1972]: 63ff).
70. In view of this historical background, the high frequency of occurrences in the CSAE is easily understood. More specifically, we can predict that the data will indicate how the ideology of ‘liberalism’ is being forcefully unsettled by the establishment of a democratic majority government in 1994. The uneasiness can be pungently sensed when the Liberal Students Association at Wits University ruefully announced a lecture on ‘Resuscitating Liberalism’.
71. Reflections about the past ‘mission’  were typically accompanied by perplexity on having lost it [70-74]. Some contextual cues, like ‘pivoted and sold’ , ‘basking’ and ‘worshipful disenfranchised’ , or ‘jealousy’  may be symptoms of a spiteful serve-them-right tone of those who opposed the liberals all along.
 the liberal students of that time had a very clear mission in life. Our mission was to oppose apartheid. That was our purpose
 Some people say that the liberals’ day was yesterday, a day when we were secure in the conducting of a holy war
 As I look at this hall I see some things haven’t changed. But the cause has changed. I don’t know what liberalism is or means now.
 The moral high ground that such liberals occupy is because they have been pivoted and sold to black people as true freedom fighters.
 they have lost the formerly serene sense of basking in the gratitude of the worshipful disenfranchised. As the bearer of a famous liberal name told me not long ago: they don’t need us any more, you know
 I think there is a jealousy here. There are liberals, so-called, who feel displaced. They’re no longer the black man’s favourite whites
72. A prominent representation of ‘university’ influence was also indicated [75-76], with hints of elitism and radicalism even as contrasted with other ‘liberals’, e.g., respecting ‘one-man-one-vote’ suffrage , which was a rallying motto central issue both for the Black Consciousness Movement and the African National Congress (Biko 1979: 126f; Mandela 1994: 239f).
 we took it for granted that being liberal meant being a university person, or ex- or child-of. Also, being white.
 It was liberal to stand up for a better deal for the blacks, even for what was then called ‘one-man-one-vote’, which post-university liberals considered pretty terroristic in itself
Elitism was clearly one theme for disparaging white liberals:
 white ‘liberals’, however good their intentions, seem oblivious to their racial socialisation. Too many, of ‘liberal’ and sometimes ‘left’ persuasion, remain in a comfortable white social cocoon
 smacks of the typical liberal naivety and paternalism born out of white privilege
But then only white people who were well-situated could have had any effective political impact during the apartheid era; and those people did after all take a stand against the very system under which they had prospered and which could be expected to sharply curb their ‘privileges’.
73. Today, their status as critical voices under the new system are being exploited to represent them having turned against ‘blacks’ , and to disqualify them as ‘hypocrites’  and even ‘racists’ .
 In the old days the Afrikaner establishment saw liberalism as a hobby of upper-class English-speakers who disliked the Afrikaners more than they disliked the Africans. Now the new liberated establishment sees liberals as upper-class English-speakers taking whacks at blacks, on every available grounds
 we must expose First World hypocrisy by speeding up the democratic revolution in South Africa, for that is what your pathetic liberal drivel is really against.
 We spent all of the afternoon complaining about racist liberals (isn’t it refreshing to find new ogres?).
Despite the sweeping changes, criticism of the new social order is subject to much the same derision as was criticism of the old:
 the proper liberal role is to make people cross with you. You’ve got to be gadfly, and if the rump of state does not get irritated
 Liberals ought to be over the moon, but in fact there is endless carping and griping. Open a paper, and there is some spokesperson of liberalism running something down, predictable and tiresome.
 the corpus of liberalism, which in the public mind becomes associated with trigger-happy and slightly hysterical denunciation.
Small wonder if some supporters of liberalism feel exasperated at public ingratitude:
 ‘Liberal’ is now a dirty word among those who were not even born when liberals fought for their rights.
In contrast to the UK data, the SA data nowhere associated ‘liberal’ with ‘conservative’, a term in the name of a political party on the far right in the apartheid context, witness such sinister data as these:
 Amnesty has also been granted to Conservative Party and Volksfront member Saint Michael Schutte for being in possession of seven AK-47 rifles
 a Conservative Party MP, DP du Plessis, repudiated fellow CP MP Koos Botha for bombing a school earmarked for black students.
