Copyright Sociological Research Online, 1997


Fisher, K. (1997) 'Locating Frames in the Discursive Universe'
Sociological Research Online, vol. 2, no. 3, <>

To cite articles published in Sociological Research Online, please reference the above information and include paragraph numbers if necessary

Received: 16/1/97      Accepted: 9/7/97      Published: 30/9/97


Scholars from a range of disciplines use the term 'frame' to mean a variety of disjointed and incompatible concepts. This paper examines a range of framing literature, from the writings of authors including Erving Goffman, Tuen van Dijk, Serge Moscovici, George Lakoff, Alan Johnson, William Gamson, David Snow, Robert Benford and Paolo Donati. Then it develops the theoretical case for defining frames as semi-structured elements of discourse which people use to make sense of information they encounter. Additionally, this paper demonstrates the need to include social system frames, which provide patterns for understanding social relations, among the presently acknowledged frame types. Frames develop in parallel with language, vary across cultures, and shape, but are distinct from other extra-linguistic discourse forms, including myths and ideologies.

Culture; Discourse; Frame Analysis; Interpretation; Knowledge; Language; Worldview


A number of scholars have found that the concept 'frame' serves a useful purpose in building a wide range of arguments. References to acts and processes of framing abound in academic literature, particularly among the social sciences, policy studies, linguistics, and fine arts genres. Even basic undergraduate social science methods textbooks advise students that 'the act of framing a question about social relations also encompasses values, beliefs, and differing perspectives' (Judd et al, 1991).

Sociologists, and in particular, Erving Goffman, have investigated framing as a process through which societies reproduce meaning. Goffman contended that some 'primary frameworks' are socially constructed concepts which people may perceive as 'natural', while other primary frameworks directly reflect physical experiences. The study of framing in this rubric is the study of representation and meaning.

Some scholars interested in the study of social movements have experimented with other research strategies which they have labelled 'frame analysis' to explore the processes by which social movements come to understand problems and to sell their perspectives to a wider audience. David Snow, Robert Benford, and William Gamson suggest that frames overlap and organise the values and beliefs of movement activists, and that individual people control frame production. These scholars express sympathy with many social movements challenging established orthodoxies, and link (in my view, inappropriately) their theories of framing with practical advise for activists. Consequently, they (especially Gamson) suggest that, in counting the appearance of frames used by social movements in relation to frames used by other parties to a policy dispute in media texts, the researcher or activist gains one measure of the influence or success of a particular movement.

In contrast, Teun van Dijk and Paolo Donati locate frames at deeper cognitive levels over which individuals, social movements and institutions hold little, if any, direct control. Van Dijk, Donati, and Anna Triandafyllidou argue that people make use of two different types of frames, discursive structural frames, which people employ to organise topics of discussion, and a higher level of frames, which people use to make sense of the information they encounter in the world. Under this perspective, the study of frames would provide insight into how people understand issues arising during public debates, though such study would not directly facilitate the assessment of social movement strategies, public opinion, or organisational behaviour.

Taken as a whole, the many branches of 'frame analysis' literature do not exhibit a consensus over some basic questions, including what frames are or how individuals and cultures make use of frames. This lack of consensus largely arises from the very different methodological aims of the various authors who have contributed to the framing literature. Nevertheless, a study of framing informs the study of how societies process information to generate meaning, and thus can in and of itself be broadly useful to many social sciences. This paper seeks to co-ordinate some of the disparate approaches to framing by first exploring the competing definitions of frames and framing processes. I argue that van Dijk and Donati have developed a useful distinction between discursive structural frames and higher level frames, which I call cultural frames. As linguists have already extensively discussed the former type of frame, I intend to concentrate on the latter, less-developed concept of cultural frames. I build the case for defining cultural frames as loose, socially-generated structures in discourse which people use to organise information, and around which groups develop ideological and policy arguments. I conclude this paper by exploring the problem of identifying the boundaries and varieties of cultural frames.

Goffman and the Sociology of Framing

In 1974, Erving Goffman proposed a new dimension of his socio-semiological methodology for studying visual images and cultural representations which he termed 'frame analysis'. His text of the same name was not among his better received, and Goffman did not later invest as much effort into developing the concept of framing as he did to other aspects of studying meaning, representation, and images. As this particular text nonetheless served to ignite one branch of the frame analysis literature, it is useful to begin by examining the arguments within it.

In Frame Analysis, Goffman argued that cultures generate 'primary frameworks', which render 'what would otherwise be a meaningless aspect of the scene into something that is meaningful' by offering a point of comparison, or a conceptual structure, through which people can digest information (Goffman, 1974: p. 21). Goffman suggested that each culture produces two types of primary frameworks: natural frameworks and social frameworks. Natural frameworks develop from 'purely physical' experiences which people understand 'to be due totally, from start to finish, to "natural" determinants', as distinct from experiences with which people associate a wilful agent who both has the power and desire to influence at least some aspect of the experience (1974: p. 22). As a result, 'success or failure in regard to these events is not imaginable; no negative or positive sanctions are involved' (1974: p. 22). In contrast, social frameworks arise from the wilful exertions of 'an intelligence, a live agency, the chief one being the human being' (1974: p. 22).

Goffman maintained that primary frameworks of both varieties develop along a continuum of systematisation, with frameworks at one end defined by highly organised sets of rules, and frameworks at the other end lacking any 'apparent articulated shape, providing only a lore of understanding' (1974: p. 21). Even so, Goffman suggested that all primary frameworks, regardless of the level of systematisation, enable users 'to locate, perceive, identify, and label a seemingly infinite number of concrete occurrences defined in' the terms of the primary framework they choose to apply (Goffman, 1974: p. 21).

Goffman contrasted this process of framing against the process of keying (or staging). Frames organise information drawn from real experiences and about people and objects and which are actually in the world. Keys, or stagings, mimic primary frameworks, but do not fully duplicate them (Goffman, 1974: p. 47). Humans, he contended, key frames and rekey keyings for a variety of reasons. Keyed events (1) allow practice for a (potential) future real performance (such as children playing post office, adults training in martial arts, or animals play fighting); (2) reaffirm cultural knowledge about the world through rituals; (3) facilitate commentary and reflection on real events, people and objects (a form of keying present in art and fictional literature); (4) and package information into transferable commodities (such as media articles, government reports, or academic studies). People can also use keys to deceive (5). Goffman broke the latter form of keying into two types: (5a) 'fabrications,' whereby one person or group dupes another or others; and (5b) 'illusions', whereby persons or groups delude themselves. Goffman contended that the different varieties of keys share seven common features: (1) they are constructed by people; (2) when people participate in keyed events willingly, they suspend their ability to notice that the event has been keyed (such as ignoring the technical people during a theatre production); (3) they can be rekeyed at any time; (4) they reduce and abstract the information related to the real framework they mimic, and, hence, can never fully replace or stand for that framework (1974: p. 79); (5) each successive rekeying of a rekeying loses a little bit more in translation, and one could eventually reach a point at which further rekeying could no longer hold meaning (1974: pp. 162; 182); (6) keys may produce layers representations of the original framework, and these layers may be shallow and superficial or deep (1974: p. 83); and (7) the process of keying is vulnerable (1974: p. 159). To illustrate this latter point, Goffman offered examples of conmen who in turn get conned, and of vacuum salespeople who perform in-house demonstrations to sell their product at the houses of customers who accept demonstrations simply to get their carpet cleaned.

This version of frame analysis contains a number of methodological imperfections. Guy Swanson criticises Goffman for overclaiming the implications of 'underdeveloped and unsystematised examples', and for brushing aside problems with the reliability of his method by 'quietly' abandoning the project of defining frameworks after the second chapter (Swanson, 1975: p. 218). Moreover, Goffman suggested that cultures construct social frameworks, but shifts between holding cultural and individual frame users accountable for the entailments of the way they frame reality, and absolving frame users of such responsibility by asserting that we are 'likely to be unaware of such organised features as the framework has and unable to describe the framework with any completeness if asked' (Goffman, 1974: p. 21). Indeed, Goffman did not convincingly distinguish between natural and social frameworks. If the presence or absence of public perception that a wilful agent caused or contributed to the production of an object or event marks this distinction, then how would one classify framings which occur when people wrongfully attribute causation to a 'wilful' actor, or when people fail to notice the intervention of a party and attribute an event to 'natural' causes?

Goffman's concept of keying magnifies the clarification problems inherent in his concept of framing. He notes that some keys very closely approximate the real thing, but offers no theoretical consideration of where one could draw the line between the two. Goffman also ignores the possibility that one group of people may look upon a particular story about the world as a social framework, while another might look upon the same story as a fabrication. Equally, Goffman's distinction between keys and frames lacks sufficient development to empower researchers to consistently distinguish illusions from natural frames, or, as Swanson observes, between keys, 'the transformations they effect and the frameworks they transform' (Swanson, 1975: p. 219). While in the 1990s, we might easily say that the old theory that the sun revolved around the earth was an illusion while the present theory of gravity-induced rotation in solar systems is a natural frame, people in previous centuries would have viewed this classification as problematic. Likewise, we could not easily apply Goffman's methodology to separate arguments in present political contests, such as the debate over whether animals deserve legal rights, into illusions, fabrications, social frameworks, and natural frameworks, as we have no objective basis from which to make these classifications.

Applications of Goffman's theory of frames to the study of gender give rise to further theoretical complications. In his 1979 book Gender Advertisements, Goffman contends that a system of gender stereotypes pervades Western cultures. Advertisers reinforce these stereotypes by incorporating them into the images which bombard people in industrialised societies during most activities of their daily lives. Goffman suggests that these stereotypes in turn frame the way people envision masculinity and femininity, and thus, at least to some extent, frame each person's sense of identity.[1] Kristin Minister, who bases her methodological discussion of Goffman's writing, extends this argument to suggest that gender itself frames our understanding of reality by defining the latitude of actions and speech options we may exercise in any given situation. She argues that people may transcend gender frames in some circumstances, but that 'even the strongest individuals ... are relatively helpless in certain kinds of communication situations that are validated by tradition' (Minister, 1991: p. 27). Minister asserts the need for the inclusion of feminist frames in oral history research. I speculate that she means that historians should interview both women and men, and take into account the way gender colours interviewees' experiences. Nevertheless, apart from including the term 'feminist frame' in her title, Minister never sets out the features that she thinks distinguishes a feminist frame from a woman's frame, and, indeed, does not explicitly define the term 'frame' in her text.

