3

The rediscovery of ‘ideology’; return of the repressed in media studies

STUART HALL

Mass communications research has had, to put it mildly, a somewhat chequered career. Since its inception as a specialist area of scientific inquiry and research—roughly, the early decades of the twentieth century—we can identify at least three distinct phases. The most dramatic break is that which occurred between the second and third phases. This marks off the massive period of research conducted within the sociological approaches of ‘mainstream’ American behavioural science, beginning in the 1940s and commanding the field through into the 1950s and 1960s, from the period of its decline and the emergence of an alternative, ‘critical’ paradigm. This paper attempts to chart this major paradigm shift in broad outline and to identify some of the theoretical elements which have been assembled in the course of the formation of the ‘critical’ approach. Two basic points about this break should be made at this stage in the argument. First, though the differences between the ‘mainstream’ and the ‘critical’ approaches might appear, at first sight, to be principally methodological and procedural, this appearance is, in our view, a false one. Profound differences in theoretical perspective and in political calculation differentiate the one from the other. These differences first appear in relation to media analysis. But, behind this immediate object of attention, there lie broader differences in terms of how societies or social formations in general are to be analysed. Second, the simplest way to characterize the shift from ‘mainstream’ to ‘critical’ perspectives is in terms of the movement from, essentially, a behavioural to an ideological perspective.

‘DREAM COME TRUE’: PLURALISM, THE MEDIA AND THE MYTH OF INTEGRATION

The ‘mainstream’ approach was behavioural in two senses. The central question that concerned American media sociologists during this period was the question of the media's effects. These effects—it was assumed—could best be identified and analysed in terms of the changes which the media were said to have effected in the behaviour of individuals exposed to their influence. The approach was also ‘behavioural’ in a more methodological sense. Speculation about media effects had to be subject to the kinds of empirical test which characterized positivistic social science. This approach was installed as the dominant one in the flowering of media research in the United States in the 1940s. Its ascendancy paralleled the institutional hegemony of American behavioural science on a world scale in the hey-day of the 1950s and early 1960s. Its decline paralleled that of the paradigms on which that intellectual hegemony had been founded. Though theoretical and methodological questions were of central importance in this change of direction, they certainly cannot be isolated from their historical and political contexts. This is one of the reasons why the shifts between the different phases of research can, without too much simplification, also be characterized as a sort of oscillation between the American and the European poles of intellectual influence.

To understand the nature of media research in the period of the behavioural mainstream hegemony, and its concern with a certain set of effects, we must understand the way it related, in turn, to the first phase of media research. For, behind this concern with behavioural effects lay a longer, less scientific and empirical tradition of thought, which offered, in a speculative mode, a set of challenging theses about the impact of the modern media on modern industrial societies. Basically European in focus, this larger debate assumed a very powerful, largely unmediated set of effects attributable to the media. The premise of this work was the assumption that, somewhere in the period of later industrial capitalist development, modern societies had become ‘mass societies’. The mass media were seen both as instruments in this evolution, and as symptomatic of its most troubling tendencies. The ‘mass society/mass culture’ debate really goes back as far, at least, as the eighteenth century. Its terms were first defined in the period of the rise of an urban commercial culture, interpreted at the time as posing a threat, because of its direct dependence on cultural production for a market, to traditional cultural values. But the debate was revived in a peculiarly intense form at the end of the nineteenth century. It is common, nowadays—and we agree with this view—largely to discount the terms in which these cultural and social problems associated with the development of industrial capitalism were debated. Nonetheless, the mass culture debate did indeed identify a deep and qualitative shift in social relations which occurred in many advanced industrial capitalist societies in this period. Although the nature of these historical transformations could not be adequately grasped or properly theorized within the terms of the ‘mass society’ thesis, these were indeed the terms which prevailed when the ‘debate’ came to the fore again at the commencement of what, nowadays, we would want to characterize as the transition to monopoly forms of advanced capitalist development.

The effects which most concerned this more speculative approach can be grouped under three rough headings. Some were defined as cultural: the displacement, debasement and trivialization of high culture as a result of the dissemination of the mass culture associated with the new media. Some were defined as political: the vulnerability of the masses to the false appeals, propaganda and influence of the media. Some were defined as social: the breakup of community ties, of gemeinschaft, of intermediary face-to-face groups and the exposure of the masses to the commercialized influences of élites, via the media. A very specific historical image came to dominate this scenario: the breakdown of European societies under the double assault of economic depression and fascism: the latter seen in terms of the unleashing of irrational political forces, in which the propaganda media had played a key role.

The Frankfurt School gave this critique its most biting philosophical elaboration. (Their work and the mass culture debate is more extensively discussed in the previous essay in this volume.) When, in the wake of fascism, the Frankfurt School was dispersed, and its members took refuge in the United States, they brought their pessimistic forebodings about mass society with them. Briefly, their message was: ‘it can happen here, too’. In a way, American behavioural science—which had already taken issue with the early versions of this mass society critique—continued, in the 1940s and 1950s, to develop a sort of displaced reply to this challenge. It argued that, though some of the tendencies of mass society were undoubtedly visible in the United States, there were strong countervailing tendencies. Primary groups had not disintegrated. Media effects were not direct, but mediated by other social processes. Essentially, to the charge that American society displayed symptoms of a sort of creeping totalitarianism, American social scientists made the optimistic response: ‘pluralism works here’.

Perhaps more important than the distinction between ‘pessimistic’ and ‘optimistic’ social predictions about media effects, were the distinctions between the theoretical and methodological approaches of the two schools. The European approach was historically and philosophically sweeping, speculative, offering a rich but over-generalized set of hypotheses. The American approach was empirical, behavioural and scientistic. In fact, hypotheses proposed within one framework were often tested, refined and found wanting in an altogether different one. It is little wonder that hypotheses and findings were not commensurable. Only those who believe that there is a given and incontrovertible set of facts, innocent of the framework of theory in which they are identified, which can be subject to empirical verification according to a universal scientific method, would have expected that to be so. But this is exactly what American behavioural science offered itself as doing. There are some intriguing transitional moments here which are worth remarking—in lieu of a fuller account. They can be encapsulated in the history of two emigrés. Lazarsfeld, a distinguished European methodologist, linked with, though not a subscribing member of, the Frankfurt School, became in fact the doyen and leading luminary of behavioural methodology in the American context. (It has been speculated that his success at the latter task may have had something to do with his early sensitization to more speculative European questions: certainly, he was a more theoretically sophisticated methodologist than his more technical colleagues.) Adorno, on the other hand, the most formidable of the Frankfurt School theorists, attempted, without any conspicuous success, to adapt his speculative critique to empirical procedures. The Authoritarian Personality (1950) was a hybrid monster of just this kind—the product of a mixed but unholy parentage.

In the approach which succeeded the European critique, the main focus was on behavioural change. If the media had ‘effects’ these, it was argued, should show up empirically in terms of a direct influence on individuals, which would register as a switch of behavour. Switches of choice—between advertised consumer goods or between presidential candidates—were viewed as a paradigm case of measurable influence and effect. The model of power and influence being employed here was paradigmatically empiricist and pluralistic: its primary focus was the individual; it theorized power in terms of the direct influence of A on B's behaviour; it was preoccupied (as so-called ‘political science’ in this mould has been ever since) with the process of decision making. Its ideal experimental test was a before/after one: its ideal model of influence was that of the campaign. Political campaign studies conceived politics largely in terms of voting, and voting largely in terms of campaign influences and the resulting voter choices. The parallel with advertising campaigns was exact. Not only was a great deal of the research funded for the purpose of identifying how to deliver specific audiences to the advertisers—loftily entitled ‘policy research’—but the commercial model tended to dominate the theory, even in the more rarified atmosphere of Academia. Larger historical shifts, questions of political process and formation before and beyond the ballot-box, issues of social and political power, of social structure and economic relations, were simply absent, not by chance, but because they were theoretically outside the frame of reference. But that was because the approach, though advanced as empirically-grounded and scientific, was predicated on a very specific set of political and ideological presuppositions. These presuppositions, however, were not put to the test, within the theory, but framed and underpinned it as a set of unexamined postulates. It should have asked, ‘does pluralism work?’ and ‘how does pluralism work?’ Instead, it asserted, ‘pluralism works’—and then went on to measure, precisely and empirically, just how well it was doing. This mixture of prophecy and hope, with a brutal, hard-headed, behaviouristic positivism provided a heady theoretical concoction which, for a long time, passed itself off as ‘pure science’.

In this model, power and influence were identical and interchangeable terms: both could be empirically demonstrated at the point of decision making. Occasionally, this reductionism was projected on to a larger canvas and the impact of the media was discussed in terms of ‘society’ as a whole. But this connection was made in a very specific way. And society was defined in a very limited manner. A largely cultural definition of society was assumed. Class formations, economic processes, sets of institutional power-relations were largely unacknowledged. What held society together it was agreed were its norms. In pluralist society, a fundamental broadly based consensus on norms was assumed to prevail throughout the population. The connection between the media and this normative consensus, then, could only be established at the level of values. This was a tricky term. In Parsons's ‘social system’ (Parsons, 1951) such values played an absolutely pivotal role; for around them the integrative mechanisms which held the social order together were organized. Yet what these values were—their content and structure—or how they were produced, or how, in a highly differentiated and dynamic modern industrial capitalist society, an inclusive consensus on ‘the core value system’ had spontaneously arisen, were questions that were not and could not be explained. Value consensus, however, was assumed. Culturally, Edward Shils (a collaborator of Parsons) argued, this broad band of values was so widely shared as to have accreted to itself the power of the sacred (Shils, 1961a, p. 117). If some groups were, unaccountably, not yet fully paid-up members of the consensus club, they were well on the way to integration within it. The core would gradually absorb the more ‘brutal’ cultures of the periphery (Shils, 1961b). Thus the democratic enfranchisement of all citizens within political society, and the economic enfranchisement of all consumers within the freeenterprise economy, would rapidly be paralleled by the cultural absorption of all groups into the culture of the centre. Pluralism rested on these three mutually reinforcing supports. In its purest form, pluralism assured that no structural barriers or limits of class would obstruct this process of cultural absorption: for, as we all ‘knew’, America was no longer a class society. Nothing prevented the long day's journey of the American masses to the centre. This must have been very good news to blacks, Hispanics, Chicanos, American Injuns, New York Italians, Boston Irish, Mexican wetbacks, California Japanese, blue-collar workers, hard-hats, Bowery bums, Southern poor-whites and other recalcitrant elements still simmering in the American melting pot. What is more (a comforting thought in the depths of the Cold War) all other societies were well on their way along the ‘modernizing’ continuum. Pluralism thus became, not just a way of defining American particularism, but the model of society as such, written into social science. Despite the theoretical form in which this ramshackle construction was advanced, and the refined methodologies by which its progress was empirically confirmed, there is no mistaking the political and ideological settlement which underpinned it. Daniel Bell assured us, in The End of Ideology (1960), that the classical problem of ‘ideology’ had at last been superseded. There would be a range of pluralistic conflicts of interest and value. But they could all be resolved within the framework of the pluralistic consensus and its ‘rules of the game’. This was essentially because, as another apologist, Seymour Lipset, forcefully put the matter:

the fundamental political problems of the industrial revolution have been solved: the workers have achieved industrial and political citizenship; the conservatives have accepted the welfare state; and the democratic left has recognized that an increase in overall state power carried with it more dangers to freedom than solutions for economic problems. (Lipset, 1963, p. 406)