 Convicted Hani murderer [Chris Hani, secretary-general of the South African Communist Party, shot down in April 1993 to disrupt the transition to democracy] and former Conservative Party politician Clive Derby-Lewis continued his testimony before the truth commission's amnesty committee
74. But ‘liberal’ was multiply associated with ‘communist’ [89-90, 93-94], although in a totally different meaning than we saw to be specific to Eastern Europe in samples [18-19] (§ 53). The apartheid regimes had indiscriminately pasted the label on all their adversaries and persecuted them under the ‘Suppression of Communism Act’ of 1950. Even Joe Slovo, long-time general secretary of the South African Communist Party and imprisoned on Robben Island together with the leaders of the African National Congress, was suspected of ‘liberal’ tendencies  despite being reputed (like Biko, § 69) to regard ‘liberals’ as ‘hypocrites’ .
 It was the hour of the raised voice of the liberals who saw themselves as the conscience of the nation. And it was also the time of the communist voices
 newspapers were seen to be unpatriotic reckless liberalistic communistic pinko <#> You know you know all those all those labels
 There are those who think that Slovo remained a true believer to the end, that the more liberal formulations of his last years were merely another tactical feint
 For Slovo, of course, all liberals are hypocrites and he is happily illiberal.
In more academic registers, the two ideologies were idiosyncratically linked in terms of their world-views [93-94], along with the ‘semi-socialist ideas of the Christian culture’  (but compare sample ):
 the vacuity of the communist vision of human nature is equally discernible in the liberal anthropology of rationalist, autonomous individuality; it cannot hope to escape the fate which befell its companion ideology. In this sense liberalism and communism are but the two sides of a single Modernist coin
 This vociferous form feeds on the radical liberal perspectives and sometimes semi-socialist ideas of the Christian culture of the educated
Nelson Mandela (1994: 285) himself certified in 1962 that ‘the Liberal Party and the Communist Party were arch-enemies’, but he too noticed how the terms were getting blurred together in order to discredit them both.
75. The academic data that collocated ‘liberal’ with ‘egalitarian’ and ‘equality’ were sternly pejorative:
 the appeal to basic rights can settle only a modest number of contentious issues, and the tendency of egalitarian liberal theory to make all contentious issues a question of right, [and to] blur the separation of the public and private spheres has even more disastrous consequences
 the dominant liberal obsession with equality leads to incoherence in theory and increasing conflict and antagonism in practice
After reviewing all these pejorative data we may feel a little wistful to hear recent calls to action like this one:
 Join the Democratic Party if you share its commitment to the values of liberalism and democracy and its vision of fighting for the freedom of every person to shape his or her own future.
76. The contexts were quite different and far less numerous for ‘liberalisation’ in the meaning we saw in the UK data. Perhaps to avoid confusion with ‘liberal’ in the traditional South African meaning, ‘neo-liberal’ is currently used for the economic meaning. I found the same usage 224 times in the UK data but, interestingly enough, not even once in the US data, where we noticed some confusion about the history of the ideology ([61-68]).
77. The data displayed the typical collocatable terms like ‘invest’ , ‘globalisation’ , ‘competition’ , and ‘privatisation’ . The ‘suspicion’ and ‘hostility’ of labour unions  provided clues about the implied meanings: ‘invest offshore’  ==> evade paying the national income taxes vitally needed for better social services; ‘join the international economy’  ==> bow to the will of multinational corporations and banks; ‘fiscal discipline’  ==> no upgrading social welfare programmes; ‘flexible labour markets’  ==> power of owners and management to fix wages; ‘improved competition’ + ‘export drive’  ==> undersell other countries by underpaying the productive workforce at home (cf. Beaugrande and Williams in press). If  was a bit more frank  was extremely so in calling these economic policies by some well-earned names.
 foreign exchange controls will be liberalised by allowing individuals to invest offshore for the first time
 the whole business of globalisation and liberalisation. We are now under enormous pressure as a country to join the international economy
 the foundation proposes a ‘multi-pillared’ strategy of legal reform, financial liberalisation, fiscal discipline, flexible labour markets, improved competition and an export drive.