This is not to say that Goffman and those who followed him did not make a useful contribution to our understanding of the framing of gender. In Gender Advertisements, Goffman challenged claims that femininity and masculinity have a solid link to nature and genetic predisposition. Instead of inheriting 'natural' gender inclinations at birth, Goffman proposed that humans inherited a capacity to learn to fit into an existing social structure (whether that be a gendered structure or a structure based on other social distinctions) by accepting traits assigned to each group within that structure as 'natural' (Goffman, 1979: p. 8). With this observation, writes Candace West, Goffman laid an important foundation for later feminist explorations of representations of gender (West, 1996: pp. 362 - 3).

While this contribution to the study of gender is important, Goffman's initial conception of frame analysis raises an irresolvable problem for the study of frames. In the absence of a clear meaning of the word frame, these writings cannot specifically explain how the concept of framing relates to the concept of gender. Even if we were to accept that such a relationship exists, these writings cannot resolve the question of whether biological features of human females and males cause them to frame the world differently, whether the social construction of separate genders in any given culture spawns gendered frames, or whether biological and cultural factors both contribute to a gendered dimension of framing. Scholars like Goffman and Berger give little consideration to how frames develop in the first place. Minister mentions the question of origin, but only with the passing comment that frames, as 'communication forms themselves conserve the values of their originators' (Minister, 1991: p. 28), and thus impede efforts to reconstruct gender roles. This comment, however, gives no insight into the processes by which originators generate frames, or, indeed, to the questions of who can produce frames, what conditions permit frame production, and what constraints influence that production.

Frames and the Study of Social Movements

A number of scholars interested in assessing the successes and failures of social movements have used Goffman's work as a basis for building an understanding of frames which is closely linked with ideology. This branch of the frame literature has tended to have a practical focus, but has also tended to blur the boundary between 'frames' and other elements of discourse.

Snow and Benford on Social Movements and Framing

David Snow and Robert Benford build on Goffman's work in their bid to account for the factors which contribute to the success or failure of social movements. Snow and Benford criticise social movement theories for taking a 'static view of participation' and for focusing on factors which give rise to grievances at the expense of studying the processes by which people define and understand grievances (Snow et al, 1986: p. 465). 'What is at issue', they argue, 'is not merely the presence or absence of grievances, but the manner in which grievances are interpreted and the generation and diffusion of those interpretations' (Snow, et al. 1986: p. 466). Snow and Benford contend that social movements not only 'frame the world in which they are acting', but that they also frame social problems (Snow et al, 1986). The way a movement frames a problem plays a significant part in the propensity of that movement to shape public policy (Snow et al, 1986: p.466; Snow and Benford, 1992).

These authors define framing as the process by which 'ordinary people make sense of public issues' (Benford, 1994: p. 1103). People work with two types of frames: (1) 'domain-specific interpretative frames', which organise sets of behaviours and individual lifestyles; and (2) 'global interpretative frames' or 'master frames' which signify meaning on a broader scope and which organise sets of domain-specific frames (Snow et al, 1986: pp. 474 - 5). Snow and Benford suggest that processes by which people make sense of issues revolve in cycles. Movements experience different opportunities to manipulate frames, and face different sets of constraints on their frame usage during different phases of protest cycles (Snow and Benford, 1992).

Frames, they suggest, function as the 'conceptual scaffolding' (Snow and Benford, 1988: p. 213) which social movements erect to construct new ideologies or to modify existing ones. 'Internal frame maintenance' can thus be equated with 'ideological work' (Snow et al, 1986: p. 478). Once erected, frames perform 'signifying work' for social movements. Activists 'assign meaning' to 'events and conditions', then use frames to transmit those meanings to members of the public. Successful framing enables social movements to 'mobilise potential adherents and constituents, to garner bystander support, and to demobilise antagonists' (Snow and Benford, 1988: p. 198). Snow and Benford do not locate frames within the discourse patterns which people learn as members of a culture; instead, they argue that individuals exercise considerable control over the framing process, and that movements can learn to successfully frame by studying the processes which constrain potential frame deployment in any given social and political context. Thus, 'hypothetically, the absence of innovative master frames may account in part for the failure of mass mobilisation when the structural conditions seem otherwise ripe' (Snow et al, 1986: p. 477, italics reflect my emphasis).

Much of Snow and Benford's writings accordingly read like manuals of the do's and don't's of successful movement organisation.[2] These authors outline three 'core framing tasks' for activists: (1) 'diagnostic framing', or clearly defining a problem and assigning blame for the problem to an agent or agencies; (2) 'prognostic framing', that is, offering solutions, and proposing specific strategies, tactics, and objectives by which these solutions may be achieved; and (3) 'motivational framing', or rallying the troops behind the cause (Snow and Benford, 1988: pp. 199 - 202). To this list of 'core tasks' they add a fourth key element, 'frame resonance'. Successful movements must appeal to the existing values and beliefs of their target population. If movements offer frames which do not resonate, or strike a chord, in the culture in which the would-be movement hopes to operate, then activists must attempt a difficult and long-term challenge of re-educating the target population to adopt new values and beliefs which will resonate with the movement's frames (Snow and Benford, 1988: pp. 198 - 9).

Snow and Benford leave key aspects of their theoretical argument ambiguous. While they assert the need for social movements to develop frames which have 'empirical credibility', that is, can be tested and verified by present and potential activists (Snow and Benford, 1988: p. 208), they ignore the question of how social scientists might test the reliability and validity of frames. They offer numerous examples of framings adopted by particular movements which, on face value, read credibly, but Snow and Benford do not address the question of how they identified the frames they discuss. Instead, these authors describe frames as 'temporally variable and subject to reassessment and renegotiation' (Snow et al, 1986: p. 476). If, during public debate, 'a value becomes discredited or loses its saliency, or a belief is popularly refuted, it may drag associated frames down along with it' (Snow et al, 1986: p. 477). Snow et al, suggest that a 'sweeping variety' of frame alignment may be periodically necessary to facilitate the cultivation of 'new values' and the expulsion of 'erroneous beliefs' thereby ameliorating 'the prospect of "misframings" or interpretative "errors" and "frame disputes"' (Goffman, 1974: pp. 301 - 38, cited in Snow et al, 1986: p. 475). Like Goffman, Snow and Benford raise the prospect that some frames are 'truthful', while others are 'misframings', but also like Goffman, they neglect the question of how researchers delineate between the two (Snow et al, 1986: p. 475). More fundamentally, Snow and Benford do not explain how their conception of frames differ from other ideological structures. While, on the one hand, they treat frames like ideological scaffolding, they also argue that frames can reprioritise the values and beliefs within ideologies. If in fact frames perform such a function, then they might more appropriately be described as ideological skeletons rather than scaffolding. Donati, however, notes that practical applications of frame analysis which blur the boundaries between ideologies and frames have generated methodological confusion (Donati, 1994: pp. 15 - 16). In a footnote, Donati suggests that this confusion 'is why Snow's procedure is unclear from his articles' (1994: p. 16).

Gamson, Frames, and Social Movements

William Gamson sets out to explain both how people think about issues and how social movements might appeal to non- activated people. Gamson describes the world of events and images as a forest which people must navigate, using both images they amass from public discourse and cultural tools which they carry with them (Gamson, 1995: p. 86). Gamson declines assessment of the dissemination of facts and information, arguing that social scientists cannot precisely define either of these terms. Instead, he argues, people think and communicate through images, which act both as reproductions and as 'a mental picture of something not real or present' (Gamson et al, 1992: p. 374). Cultures do not assign fixed meanings to images; instead, people negotiate the image meanings (Gamson, 1995: p. 87). People may draw widely divergent conclusions about an image (thus, white Americans could see OJ Simpson as a rich man able to buy his way out of punishment, while black Americans viewed OJ Simpson as a victim of racism in the USA judicial system), but even if people do not share the same meanings, they can make use of public discourse 'with some assurance that potential challengers will understand the references and allusions' (Gamson, 1995: p. 86).

Images lack fixed meaning, he argues, because people possess the power to frame, and later reframe, them. Frames, for Gamson, are central ideas which define 'what is at issue' by addressing the 'pattern- organising aspect of meaning' (Gamson and Wolfsfeld, 1993: p. 118). Gamson identifies two levels of framing activity: the generation of 'cultural themes' (or 'frames') and 'issue cultures' (which Gamson later terms 'packages'). Cultural themes, which Gamson defines as 'akin to such terms as ideology, values, belief systems, and weltanschauung', 'transcend specific issues and suggest larger world views' (Gamson, 1988: p. 220). Frames develop dialectically, with established, conventionalised frames attracting adversarial counter-frames (1988: p. 221). Packages contain the internal structure of a frame, but are applied to a particular issue or event (1988: pp. 221 - 2).

Gamson argues that social movements employ three kinds of frames: 'aggregate frames'; 'consensus frames'; and 'collective action frames' (Gamson, 1992; 1995). Aggregate frames define issues as social problems, and assign responsibility to people who hear the message of the frame to take action about the problem as individuals. Consensus frames (or 'consensus strategies') define a social problem as one which can only be resolved by collective action. Consensus frames construct a strong sense of identity for the people who will act collectively, but leave the identity of the party or parties responsible for causing the problem indistinct. Collective action frames, in contrast, define a problem as both intrinsically wrong and caused by an identifiable actor (an 'injustice frame'), and also establish an adversarial relationship between 'us', members of a social movement who can resolve the problem through collective active (an 'agency frame' defines 'us'), and 'them', the agents responsible for the problem (an 'identity frame' demarcates 'us' from 'them'). Collective action frames can only form if people perceive an issue through all three component frames (injustice, agency, and identity).[3]

Of the three component frames, the injustice frame is most vulnerable to reframing by the people or institutions targeted by social movements employing collective action frames. Two general strategies dilute injustice frames: opponents of a social movement can reify the cause of the problem, shifting the focus of anger to an agentless entity, like 'the system', 'society', 'life', or 'human nature'; or they can manipulate the injustice frame by redirecting people's anger towards another target agent (Gamson, 1995: p. 91).