The installation of pluralism as the model of modern industrial social order represented a moment of profound theoretical and political closure. It was not, however, destined to survive the testing times of the ghetto rebellions, campus revolts, counter-cultural upheavals and anti-war movements of the late 1960s. But, for a time, it prevailed. It became a global ideology, backed by the credentials of social science. It was exported with a will around the globe. Some of its force arose from the fact that what, in theory, ought to be the case, could be shown so convincingly and empirically to be, in fact, the case. The American Dream had been empirically verified. A whole number of decisive interventions in developing countries were made in the name of hastening them along this modernizing pathway. It is sometimes asked what a moment of political settlement and theoretical hegemony looks like: this would certainly be one good candidate.

The media were articulated to this general social scientific model in, principally, two ways. In the campaign/decision-making framework, its influences were traced: directly, in behaviour changes amongst individuals; indirectly, in its influences on opinion which led, in a second step, to empirically observable behavioural differences. Here, media messages were read and coded in terms of the intentions and biases of the communicators. Since the message was assumed as a sort of empty linguistic construct, it was held to mirror the intentions of its producers in a relatively simple way. It was simply the means by which the intentions of communicators effectively influenced the behaviour of individuals receivers. Occasionally, moves were announced to make the model of media influence more fully societal. But these, largely, remained at the level of unfulfilled programmatic promises. The methods of coding and processing a vast corpus of messages in an objective and empirically-verifiable way (content analysis) were vastly sophisticated and refined. But, conceptually, the media message, as a symbolic sign vehicle or a structured discourse, with its own internal structuration and complexity, remained theoretically wholly undeveloped.

At the broader level, the media were held to be largely reflective or expressive of an achieved consensus. The finding that, after all, the media were not very influential was predicated on the belief that, in its wider cultural sense, the media largely reinforced those values and norms which had already achieved a wide consensual foundation. Since the consensus was a ‘good thing’, those reinforcing effects of the media were given a benign and positive reading. The notion of selective perception was subsequently introduced, to take account of the fact that different individuals might bring their own structure of attention and selectivity to what the media offered. But these differential interpretations were not related back either to a theory of reading or to a complex map of ideologies. They were, instead, interpreted functionally. Different individuals could derive different satisfactions and fulfil different needs from the different parts of the programming. These needs and satisfactions were assumed to be universal and trans-historical. The positive assumption arising from all this was, in sum, that the media—though open to commercial and other influences—were, by and large, functional for society, because they functioned in line with and strengthened the core value system of society. They underwrote pluralism.

DEVIANTS AND THE CONSENSUS

We can identify two kinds of breaks within this theoretical synthesis which began to occur towards the closing years of the paradigm's dominance, but before it was more profoundly challenged from outside its confines. The first may be summed up as the problematizing of the term ‘consensus’ itself. As we suggested, the presumption of an integral and organic consensus did leave certain empirically identifiable groups beyond the pale. Since, at first, these groups were not conceived to be organized around conflicting structural or ideological principles, they were defined exclusively in terms of their deviation from the consensus. To be outside the consensus was to be, not in an alternative value system, but simply outside of norms as such: normless—therefore, anomic. In mass society theory, anomic was viewed as a condition peculiarly vulnerable to over-influence by the media. But when these deviant formations began to be studied more closely, it became clear that they did often have alternative foci of integration. These enclaves were then defined as ‘sub-cultural’. But the relation of sub-cultures to the dominant culture continued to be defined culturally. That is, sub-cultural deviation could be understood as learning or affiliating or subscribing to a ‘definition of the situation’ different or deviant from that institutionalized within the core value system. The career deviant in a sub-culture had subscribed positively to, say, a definition of drug-taking which the dominant consensus regarded as outside the rules (with the exception of alcohol and tobacco which, unaccountably, were given a high and positive premium within the American central value system). For a time, these different ‘definitions of the situation’ were simply left lying side by side. Sub-cultural theorists set about investigating the rich underlife of the deviant communities, without asking too many questions about how they connected with the larger social system. Robert Merton is one of the few sociologists who, from a position within the structural functionalist or ‘anomie’ perspective, took this question seriously (Merton, 1957).

But this theoretical pluralism could not survive for long. For it soon became clear that these differentiations between ‘deviant’ and ‘consensus’ formations were not natural but socially defined—as the contrast between the different attitudes towards alcohol and cannabis indicated. Moreover, they were historically variable: sub-cultural theorists were just old enough to recall the days of Prohibition, and could contrast them with the period when the positive definitions of American masculinity appeared to require a steady diet of hard liquor and king-sized filter-tips. What mattered was the power of the alcoholtakers to define the cannabis-smokers as deviant. In short, matters of cultural and social power—the power to define the rules of the game to which everyone was required to ascribe—were involved in the transactions between those who were consensus-subscribers and those who were labelled deviant. There was what Howard Becker, one of the early ‘appreciators’ of deviance, called a ‘hierarchy of credibility’ (Becker, 1967). Moreover, such ‘definitions’ were operational. Deviants were positively identified and labelled: the labelling process served to mobilize moral censure and social sanction against them. This had—as those who now recalled the forgotten parts of Durkheim's programme acknowledged—the consequence of reinforcing the internal solidarity of the moral community. As Durkheim puts it: ‘Crime brings together upright consciences and concentrates them’ (Durkheim, 1960, p. 102). But it also served to enforce greater conformity to society's ‘rules’ by punishing and stigmatizing those who departed from them. Beyond the limit of moral censure were, of course, all those sterner practices of legal processing and enforcement which punished, on behalf of society, deviant infractors. The question then arose: who had the power to define whom? And, more pertinently, in the interest of what was the disposition of power between definers and defined secured? In what interest did the consensus ‘work’? What particular type of special order did it sustain and underpin?

In fact, what was at issue here was the problem of social control, and the role of social control in the maintenance of the social order. But this was no longer simply that form of social order expressively revealed in the spontaneous ‘agreement to agree on fundamentals’ of the vast majority: it was not simply the ‘social bond’ which was enforced. It was consent to a particular kind of social order; a consensus around a particular form of society: integration within and conformity to the rules of a very definite set of social, economic and political structures. It was for these—in a direct or indirect sense—that the rules could be said to ‘work’. Social order now looked like a rather different proposition. It entailed the enforcement of social, political and legal discipline. It was articulated to that which existed: to the given dispositions of class, power and authority: to the established institutions of society. This recognition radically problematized the whole notion of ‘consensus’.

What is more, the question could now be asked whether the consensus did indeed spontaneously simply arise or whether it was the result of a complex process of social construction and legitimation. A society, democratic in its formal organization, committed at the same time by the concentration of economic capital and political power to the massively unequal distribution of wealth and authority, had much to gain from the continuous production of popular consent to its existing structure, to the values which supported and underwrote it, and to its continuity of existence. But this raised questions concerning the social role of the media. For if the media were not simply reflective or ‘expressive’ of an already achieved consensus, but instead tended to reproduce those very definitions of the situation which favoured and legitimated the existing structure of things, then what had seemed at first as merely a reinforcing role had now to be reconceptualized in terms of the media's role in the process of consensus formation.

A second break, then, arose around the notion of ‘definitions of the situation’. What this term suggested was that a pivotal element in the production of consent was how things were defined. But this threw into doubt the reflexive role of the media—simply showing things as they were—and it put in question the transparent conception of language which underpinned their assumed naturalism. For reality could no longer be viewed as simply a given set of facts: it was the result of a particular way of constructing reality. The media defined, not merely reproduced, ‘reality’. Definitions of reality were sustained and produced through all those linguistic practices (in the broad sense) by means of which selective definitions of ‘the real’ were represented. But representation is a very different notion from that of reflection. It implies the active work of selecting and presenting, of structuring and shaping: not merely the transmitting of an already existing meaning, but the more active labour of making things mean. It was a practice, a production, of meaning: what subsequently came to be defined as a ‘signifying practice’. The media were signifying agents. A whole new conception of the symbolic practices through which this process of signification was sustained intervened in the innocent garden of ‘content analysis’. The message had now to be analysed, not in terms of its manifest ‘message’, but in terms of its ideological structuration. Several questions then followed: how was this ideological structuration accomplished? How was its relation to the other parts of the social structure to be conceptualized? In the words of Bachrach and Baratz, did it matter that the media appeared to underwrite systematically ‘a set of predominant values, beliefs, rituals, and institutional procedures (“rules of the game”) that operate systematically and consistently to the benefit of certain persons and groups at the expense of others?’ (Bachrach and Baratz, 1970, pp. 43–4). In this move to take seriously the power of the media to signify reality and to define what passed as ‘the real’, the so-called ‘end of ideology’ thesis was also radically problematized.