 union movements, which now view privatisation with suspicion and hostility and which have expressed misgivings about trade liberalisation
 Any finance obtained from the World Bank would be on policy terms dictated by it. The usual neo-liberal conditions — privatisation, liberalisation of trade, monetary restraint or high interest rates, deregulation, cutting state subsidies, and so on.
 the global community we are supposed to be part of. The shift to the left is primarily a resounding rejection of neo-liberal society, in its uncaring, alienated politics, as well as its free market economics of unabashed profiteering
We might well speculate that the future well-being of the democratic South Africa will depend chiefly on whether the old ‘liberalism’ of egalitarian human rights can withstand the ‘enormous pressure’ from the ‘neo-liberalism’ of ‘uncaring, alienated politics’ and ‘unabashed profiteering’, whilst the huge majority of black Africans continue to live in poverty and hopelessness.
F. Further issues for analysis
78. The data presented above are but a small sample of 103 passages out of the total of 1,534 data samples I examined — not even one tenth, although still far more real data than you will find in many publications by linguists, especially by ones who idealise language. To work as a discourse analyst confronting plentiful corpus data, you are compelled to be selective and to rely in early stages on your own expectations and intuitions about which data might be more relevant or interesting. Later, the parameters you are finding will send you back to the larger sets in search of more specific or subtle indicators (cf. § 88).
79. Obviously, the methods for doing a ‘critical discourse analysis’ of corpus data are far from established yet. Even when we have examined a fairly large set of attestations, we cannot be certain whether our own interpretations of key items and collocations are genuinely representative of the large populations who produced the data. But we can be fairly confident of accessing a range of interpretative issues that is both wider and more precise than we could access by relying on our own personal usages and intuitions (§ 26). Moreover, when we observe our own ideological position in contest with others, we are less likely to overlook it or take it for granted.
80. Nor can I point to any consensus about which among the many analytic techniques of previous linguistics we might apply. Do we focus on morphology, syntax, semantics, or pragmatics, or some combination of these (Beaugrande 1996)? Or, do we focus on less familiar factors like ‘colligability’ and ‘collocability’ (§ 16), which cut across all of those in ways we are just beginning to grasp (Beaugrande 1998b)? Or again, do we focus on the ‘discursive structures, strategies, and moves’ whereby ideological discourse enacts, ‘at all levels of the text, the positive presentation of the ingroup and the negative presentation of the outgroup’ by means of ‘polarisation’, ‘metaphor’, ‘hyperbolic emphasis’, ‘negative lexicalisation’, ‘rhetorical mitigation’, ‘euphemisation’, and so on (van Dijk 1998b)? Moreover, what are the implications of using key words as the units of search and analysis when our major interest is in contexts and in the positions of speakers (Lemke 1998)? We plainly need some means for searching and analysing relations among meanings or belief-systems as well, but the necessary software would demand a highly innovative design.
81. To be consistent with the principles described in section C, we cannot settle these questions from the top down, but only from our ongoing bottom-up work with plentiful data (§ 18). Not merely do corpus data in no way manifest the disorder projected by ‘linguisticism’ since Saussure (§ 14), they manifest an amazingly delicate order despite their enormous quantity and variety. To deal with all that, we shall need to pursue multiple inquiries guided and co-ordinated by some evolving dialectical agenda, where a ‘critical’ ideology could strategically guide our focus toward data that are indicative of power and domination but also of solidarity and equality (§ 31).
82. Consider in this connection ‘the principle observation of corpus linguistics in the last decade’: ‘meaning affects the structure profoundly’ (Sinclair 1991: 496). One significant reflex of this is that the collocability and colligability of any one item, such as a Noun or a Verb, are unequally distributed among its alternative forms, such as Singular and Plural (Sinclair 1991: 494ff). I in turn found that the Adjective ‘liberal’ is far more variable and contested in its collocations than the Noun ‘liberalism’, whilst the Verb ‘liberalise’ or (in US data) ‘liberalize’ is by far the most restricted (cf. § 49, 76). The (invariably Third Person) Subjects of the Verb were nearly always governments, institutions, or whole countries, all representing enough power to make resistance seem futile. The Objects were nearly always abstractions, coming mainly from a single semantic domain with the headword ‘economy’, such as ‘prices’, ‘trade’, ‘markets’, ‘financial system’, and ‘tariffs’; ‘laws’ appeared at a much lower frequency. A few data signalled ongoing evolutions in usage, such as using the Verb in Intransitive colligations, e.g.: ‘two regimes which, having liberalised under pressures’; and ‘steps were taken to explore whether Tanzania should liberalise politically as well as economically’. Clear innovations included the ‘closet liberalisers’ in the British Parliament, and the way Poland and Hungary were said to ‘run import regimes that out-liberalise those of many OECD countries’. I can see no rational way to account for all these emergent constraints except by reference to the ideological status of the key words.