Gamson differs from other social movement scholars by emphasising the significance of studying media discourse, which he defines 'as a tool for analysing meaning' (Gamson, 1995). Gamson argues that people understand issues by employing a combination of three lenses to filter out relevant images. In his 1988 work, Gamson identifies the lenses of his 'value-added model' as the original discourse produced by interested parties about a topic, media translations of debates between proponents of different original discourses, and 'cultural tools' which people use to make sense of both original discourse and media translations of that discourse (Gamson, 1988). By 1992, Gamson altered his value-added model to involve the lenses of people's own experiences, common knowledge in a given culture, and 'media generated images' (Gamson et al, 1992: p. 374), though he also concedes that 'no resource is purely personal or cultural' (1995: 87). Gamson argues that people cultivate a more substantial level of understanding, and generate frames with a 'special robustness' when they combine the three resources of the value-added model to make sense of images (Gamson, 1995: pp. 88 - 9). Likewise, when people depend primarily on one of these three sources, they experience difficulty applying their understanding through another lens (1995: p. 104). Gamson sees the media as not only a central tool in the production of cultural meaning, but also as one of the focal points in which social movements wage their contests to change public policy or perceptions. 'Movement activists are media junkies', he writes, and 'media discourse provides them with "weekly, daily, sometimes hourly triumphs and defeats"' (1995: p. 85).

Gamson notes that the media periodically advance the cause of social movements. Journalism depends on story- telling and narratives involving 'motivated actors', rather than reports of structural causes of problems. As a consequence, the media adopt textual formats conducive to injustice frames (Gamson, 1995: pp. 92 - 3). By simply taking a social movement seriously, and treating it as a relevant actor in a dispute, the media can raise the profile of activists and expand their influence, even if journalists are not particularly sympathetic (1995: p. 99; Ruzza and Schmidtke, 1993 reach a similar conclusion). For this reason, Gamson and Wolfsfeld (1993) outline strategies which social movements can adopt to transmit their messages through the media.

Nevertheless, Gamson, warns, 'every silver lining has a cloud' - media coverage can exert a double-edged impact on social movements (Gamson, 1992: p. 71; 1995: p. 92). To begin with, many individual mediums in the West seldom take a definitive editorial position with regard to issues of concern to activists, and regularly generate inconsistent meanings in their coverage of any particular issue. As a result, Gamson maintains, 'it would be difficult to find any coherent frame at all, let alone an injustice frame' in media portrayals of many issues (1995: p. 94). The more dependent journalists are on official sources, suggest Gamson and Stuart, the more likely they are to give credence to official perspectives at the expense of the views of social movements (Gamson & Stuart, 1992). Moreover, while many mediums may favour the story-line of injustice frames, the US media 'systematically discourages the idea that ordinary citizens can alter the conditions and terms of their daily lives through their own actions' on many (though not all) issues (1995: 97).[4] At the bottom line, Gamson argues, the Western media 'often obstructs and only rarely and unevenly contributes to the development of collective action frames. The good news for movement activists is that media discourse is only one resource' (1995: 104).

Gamson's writings do make a significant contribution to the study of public opinion and social representations of issues. Philo Wasburn praises Gamson for studying public opinion by analysing transcripts from discussions held by small groups of ordinary people. Gamson's data, Wasburn contends, reveals that people are not 'the dolts social scientists long have arrogantly assumed them to be' (Wasburn, 1994: p. 923).

Gamson also offers some useful advice for the study of framing competitions in the media. Gamson notes that the journalists tend to strive for a good story, rather than to represent all sides on a conflict. The frames proposed by some parties in a framing contest may not appear regularly or at all in media texts. Gamson lists an array of sources he and his researchers consulted to develop a more comprehensive list of frames.

Nevertheless, other aspects of Gamson's discussion of framing are less persuasive. At an abstract level, Gamson's arguments regarding collective action frames, and their three component frames, look convincing, though Gamson draws the same theoretical problems experienced by Snow and Benford by linking his frames with ideologies. As Gamson applies his concept of framing to particular issues, his definition of frames becomes less clear. In an appendix to Talking Politics (1992) entitled Working frames and codes, Gamson sets out the principal 'frames' he and his researchers have identified with regard to four policy issues. In the case of troubled industry, for example, Gamson lists four frames: partnership; capital flight; foreign invasion; and free enterprise, which has two variations, an anti-union and an anti-government variation. These frame categories pose two problems. First, Gamson leaves unclear if these frames are 'cultural themes', 'packages', or some other form of frame. Donati criticises Gamson's concept of the package, arguing that 'it may lead to vagueness in distinguishing different frames, especially because ... the relationships between the different objects of a "package" are not logical - and therefore unique and fixed - kind, but rather depend on the specific construction of each discourse' (Donati, 1992: p. 144). Second, in the case of the free enterprise frame, the variations are not mutually exclusive. A person could simultaneously argue that business operates best free from interference from both big government and unions.

Gamson's methodology raises other complications in his coding of the frames. In the previously mentioned appendix, he explains his coding sheets, which contain series of three digit numbers, the first of which represents the frame or variation, the second which indicates if the variation is strong or weak, and the third which 'refers to the specific idea element within the frame' (Gamson, 1992: p. 218). Some examples of his coding lines include:

115 Saving jobs is more important than any abstract principles involved.
116 Opponents of government aid are kicking a victim when he's down.
117 Opponents of government aid are antibusiness obstructionists (1992: p. 220).

A strategy of breaking down frames into 'specific idea elements', however, can easily become tedious and unuseful. To begin with, the people who produce the texts the frame analyst proposes to study do not necessarily explicitly spell out or distinguish their 'idea elements'. While the difference between elements 116 and 117 may be important in one text, that distinction may not be so clear in the next. Moreover, even when people share the same sentiment about an issue, they do not always use the same words to discuss it. In attempting to develop a coding list as Gamson proposes, the researcher must either rely on the wording of a small number of people who are not necessarily representative of members of a particular group, or develop abstract terms to encompass a wider range of wordings. In choosing either option, the researcher must code idea elements against a list which is at least partly artificial. I have my doubts about whether any random set of people who believe that their government should intervene to prevent the failure of large businesses would share a common conception of 'anitbusiness obstructionists', or agree on who precisely is 'victimised' when large companies fail.

Gamson's approach lays the groundwork for the micro-analysis strategy proposed by Hank Johnston. If precise measures of idea elements provided a useful way to analyse framing, then one could get reliable results by scanning the texts into a computerised text analysis package and producing lists of regularly used words and phrases; yet a general reliance on micro- analysis may not prove inappropriate for cultural frame analysis, as I will now explain in reviewing Johnston's work.

Johnston on Frame Methodology

Hank Johnston shares Snow, Benford, and Gamson's concern that the study of social movements include an emphasis on how people conceptualise grievances. Also like these authors, Johnston contends that frames shape ideological arguments and that individual actors in social movements hold the power to transform framings of problems and solutions. Johnston distinguishes himself from other social movement scholars who have applied frame analysis by emphasising questions of reliability. He suggests that framing occurs inside the 'black box' of the mind, and that a myriad of factors present in any particular context influence the way people frame an issue or event (Johnston, 1995: pp. 218 - 9). In particular, Johnston argues, the role an individual producer of text performs in a social movement organisation, as well as the social roles that individual plays in broader society, influence the framing processes that person undertakes. Ultimately, however, he argues that no researcher can peer inside the black box of the brain and get a clear picture of what a person is thinking. 'Traditional frame analysis', as he sees it, relies on inferential assumptions about mental activities, and, consequently, yields 'too much loose interpretation taking place too far from the data' (1995: p. 241).

Johnston thus rejects 'macro- discourse analysis', and advocates a text-dependent micro-discourse approach emphasising analysis of words and phrases which mark frames, as well as visual interpretative clues people generate while producing oral text. 'The kind of micro-frame analysis that I am suggesting', he writes, 'involves approximating the hierarchical relationships that constitute frames and scripts appropriate to typical or crucial situations in a movement career' (1995: p. 237). Johnston contends that his approach achieves three methodological ends: it provides a 'systematic approach to the content of social movement frames' which enables other researchers to duplicate findings; it minimises the risk of 'outright misinterpretation'; and, 'regarding validity, it confronts head-on the fundamental problem in analysing textual materials; namely, their infusion with cultural, organisational, and interactional considerations that always - in varying degrees - bend and shape what gets said' (1995: pp. 245 - 6). Acknowledging that his approach requires more cumbersome examination of detail than other forms of frame analysis, Johnston contends that the key to micro-frame examination lies in collecting a small set of texts from 'critical junctures' in the life of a social movement for study (1995: p. 229).

Johnston's arguments raise some problems. First, he does not clearly address the question of where frames develop in discourse. Some authors, notably van Dijk, Donati, and Triandafyllidou, develop detailed arguments for defining separate levels of framing at discursive and deep structural levels (I address these arguments in Section 2). Johnston, in contrast, lumps all framing into a single set of social movement frames (1995: p. 241), and suggests that these frames, or 'higher- level concepts are the basis of sharing and co-ordination for many social movement actions, and it is through their systematic reconstruction and verification that the real contributions of micro-frame analysis lies' (1995: p. 237). Scholars like Gamson, van Dijk, and Donati, however, do not have the same concept in mind when they discuss framing, and Johnston cannot achieve the 'systematic reconstruction and verification' of frames or the high degree of validity he asserts by simply willing away the differences.