In part, what was involved in these questions was a return of the problem of power to the powerless universe of mainstream pluralism, but also, a shift in the very conception of power. Pluralism, as Lukes has suggested (Lukes, 1976), did retain a model of power, based around the notion of ‘influence’. A influenced B to make decision X. Certainly, this was a form of power. Pluralism qualified the persistence of this form of power by demonstrating that, because, in any decisionmaking situation, the As were different, and the various decisions made did not cohere within any single structure of domination, or favour exclusively any single interest, therefore power itself had been relatively ‘pluralized’. The dispersal of power plus the randomness of decisions kept the pluralist society relatively free of an identifiable power-centre. (Various gaps in this randompower model were unconvincingly plugged by the discreet deployment of a theory of ‘democratic élitism’ to up-date the ‘pure’ pluralist model and make it square more with contemporary realities). Lukes observes that this is a highly behaviouristic and one-dimensional model of power. But the notion of power which arose from the critique of consensus-theory, and which Bachrach and Baratz, for example, proposed, was of a very different order: ‘Power is also exercised when A devotes energies to creating or reinforcing social and political values and institutional practices that limit the scope of the political process to public consideration of only those issues which are comparatively innocuous to A’ (Bachrach and Baratz, 1970, p. 7),—a modest way of putting the ideological question. Lukes puts this twodimensional model even more clearly when he refers to that power exercised ‘by influencing, shaping and determining [an individual's] very wants’ (Lukes, 1975, p. 16). In fact, this is a very different order of question altogether—a three-dimensional model, which has thoroughly broken with the behaviourist and pluralist assumptions. It is the power which arises from ‘shaping perceptions, cognitions and preferences in such a way that they [i.e. social agents] accept their role in the existing order of things, either because they can see or imagine no alternative to it, or because they see it as natural and unchangeable, or because they value it as divinely ordained or beneficial’ (Lukes, 1975, p. 24). This is an ‘ideological’ model of power, by whatever other name it is called. The move from the pluralist to the critical model of media research centrally involved a shift from a one-to the two-and threedimensional models of power in modern societies. From the viewpoint of the media, what was at issue was no longer specific message-injunctions, by A to B, to do this or that, but a shaping of the whole ideological environment: a way of representing the order of things which endowed its limiting perspectives with that natural or divine inevitability which makes them appear universal, natural and coterminous with ‘reality’ itself. This movement—towards the winning of a universal validity and legitimacy for accounts of the world which are partial and particular, and towards the grounding of these particular constructions in the taken-for-grantedness of ‘the real’—is indeed the characteristic and defining mechanism of ‘the ideological’.

THE CRITICAL PARADIGM

It is around the rediscovery of the ideological dimension that the critical paradigm in media studies turned. Two aspects were involved: each is dealt with separately below. How does the ideological process work and what are its mechanisms? How is ‘the ideological’ to be conceived in relation to other practices within a social formation? The debate developed on both these fronts, simultaneously. The first, which concerned the production and transformation of ideological discourses, was powerfully shaped by theories concerning the symbolic and linguistic character of ideological discourses—the notion that the elaboration of ideology found in language (broadly conceived) its proper and privileged sphere of articulation. The second, which concerned how to conceptualize the ideological instance within a social formation, also became the site of an extensive theoretical and empirical development.

In our discussion of these two supporting elements of the critical paradigm, I shall not be concerned with identifying in detail the specific theoretical inputs of particular disciplines—linguistics, phenomenology, semiotics, psychoanalysis, for example—nor with the detailed internal arguments between these different approaches. Nor shall I attempt to offer a strict chronological account of how the succession of concepts and disciplines were integrated in sequences into the paradigm. I shall rather be concerned exclusively with identifying the broad lines through which the reconceptualization of ‘the ideological’ occurred, and the integration of certain key theoretical elements into the general framework of the paradigm as such.

Cultural inventories

I shall first examine how ideologies work. Here we can begin with the influence of the Sapir-Whorf hypothesis in linguistic anthropology: an idea which, though never picked up in detail, suggests some important continuities between the new paradigm and some previous work, especially in social anthropology. The Sapir-Whorf hypothesis suggested that each culture had a different way of classifying the world. These schemes would be reflected, it argued, in the linguistic and semantic structures of different societies. Lévi-Strauss worked on a similar idea, though he gradually became less interested in the cultural specificity of each society's classification system, and more involved with outlining the universal ‘laws’ of signification—a universal transformational cultural ‘grammar’, common to all cultural systems—associated with the cognitive function, the laws of the mind, and with thinking as such. Lévi-Strauss performed such an analysis on the cultural systems and myths of so-called ‘primitive’ societies—‘societies without history’, as he called them. These examples were well fitted to his universalism, since their cultural systems were highly repetitive, consisting often of the weaving together of different transformations on the same, very limited classificatory ‘sets’. Though the approach did not, clearly, hold so well for societies of more continuous and extensive historical transformation, the general idea proved a fruitful one: it showed how an apparently ‘free’ construction of particular ideological discourses could be viewed as transformations worked on the same, basic, ideological grid. In this, Lévi-Strauss was following Saussure's (1960) call for the development of a general ‘science of signs’—semiology: the study of ‘the life of signs at the heart of social life’ (Lévi-Strauss, 1967, p. 16). Potentially, it was argued, the approach could be applied to all societies and a great variety of cultural systems. The name most prominently associated with this broadening of ‘the science of signs’ was that of Roland Barthes, whose work on modern myths, Mythologies, is a locus classicus for the study of the intersection of myth, language and ideology. The further extrapolation —that whole societies and social practices apart from language could also be analysed ‘on the model of a language’—was subsequently much developed, especially in Marxist structuralism: though the germ of the idea was to be found in Lévi-Strauss, who analysed kinship relations in primitive societies in just this way (i.e. on a communicative model—the exchange of goods, messages and women) (Lévi-Strauss, 1969).

The structuralist strand is, clearly, the most significant one, theoretically, in this development. But we should note that similar pointers could be found in theoretical approaches far removed from the universe of structuralism. It was also present in the ‘social construction of reality’ approach, developed by Berger and Luckmann (1966). Interactionist deviancy theory, which we earlier suggested first identified the question of ‘the definition of the situation’ and ‘who defines whom?’ also moved, though more tentatively, in the same direction. David Matza's book, Becoming Deviant, concluded with a strange and wayward section, intriguingly entitled ‘Signification’ (Matza, 1969). Also relevant was the work of the ethnomethodologists, with their concern for the strategies involved in the understandings of everyday situations, the form of practical accounting by means of which societal members produced the social knowledge they used to make themselves understood, and their increasing attention to conversational strategies.

In the structuralist approach, the issue turned on the question of signification. This implies, as we have already said, that things and events in the real world do not contain or propose their own, integral, single and intrinsic meaning, which is then merely transferred through language. Meaning is a social production, a practice. The world has to be made to mean. Language and symbolization is the means by which meaning is produced. This approach dethroned the referential notion of language, which had sustained previous content analysis, where the meaning of a particular term or sentence could be validated simply by looking at what, in the real world, it referenced. Instead, language had to be seen as the medium in which specific meanings are produced. What this insight put at issue, then, was the question of which kinds of meaning get systematically and regularly constructed around particular events. Because meaning was not given but produced, it followed that different kinds of meaning could be ascribed to the same events. Thus, in order for one meaning to be regularly produced, it had to win a kind of credibility, legitimacy or taken-for-grantedness for itself. That involved marginalizing, down-grading or de-legitimating alternative constructions. Indeed, there were certain kinds of explanation which, given the power of and credibility acquired by the preferred range of meanings were literally unthinkable or unsayable (see Hall et al., 1977). Two questions followed from this. First, how did a dominant discourse warrant itself as the account, and sustain a limit, ban or proscription over alternative or competing definitions? Second, how did the institutions which were responsible for describing and explaining the events of the world—in modern societies, the mass media, par excellence—succeed in maintaining a preferred or delimited range of meanings in the dominant systems of communication? How was this active work of privileging or giving preference practically accomplished?

This directed attention to those many aspects of actual media practice which had previously been analysed in a purely technical way. Conventional approaches to media content had assumed that questions of selection and exclusion, the editing of accounts together, the building of an account into a ‘story’, the use of particular narrative types of exposition, the way the verbal and visual discourses of, say, television were articulated together to make a certain kind of sense, were all merely technical issues. They abutted on the question of the social effects of the media only in so far as bad editing or complex modes of narration might lead to incomprehension on the viewer's part, and thus prevent the pre-existing meaning of an event, or the intention of the broadcaster to communicate clearly, from passing in an uninterrupted or transparent way to the receiver. But, from the viewpoint of signification, these were all elements or elementary forms of a social practice. They were the means whereby particular accounts were constructed. Signification was a social practice because, within media institutions, a particular form of social organization had evolved which enabled the producers (broadcasters) to employ the means of meaning production at their disposal (the technical equipment) through a certain practical use of them (the combination of the elements of signification identified above) in order to produce a product (a specific meaning) (see Hall, 1975). The specificity of media institutions therefore lay precisely in the way a social practice was organized so as to produce a symbolic product. To construct this rather than that account required the specific choice of certain means (selection) and their articulation together through the practice of meaning production (combination). Structural linguists like Saussure and Jacobson had, earlier, identified selection and combination as two of the essential mechanisms of the general production of meaning or sense. Some critical researchers then assumed that the description offered above—producers, combining together in specific ways, using determinate means, to work up raw materials into a product—justified their describing signification as exactly similar to any other media labour process. Certain insights were indeed to be gained from that approach. However, signification differed from other modern labour processes precisely because the product which the social practice produced was a discursive object. What differentiated it, then, as a practice was precisely the articulation together of social and symbolic elements—if the distinction will be allowed here for the purposes of the argument. Motor cars, of course, have, in addition to their exchange and use values, a symbolic value in our culture. But, in the process of meaning construction, the exchange and use values depend on the symbolic value which the message contains. The symbolic character of the practice is the dominant element although not the only one. Critical theorists who argued that a message could be analysed as just another kind of commodity missed this crucial distinction (Garham, 1979; Golding and Murdock, 1979).

The politics of signification

As we have suggested, the more one accepts that how people act will depend in part on how the situations in which they act are defined, and the less one can assume either a natural meaning to everything or a universal consensus on what things mean—then, the more important, socially and politically, becomes the process by means of which certain events get recurrently signified in particular ways. This is especially the case where events in the world are problematic (that is, where they are unexpected); where they break the frame of our previous expectations about the world; where powerful social interests are involved; or where there are starkly opposing or conflicting interests at play. The power involved here is an ideological power: the power to signify events in a particular way.