83. Whilst engaging with such data, you may detect a dual evolution in your interpretative work: both toward a convergence of meanings for a give key word among sets of data; and toward a divergence between sets so loosely associated as to make a common core meaning quite difficult to formulate. What might be the common core shared by the set of economic contexts (these from the UK data) such as ‘open and liberal economic policies’ or ‘the money-capital concept linked to the classical liberal ideology’, and by the set of ironic goodwill contexts such as ‘liberal Christian god’ or ‘dressing up their doctrine with a liberal sprinkling of post-Daisy Age happy harmonies’, or again by the set of plenitude contexts such as ‘a liberal supply of live food including small insects’ or ‘a liberal application of face powder on top of foundation’?
84. Or might the expectation that a given expression has a stable and determinate core meaning be unduly proximate to the ‘linguisticist’ vision of language being a stable and determinate system (§ 10)? Corpus data may be telling us that the meaning of an expression typically requires a band of fluctuation and indeterminacy in order to adapt so effortlessly to emergent constraints during discourse. Contestations over meanings are one implication of this band, as when the connotations of ‘liberal’ with generosity and freedom can be handily exploited for policies of selfishness and compulsion.
85. Still, the fluctuation could plausibly be constrained by the attractions of analogies among the meanings. We might consider is whether the meaning of ‘liberal’ as ‘wishy-washy’  (§ 55) may stand in some analogy to the officially laissez-faire policies of the core economic meaning, or is merely an empty projection of ‘negative lexicalisation’ (in the sense of van Dijk, § 80) to dismiss everyone to the left of the New Right, whatever their ideologies (§ 66). If the analogy is implied, then it is itself just manipulative as the ‘free’ in ‘free market’. The massive evidence marshalled by Martin and Schumann (1996) among others demonstrates that economic ‘liberalism’ in its current form is far from laissez-faire; multinational corporations, the World Bank, and the International Monetary Fund, are literally forcing their ‘free market’ ideology onto whole populations and governments with the conviction that
continual wage reductions, longer work-hours, cuts in social welfare, even the compete abandonment of social welfare in the US, are supposed to make the nations ‘fit’ for global competition. (Martin and Schumann 1996: 312, my translation)
The two financial journalists are moved to ask: ‘how much free market can democracy stand?’
Now what the founders of the post-war welfare states had learned from bitter experience is becoming ever more clearly visible: free-market economy and democracy are by no means inseparable blood-brothers, who harmoniously nourish prosperity for everyone. Instead, the two central guiding models of the old industrial states of West stand in constant contradiction to each other. (1996: 311, my translation)
We could run extensive data queries to detect the means whereby public discourses in countries like the UK and the US carry a heavy ‘ideological investment’ (Fairclough) in papering over the ‘contradiction’, e.g.. by thematising the ‘virtues of the economic system that underpins liberal democracy’  (§ 56).
86. We could also run multiple data queries to observe and describe the emergence of a set of evasive terms within the ideological proximity of economic neo-liberalism to designate and yet obscure the complex and devious tactics for profiteering at the expense of ordinary workers: ‘rationalising’, ‘privatising’, ‘restructuring’, re-engineering’, ‘downsizing’, ‘outsourcing’, ‘multiskilling’, ‘increasing productivity’, and so on, all mean: fewer jobs with harder work for lower wages and no benefits at all, whilst the gains for owners and shareholders soar. The New York Times calculated from the statistics of the US Department of Labour that, between 1979 and 1995, 43 million workers lost their jobs; and two thirds of those who found other jobs were forced to accept much poorer working conditions (reported in the International Herald Tribune, 6 March 1996).