Second, contrary to Johnston's assertion, micro-analysis does not avoid inferential assumptions. As van Dijk illustrates, the discursive level of discourse represents more than the sum of the individual parts of sentences and fragments in text. At some point, the researcher must draw inferential conjectures about 'black box' operations of the human mind to identify the markers of the frame in the text. (Indeed, in the applications of frame analysis which I have read, researchers have tended to study texts produced in the culture(s) in which they were raised, and, in the few exceptional cases, researchers have studied texts produced in languages which they speak. As active users of the discourse they study, they may thus be able to make more accurate inferences about frame employment than researchers who do not actively engage with the discourse they study.) Johnston himself makes a large inferential leap in assuming that text markers of the topic of discourse also mark higher level frames as well. (I will shortly address the work of linguists who reject this assumption).

Third, Johnston makes ambiguous arguments about both the degree of control which individuals and social movements exercise over discourse structures, and the degree to which environmental factors influence discourse. On the one hand, Johnston contends that two factors produce frames, 'cultural currents of an international scope', and the strategic planning in which activists engage as they respond to situational events and conditions in their political environment. Thus, he argues, 'the notion of frame is similar to a Weberian ideal type' (Johnston, 1995: p. 218). Social movements, and other participants in framing debates, such as the media, influence each other's framing of grievances, and 'one would expect to find parallels between the structure of media texts at these upper levels and the structures of movement documents and the speech of participants' (1995: p. 241). 'The final test of whether a social movement frame has been correctly described', he maintains, 'is if these reconstructions help the analyst understand why individual participants and social movement organisations act the way they do' (1995: p. 235).

On the other hand, Johnston contends that 'social movement frames are cognitive structures' (Johnston, 1995: p. 220), and that 'what we know about the organisation of interpretative frames also comes from cognitive science' ((1995: p. 235). Johnston does not specifically reference this claim, and many influential cognitive researchers do not share his conclusions. Scholars like Moscovici do not ascribe the power to change social representations to individuals. In reviewing theories of language acquisition and development, Trevor Harley argues that 'it is a simplification to pick out biological, cognitive, and social factors in isolation and say that this is the driving force of language acquisition. Instead, language acquisition occurs because of a complex interaction between these processes' ((1995: p. 349)[5], and, by extension, it would be unwise to assert that any of these sets of factors in isolation regulates frame usage without systematically analysed data, which Johnston does not offer.

This is not to say that neither individuals nor environmental factors influence framing. As I will elaborate in the next section, I think that exceptional individuals can wilfully affect framing, however, instances of individual-initiated changes to cultural frame structures rarely occur. Likewise, as George Lakoff (1987) notes, some events can transform our conceptual categories, and thereby impact other levels of discourse. The development of nuclear weapons, for example, resulted in the possibility that a tactical manoeuvre could literally annihilate a region, and raised the spectre that sufficiently motivated people could exterminate the human species. As a result, argues Lakoff, war metaphors in the nuclear age, unlike earlier war metaphors from previous eras, reflect the possibility of absolute destruction. Even Lakoff, however, maintains that cultural frames do not change in response to most fluctuations on the political or social landscapes.

Separating Frames and Ideology

A more disparate collection of scholars locate frames in the discursive system which people use in daily communication, and argue that the study of frames gives insight into how people understand and negotiate their world. This group of frame scholars, which includes social psychologists, linguists, and sociologists, defines the framing process as distinct from the formation and manipulation of ideologies. As a consequence, these authors have been more concerned with how people think about a range of issues at any given time, rather than how activists might respond to a particular social problem.

Moscovici and the Psychology of Social Representation

Social psychologists have striven to explain both how people and cultures make sense of the world by categorising beings, objects, and experiences, and how people develop systems of representing their shared understanding of the world. Serge Moscovici calls this pursuit the study of 'social representations'. Social representations 'conventionalise' people's knowledge through two 'mechanisms': (1) 'objectification' - which identifies relationships between images and ideas people store in their minds and objects they encounter in the physical world; and (2) 'anchoring' - which groups objectified relationships into broadly encompassing categories, and tethers, or 'anchors', these categories together to form patterns of general knowledge (Moscovici, 1984: p. 29). Consequently, representations provide a structure, 'environment', or frame, into which people can fit new information, and associate their information with a known category, and thereby 'make something unfamiliar, or unfamiliarity itself, familiar' (Moscovici, 1984: p. 24).

Moscovici emphasises that social representations arise from cognitive processes and manifest differently in each culture, but leaves the role of individuals and social groups in this production ambiguous. On the one hand, he contends, 'the real speakers, the real creators of language patterns, are groups: classes, nations, professions, and so forth' (Moscovici, 1971: p. xiii), and thus, at least initially, 'individuals and groups create representations in the course of communication and co- operation' (Moscovici, 1984: p. 13). Moscovici expands his discussion of individual and collective agency in the construction of representations by contrasting 'scientific' assumptions that reality and objects shape language and cognition with 'primitive' assumptions that language and cognition shape reality. Moscovici declines to adopt the 'scientific' assumption, and instead concludes that both sets of assumptions contain some validity, with the former reflecting discomfort with the inability of science to fully account for the non-logical operations of the thinking process, and the latter reflecting discomfort with the inability of primitive societies to fully account for events in the natural world. In this respect, suggests Moscovici, scientific and primitive understandings of the source of representations revolve around the same axis, or represent two sides of the same coin.

Nevertheless, Moscovici also rejects the proposition that people direct or control the operation of systems of social representation. The process of conventionalising knowledge, he suggests, does not provide individuals with the opportunity to think 'freely' about new information. Rather, social representations confine individuals to weigh up information in relation to a set of finite model categories in the general knowledge of his or her culture. Individuals may select the model from this set which they feel most closely resembles the previously uncategorised information, but the process of conventionalisation then ensures that 'all new elements adhere to this model and merge into it' or are dismissed from the individual's mind (7). Moreover, Moscovici maintains, 'representations are prescriptive, that is they impose themselves upon us with an irresistible force. This force is a combination of a structure which is present before we have even begun to think, and of a tradition which decrees what we should think' (Moscovici, 1984: p. 9). In this respect, 'how we think is not distinct from what we think' (1984: 67). Indeed, Moscovici contends that after cultures conventionalise information, people not only forget that they played any role in the conventionalisation at the outset, but also engage in behaviour conditioned by the system of representations to which they contributed. He writes:

Far from reflecting either behaviour or the social structure, a representation often conditions and even responds to them. This is so, not because it has a collective origin, or because it refers to a collective object, but because, as such, being shared by all and strengthened by tradition, it constitutes a social reality sui generis. The more its origin is forgotten, and its conventional nature ignored, the more fossilised it becomes' (Moscovici, 1984: 13).

As a result, social representations 'lead a life of their own, circulate, merge, attract and repel each other, and give birth to new representations, while old ones die out' (Moscovici, 1984: p. 13).

With this argument, Moscovici raises a dilemma for his discussion of the origin and development of social representations. Representations cannot exist, let alone possess a 'life of their own', without thinking beings who continually engage in the cognitive processes that give rise to and sustain them. In describing social representations with an animal life metaphor, Moscovici also creates a need to explain why certain representations 'merge' or 'attract' to reproduce, to account for the gestation of new representations independent of human agency, and to explain why and under what circumstance some representations die out. Indeed, by describing entrenched representations as 'fossilised', Moscovici also opens the possibility of regeneration after death - and perhaps the nightmare of less than scrupulous scholars pouring through old texts to unearth fossilised representations, locating their cognitive DNA, and developing a cognitive Jurassic Park. If representations truly 'impose themselves upon us with an irresistible force', then Moscovici cannot also credibly claim that 'individuals and groups create representations in the course of communication and co-operation', unless he specifies that, at one time, people created a set of representations which have reproduced themselves since that point, though Moscovici gives no indication of when that point may have been.

If people and groups can still wilfully add to the cocktail of representations, then people retain the power to transform the cocktail, even if they remain in part constrained by the base mixture which they seek to alter. Indeed, Moscovici does suggest that the set of representations in contemporary cultures may respond more readily to intervention by groups which gain a broad base for cultural support for a new representation than did cultures of previous centuries. Advances in scientific research, the rapid development of new technologies, the increased mobility of people, and the accelerated exchange of information between cultures made possible by the mass media and information technologies, generate 'a continual need to re-constitute "common sense"' (Moscovici, 1984: pp. 18 - 19). As a result, cultures have less time to conventionalise new representations before still newer representations challenge the domain of 'common sense'. One could extrapolate from this argument that social representation might now operate less prescriptively than in previous centuries. The relationships between language users and representations are thus not so cut-and-dried as Moscovici implicates through the majority of his argument.

Language and Framing - The Writings of Van Dijk

Linguist Teun van Dijk identifies procedures in his study of macrostructures (which he later renames discursive structures) as frame analysis. Van Dijk argues that language, discourse, and social behaviour arise from the cognitive processes by which people perceive, interpret, organise, and represent their knowledge of the world, that is, the way people construct a 'social "reality"' (van Dijk, 1977: p. 99; 1980: p. 2). Van Dijk distinguishes between the study of language (a domain including lexicons, semantics, grammars, syntax, and macro/discursive structures), which he defines as the study of the parts people use to construct a text, and the study of discourse (a domain of 'semantic superstructures' or 'schematic forms', such as narratives, myths, arguments, or scientific reports), which he sees as the study of types of texts and the way people use these different types to perform specific communicative purposes (1977: p. 153). Van Dijk postulates that framing processes may occur at both the macrostructural and superstructural levels, though he expresses doubts about the potential operation of the latter. He suggests that if superstructural framing occurs, then that process operates independently of discursive structural framing.

Van Dijk argues that language represents meaning on two levels, a micro level, and a macro or global level. Grammatical and syntactic structures, as well as words themselves, connote meaning at the micro-level; however, the sum of micro-meanings does not always total the whole meaning of a text (1977: p. 143). Consider the following two passages.