To give an obvious example: suppose that every industrial dispute could be signified as a threat to the economic life of the country, and therefore against ‘the national interest’. Then such significations would construct or define issues of economic and industrial conflict in terms which would consistently favour current economic strategies, supporting anything which maintains the continuity of production, whilst stigmatizing anything which breaks the continuity of production, favouring the general interests of employers and shareholders who have nothing to gain from production being interrupted, lending credence to the specific policies of governments which seek to curtail the right to strike or to weaken the bargaining position and political power of the trade unions. (For purposes of the later argument, note that such significations depend on taking for-granted what the national interest is. They are predicated on an assumption that we all live in a society where the bonds which bind labour and capital together are stronger, and more legitimate, than the grievances which divide us into labour versus capital. That is to say, part of the function of a signification of this kind is to construct a subject to which the discourse applies: e.g. to translate a discourse whose subject is ‘workers versus employers’ into a discourse whose subject is the collective ‘we, the people’). That, on the whole, industrial disputes are indeed so signified is a conclusion strongly supported by the detailed analyses subsequently provided by, for example, the Glasgow Media Group research published in Bad News (1976) and More Bad News (1980). Now, of course, an industrial dispute has no singular, given meaning. It could, alternatively, be signified as a necessary feature of all capitalist economies, part of the inalienable right of workers to withdraw their labour, and a necessary defence of working-class living standards—the very purpose of the trade unions, for which they have had to fight a long and bitter historic struggle. So, by what means is the first set of significations recurrently preferred in the ways industrial disputes are constructed in our society? By what means are the alternative definitions which we listed excluded? And how do the media, which are supposed to be impartial, square their production of definitions of industrial conflict which systematically favour one side in such disputes, with their claims to report events in a balanced and impartial manner? What emerges powerfully from this line of argument is that the power to signify is not a neutral force in society. Significations enter into controversial and conflicting social issues as a real and positive social force, affecting their outcomes. The signification of events is part of what has to be struggled over, for it is the means by which collective social understandings are created—and thus the means by which consent for particular outcomes can be effectively mobilized. Ideology, according to this perspective, has not only become a ‘material force’, to use an old expression—real because it is ‘real’ in its effects. It has also become a site of struggle (between competing definitions) and a stake—a prize to be won—in the conduct of particular struggles. This means that ideology can no longer be seen as a dependent variable, a mere reflection of a pre-given reality in the mind. Nor are its outcomes predictable by derivation from some simple determinist logic. They depend on the balance of forces in a particular historical conjuncture: on the ‘politics of signification’.

Central to the question of how a particular range of privileged meanings was sustained was the question of classification and framing. Lévi-Strauss, drawing on models of transformational linguistics, suggested that signification depended, not on the intrinsic meaning of particular isolated terms, but on the organized set of interrelated elements within a discourse. Within the colour spectrum, for example, the range of colours would be subdivided in different ways in each culture. Eskimos have several words for the thing which we call ‘snow’. Latin has one word, mus, for the animal which in English is distinguished by two terms, ‘rat’ and ‘mouse’. Italian distinguishes between legno and bosco where English only speaks of a ‘wood’. But where Italian has both bosco and foresta, German only has the single term, wald. (The examples are from Eco's essay, ‘Social life as a sign system’ (1973)). These are distinctions, not of Nature but of Culture. What matters, from the viewpoint of signification, is not the integral meaning of any single colour-term,—mauve, for example—but the system of differences between all the colours in a particular classificatory system; and where, in a particular language, the point of difference between one colour and another is positioned. It was through this play of difference that a language system secured an equivalence between its internal system (signifiers) and the systems of reference (signifieds) which it employed. Language constituted meaning by punctuating the continuum of Nature into a cultural system; such equivalences or correspondences would therefore be differently marked. Thus there was no natural coincidence between a word and its referent: everything depended on the conventions of linguistic use, and on the way language intervened in Nature in order to make sense of it. We should note that at least two, rather different epistemological positions can be derived from this argument. A Kantian or neo-Kantian position would say that, therefore, nothing exists except that which exists in and for language or discourse. Another reading is that, though the world does exist outside language, we can only make sense of it through its appropriation in discourse. There has been a good deal of epistemological heavy warfare around these positions in recent years.

What signified, in fact, was the positionality of particular terms within a set. Each positioning marked a pertinent difference in the classificatory scheme involved. To this Lévi-Strauss added a more structuralist point: that it is not the particular utterance of speakers which provides the object of analysis, but the classificatory system which underlies those utterances and from which they are produced, as a series of variant transformations. Thus, by moving from the surface narrative of particular myths to the generative system or structure out of which they were produced, one could show how apparently different myths (at the surface level) belonged in fact to the same family or constellation of myths (at the deep-structure level). If the underlying set is a limited set of elements which can be variously combined, then the surface variants can, in their particular sense, be infinitely varied, and spontaneously produced. The theory closely corresponds in certain aspects to Chomsky's theory of language, which attempted to show how language could be both free and spontaneous, and yet regular and ‘grammatical’. Changes in meaning, therefore, depended on the classificatory systems involved, and the ways different elements were selected and combined to make different meanings. Variations in the surface meaning of a statement, however, could not in themselves resolve the question as to whether or not it was a transformation of the same classificatory set.

This move from content to structure or from manifest meaning to the level of code is an absolutely characteristic one in the critical approach. It entailed a redefinition of what ideology was—or, at least, of how ideology worked. The point is clearly put by Veron:

If ideologies are structures…then they are not ‘images’ nor ‘concepts’ (we can say, they are not contents) but are sets of rules which determine an organization and the functioning of images and concepts…. Ideology is a system of coding reality and not a determined set of coded messages…in this way, ideology becomes autonomous in relation to the consciousness or intention of its agents: these may be conscious of their points of view about social forms but not of the semantic conditions (rules and categories or codification) which make possible these points of view…. From this point of view, then, an ‘ideology’ may be defined as a system of semantic rules to generate messages…it is one of the many levels of organization of messages, from the viewpoint of their semantic properties…(Veron, 1971, p. 68)

Critics have argued that this approach forsakes the content of particular messages too much for the sake of identifying their underlying structure. Also, that it omits any consideration of how speakers themselves interpret the world—even if this is always within the framework of those shared sets of meanings which mediate between individual actors/speakers and the discursive formations in which they are speaking. But, provided the thesis is not pushed too far in a structuralist direction, it provides a fruitful way of reconceptualizing ideology. Lévi-Strauss regarded the classificatory schemes of a culture as a set of ‘pure’, formal elements (though, in his earlier work, he was more concerned with the social contradictions which were articulated in myths, through the combined operations on their generative sets). Later theorists have proposed that the ideological discourses of a particular society function in an analogous way. The classificatory schemes of a society, according to this view, could therefore be said to consist of ideological elements or premises. Particular discursive formulations would, then, be ideological, not because of the manifest bias or distortions of their surface contents, but because they were generated out of, or were transformations based on, a limited ideological matrix or set. Just as the myth-teller may be unaware of the basic elements out of which his particular version of the myth is generated, so broadcasters may not be aware of the fact that the frameworks and classifications they were drawing on reproduced the ideological inventories of their society. Native speakers can usually produce grammatical sentences in their native language but only rarely can they describe the rules of syntax in use which make their sentences orderly, intelligible to others and grammatical in form. In the same way, statements may be unconsciously drawing on the ideological frameworks and classifying schemes of a society and reproducing them—so that they appear ideologically ‘grammatical’—without those making them being aware of so doing. It was in this sense that the structuralists insisted that, though speech and individual speech-acts may be an individual matter, the language-system (elements, rules of combination, classificatory sets) was a social system: and therefore that speakers were as much ‘spoken’ by their language as speaking it. The rules of discourse functioned in such a way as to position the speaker as if he or she were the intentional author of what was spoken. The system on which this authorship depended remained, however, profoundly unconscious. Subsequent theorists noticed that, although this de-centered the authorial ‘I’ making it dependent on the language systems speaking through the subject, this left an empty space where, in the Cartesian conception of the subject, the allencompassing ‘I’ had previously existed. In theories influenced by Freudian and Lacanian psychoanalysis (also drawing on LéviStrauss), this question of how the speaker, the subject of enunciation, was positioned in language became, not simply one of the mechanisms through which ideology was articulated, but the principal mechanism of ideology itself (Coward and Ellis, 1977). More generally, however, it is not difficult to see how Lévi-Strauss's proposition—‘speakers produce meaning, but only on the basis of conditions which are not of the speaker's making, and which pass through him/her into language, unconsciously’—could be assimilated to the more classic Marxist proposition that ‘people make history, but only in determinate conditions which are not of their making, and which pass behind their backs’. In later developments, these theoretical homologies were vigorously exploited, developed—and contested.

Historicizing the structures

Of course, in addition to the homologies with Lévi-Strauss's approach, there were also significant differences. If the inventories from which particular significations were generated were conceived, not simply as a formal scheme of elements and rules, but as a set of ideological elements, then the conceptions of the ideological matrix had to be radically historicized. The ‘deep structure’ of a statement had to be conceived as the network of elements, premises and assumptions drawn from the longstanding and historically-elaborated discourses which had accreted over the years, into which the whole history of the social formation had sedimented, and which now constituted a reservoir of themes and premises on which, for example, broadcasters could draw for the work of signifying new and troubling events. Gramsci, who referred, in a less formal way, to the inventory of traditional ideas, the forms of episodic thinking which provide us with the taken-for-granted elements of our practical knowledge, called this inventory ‘common sense’.

What must be explained is how it happens that in all periods there coexist many systems and currents of philosophical thought, how these currents are born, how they are diffused, and why in the process of diffusion they fracture along certain lines and in certain directions…it is this history which shows how thought has been elaborated over the centuries and what a collective effort has gone into the creation of our present method of thought which has subsumed and absorbed all this past history, including all its follies and mistakes. (Gramsci, 1971, p. 327)

In another context, he argued:

Every social stratum has its own ‘common sense’ and its own ‘good sense’, which are basically the most widespread conception of life and of men. Every philosophical current leaves behind a sedimentation of ‘common sense’: this is the document of its historical effectiveness. Common sense is not something rigid and immobile, but is continually transforming itself, enriching itself with scientific ideas and with philosophical opinions which have entered ordinary life…. Common sense creates the folklore of the future, that is as a relatively rigid phase of popular knowledge at a given place and time. (Gramsci, 1971, p. 326)

The formalist conception of the ‘cultural inventory’ suggested by structuralism was not, in my view, available as a theoretical support for the elaboration of an adequate conception of ideology until it had been thoroughly historicized in this way. Only thus did the preoccupation, which Lévi-Strauss initiated, with the universal ‘grammars’ of culture begin to yield insights into the historical grammars which divided and classified the knowledge of particular societies into their distinctive ideological inventories.