87. Then irony is consummated when neo-liberalism adopts the strategy of claiming to be no ideology at all. The European Union Commissioner for Competitiveness, Karl van Miert, recently defended privatisation by declaring: ‘the decision to liberalise certain branches which offer public services is by no means ideological, but the expression of a natural adaptation to economic and technical developments’ (Le Monde diplomatique, January 1996, my translation and italics). His ‘choice of words’, Martin and Schumann (1997: 190f) comment, ‘already betrays the unexamined ideology that is always recognisable when politicians invoke “nature” whilst handing out state properties, tax revenues, and economic privileges’.
88. The echoes in Fairclough’s concept of ‘naturalisation’ (§ 41) are highly serendipitous here. We can return to our data to look for more subtle choices of language intended to make ‘liberalism’ seem natural if not virtually inevitable. I would highlight what is underlined in these data:
[1a] Driven by technological progress
[2a] financial polices that are consistent over time
[7a] a movement towards ever higher technological capability […] has led […] inexorably towards capitalism and liberal democracy
[10a] He will […] push ahead with liberal policies and go faster with privatisation
[29a] He wants a market economy instituted rapidly
[30a] If Russia liberalizes prices, Ukraine has no choice but to follow
[12a] That conservative-liberal philosophy remained, and still remains, powerful
[34a] The bourgeoisie is assumed to be necessarily liberal
[64a] the Western idea of economic and political liberalism has triumphed over alternatives
When efforts are made to ‘persuade voters that liberal trade is in their interests’  (§ 49), the contradiction is to be actually reproduced in the awareness of the population. They are to see themselves only as shoppers wanting cheaper goods, and to forget being also workers who will get lower wages to shop with when ‘protection at home’ has been abolished.
89. However, we also need to consider the prospect that ideological terms may not appear in a corpus with a frequency corresponding to their influence, because they have been quietly incorporated into the state policies. The radically ‘conservative’ Reagan-Bush government and their supporters in the business and financial communities did not have to confuse the voters by expressly advocating ‘liberal’ policies, leaving their New Right allies free to hurl the term ‘liberals’ at the advocates of the rights of women and minorities and of social welfare programmes (§ 66). Yet on an international plane, ‘liberal’ remains in vogue as an agreeable label for of economic policies that could be more honestly called ‘unabashed profiteering’ (§ 77). Here too, instability in the meaning is an ideological advantage.
90. Still another advantage of the instability is to hinder the consolidation of an effective counter-ideology. What can we rationally call the opposite of ‘liberal’ policies when the usual converse ‘conservative’ is not merely unattractive for critical thinkers and ecologically oriented intellectuals, but is also becoming hardly distinguishable from ‘liberal’ anyway, as our data indicated (§ 52)? We certainly wouldn’t declare ourselves ‘illiberal’ which the Random House Webster’s variously defines as ‘narrow-minded, bigoted, miserly, lacking culture or refinement’ (p. 670).
91. Now, if ideological terms are so strategic for the advancement of the ideology itself, a lack of accredited terms for counter-ideologies is a serious problem. I have suggested the problem to be a general one, also holding for consumerism, individualism, capitalism, and free-marketism, now that the terms ‘socialism’ and ‘communism’ are losing favour (§ 32). The problem is indeed one major motive for advocating a ‘critical’ enterprise that formulates and promotes an explicit counter-ideology such as ‘ecologism’ (§ 8, 14, 31), whose name can profit from the serendipitous echoes with the ‘ecology movement’ (e.g. Myers et al. 1993) and also with Neisser’s (1976: 2) concept of ‘ecological validity’ (§ 8).
92. I certainly couldn’t deny that the sincere ambition of critical discourse analysts to transcend ideology is admirable, since their goal is to overcome practices of power and domination, as we have seen in section D. But I feel uneasy recalling the ominous precedents of various groups, including scientists and economists, whose self-serving denials to hold any ideology constituted one major motive for critical discourse analysis in the first place. Or, if we agree to call the ‘ideology’ (in my sense) of critical discourse analysis by some other name — say, ‘position’, ‘agenda’, or ‘programme’ — we risk being scolded for terminological squeamishness and also risk obscuring the oppositional status of our programme vis-à-vis the ideologies of domination. Besides, critical discourse analysis currently does have power, as manifested in being a favoured domain for books, journals, and even academic job descriptions — and also a conspicuous target for continual attacks such as Widdowson’s (§ 42f).