Mr. Grace hesitated. 'I feel a universal religion would have to be so very cut and dried that one would lose all one's freedom of belief. Religion is a kind of scientific research and every faith and creed is a part of the truth. "No creed does more than shadow imperfectly forth some one side of the truth". I can't remember who said that, but it was well said'. 'Father Hall in John Inglesant,' said Sal in a low voice. 'I can always depend on Sal,' said Mr. Grace, smiling at his second daughter. 'Because I never went to school,' explained Sal. 'Oh, what rot!' exclaimed Addie. 'You're always down on schools because you didn't go to one yourself. School taught Tilly and I all sorts of important things we couldn't have learnt at home'. 'What a pity it didn't teach you grammar,' said Sal, smiling (Stevenson, 1946).

'C++ will enable us to develop a windows application for MDSx' he said. 'Really,' she exclaimed, 'is that because the sun is shining brightly today?' 'Yes,' he replied, 'and Sarah has just found a new job.'

Both passages contain sentences which independently make sense. Passage B contains no grammatical errors, while passage A contains one. Both passages involve three topic shifts. An English dictionary and a list of English grammatical rules and exceptions, however, would not provide sufficient input for a computer program to identify why Passage A is coherent while Passage B is not. Language users are able to construct texts like Passage A while avoiding texts like Passage B, van Dijk argues, because languages also contain discursive structures, formal linguistic elements which reduce the potential meanings of words and grammatical structures to allow the formation of coherent of text. People learn to manipulate discursive structures through the same cognitive processes by which they learn to use other features of language (van Dijk, 1985).

Van Dijk contends that discursive structures perform four tasks: 1) enabling the receiver of a language message to select some elements for interpretation and to delete others as not relevant to the global meaning of that message;[6] 2) enabling receivers to organise elements of messages hierarchically, and to mark those elements requiring 'strong' or 'weak' deletion or selection; 3) allowing receivers to generalise the meaning of the message; and 4) equipping receivers to derive a global fact from the message (van Dijk, 1980: pp. 46 - 7). Discursive structures can achieve the fourth task because they contain frames, or structures which define the basic elements of a category of objects, concepts, or action-event sequences. Thus, van Dijk suggests, 'the sequence bought wood, stones and concrete; I laid foundations; I erected walls, I made a roof... may for example be subsumed under a proposition like "I built (a house)"' - or a house frame (van Dijk, 1977: p. 146). Passage A makes sense because it conforms to the expectations we have about the different roles participants can play in a conversation (in this case, making an argument, aiding another speaker to recall a fact, bantering between speakers). A conversation frame organises the 'global meaning' of this passage. Passage B contains no frame, and consequently lacks coherence.

Discursive structural frame analysis includes two main steps. First, one must identify the frame. One does this by identifying the main elements of the text and by considering the vantage, or 'point of view', from which the speaker examines those elements (van Dijk, 1980: p. 3). The frame itself will be the smallest common denominator that subsumes all of the main elements. Second, one maps the elements of the text as they are organised by the frame in accordance with the four tasks of discursive structures outlined in the previous paragraph. I will illustrate such a procedure with the following letter to the editor of Newsweek.

Is President Clinton planning to lift the ban on gays in the military because of a genuine commitment to a campaign promise, or is he just creating a smoke screen to hide his failure to introduce an economic plan? Where are your priorities, Mr. President? As your campaign slogan put it, 'It's the economy, stupid!' (Sakach, 1993).

At a lexical and grammatical letter, one could say that this letter is about President Clinton's campaign promises. Clinton is the subject of the first sentence, and is directly addressed by the author in the second and third sentences. The campaign promises are the predicate nominatives of the first and longest sentence, and feature again in the final sentence. Nevertheless, the author does not tailor his criticism to suggest that he has identified a unique problem of Bill Clinton. Likewise, the specific campaign promise mentioned in the first line, the promise 'to lift the ban on gays in the military', does not shape the meaning of the text. Had George Bush won the 1992 election, one could have easily substituted the name 'President Bush' and the promise 'to eliminate federal funding of abortion' in place of the name and promise the letter now contains and still derive the same meaning from this text. The smallest common denominator which subsumes all of the main elements of this letter is a democratic politician frame (politicians principally should be interested in the macro issues of government, like the economy; politicians principally are interested in their public image; politicians get side-tracked by issues which raise public interest but are not really important; politicians compete for office; politicians make lots of promises, some of which they do not intend to keep, to win votes; politicians don't always keep the promises they may originally intend to keep; politicians try to cover up their failings to look good to the voting public; politicians can't be fully trusted). The reader thus can infer from this letter that Bill Clinton is a typical politician, and that it is no surprise that the economy has lost out to another issue.

Having identified the frame, a researcher performing a discursive structural frame analysis on this text could then begin the four steps of mapping the text segments in terms of that frame.

  1. Select-Delete: The democratic politician frame enables the reader to select the economy and the role the president should play in making economic policy as the foci of this text, and to delete the reference to the ban on homosexual soldiers.

  2. Strong/Weak Selection and Deletion: The frame organises the segments of the text into the highly important sections (strong select); sections which provide detail relevant to the context in which the reader would encounter the text, but which are not central to the meaning (weak select); sections which provide transitions between the selected elements and weave the text into a cohesive whole (weak delete); and sections unimportant in understanding the text (strong delete). One could thus map this text as follows:

    Where are your priorities(strong select)
    to introduce an economic plan(strong select)
    Mr. President(strong select)
    It's the economy stupid(strong select)
    a smoke screen to hide(weak select)
    his failure(weak select)
    As your campaign slogan put it(weak select)
    to a campaign promise(weak select)
    President Clinton(weak select)
    a genuine commitment(weak delete)
    Is planning because of(weak delete)
    or is he just creating(weak delete)
    to lift the ban(weak delete)
    on gays(strong delete)
    in the military(strong delete)

  3. Generalised Meaning: This passage generalises the tendency of a politician to get side-tracked and fail to address the important issues like the economy.

  4. Global Fact: The democratic politician frame applied in this text generates the global fact that President Clinton should focus his attention on the economy.

Discursive structural frame analysis, as developed by van Dijk (and also applied by Donati and Triandafyllidou), facilitates the analysis of how people construct individual texts to achieve a particular communication goal. One could use this methodology to compare of framing strategies of people who attempt the same communication goal in producing the same type of text, or to compare uses of a particular discursive frame by different people.

Van Dijk engages speculation that a second framing process may operate at the superstructural level. Van Dijk defines a superstructure as 'the schematic form that organises the global meaning of a text' (van Dijk, 1980: pp. 108 - 9). Present formal languages do not have associated superstructural discourses for much the same reason that they do not have discursive structures: superstructures leave many relationships implicit, while formal languages make all relationships explicit (1980: p. 29). Superstructures, van Dijk maintains, have communicative functions, and may also have a clear format or internal structure.

Some superstructures, like narratives or scientific reports, have reasonably fixed structural forms. Other superstructures, like a press article, have less fixed, more superficial, and more duplicitous forms, while some superstructures may lack a consistent form (1980: pp. 118 - 23). One superstructure may also contain one or more others (a narrative within a narrative, an advertisement within a film, an argument within an advertisement, etc.). Discursive structures provide a semantic link between superstructures and grammatical structures (1980: p. 130).

Superstructures 'further organise the macrostructure of a text, by assigning sequences of macropropositions to schematic categories,' but 'also play a role in the formation of macrostructures themselves' by constraining 'the application of the macrorules' (1980: p. 128). One might argue that in constraining discourse structures, superstructures also frame them; however, van Dijk has reservations about defining frames in this manner. As he explains, some scholars have viewed superstructures:

as examples of typical frames or scripts that define the stereotypical knowledge of language users. Such a frame would then consist of the major categories and formation rules of the superstructural schemata. We have, however, some doubts about this frame-like nature of superstructures. Not only are they very much different from the `real-life' examples for frames, such as `eating in a restaurant' or `going to the movies,' but also they may be much more implicit than the traditional knowledge about social episodes (van Dijk, 1980: p. 131).

While these reservations may in and of themselves be convincing, van Dijk overlooks the possibility the domain of discourse, as he defines it, may have higher levels beyond superstructures.

Even if superstructures may not be said to properly frame discursive structures, I argue that a higher level of discourse, a cultural frame, nonetheless modifies superstructures. I also take issue with van Dijk's characterisation of discursive structural frames as '"real-life" examples of frames'. Such frames, as van Dijk argues, depend on the point of view of the speaker, and discursive structural frames select and hierarchically organise some information, while deleting other information. Such frames would be more properly described as socially constructed rather than 'real-life' (in some absolute sense) frames. If, for example, a nomadic community changes to a settled community dependent upon an economic system requiring limited movement of most families, the concept of house, and, consequently, both the 'house frame' and the way people experience 'real life' when they use the reconstructed house frame, would change. Likewise, in the contemporary USA or UK, the house frame can connote different meaning for property speculators and homeless persons. Supra-superstructural (or cultural) frames thus need not be significantly more abstract or more dependent on context than discursive structural frames.

Frames as Metaphors - The Lakoff and Johnson Approach

Unlike van Dijk, linguists George Lakoff and Alan Johnson do not identify a hard distinction between language and discourse, but rather see a fluid continuum of forms of human communication. Lakoff and Johnson argue that 'human thought processes are largely metaphorical' (Lakoff and Johnson, 1980: p. 6). Individuals use metaphors at the micro-level of language to understand and pass on their understanding of 'one kind of thing in terms of another' (1980: p. 5). At the same time, cultures also develop systems of 'conceptual metaphors' which organise the experiences and values of all members of that culture in a systematic and coherent way by highlighting some aspects of the relationship between the compared items, while hiding other aspects. While individuals may from time to time critically assess a conceptual metaphor, people within any given culture will employ these metaphors on a regular basis to talk about a wide array of subjects, spanning complex theoretical debates to the routine and mundane details of daily life (1980: p. 3). Lakoff and Johnson do not refer to the work of van Dijk, but their conceptual metaphors could be said to operate as cultural frames at a level which could organise the superstructures which van Dijk describes.