The structural study of myth suggested that, in addition to the ways in which knowledge about the social world was classified and framed, there would be a distinctive logic about the ways in which the elements in an inventory could yield certain stories or statements about the world. It was, according to Lévi-Strauss, the ‘logic of arrangement’ rather than the particular contents of a myth which ‘signified’. It was at this level that the pertinent regularities and recurrencies could best be observed. By ‘logic’ he did not, certainly, mean logic in the philosophical sense adopted by western rationalism. Indeed, his purpose was to demonstrate that Western rationalism was only one of the many types of discursive arrangement possible; no different intrinsically, in terms of how it worked, from the logic of so-called pre-scientific thinking or mythic thought. Logic here simply meant an apparently necessary chain of implication between statement and premise. In western logic, propositions are said to be logical if they obey certain rules of inference and deduction. What the cultural analyst meant by logic was simply that all ideological propositions about the social world were similarly premised, predicated or inferenced. They entailed a framework of linked propositions, even if they failed the test of logical deduction. The premises had to be assumed to be true, for the propositions which depended on them to be taken as true. This notion of ‘the entailment of propositions’, or, as the semanticists would say, the embeddedness of statements, proved of seminal value in the development of ideological analysis. To put the point in its extreme form, a statement like ‘the strike of Leyland tool-makers today further weakened Britain's economic position’ was premised on a whole set of taken-for-granted propositions about how the economy worked, what the national interest was, and so on. For it to win credibility, the whole logic of capitalist production had to be assumed to be true. Much the same could be said about any item in a conventional news bulletin, that, without a whole range of unstated premises or pieces of taken-for-granted knowledge about the world, each descriptive statement would be literally unintelligible. But this ‘deep structure’ of presuppositions, which made the statement ideologically ‘grammatical’, were rarely made explicit and were largely unconscious, either to those who deployed them to make sense of the world or to those who were required to make sense of it. Indeed, the very declarative and descriptive form of the statement rendered invisible the implied logic in which it was embedded. This gave the statement an unchallenged obviousness, and obvious truthvalue. What were in fact propositions about how things were, disappeared into and acquired the substantive affirmation of merely descriptive statements: ‘facts of the case’. The logic of their entailment being occluded, the statements seemed to work, so to speak, by themselves. They appeared as proposition-free—natural and spontaneous affirmations about ‘reality’.

The reality effect

In this way, the critical paradigm began to dissect the so-called ‘reality’ of discourse. In the referential approach, language was thought to be transparent to the truth of ‘reality itself’—merely transferring this originating meaning to the receiver. The real world was both origin and warrant for the truth of any statement about it. But in the conventional or constructivist theory of language, reality came to be understood, instead, as the result or effect of how things had been signified. It was because a statement generated a sort of ‘recognition effect’ in the receiver that it was taken or ‘read’ as a simple empirical statement. The work of formulation which produced it secured this closing of the pragmatic circle of knowledge. But this recognition effect was not a recognition of the reality behind the words, but a sort of confirmation of the obviousness, the taken forgrantedness of the way the discourse was organized and of the underlying premises on which the statement in fact depended. If one regards the laws 01 a capitalist economy as fixed and immutable, then its notions acquire a natural inevitability. Any statement which is so embedded will thus appear to be merely a statement about ‘how things really are’. Discourse, in short, had the effect of sustaining certain ‘closures’, of establishing certain systems of equivalence between what could be assumed about the world and what could be said to be true. ‘True’ means credible, or at least capable of winning credibility as a statement of fact. New, problematic or troubling events, which breached the taken-for-granted expectancies about how the world should be, could then be ‘explained’ by extending to them the forms of explanation which had served ‘for all practical purposes’, in other cases. In this sense, Althusser was subsequently to argue that ideology, as opposed to science, moved constantly within a closed circle, producing, not knowledge, but a recognition of the things we already knew. It did so because it took as already established fact exactly the premises which ought to have been put in question. Later still, this theory was to be complemented by psychoanalytic theories of the subject which tried to demonstrate how certain kinds of narrative exposition construct a place or position of empirical knowledge for each subject at the centre of any discourse—a position or point of view from which alone the discourse ‘makes sense’. It, accordingly, defined such narrative procedures, which established an empirical pragmatic closure in discourse, as all belonging to the discourse of ‘realism’.

More generally, this approach suggested, discourses not only referenced themselves in the structure of already objectivated social knowledge (the ‘already known’) but established the viewer in a complicitous relationship of pragmatic knowledge to the ‘reality’ of the discourse itself. ‘Point of view’ is not, of course, limited to visual texts—written texts also have their preferred positions of knowledge. But the visual nature of the point-of-view metaphor made it particularly appropriate to those media in which the visual discourse appeared to be dominant. The theory was therefore most fully elaborated in relation to film: but it applied, tout court, to television as well—the dominant medium of social discourse and representation in our society. Much of television's power to signify lay in its visual and documentary character—its inscription of itself as merely a ‘window on the world’, showing things as they really are. Its propositions and explanations were underpinned by this grounding of its discourse in ‘the real’—in the evidence of one's eyes. Its discourse therefore appeared peculiarly a naturalistic discourse of fact, statement and description. But in the light of the theoretical argument sketched above, it would be more appropriate to define the typical discourse of this medium not as naturalistic but as naturalized: not grounded in nature but producing nature as a sort of guarantee of its truth. Visual discourse is peculiarly vulnerable in this way because the systems of visual recognition on which they depend are so widely available in any culture that they appear to involve no intervention of coding, selection or arrangement. They appear to reproduce the actual trace of reality in the images they transmit. This, of course, is an illusion—the ‘naturalistic illusion’—since the combination of verbal and visual discourse which produces this effect of ‘reality’ requires the most skilful and elaborate procedures of coding: mounting, linking and stitching elements together, working them into a system of narration or exposition which ‘makes sense’.

This argument obviously connects with the classical materialist definition of how ideologies work. Marx, you will recall, argued that ideology works because it appears to ground itself in the mere surface appearance of things. In doing so, it represses any recognition of the contingency of the historical conditions on which all social relations depend. It represents them, instead, as outside of history: unchangeable, inevitable and natural. It also disguises its premises as already known facts. Thus, despite its scientific discoveries, Marx described even classical political economy as, ultimately, ‘ideological’ because it took the social relations and the capitalist form of economic organization as the only, and inevitable, kind of economic order. It therefore presented capitalist production ‘as encased in eternal natural laws independent of history’. Bourgeois relations were then smuggled in ‘as the inviolable laws on which society in the abstract is founded’. This eternalization or naturalization of historical conditions and historical change he called ‘a forgetting’. Its effect, he argued, was to reproduce, at the heart of economic theory, the categories of vulgar, bourgeois common sense. Statements about economic relations thus lost their conditional and premised character, and appeared simply to arise from ‘how things are’ and, by implication, ‘how they must forever be’. But this ‘reality-effect’ arose precisely from the circularity, the presupposition-less character, the selfgenerating and self-confirming nature, of the process of representation itself.

The ‘class struggle in language’

Later, within the framework of a more linguistic approach, theorists like Pêcheux were to demonstrate how the logic and sense of particular discourses depended on the referencing, within the discourse, of these preconstructed elements (Pêcheux, 1975). Also, how discourse, in its systems of narration and exposition, signalled its conclusions forward, enabling it to realize certain potential meanings within the chain or logic of its inferences, and closing off other possibilities. Any particular discursive string, they showed, was anchored within a whole discursive field or complex of existing discourses (the ‘inter-discourse’); and these constituted the presignifieds of its statements or enunciations. Clearly, the ‘pre-constituted’ was a way of identifying, linguistically, what, in a more historical sense, Gramsci called the inventory of ‘common sense’. Thus, once again, the link was forged, in ideological analysis, between linguistic or semiological concerns, on the one hand, and the historical analysis of the discursive formations of ‘common sense’ on the other. In referencing, within its system of narration, ‘what was already known’, ideological discourses both warranted themselves in and selectively reproduced the common stock of knowledge in society.

Because meaning no longer depended on ‘how things were’ but on how things were signified, it followed, as we have said, that the same event could be signified in different ways. Since signification was a practice, and ‘practice’ was defined as ‘any process of transformation of a determinate raw material into a determinate product, a transformation effected by a determinate human labour, using determinate means (of “production”)’ (Althusser, 1969, p. 166), it also followed that signification involved a determinate form of labour, a specific ‘work’: the work of meaningproduction, in this case. Meaning was, therefore, not determined, say, by the structure of reality itself, but conditional on the work of signification being successfully conducted through a social practice. It followed, also, that this work need not necessarily be successfully effected: because it was a ‘determinate’ form of labour it was subject to contingent conditions. The work of signification was a social accomplishment—to use ethnomethodological terminology for a moment. Its outcome did not flow in a strictly predictable or necessary manner from a given reality. In this, the emergent theory diverged significantly, both from the reflexive or referential theories of language embodied in positivist theory, and from the reflexive kind of theory also implicit in the classical Marxist theory of language and the superstructures.

Three important lines of development followed from this break with early theories of language. Firstly, one had to explain how it was possible for language to have this multiple referentiality to the real world. Here, the polysemic nature of language—the fact that the same set of signifiers could be variously accented in those meanings—proved of immense value. Vološinov put this point best when he observed:

Existence reflected in the sign is not merely reflected but refracted. How is this refraction of existence in the ideological sign determined? By an intersecting of differently oriented social interests in every ideological sign. Sign becomes an arena of class struggle. This social multi-accentuality of the ideological sign is a very crucial aspect…. A sign that has been withdrawn from the pressures of the social struggle—which, so to speak, crosses beyond the whole of the class struggle—inevitably loses force, degenerates into allegory, becoming the object not of a live social intelligibility but of a philological comprehension. (Vološinov, 1973, p. 23)

The second point is also addressed as an addendum, in Vološinov's remark. Meaning, once it is problematized, must be the result, not of a functional reproduction of the world in language, but of a social struggle—a struggle for mastery in discourse—over which kind of social accenting is to prevail and to win credibility. This reintroduced both the notion of ‘differently oriented social interests’ and a conception of the sign as ‘an arena of struggle’ into the consideration of language and of signifying ‘work’.