93. At all events, I hope to have made a reasonable case for broadening our base to explore the ideological aspects of discourse made available by large corpus data before we can decide such questions as whether or not all discourse implies ideology (§ 33ff). The various debaters of this question would presumably agree that the ideological constraints on and in discourse are substantially more pervasive and subtle than most people, including most linguists, remotely suspect (cf. § 36ff). If so, then surely we should look at extensive discourse data with an open mind respecting its potentially ideological character.
94. Our methods will of course demand extensive refinement for such tasks as sorting out corpus data according to ideologically relevant factors, such as political settings and historical periods, along with sources, registers, genres, audiences, and perhaps even individual speakers or writers. For example, we might well find significant distinctions between public speeches or editorials versus casual conversations in homes, pubs, or grandstands, even for the same groups of speakers. If an ideology encourages consistency on some issues, it may encourage inconsistency on others in order to avoid drawing distasteful conclusions and making painful admissions (cf. Potter and Wetherell 1987). Psychotherapy and psychoanalysis labour in a veritable minefield of such avoidances.
95. If, as so much work in critical discourse analysis implies, ideology is essential a construct for power and closure, a paradox might be conjectured for the enterprise of developing an ideology of solidarity and equality wherein concurrent alternatives are accepted with openness and respect. I would again draw an analogy to a scientific theory, which in classical science wages war against all others until a newer theory discredits it (cf. 8). I would submit that this adversarial and confrontational notion of science should itself be regarded as a discredited theory or falsified paradigm. The significant problems of the post-modern, multicultural society can be solved only by ‘post-classical science’ that integrates alternatives (Beaugrande 1997). So too should ‘ecologism’ as an integrative ideology for both society and science transcend the combative proclivities of the ideologies of earlier ages with their horrifying rosters of atrocities. Only this integration can provide the consensus and strength to effectively deconstruct such anachronistic and recidivist ideologies as an ‘economic neo-liberalism’ propelling global society backwards into the primitive stages of robber-baron capitalism, non-unionised labour, subsistence wages, chronic unemployment, and endemic poverty. The alternatives are as clear as our imperatives: we cannot choose not to choose between power and solidarity, between inequality and equality.
* I am much indebted to Jay Lemke, Jim Martin, Norm Fairclough, Ruth Wodak, and Teun van Dijk for helpful discussions of this work.
1 On the origins of these terms, compare Firth (1968); Greenbaum (1974).
2 All three data sets were kindly extracted by Linda Pearce Williams from the Bank of English in Birmingham during her visit early in 1998, and from the Corpus of South African English at her home University in Port Elizabeth. I also wish to thank Stephen Bullon at COBUILD for completing some data lines for me.
3 ‘Process-Verb’ in the sense of functional grammar (Halliday 1994; Beaugrande 1997), i.e., not an Auxiliary or a Modal Verb.
4 This line appeared in a UK source, but quoting a US source.
5 My sketch of South African politics is purely for the orientation of unfamiliar readers, and has drawn upon Plaatje (1982 [orig. 1916]); La Guma (1972); Biko (1978); Paton (1981); Kuzwayo (1985); Mandela (1994); and on the recent South African press (e.g. Mail and Guardian) as well as the corpus itself. Whether these representations are currently accredited and by whom is a question far beyond the scope of the paper; the continuing flood of discoveries, admissions, confessions and so on has plunged South Africa’s sense of its own history into a ferment which may not settle down for many years.
Beaugrande, R. de (1988a) Critical Discourse: A Survey of Contemporary Literary Theorists. Norwood, NJ.: Ablex.
Beaugrande, R. de (1988b) ‘Achieving Feminist discourse: The difficult case of Luce Irigaray’, College English 50/3: 1-20.
Beaugrande, R. de (1991) Linguistic Theory: The Discourse of Fundamental Works. London: Longman.