Lakoff and Johnson identify some distinct sets within the realm of conceptual metaphors. One set, orientational metaphors, enable people to understand constructs in the same way in which they understand space. In Western societies, for example, people categorise political groups along an ideological spectrum, in which 'centrists' occupy 'the middle ground', conservative forces lie to the right, and liberal and socialist forces lie to the left. Conflicting parties can reach compromise by meeting half way, by giving ground, or by moving towards a compromise. People may also evaluate objects, events and other people by positioning them along a vertical continuum, though Lakoff and Johnson stress that people assign greater value to different ends of different vertical spectrums. Thus 'being down-to-earth' or 'keeping your feet on the ground' may be more desirable than being 'starry-eyed', 'drifting up into dream land', or 'walking with your head in the clouds'. At the same time, 'staying on the moral highground', earning a 'top mark' in school, and 'keeping your spirits up' are more desirable than 'slithering through the gutter', 'falling to the bottom of the class', or 'feeling down in the dumps'.

Lakoff and Johnson also identify a set of container metaphors, whereby people understand issues and events in terms of the physical process of containing something. People may thus be said to 'bottle up' their feelings, or 'pour their hearts out'. Bureaucrats may address a fiscal crisis by trying to 'close the flood gates', while a football team may be 'bursting at the seams' with enthusiasm.

Other conceptual metaphors do not fit into a category per se, but they nonetheless fundamentally shape the way we think about a particular concept. Lakoff and Johnson provide the example 'argument is war' to illustrate this point. In the United States, people regularly construct sentences employing war metaphors, such as: 'the opposition laid a political minefield which the government must navigate to pass the bill'; 'the management torpedoed union hopes'; 'I hear you have some complaints. OK, fire away'; 'the state attorneys thought that they had found a smoking gun, but by the end of the cross-examination, the defence had blasted a hole through the heart of the prosecution'; 'the speaker takes no prisoners'. Lakoff and Johnson contend that the 'argument is war' metaphor is so deeply ingrained in US culture, that people cannot engage in an argument without becoming verbal combatants. They write:

It is important to see that we don't just talk about arguments in terms of war. We can actually win or lose arguments. We see the person we are arguing with as an opponent. We attack his positions and we defend our own (Lakoff and Johnson, 1980: p. 4).

People who do not wish to assume the combatant mantle while engaged in a contest over knowledge have to change the construction of the contest. One might say 'I think we should agree to differ', or 'I don't want to fight with you. Why don't we just sit down and talk this out like two civilised human beings'. Lakoff and Johnson further suggest that people could not casually change the metaphoric construction of an argument to a new form, such as 'argument is dance', without constructing a new concept altogether (1980: pp. 4 - 5).

In their 1980 book, Lakoff and Johnson point out that sets of conceptual metaphors vary across cultures. As one example, they compare discussions with US students and students from non-Western countries about finding solutions to problems. They observe that their US students framed difficulties in terms of mathematical formulas. For this group, the solution metaphor implied that an answer could be found much as one might resolve a mathematical problem by working through steps toward a solution. Some non-Western students, on the other hand, viewed problems like a solution of chemicals in a test tube. One could add ingredients to make the solution more manageable, or to dilute the influence of one element or compound, but one could not eliminate all troublesome contents of the solution.

In more recent work, however, Lakoff (1987) suggests that some conceptual metaphors arise from biological features of the human body which impact our experiences, and thus have greater potential to exist across many cultures, if not universally. Lakoff notes, for example, that people in the US can use metaphors relating to body positions they assume while physically or mentally unwell, such as 'she has been laid low by flu', or 'since he wife died, he has been feeling really down'. Likewise, Lakoff suggests that some emotions cause physical reactions in the human body which we in turn use as conceptual metaphors to talk about these emotions. When people feel enraged, Lakoff contends, their body heat rises and blood rushes to their heads. For this reason, he maintains, people use heat and blood metaphors to describe anger. As examples, people might say: 'you shouldn't get so hot under the collar'; 'she just needs to let off some steam'; 'he is hot-headed'; 'this really makes my blood boil'; 'when I told my boss about the accident, he erupted into a tirade'; 'he's seeing red'. Lakoff also suggests that people may adopt conceptual metaphors based on behaviour we observe in animals. When talking about anger, people in the US will make such statements as: 'he came with teeth bared'; 'she has the claws out'; 'it got their hackles up'; 'I am trying to soothe ruffled feathers'. Whether people actually do experience anger or any other emotion in the same way across all cultures as they do in European and European-based cultures, however, is open to dispute. Study of physical experiences of emotion lie outside the scope of this study, and the popular cultures of the US and Britain are too similar to serve as adequate points of comparison for me to use the press data to comment on this issue.

Nevertheless, Lakoff and Johnson's discussion of conceptual metaphors closely parallels Donati's discussion of object analogy frames (which I will address in the section on Donati and Triandafyllidou), with the main difference being that Lakoff and Johnson offer a more extensively developed argument. Conceptual metaphors, or object analogy frames, share common features with other forms of cultural frames: (1) they are a part of the discourse which develops in any given culture, and people learn these frames as they learn to competently participate in that culture; (2) they highlight some aspects of an event or issue to which people apply a frame, while hiding others; (3) they organise experiences, values and beliefs of all members of a culture in a systematic and coherent way; and (4) they are accessible and useful to people in the same culture who ascribe to a wide array of ideologies.

Swidler and the Study of Institutions

Ann Swidler argues that higher levels of discourse lie along a continuum, with ideology ('a highly articulated, self-conscious belief and ritual system, aspiring to offer a unified answer to problems of social action') at one end, tradition ('articulated cultural beliefs and practices, but ones taken for granted so that they seem inevitable parts of life') at the centre, and common sense ('the set of assumptions so unselfconscious as to seem a natural, transparent, undeniable part of the structure of the world' - a definition which resembles Donati's concept of a frame) at the other end (Swidler, 1986: p. 279). She does not discuss the processes by which people may reconstruct information along this continuum in detail, but she does lay the ground from which Donati builds his argument that frames operate separately from ideologies.

Swidler herself expresses a greater interest in cultural study than frame analysis. Swidler contends that a culture facilitates the production of customised realities by providing people with a 'tool kit' or 'repertoire' of 'tacit' cultural elements, including attitudes and styles, and 'explicit cultural materials', including rituals and beliefs, from which they can construct interpretations of events (Swidler, 1986: pp. 277; 281 - Donati borrows from this idea).[7] Institutions, semiotic codes, and political contexts constrain the process of reality construction. To change the encoding of meaning about a particular issue, social movements confront institutions which have invested in producing the established meaning. In defining themselves in opposition to these institutions, social movements have to develop 'tactics and internal cultures' that challenge the existing structures (Swidler, 1995: p. 37). This structural constraint likewise tends to channel social movements to adopt key elements of the language used in the institutions they confront. She writes:

Where the state enshrines 'rights' as the crucial legal claim that trumps all others, both individuals and social movements will conceive of the claims they make as 'rights'. And where legal claims are tied to group identities, as they long have been for American Indians and increasingly have become for women, the disabled, and members of many ethnic and racial groups, identity becomes a central focus for social movements' (Swidler, 1995: p. 37).

Additionally, the semiotic codes and systems of political power that hold cultures together 'as a global, collective property' constrain the capacity of people to construct meaning. Though systems of representation and systems of power may not set solid boundaries in which people think, they do define individuals' 'knowledge of how others will interpret their actions' (Swidler, 1995: p. 39). Thus, semiotic and power systems affect the creation of meaning, even if people reject the composition of those systems.

Swidler criticises Snow and Benford for suggesting that social movements should strive to create 'frame alignment' among members and potential activists. She contends that people exercise little direct control over systems of meaning in culture, and thus, an achievable conception of frame alignment 'is not just a matter of individuals getting their frames in sync. Rather, individuals develop common scripts [my emphasis] in response to the features of the institutions they confront' (Swidler, 1995: p. 38). Social movements challenge institutional constraints by waging a 'battle for symbolic encoding' (1995: p. 34). If the movement wins, it 'can reshape the world more effectively through redefining its terms rather than rearranging its sanctions' (1995: p. 34).

It should be noted that Swidler uses the term 'script' more loosely than Donati and Triandafylliduo, who employ this term interchangeably with the level of framing which I call cultural framing. Swidler, on the other hand, does not always specify the point on her continuum of higher-level discourse to which she refers when discussing scripts, or, for that matter, the influence of institutions on discourse. In her passage on legal claims, cited previously, for example, Swidler leaves the question of whether the constructs 'rights' and 'group identity' fall into the category of ideology, tradition, or common sense -or if there are separate ideological, traditional, and common sense forms of such constructs. This point is complicated by the fact that other scholars who have addressed the influence of institutions on social movements (for examples, see Martin Rein, 1983; or Georgia Duerst-Lahti and Rita Mae Kelly, 1995) have used the term 'framework' to reference the influence of institutional cultures on the way people in organisations approach the process of policy design. While institutions, as Swidler, Rein and Kelly suggest, may well influence ideological debates, one could not necessarily extrapolate that institutions therefore also influence cultural framing. Nevertheless, Swidler does contribute to the study of framing through her discussion of 'cultural tool kits' and a higher-level discourse continuum in which ideology and common sense do not overlap.

Donati and Triandafyllidou

Anna Triandafyllidou and Paolo Donati apply frame analysis, or discourse analysis, as both prefer to call this approach, to study sustainable development. Donati and Triandafyllidou suggest that discourse functions on two bifurcated structural levels, a 'surface structure' and a 'deep structure'. The first sub-level of the 'surface structure' contains 'codes', 'hypercodes', and 'signature elements'. Codes are marked lexicons and text segments which signify a deeper meaning beyond the basic lexicon of the word or words in a phrase in isolation, and signature elements alert the reader/listener to interpret a text in terms of a particular frame. Hypercodes are collections of words which give meaning to codes. The 'surface structure' also contains 'discursive structures', which organise the meaning of passages of texts (both cite van Dijk as the originator of this term). The 'deep structure' includes 'narrative structures', or storylines and plots, and 'ideologies'. Framing, contend these scholars, occurs at both the discursive structural level and at the deep structural level.