Althusser, who transposed some of this kind of thinking into his general theory of ideology, tended to present the process as too uni-accentual, too functionally adapted to the reproduction of the dominant ideology (Althusser, 1971). Indeed, it was difficult, from the base-line of this theory, to discern how anything but the ‘dominant ideology’ could ever be reproduced in discourse. The work of Vološinov and Gramsci offered a significant correction to this functionalism by reintroducing into the domain of ideology and language the notion of a ‘struggle over meaning’ (which Vološinov substantiated theoretically with his argument about the multi-accentuality of the sign). What Vološinov argued was that the mastery of the struggle over meaning in discourse had, as its most pertinent effect or result, the imparting of a ‘supraclass, eternal character to the ideological sign, to extinguish or drive inward the struggle between social value judgements which occurs in it, to make the sign uni-accentual’ (1973, p. 23). To go back for a moment to the earlier argument about the realityeffect: Vološinov's point was that uni-accentuality—where things appeared to have only one, given, unalterable and ‘supraclass’ meaning—was the result of a practice of closure: the establishment of an achieved system of equivalence between language and reality, which the effective mastery of the struggle over meaning produced as its most pertinent effect. These equivalences, however, were not given in reality, since, as we have seen, the same reference can be differently signified in different semantic systems; and some systems can constitute differences which other systems have no way of recognizing or punctuating. Equivalences, then, were secured through discursive practice. But this also meant that such a practice was conditional. It depended on certain conditions being fulfilled. Meanings which had been effectively coupled could also be un-coupled. The ‘struggle in discourse’ therefore consisted precisely of this process of discursive articulation and disarticulation. Its outcomes, in the final result, could only depend on the relative strength of the ‘forces in struggle’, the balance between them at any strategic moment, and the effective conduct of the ‘politics of signification’. We can think of many pertinent historical examples where the conduct of a social struggle depended, at a particular moment, precisely on the effective disarticulation of certain key terms, e.g. ‘democracy’, the ‘rule of law’, ‘civil rights’, ‘the nation’, ‘the people’, ‘Mankind’, from their previous couplings, and their extrapolation to new meanings, representing the emergence of new political subjects.

The third point, then, concerned the mechanisms within signs and language, which made the ‘struggle’ possible. Sometimes, the class struggle in language occurred between two different terms: the struggle, for example, to replace the term ‘immigrant’ with the term ‘black’. But often, the struggle took the form of a different accenting of the same term: e.g. the process by means of which the derogatory colour ‘black’ became the enhanced value ‘Black’ (as in ‘Black is Beautiful’). In the latter case, the struggle was not over the term itself but over its connotative meaning. Barthes, in his essay on ‘Myth’, argued that the associative field of meanings of a single term—its connotative field of reference—was, par excellence, the domain through which ideology invaded the language system. It did so by exploiting the associative, the variable, connotative, ‘social value’ of language. For some time, this point was misunderstood as arguing that the denotative or relatively fixed meanings of a discourse were not open to multiple accentuation, but constituted a ‘natural’ language system; and only the connotative levels of discourse were open to different ideological inflexion. But this was simply a misunderstanding. Denotative meanings, of course, are not uncoded; they, too, entail systems of classification and recognition in much the same way as connotative meanings do; they are not natural but ‘motivated’ signs. The distinction between denotation and connotation was an analytic, not a substantive one (see Camargo, 1980; Hall, 1980a). It suggested, only, that the connotative levels of language, being more open-ended and associative, were peculiarly vulnerable to contrary or contradictory ideological inflexions.

Hegemony and articulation

The real sting in the tail did not reside there, but in a largely unnoticed extension of Vološinov's argument. For if the social struggle in language could be conducted over the same sign, it followed that signs (and, by a further extension, whole chains of signifiers, whole discourses) could not be assigned, in a determinate way, permanently to any one side in the struggle. Of course, a native language is not equally distributed amongst all native speakers regardless of class, socio-economic postion, gender, education and culture: nor is competence to perform in language randomly distributed. Linguistic performance and competence is socially distributed, not only by class but also by gender. Key institutions—in this respect, the familyeducation couple—play a highly significant role in the social distribution of cultural ‘capital’, in which language played a pivotal role, as educational theorists like Bernstein and social theorists like Bourdieu have demonstrated. But, even where access for everyone to the same language system could be guaranteed, this did not suspend what Vološinov called the ‘class struggle in language’. Of course, the same term, e.g. ‘black’, belonged in both the vocabularies of the oppressed and the oppressors. What was being struggled over was not the ‘class belongingness’ of the term, but the inflexion it could be given, its connotative field of reference. In the discourse of the Black movement, the denigratory connotation ‘black=the despised race’ could be inverted into its opposite: ‘black=beautiful’. There was thus a ‘class struggle in language’; but not one in which whole discourses could be unproblematically assigned to whole social classes or social groups. Thus Vološinov argued:

Class does not coincide with the sign community i.e. with the community which is the totality of users of the same set of signs for ideological communication. Thus various different classes will use one and the same language. As a result, differently oriented accents intersect in every ideological sign. Sign becomes an arena of class struggle. (Vološinov, 1973, p. 23)

This was an important step: the ramifications are briefly traced through below. But one could infer, immediately, two things from this. First, since ideology could be realized by the semantic accenting of the same linguistic sign, it followed that, though ideology and language were intimately linked, they could not be one and the same thing. An analytic distinction needed to be maintained between the two terms. This is a point which later theorists, who identified the entry of the child into his/her linguistic culture as one and the same mechanism as the entry of the child into the ideology of its society neglected to show. But the two processes, though obviously connected (one cannot learn a language without learning something of its current ideological inflexions) cannot be identified or equated in that perfectly homologous way. Ideological discourses can win to their ways of representing the world already-languaged subjects, i.e. subjects already positioned within a range of existing discourses, fully-social speakers. This underlined the necessity to consider, instead, the ‘articulation’ of ideology in and through language and discourse.

Second, though discourse could become an arena of social struggle, and all discourses entailed certain definite premises about the world, this was not the same thing as ascribing ideologies to classes in a fixed, necessary or determinate way. Ideological terms and elements do not necessarily ‘belong’ in this definite way to classes: and they do not necessarily and inevitably flow from class position. The same elementary term, ‘democracy’ for example, could be articulated with other elements and condensed into very different ideologies: democracy of the Free West and the German Democratic Republic, for example. The same term could be disarticulated from its place within one discourse and articulated in a different position: the Queen acknowledging the homage of ‘her people’, for example, as against that sense of ‘the people’ or ‘the popular’ which is oppositional in meaning to everything which connotes the élite, the powerful, the ruler, the power bloc. What mattered was the way in which different social interests or forces might conduct an ideological struggle to disarticulate a signifier from one, preferred or dominant meaning-system, and rearticulate it within another, different chain of connotations. This might be accomplished, formally, by different means. The switch from ‘black=despised’ to ‘black = beautiful’ is accomplished by inversion. The shift from ‘pig=animal with dirty habits’ to ‘pig=brutal policeman’ in the language of the radical movements of the 1960s to ‘pig=male-chauvinist pig’ in the language of feminism, is a metonymic mechanism—sliding the negative meaning along a chain of connotative signifiers. This theory of the ‘no necessary class belongingness’ of ideological elements and the possibilities of ideological struggle to articulate/disarticulate meaning, was an insight drawn mainly from Gramsci's work, but considerably developed in more recent writings by theorists like Laclau (1977).

But the ‘struggle over meaning’ is not exclusively played out in the discursive condensations to which different ideological elements are subject. There was also the struggle over access to the very means of signification: the difference between those accredited witnesses and spokesmen who had a privileged access, as of right, to the world of public discourse and whose statements carried the representativeness and authority which permitted them to establish the primary framework or terms of an argument; as contrasted with those who had to struggle to gain access to the world of public discourse at all; whose ‘definitions’ were always more partial, fragmentary and delegitimated; and who, when they did gain access, had to perform with the established terms of the problematic in play.

A simple but recurrent example of this point in current media discourse is the setting of the terms of the debate about black immigrants to Britain as a problem ‘about numbers’. Liberal or radical spokesmen on race issues could gain all the physical access to the media which they were able to muster. But they would be powerfully constrained if they then had to argue within the terrain of a debate in which ‘the numbers game’ was accepted as the privileged definition of the problem. To enter the debate on these terms was tantamount to giving credibility to the dominant problematic: e.g. ‘racial tension is the result of too many black people in the country, not a problem of white racialism’. When the ‘numbers game’ logic is in play, opposing arguments can be put as forcefully as anyone speaking is capable of: but the terms define the ‘rationality’ of the argument, and constrain how the discourse will ‘freely’ develop. A counter argument—that the numbers are not too high—makes an opposite case: but inevitably, it also reproduces the given terms of the argument. It accepts the premise that the argument is ‘about numbers’. Opposing arguments are easy to mount. Changing the terms of an argument is exceedingly difficult, since the dominant definition of the problem acquires, by repetition, and by the weight and credibility of those who propose or subscribe it, the warrant of ‘common sense’. Arguments which hold to this definition of the problem are accounted as following ‘logically’. Arguments which seek to change the terms of reference are read as ‘straying from the point’. So part of the struggle is over the way the problem is formulated: the terms of the debate, and the ‘logic’ it entails.

A similar case is the way in which the ‘problem of the welfare state’ has come, in the era of economic recession and extreme monetarism, to be defined as ‘the problem of the scrounger’, rather than the ‘problem of the vast numbers who could legally claim benefits, and need them, but don't’. Each framework of course, has real social consequences. The first lays down a base-line from which public perceptions of the ‘black problem’ can develop—linking an old explanation to a new aspect. The next outbreak of violence between blacks and whites is therefore seen as a ‘numbers problem’ too—giving credence to those who advance the political platform that ‘they should all be sent home’, or that immigration controls should be strengthened. The definition of the welfare state as a ‘problem of the illegal claimant’ does considerable duty in a society which needs convincing that ‘we cannot afford welfare’, that it ‘weakens the moral fibre of the nation’, and therefore, that public welfare spending ought to be drastically reduced. Other aspects of the same process—for example, the establishment of the range of issues which demand public attention (or as it is more commonly known, the question of ‘who sets the national agenda?’)—were elaborated as part of the same concern with extending and filling out precisely what we could mean by saying that signification was a site of social struggle.