Beaugrande, R. de (1996) ‘The “pragmatics” of doing language science: The “warrant” for large-corpus linguistics’, Journal of Pragmatics 25: 503-535.
Beaugrande, R. de (1997) New Foundations for a Science of Text and Discourse. Greenwich, CT: Ablex.
Beaugrande, R. de (1998a) ‘Performative speech acts in linguistic theory: The rationality of Noam Chomsky’, Journal of Pragmatics.
Beaugrande, R. de (1998b) ‘Corpus and controversy: Ideal language versus real language’, for International Journal of Corpus Linguistics.
Beaugrande R. de and Williams, L. (in press) ‘Discourse analysis and “democracy”: Some evidence from the Corpus of South African English’, Discurso y Sociedad 1.
Biko, S. (1978) I Write What I Like. London: Heinemann.
Billig, M. (1991) Ideology and Opinions. London: Sage.
Bolinger, D.L. (1970) ‘Getting the words in’, American Speech 45, 78-84.
Cameron, D. (1992) Feminism and Linguistic Theory: Second Revised Edition. London: Macmillan.
Chomsky, N. (1965) Aspects of the Theory of Syntax. Cambridge: M.I.T. Press
Chomsky, N. (1977) Language And Responsibility. New York: Pantheon.
Christie, F. and Martin, J.R. (eds.) (1997) Genre and Institutions: Social Processes in the Workplace and School. London: Cassell.
Dijk, T. van (1993) Elite Discourse And Racism. London: Sage.
Dijk, T. van (1998a) Ideology: A Multidisciplinary Study. London: Sage.
Dijk, T. van (1998b) ‘Critical discourse analysis’. Ms. supplied by author.
Fairclough, N. (1992) Discourse and Social Change. Cambridge: Polity.
Fairclough, N. (1995) Critical Discourse Analysis. London: Longman.
Fairclough, N. (1996) ‘A reply to H. Widdowson’s “Discourse analysis - a critical view”’, Language and Literature 5/1: 49-56.
Fairclough, N. (1998) Response via e-mail to a previous draft of this paper.
Firth, J.R. (1968) Selected Papers of J.R. Firth 1952-1959, ed. F.R. Palmer. London: Longman
Fish, S.E. (1994) There is No Such Thing as Free Speech. Oxford: Oxford UP.
Fowler, R.,. Hodge, R., Kress, G., and Trew, T. (1979) Language and Control. London: Routledge & Kegan Paul.
Francis, G. and Sinclair, J.McH. (1994) ‘I bet he drinks Carling Black Label: A riposte to Owen on corpus grammar’, Applied Linguistics 15: 190-200.
Gee, J. (1990) Social Linguistics and Literacies. London: Falmer.
Geertz, C. (1973) The Interpretation of Cultures. New York: Basic Books.
Gell-Mann, M. (1994) The Quark and The Jaguar: Adventures in the Simple and the Complex. London: Little, Brown.
Gilbert, N. and Mulkay, M. (1984) Opening Pandora’s Box: A Sociological Analysis of Scientists’ Discourse. Cambridge: Cambridge UP.
Greenbaum, S. (1974) Some verb-intensifier collocations in American and British English. American Speech 49: 79-89.
Greenbaum, S. (ed.) (1996) Comparing English Worldwide: The International Corpus of English. Oxford: Clarendon.
Hall, S. (1985) ‘Signification, representation ideology: Althusser and the post-structuralist debate’, Critical Studies in Mass Communication 2: 91-114.
Halliday, M.A.K. (1994) An Introduction to Functional Grammar: Second Revised Edition. London: Arnold.
Hjelmslev, L. (1969 [orig. 1943]) Prolegomena to a Theory of Language. Madison: University of Wisconsin Press.
Hodge, R. and Kress, G. (1988) Social Semiotics. Ithaca: Cornell University Press.
Kotthoff, A. and R. Wodak (eds.) (1998) Communicating Gender in Context. Amsterdam: Benjamins.
Kuhn, T. (1970) The Structure of Scientific Revolutions. Chicago: Chicago UP.
Kuzwayo, E. (1985) Call Me Woman. Randburg, SA: Ravan.