Donati (1994) and Triandafyllidou (1995) illustrate the process of discursive frame analysis on texts collected from their research on sustainable development. Donati counts instances in which derivations of the term sustainable appear in his collections of texts (Donati, 1994: p. 20). Their analysis otherwise does not differ significantly from the approach outlined by van Dijk, though neither develop their discussion of surface structural framing to the degree with which van Dijk approaches this subject. Their contribution to the study of frame analysis arises in their discussion of deep structural frames, on which I will now concentrate.

Deep structural frames, they contend, function metaphorically. Indeed, writes Triandafyllidou, 'the term "discourse" [itself] ... is used as a metaphor to denote the social "dialogue" that takes place among political or social actors, groups and institutions' (Triandafyllidou, 1995: p. 3). Both also see deep structural frames as patterns of organised information by which people make sense of the world. These 'patterns', 'schemas', or 'frames' form part of the 'discursive universe' (Triandafyllidou, 1995: p. 3) in which people interact with each other. People learn frames as they learn to fluently use a language and as they learn the narrative structures and ideologies present in the cultures which use that language. When people encounter new information or a new experience, they make sense of that information or experience by fitting it into an existing frame. Nevertheless, people will generally be able to fit any given collection of information into multiple frames (Donati, 1992: p. 141), though, at the same time, people will also tend to selectively perceive information, focusing on details which most readily fit into the frames they know (Triandafyllidou, 1995: p. 7). Additionally, as people share framed information, they need not refer to all aspects of a frame directly to communicate which frame they have adopted to make sense of the information. Instead, people need only make reference to one dimension of a pattern to enable hearers or readers of their text to recall the whole frame (Donati 1992: p. 149). Discourse analysis (or deep structural frame analysis), then, entails the study of the process by which people understand their world by organising experiences and information in terms of these familiar patterns, or frames (Triandafyllidou, 1995: p. 7; Donati 1992: p. 140).

These two authors differ on subtle points in their definitions of frames. Triandafyllidou argues that deep structural frames operate in hierarchies, with higher level, or 'master frames', subsuming and shaping lower level deep structural frames. She identifies the concept of sustainable development as an example of a master frame (Triandafyllidou, 1995). Triandafyllidou does not elaborate on this suggestion, and leaves open questions of the relationships between lower level deep structural frames, master frames, and other deep structural discourse forms. Donati, in contrast, suggests that loose and fluid relationships exist between deep structural frames. For him, these frames include analogies with objects (x is a 'carefully crafted watch', where x may equal the economy, the public education system, the environment, etc.), and action/event sequences ('no gain without cost'), not concepts, like sustainable development.

These two authors also draw different conclusions from deep frame analysis. Triandafyllidou suggests that discourse analysis facilitates the identification of people who support and oppose sustainable development (Triandafyllidou, 1995). Donati, on the other hand, distinguishes content analysis, which can involve the identification of codes in texts which a research may define as an indicator of what the author or speaker thinks about an issue, from deep discourse (or frame) analysis (Donati, 1992: p. 153). For Donati, deep frame analysis reveals how people understand an issue, and can give insights into why people might favour or reject an argument (1992: pp. 153 - 4), but does not allow classification of any given speaker as for or against an argument (Donati, 1994: p. 16). Donati's argument provides the base from which I now build my own conception of cultural framing.

Defining Cultural Frames and the Framing Process

I define cultural frames as socio- culturally and cognitively generated patterns which help people to understand their world by shaping other forms of deep structural discourse. In this respect, cultural frames, like van Dijk's superstructures, would be a part of the 'memory of language users' (van Dijk, 1980: p. 131). The array of cultural frames in any given culture need not be consistent with each other. Cultural frames provide people with the tools from which they can construct meaning from information they encounter to make sense of their world, but, as suggests, cultural frames themselves are not 'finished constructions' (Donati, 1992: p. 139).

Like Swidler and Donati, I define cultural frames as distinct from ideologies.[8] Like the concept of 'frame', the meaning of ideology is the subject of dispute among social scientists. While some scholars, like Swidler, define ideology as 'a self-conscious belief and ritual system, aspiring to offer a unified answer to problems of social action' which does not include any necessary reference to social hierarchy (Swidler, 1986: p. 279), others, like Anthony Giddens, see ideologies as the 'shared ideas or beliefs which serve to justify the interests of dominant groups' (Giddens, 1989: p. 727). Giddens elaborates that he believes 'the concept of ideology connects closely with that of power, since ideological systems serve to legitimise the differential power which groups hold' (1989: p. 727). Cultural frame analysis does not require a researcher to engage with this debate over the meaning of ideology, as cultural frames cut across ideological boundaries, regardless of which definition one adopts. Ardent socialists, radical animal rights activists, conservative business people, and religious fundamentalists can all make use of such frames as 'anger is heat', 'argument is war', or 'no gain without cost'. Some frames work better with some ideologies than with others, but ideologies do not generate frames, and frames do not depend on any specific ideology to remain viable in a culture.

When one distinguishes frames from ideologies, Donati observes, then frame analysis gains 'the uncommon advantage of not treating ideology as a monolithic and fully structured element', thus emphasising 'the weakness rather than strength of ideologies' (Donati, 1992: p. 139). Social actors can reconstruct their ideological arguments by swapping cultural frames. The new frame provides guidance for re-arranging values and beliefs into new arguments. The new cultural frame does not formally restructure arguments, but rather operates more like an interior designers' eye, allowing social actors to observe sequences of things which work in a given space and sequences which do not work. Changing cultural frame thus may not be so dramatic a move for a group of people as changing ideology without changing framing as well.

I share the view of the majority of authors reviewed here that frames function metaphorically, stimulating people to focus on particular elements of representations while ignoring others. In this respect, as Donati observes, frames operate as 'an instrument for defining reality' as opposed to 'an instrument for describing reality' (Donati, 1992: p. 142). Nevertheless, I share Donati's view that 'a frame is not necessarily literally outlined in the text' (1992: p. 149); rather, mention of 'some elements - sometimes even one - is usually enough to `suggest' or to recall the whole set' (Donati, 1992: p. 141). Elements of cultural frames thus operate metonymically. Lakoff and Johnson note that metonymies themselves shape understanding. If one part can stand for the whole, then people could theoretically select from the whole range of parts for the basis of the metonym. The part people actually use reflects the aspect of the whole on which people place the greatest focus (Lakoff and Johnson, 1980: p. 36).

Donati offers two useful 'rules of thumb' for identifying cultural frames. First, a frame should represent a widely comprehended category which encompasses a broader range of meaning than the object of framing; hence, 'the principle is that the frame should work as a guiding model for what is to be understood' (Donati, 1992: p. 146). Second, the researcher should be able to interchange the cultural frame for the topic of the text without changing the meaning of the text (1992: p. 147).

Cultural Frame Boundaries

Some scholars, including Levi- Strauss, have argued that humans have evolved a natural capacity to use myths and language, and that all cultures select from a set of myth types and language elements which can exist universally across cultures (Levi-Strauss, 1985: p. 191). I find such claims intellectually dissatisfying. The present literature on frame analysis does not address questions of the origin of frames or of the conditions which give rise to particular frames, but I would reject any assertion that the humans, as a species, possess a box containing a finite set of identifiable frames from which cultures borrow selectively. Each of the frames discussed in the published frame analysis literature entails assumptions found in some cultures, but not in others. For the 'x is a carefully crafted watch' frame identified by Donati to be meaningful, for example, one must assume that time is linear and can be precisely measured by a crafted instrument. The 'house' frame described by van Dijk (1977: p. 146) likewise entails assumptions about the material used to construct a house, as well as about the permanency of a house as a structure.

One might be tempted to speculate that the 'injustice' frames studied by Gamson could be more universalised, as groups of people in many cultures have confronted events which they perceive as unjust. Nevertheless, a universal 'injustice' frame would presuppose some universal concept of justice against which injustice could be defined. While Kantian philosophers might accept such a presupposition, the groups participating in many social policy contests advocate incompatible visions of social justice. Even the examples which Gamson provides of social movements using injustice frames (to argue against nuclear power, against Israeli occupation of the West Bank, against Arab terrorism in Israel, or in favour of affirmative action) do not contain common assumptions of the meaning of justice. The intensity of debates within states, between member states of regional associations, and between players in international organisations over questions of justice support the argument that humans do not possess a universal concept of justice.

At the same time, I do believe that a cultural frame can work in more than one society or linguistic environment. Different groups of people who share some similar experiences and hold some similar beliefs may independently develop the same cultural frame to make sense of some issues, and one group also may adopt a frame developed by another. A group which separates from one society, and then proceeds to form a new society will likely retain many cultural frames from its parent society, discard others, and develop or adopt other new cultural frames. Indeed, the press, other media actors, corporate advertisers, performers in the better funded branches of entertainment industries, and recently developed information technologies have accelerated contact between cultures. The exposure of societies to cultural frames they may not have previously used increases the propensity for multiple cultures to use some similar frames. Within any cultural context, some frames will provide more widely accepted ways for making sense of the world than others, and cultural frames acceptable within one social context may not prove acceptable in another.

I do not intend to fully explore the factors which may bound cultural frames in this paper, and make no pretence of offering more than a cursory explanation of some of the factors. Nevertheless, I do note that cultural frame use does not necessarily reflect the desire or unconstrained choice of people within a given culture. Moscovici and the other socio-psychologists in his edited volumes (Moscovici, 1971; 1984) belong to a group of scholars which rejects the notion that people can think or act - let alone create new frames - unfettered by prevailing cultural assumptions. Berger argues that 'capitalism survives by forcing the majority, whom it exploits, to define their own interests as narrowly as possible', and that 'publicity is the life of this culture - in so far as without publicity capitalism could not survive - and at the same time publicity is its dream' (Berger, 1972: p. 154). Some cultural frames work better within a capitalist environment than others, and one could logically extend Berger's argument to postulate that discourse marketed under the global capitalist structure would favour those cultural frames which worked best within this economic rubric.