The fact that one could not read off the ideological position of a social group or individual from class position, but that one would have to take into account how the struggle over meaning was conducted, meant that ideology ceased to be a mere reflection of struggles taking place or determined elsewhere (for example, at the level of the economic struggle). It gave to ideology a relative independence or ‘relative autonomy’. Ideologies ceased to be simply the dependent variable in social struggle: instead, ideological struggle acquired a specificity and a pertinence of its own—needing to be analysed in its own terms, and with real effects on the outcomes of particular struggles. This weakened, and finally overthrew altogether, the classic conception of ideas as wholly determined by other determining factors (e.g. class position). Ideology might provide sets of representations and discourses through which we lived out, ‘in an imaginary way, our relation to our real conditions of existence’ (Althusser, 1969, p. 233). But it was every bit as ‘real’ or ‘material’, as so-called nonideological practices, because it affected their outcome. It was ‘real’ because it was real in its effects. It was determinate, because it depended on other conditions being fulfilled. ‘Black’ could not be converted into ‘black= beautiful’ simply by wishing it were so. It had to become part of an organized practice of struggles requiring the building up of collective forms of black resistance as well as the development of new forms of black consciousness. But, at the same time, ideology was also determining, because, depending on how the ideological struggle was conducted, material outcomes would be positively or negatively affected. The traditional role of the trade unions is to secure and improve the material conditions of their members. But a trade-union movement which lost the ideological struggle, and was successfully cast in the folk-devil role of the ‘enemy of the national interest’, would be one which could be limited, checked and curtailed by legal and political means: one, that is, in a weaker position relative to other forces on the social stage; and thus less able to conduct a successful struggle in the defence of working-class standards of living. In the very period in which the critical paradigm was being advanced, this lesson had to be learned the hard way. The limitations of a trade-union struggle which pursued economic goals exclusively at the expense of the political and ideological dimensions of the struggle were starkly revealed when obliged to come to terms with a political conjuncture where the very balance of forces and the terms of struggle had been profoundly altered by an intensive ideological campaign conducted with peculiar force, subtlety and persistence by the radical Right. The theory that the working class was permanently and inevitably attached to democratic socialism, the Labour Party and the trade-union movement, for example, could not survive a period in which the intensity of the Thatcher campaigns preceding the General Election of 1979 made strategic and decisive inroads, precisely into major sectors of the working-class vote (Hall, 1979; Hall, 1980b). And one of the key turning-points in the ideological struggle was the way the revolt of the lower-paid public-service workers against inflation, in the ‘Winter of Discontent’ of 1978–9, was successfully signified, not as a defence of eroded living standards and differentials, but as a callous and inhuman exercise of overweening ‘trade-union power’, directed against the defenceless sick, aged, dying and indeed the dead but unburied ‘members of the ordinary public’.

Ideology in the social formation

This may be a convenient point in the argument to turn, briefly, to the second strand: concerning the way ideology was conceived in relation to other practices in a social formation. Many of the points in this part of the argument have already been sketched in. Complex social formations had to be analysed in terms of the economic, political and ideological institutions and practices through which they were elaborated. Each of these elements had to be accorded a specific weight in determining the outcomes of particular conjunctures. The question of ideology could not be extrapolated from some other level—the economic, for example—as some versions of classical Marxism proposed. But nor could the question of value-consensus be assumed, or treated as a dependent process merely reflecting in practice that consensus already achieved at the level of ideas, as pluralism supposed. Economic, political and ideological conditions had to be identified and analysed before any single event could be explained. Further, as we have already shown, the presupposition that the reflection of economic reality at the level of ideas could be replaced by a straightforward ‘classdetermination’, also proved to be a false and misleading trail. It did not sufficiently recognize the relative autonomy of ideological processes, or the real effects of ideology on other practices. It treated classes as ‘historical givens’—their ideological ‘unity’ already given by their position in the economic structure—whereas, in the new perspective, classes had to be understood only as the complex result of the successful prosecution of different forms of social struggle at all the levels of social practice, including the ideological. This gave to the struggle around and over the media—the dominant means of social signification in modern societies—a specificity and a centrality which, in previous theories, they had altogether lacked. It raised them to a central, relatively independent, position in any analysis of the question of the ‘politics of signification’.

Though these arguments were cast within a materialist framework, they clearly departed radically from certain conventional ways of putting the Marxist question. In their most extended text on the question, The German Ideology, Marx and Engels had written, The ideas of the ruling classes are in every epoch the ruling ideas i.e. the class which is the ruling material force is at the same time its ruling intellectual force’ (p. 64). The passage is, in fact, more subtle and qualified than that classic and unforgettable opening suggests. But, in the simple form in which it appeared, it could no longer—for reasons partly sketched out earlier—be sustained. Some theorists took this to mean that any relationship between ruling-class and dominant ideas had therefore to be abandoned. My own view is that this threw the baby out with the bath water, in two senses. It was based on the unsupported, but apparently persuasive idea that, since ‘ideas’ could not be given a necessary ‘class belongingness’, therefore there could be no relation of any kind between the processes through which ideologies were generated in society and the constitution of a dominant alliance or power bloc based on a specific configuration of classes and other social forces. But clearly it was not necessary to go so far in breaking the theory of ideology free of a necessitarian logic. A more satisfactory approach was to take the point of ‘no necessary class belongingness’: and then to ask under what circumstances and through what mechanisms certain class articulations of ideology might be actively secured. It is clear, for example, that even though there is no necessary belongingness of the term ‘freedom’ to the bourgeoisie, historically, a certain class articulation of the term has indeed been effectively secured, over long historical periods: that which articulated ‘freedom’ with the liberty of the individual, with the ‘free’ market and liberal political values, but which disarticulated it from its possible condensations in a discourse predicated on the ‘freedom’ of the worker to withdraw his labour or the ‘freedom’ of the ‘freedom-fighter’. These historical traces are neither necessary nor determined in a final fashion. But such articulations have been historically secured. And they do have effects. The equivalences having been sustained, they are constantly reproduced in other discourses, in social practices and institutions, in ‘free societies’. These traditional couplings, or ‘traces’ as Gramsci called them, exert a powerful traditional force over the ways in which subsequent discourses, employing the same elements, can be developed. They give such terms, not an absolutely determined class character, but a tendential class articulation. The question as to how the articulation of ideological discourses to particular class formations can be conceptualized, without falling back into a simple class reductionism, is a matter on which important work has since been done (the work of Laclau referred to earlier here is, once again, seminal).

Second, to lose the ruling-class/ruling-ideas proposition altogether is, of course, also to run the risk of losing altogether the notion of ‘dominance’. But dominance is central if the propositions of pluralism are to be put in question. And, as we have shown, the critical paradigm has done a great deal of work in showing how a non-reductionist conception of dominance can be worked out in the context of a theory of ideology. However, important modifications to our way of conceiving dominance had to be effected before the idea was rescuable. That notion of dominance which meant the direct imposition of one framework, by overt force or ideological compulsion, on a subordinate class, was not sophisticated enough to match the real complexities of the case. One had also to see that dominance was accomplished at the unconscious as well as the conscious level: to see it as a property of the system of relations involved, rather than as the overt and intentional biases of individuals; and to recognize its play in the very activity of regulation and exclusion which functioned through language and discourse before an adequate conception of dominance could be theoretically secured. Much of this debate revolved around the replacement of all the terms signifying the external imposition of ideas or total incorporation into ‘ruling ideas’ by the enlarged concept of ‘hegemony’. Hegemony implied that the dominance of certain formations was secured, not by ideological compulsion, but by cultural leadership. It circumscribed all those processes by means of which a dominant class alliance or ruling bloc, which has effectively secured mastery over the primary economic processes in society, extends and expands its mastery over society in such a way that it can transform and re-fashion its ways of life, its mores and conceptualization, its very form and level of culture and civilization in a direction which, while not directly paying immediate profits to the narrow interests of any particular class, favours the development and expansion of the dominant social and productive system of life as a whole. The critical point about this conception of ‘leadership’—which was Gramsci's most distinguished contribution—is that hegemony is understood as accomplished, not without the due measure of legal and legitimate compulsion, but principally by means of winning the active consent of those classes and groups who were subordinated within it.

From the ‘reflection of consensus’ to the ‘production of consent’

This was a vital issue—and a critical revision. For the weakness of the earlier Marxist positions lay precisely in their inability to explain the role of the ‘free consent’ of the governed to the leadership of the governing classes under capitalism. The great value of pluralist theory was precisely that it included this element of consent—though it gave to it a highly idealist and power-free gloss or interpretation. But, especially in formally democratic class societies, of which the US and Britain are archetypal cases, what had to be explained was exactly the combination of the maintained rule of powerful classes with the active or inactive consent of the powerless majority. The ruling-class/ruling-ideas formula did not go far enough in explaining what was clearly the most stabilizing element in such societies—consent. ‘Consensus theory’ however, gave an unproblematic reading to this element—recognizing the aspect of consent, but having to repress the complementary notions of power and dominance. But hegemony attempted to provide the outlines, at least, of an explanation of how power functioned in such societies which held both ends of the chain at once. The question of ‘leadership’ then, became, not merely a minor qualification to the theory of ideology, but the principal point of difference between a more and a less adequate explanatory framework. The critical point for us is that, in any theory which seeks to explain both the monopoly of power and the diffusion of consent, the question of the place and role of ideology becomes absolutely pivotal. It turned out, then, that the consensus question, in pluralist theory, was not so much wrong as incorrectly or inadequately posed. As is often the case in theoretical matters, a whole configuration of ideas can be revealed by taking an inadequate premise and showing the unexamined conditions on which it rested. The ‘break’ therefore, occurred precisely at the point where theorists asked, ‘but who produces the consensus?’ ‘In what interests does it function?’ ‘On what conditions does it depend?’ Here, the media and other signifying institutions came back into the question—no longer as the institutions which merely reflected and sustained the consensus, but as the institutions which helped to produce consensus and which manufactured consent.