Labov, W. (1970). ‘The logic of non-standard English’, in J. Alatis (ed.) 20th Georgetown University Round Table on Languages and Linguistics 1969, pp. 1-43. Washington D.C: Center for Applied Linguistics.
La Guma, A. (1972) In the Fog of the Season’s End. London: Heinemann
Lemke, J. (1995) Textual Politics. London: Longman.
Lemke, J. (1998) Response via e-mail to a previous draft of this paper.
Leeuwen, T. van and Wodak, R. (1998) ‘Legitimizing immigration control: A discourse historical study’, Discourse Studies 1.
Mandela, N.R.. (1994) Long Walk to Freedom. London: Little, Brown.
Mannheim, K. (1936) Ideology and Utopia. New York: Harcourt Brace.
Martin, H.P. and Schumann, H. (1996). Die Globaliserungsfalle: Der Angriff auf Demokratie und Wohlstand. Reinbeck: Rowohlt.
Mey, J. (1979 [orig. 1974]) ‘Zur kritischen Sprachtheorie’, in Jakob Mey (ed.) Pragmalinguistics: Theory and Practice, pp. 412-434. The Hague: Mouton.
Myers, Norman et al. (1993) Gaia: An Atlas of Planet Management. New York: Doubleday.
Neisser, U. (1976) Cognition and Reality. San Francisco: Freeman.
Paton, Alan. (1981) Ah, But Your Land Is Beautiful. Harmondsworth: Penguin
Pêcheux, M. (1982) Language, Semantics, and Ideology. London: Macmillan.
Plaatje, S. (1982 [orig. 1916]) Native Life in South Africa. Randburg, SA: Ravan.
Potter, J., and Wetherell, M. (1987) Discourse and Social Psychology. London: Sage.
Saussure, F. de. (1966 [orig. 1916]) Course in General Linguistics, transl. W. Baskin. New York: McGraw-Hill.
Shils, E. (1958) ‘Ideology and civility: On the politics of the intellectual’, Sewanee Review 66: 450-480.
Sinclair, J.McH. (1991) ‘Shared knowledge’, in J. Alatis (ed.) Georgetown University Round Table on Languages and Linguistics 1991, pp. 489-500. Washington, D.C.: Georgetown University Press.
Sinclair, J.McH. (1996) ‘What do we know about language, how do we get to know it, and what has all that got to do with language teaching?’ Paper at the International Conference on Analysis and Description: Applications to Language Teaching, at Lignan College and at the Hong Kong University of Science and Technology, June 1996.
Sinclair, J.McH. (1997) ‘Corpus linguistics at the millennium’, in J. Kohn, B. Ruschoff, and D. Wolff (eds.) New Horizons in CALL: Proceedings of EUROCALL 96 in Szombathely, Hungary, pp. 1-10.
Stark, W. (1958) The Sociology of Knowledge. London: Routledge and Kegan Paul.
Vološinov, V.N. (1973 [orig. 1929]) Marxism and the Philosophy Of Language. New York: Seminar.
Widdowson H.G. (1988) ‘Language spread in modes of use’, in P.H. Lowenberg (ed.) Georgetown University Round Table on Languages and Linguistics 1987: Language Spread. and Language Policy, pp. 342-360. Washington, D.C.: Georgetown University Press.
Widdowson H.G. (1991a) ‘Discourses of enquiry and conditions of relevance’, in J. Alatis (ed.) Georgetown University Round Table on Languages and Linguistics 1990. Washington, D.C.: Georgetown University Press.
Widdowson H.G. (1991b) Types of equivalence. Paper at the Triangle Seminar, Paris January 1991. Manuscript.
Widdowson H.G. (1995) ‘Discourse analysis - a critical view’, Language and Literature 4/3: 157-172.
Wodak, R. (1996) Disorders of Discourse. London: Longman.
Wodak, R. (ed.) (1997) Gender and Discourse. London: Sage.
Wodak, R. et al. (1998) Diskursive Konstruktion nationaler Identität. Frankfurt a.M.: Suhrkamp.
Zima, P. (1981) ‘Les mécanismes discursifs de l’idéologie’, Revue de l’Institut de Sociologie 4.