This approach could also be used to demonstrate how media institutions could be articulated to the production and reproduction of the dominant ideologies, while at the same time being ‘free’ of direct compulsion, and ‘independent’ of any direct attempt by the powerful to nobble them. Such institutions powerfully secure consent precisely because their claim to be independent of the direct play of political or economic interests, or of the state, is not wholly fictitious. The claim is ideological, not because it is false but because it does not adequately grasp all the conditions which make freedom and impartiality possible. It is ideological because it offers a partial explanation as if it were a comprehensive and adequate one—it takes the part for the whole (fetishism). Nevertheless, its legitimacy depends on that part of the truth, which it mistakes for the whole, being real in fact, and not merely a polite fiction.

This insight was the basis for all of that work which tried to demonstrate how it could be true that media institutions were both, in fact, free of direct compulsion and constraint, and yet freely articulated themselves systematically around definitions of the situation which favoured the hegemony of the powerful. The complexities of this demonstration cannot be entered into here and a single argument, relating to consensus, will have to stand. We might put it this way. Formally, the legitimacy of the continued leadership and authority of the dominant classes in capitalist society derives from their accountability to the opinions of the popular majority—the ‘sovereign will of the people’. In the formal mechanisms of election and the universal franchise they are required to submit themselves at regular intervals to the will or consensus of the majority. One of the means by which the powerful can continue to rule with consent and legitimacy is, therefore, if the interests of a particular class or power bloc can be aligned with or made equivalent to the general interests of the majority. Once this system of equivalences has been achieved, the interests of the minority and the will of the majority can be ‘squared’ because they can both be represented as coinciding in the consensus, on which all sides agree. The consensus is the medium, the regulator, by means of which this necessary alignment (or equalization) between power and consent is accomplished. But if the consensus of the majority can be so shaped that it squares with the will of the powerful, then particular (class) interests can be represented as identical with the consensus will of the people. This, however, requires the shaping, the education and tutoring of consent: it also involves all those processes of representation which we outlined earlier.

Now consider the media—the means of representation. To be impartial and independent in their daily operations, they cannot be seen to take directives from the powerful, or consciously to be bending their accounts of the world to square with dominant definitions. But they must be sensitive to, and can only survive legitimately by operating within, the general boundaries or framework of ‘what everyone agrees’ to: the consensus. When the late Director General of the BBC, Sir Charles Curran remarked that ‘the BBC could not exist outside the terms of parliamentary democracy’, what he was pointing to was the fact that broadcasting, like every other institution of state in Britain, must subscribe to the fundamental form of political regime of the society, since it is the foundation of society itself and has been legitimated by the will of the majority. Indeed, the independence and impartiality on which broadcasters pride themselves depends on this broader coincidence between the formal protocols of broadcasting and the form of state and political system which licenses them. But, in orienting themselves in ‘the consensus’ and, at the same time, attempting to shape up the consensus, operating on it in a formative fashion, the media become part and parcel of that dialectical process of the ‘production of consent’—shaping the consensus while reflecting it—which orientates them within the field of force of the dominant social interests represented within the state.

Notice that we have said ‘the state’, not particular political parties or economic interests. The media, in dealing with contentious public or political issues, would be rightly held to be partisan if they systematically adopted the point of view of a particular political party or of a particular section of capitalist interests. It is only in so far as (a) these parties or interests have acquired legitimate ascendancy in the state, and (b) that ascendancy has been legitimately secured through the formal exercise of the ‘will of the majority’ that their strategies can be represented as coincident with the ‘national interest’—and therefore form the legitimate basis or framework which the media can assume. The ‘impartiality’ of the media thus requires the mediation of the state—that set of processes through which particular interests become generalized, and, having secured the consent of ‘the nation’, carry the stamp of legitimacy. In this way a particular interest is represented as ‘the general interest’ and ‘the general interest as “ruling”’. This is an important point, since some critics have read the argument that the operations of the media depend on the mediation of the state in too literal a way—as if it were merely a matter of whether the institution is state-controlled or not. The argument is then said to ‘work better for the BBC than for ITV’. But it should be clear that the connections which make the operations of the media in political matters legitimate and ‘impartial’ are not institutional matters, but a wider question of the role of the State in the mediation of social conflicts. It is at this level that the media can be said (with plausibility—though the terms continue to be confusing) to be ‘ideological state apparatuses’. (Althusser, however, whose phrase this is, did not take the argument far enough, leaving himself open to the charge of illegitimately assimilating all ideological institutions into the state, and of giving this identification a functionalist gloss).

This connection is a systemic one: that is, it operates at the level where systems and structures coincide and overlap. It does not function, as we have tried to show, at the level of the conscious intentions and biases of the broadcasters. When in phrasing a question, in the era of monetarism, a broadasting interviewer simply’ takes it for granted that rising wage demands are the sole cause of inflation, he is both ‘freely formulating a question’ on behalf of the public and establishing a logic which is compatible with the dominant interests in society. And this would be the case regardless of whether or not the particular broadcaster was a lifelong supporter of some left-wing Trotskyist sect. This is a simple instance; but its point is to reinforce the argument that, in the critical paradigm, ideology is a function of the discourse and of the logic of social processes, rather than an intention of the agent. The broadcaster's consciousness of what he is doing—how he explains to himself his practice, how he accounts for the connection between his ‘free’ actions and the systematic inferential inclination of what he produces—is indeed, an interesting and important question. But it does not substantially affect the theoretical issue. The ideology has ‘worked’ in such a case because the discourse has spoken itself through him/her. Unwittingly, unconsciously, the broadcaster has served as a support for the reproduction of a dominant ideological discursive field.

The critical paradigm is by no means fully developed; nor is it in all respects theoretically secure. Extensive empirical work is required to demonstrate the adequacy of its explanatory terms, and to refine, elaborate and develop its infant insights. What cannot be doubted is the profound theoretical revolution which it has already accomplished. It has set the analysis of the media and media studies on the foundations of a quite new problematic. It has encouraged a fresh start in media studies when the traditional framework of analysis had manifestly broken down and when the hard-nosed empirical postivisim of the halcyon days of ‘media research’ had all but ground to a stuttering halt. This is its value and importance. And at the centre of this paradigm shift was the rediscovery of ideology and the social and political significance of language and the politics of sign and discourse: the re-discovery of ideology, it would be more appropriate to say—the return of the repressed.

REFERENCES

Adorno, T.W. et al. (1950) The Authoritarian Personality, New York, Harper & Bros.

Althusser, L. (1969) For Marx, London, Allen Lane.

Althusser, L. (1971) ‘Ideology and ideological state apparatuses’, in Lenin and Philosophy and Other Essays, London, New Left Books.

Bachrach, P. and Baratz, M. (1970) Power and Poverty, Theory and Practice, Oxford, Oxford University Press.

Barthes, R. (1972) Mythologies, London, Jonathan Cape.

Becker, H. (1967) ‘Whose side are we on?’, Social Problems, 14, Winter.

Bell, D. (1960) The End of Ideology, New York, Free Press.

Berger, P. and Luckmann, T. (1966) The Social Construction of Reality, Harmondsworth, Penguin.

Camargo, M. (1980) ‘Ideological dimension of media messages’, in Hall, S. et al. (eds) Culture, Media, Language, London, Hutchinson.

Coward, R. and Ellis, J. (1977) Language and Materialism, London, Routledge & Kegan Paul.

Durkheim, E. (1960) The Division of Labour in Society, Glencoe, Free Press.

Eco, U. (1973) ‘Social life as a sign system’, in Robey, D. (ed.) Structuralism: An Introduction, Oxford, Clarendon Press.

Garnham, N. (1979) ‘Contribution to a political economy of mass communication’, Media, Culture and Society, 1 (2), April.

Glasgow University Media Group (1976) Bad News, London, Routledge & Kegan Paul.

Glasgow University Media Group (1980), More Bad News, London, Routledge & Kegan Paul.

Golding, P. and Murdock, G. (1979) ‘Ideology and mass communication: the question of determination’, in Barrett, M., Corrigan, P., Kuhn A. and Wolff J. (eds) Ideology and Cultural Reproduction, London, Croom Helm.

Gramsci, A. (1971) Selections from the Prison Notebooks, London, Lawrence & Wishart.

Hall, S. (1975) ‘Encoding and decoding in the television discourse’, Education and Culture, 6, Council of Europe, Strasbourg.

Hall, S. (1979) ‘The great right moving show’, Marxism Today, January.

Hall, S. (1980a) ‘Encoding and decoding’ (revised extract), in Hall, S. et al. (eds) Culture, Media, Language, London, Hutchinson.

Hall, S. (1980b) ‘Popular democratic vs authoritarian populism’, in Hunt, A. (ed.) Marxism and Democracy, London, Lawrence & Wishart.

Hall, S., Connell, I. and Curti, L. (1977) ‘The “unity” of current affairs television’, Cultural Studies, 9.

Laclau, E. (1977) Politics and Ideology in Marxist Theory, London, New Left Books.

Lévi-Strauss, C. (1969) The Elementary Structures of Kinship, London, Eyre & Spottiswoode.

Lévi-Strauss, C. (1967) The Scope of Anthropology, London, Jonathan Cape.

Lipset, S. (1963) Political Man, London, Heinemann.

Lukes, S. (1975) Power: A Radical View, London, Macmillan.

Marx, K. and Engels, F. (1970), The German Ideology, London, Lawrence & Wishart.

Matza, D. (1969) Becoming Deviant, New Jersey, Prentice Hall.

Merton, R. (1957) Social Theory and Social Structure, New York, Free Press.

Parsons, T. (1951) The Social System, New York, Free Press.

Pêcheux, P. (1975) Les Verités de la Police, Paris, Maspero.

Saussure, F. de (1960) Course in General Linguistics, London, P. Owen.

Shils, E. (1961a) ‘Centre and periphery’, in The Logic of Personal Knowledge, London, Routledge & Kegan Paul.

Shils, E. (1961b) ‘Mass society and its culture’, in Jacobs, N. (ed.) Culture for the Millions, New York, Van Nostrand.

Veron, E. (1971) ‘Ideology and the social sciences’, Semiotica, III (2), Mouton.

Vološinov, V.N. (1973) Marxism and the Philosophy of Language, New York, Seminar Press.