While capitalist systems would not necessarily preclude cultural frames inconsistent with elements of capitalist ideology, these structures could impact the development of the less harmonious cultural frames. Table 1 displays the principal cultural frames identified in popular press coverage of military gender and sexuality policies in Britain and the United States (Fisher, 1997). Three of these frames tell self-contained stories. The fourth, 'militarism harms society', differs from the other three in four ways: it is more vague; it is a negative statement (while the 'not for everybody' frame need not necessarily have negative implications); it prescribes change - but the change prescribed is unclear; it is most clearly identified with a philosophical position. Of the four, this frame is least compatible with capitalism. This frame additionally faces a second obstacle. The political structures in both Britain and the United States presume the legitimacy of the use of force to achieve political objectives (Schlesinger, 1991). While not all people residing in either country agree with this presumption, those who question the legitimacy of violence as a policy tool are in the minority. In directly raising questions about the use of force as well as the more ruthless aspects of capitalist systems, this frame confronts rather than affirming dominant values in British and US culture, and, hence, can neither assume the form of a fully self-contained story about common knowledge nor fully shake its association with a philosophical perspective.[9] Nonetheless, 'militarism harms society' can still function as a cultural frame for two reasons: (1) its entailments are widely known among residents of Britain and the USA, even if people identify this frame as a frame of a minority (of courageous reformers, or, alternatively, of hippies, radicals, draft dodgers, and small religious sects); and (2) it does not have ties to a specific ideology, but instead works well with any ideology which questions to legitimacy of state- sponsored violence.

At the same time, I do not exonerate individuals or social groups within any social system of responsibility for the array of cultural frames on offer in that society's 'tool kit' at any one time. While most individuals and groups will make use of existing cultural frames exclusively, exceptional individuals and groups will significantly modify frames, or develop new ones.[10] Cultural frame developers may have to abide by some constraints within the dominant systems (whether patriarchy, capitalism, liberalism, or others) of their culture at the time when they generate the cultural frame. Even so, cultural frame developers do retain a degree of personal control, and hence some personal responsibility, for their ideas. Nevertheless, individuals and small groups of people, no matter how clever, will not succeed as frame creators or modifiers - and cannot impose cultural frames upon the majority - without collective consent. As observes, 'representations, obviously, are not created by individuals in isolation' (Moscovici, 1984: p. 13). Even so, I disagree with Moscovici's conclusion that collective responsibility for consent ends with the process of conventionalising representations. By selecting a cultural frame to understand an issue or event, individuals maintain the saliency of that particular frame, whether or not they consciously acknowledge the consequences of their choice. A cultural frame may fade unremarked as people cease to select it, however, people can also remember - or be reminded - that they have the power to select, deselect, or change a cultural frame during a framing dispute (even if that power is partially checked by cultural and systemic constraints).

Types of Cultural Frames

Having situated frames within the study of discourse and culture, one can then address the question of whether all cultural frames function similarly, or whether these frames cluster into categories. Donati groups frames into 'two fundamental types': 'those which highlight analogies with objects (especially mechanical objects, such as "Carefully crafted watch"), and those which highlight analogies with event/action sequences (also called "scripts", such as "No gain without cost")'. Donati contends that his experience working with frame analysis leads him to suspect that people more regularly use event/action sequence frames in political discussions (Donati, 1992: p. 152).

While I find Donati's explanation of event/action sequence frames convincing, I think that his discussion of object analogy frames could be more fully explored. I propose that object analogy frames, such as 'x is a carefully crafted watch', neither possess a single, fixed form, nor do they always function in isolation. In any given linguistic community, a relatively small range of objects gain status as conceptual metaphors, as Lakoff and Johnson define this term, and these objects serve as the focus of object analogy frames. One can use a metaphor in speech without using a frame. For this reason, statements such as 'the economy is a carefully crafted VCR' or 'the environment is a carefully crafted radio' make no apparent sense. Timepieces, on the other hand, function as conceptual metaphors in many English-speaking societies, and one can thus find an array instances in which one item is compared with a time piece ['a promise to turn back the hands of time' - a phrase with positive implications if used in conjunction with makers of anti- ageing cream, and negative implications if used to describe a politician or business person accused of lacking a vision for the future; 'the body is a finely made clock' (hear the ticking of the biological clock, seek medical treatment to get a body part working like clockwork again, the heart is a ticker); 'x is a time-bomb waiting to explode', and so forth]. Object analogy frames which share the same object as the focus of comparison also share many entailments.

One source of the ambiguity in much of the frame analysis literature may arise from the fact that the word frame itself is a conceptual metaphor in many English-speaking societies. Politicians debate how to frame problems, advertisers frame products to increase consumption, and some people accused of crimes before courts claim to have been framed. Though I think the word frame still provides a useful focus for discussing the process by which people construct meaning, scholars who employ frame analysis should also recognise that the term 'frame' itself frames the understanding which readers will draw from the results of the analysis.

Additionally, I find Donati's list of fundamental types of cultural frames incomplete, and argue that at least one other category exists. Consider the frames which surface in popular press coverage on military gender and sexuality policies in the US and UK (identified by Fisher, 1997), listed in Table 1 below.

Table 1: Principal Frames in ihe USA and Uk Military Gender and Sexuality Debates
Some Institutions/Tasks Are Not For Everybody
Everyone Deserves an Equal Chance
Women/Gays Are Important Actors in Society
Militarism Harms Society

These cultural frames neither make reference to objects nor to event/action sequences, but they do offer flexible, 'common sense' explanations of why things are the way they are which could be applied to a wide range of issues. I label these frames social system frames, as they highlight analogies with models of social relations. Social system frames have more explicit value associations and political implications than analogies to event/action sequences or objects, but like the other types of cultural frames identified by Donati, social system frames do not have any fixed ideological connections, a feature which emerges when one details the entailments of these frames.

The 'some institutions are not for everybody' frame posits that fundamental differences divide people in a society into groups. Some groups of people are better suited to perform certain tasks than others; some groups are not suited to perform particular tasks. Society takes foolish risks if it does not channel people into tasks for which they are suited or if it allows people to perform tasks and access institutions for which they are not suited. Five types of feature(s) qualify or disqualify a group from performing a task or participating in an institution: (1) biological features ('Governments should take action to boost declining white birth rates, especially as other ethnic groups are reproducing themselves above the rate of replacement'; 'People with epilepsy should not be allowed to drive vehicles'); (2) spiritual/moral features ('Jesus did not select any women apostles because God did not intend for women to be priests'; 'Child custody hearings should generally award custody to the mother'); (3) birthright ('Tax money should be spent helping the lawful citizens of this country, and the government must step up efforts to expel illegal aliens, benefit tourists, and bogus asylum seekers who are draining the public coffers'); (4) results of behavioural choices ('Prisoners convicted of felonies should not have the right to vote'; 'People who have molested children should not work in day care centres'); or (5) results of experiences a person has or has not had ('Colin Powell grew up in an affluent neighbourhood, attended predominantly white schools, and held a military career. He does not represent the majority of black Americans'; 'Because H. Ross Perot has never held elected office, he has not been corrupted by the lobbyists and professional politicians. He is therefore the kind of man we need to clean up the sleaze in Washington'). One can use the 'not for everybody' frame to imply that more than one of these features occurs at a time, as occurs in the following examples: 'Only people over 35 have enough worldly experience to serve as President of the USA' (biological feature; result of experience); 'Our company hires women for the factory floor. Their small hands suit them to the detail of the work, and motherhood prepares them to cope with the boredom on the assembly-line' (biological feature; results of experience).

The 'everyone deserves an equal chance' frame declares that societies benefit from differences among their members. This frame rejects rigid hierarchies based on difference, as well as assumptions of essential difference between groups of people. Instead, this frame posits that all members of society are entitled to the same opportunities, and that it is immoral to deny equal opportunity ('Heating or eating - this is the choice which government policy forces many elderly people to face in winter'; 'Meat is murder'; 'We must reinvest in state run primary schools and stop relying on the private sector to educate our children. We are not true to our democratic ideals if we allow the income of prospective parents to determine the place of future citizens in our society'; 'We should demand that governments which receive financial aid from our taxpayers give women the right to vote. No just society can ignore the voices of half of its members'). The equal chance frame does not offer any criteria for weighing competing claims of different groups. As a consequence, in some recent political controversies, activists on different sides have been able to apply this social system frame to build conflicting arguments. Consider these contrasting positions on affirmative action.

Contextual uses of this cultural frame illustrate the significance of two of Donati's observations: that people frame an object around which an issue revolves rather than an issue itself, and that the study of cultural frames involves identifying how people understand an issue, rather than determining if they are 'for' or 'against' a proposition.

People do not always need the same level of explanation of how their world works to feel satisfied with their understanding of a particular set of information. If people feel compelled to make or defend an argument, or if people feel that an issue is highly relevant to themselves, they may well be more likely to select a fully scripted social system frame, like one of the first two frames listed in Table 1. However, not all information and experiences compel people to make arguments or confront people with dilemmas in their own lives. Consequently, people may also use a more general and loose social system frame to understand an issue. The third social system frame in Table One, 'women/gays are important actors in society' is an example of such a loosely scripted social system frame. This frame does have entailments. It assumes the legitimacy of homosexuality, a premise not accepted by some people who have applied the 'not for everybody frame' to military sexuality debates. Even so, this frame has less clear value and political implications than the first two listed frames.

A person's choice to employ a loosely scripted cultural frame, however, does not necessarily means that the issue to which she or he applies this frame holds little relevance for her or him. The fourth soical system frame in Table 1, 'militarism harms society' likewise lacks the same completeness of script found in the first two frames. This frame reflects constraints within the cultural context of US and British society on the range of explanations people can use to make sense of their world.


Discursive structural frames and cultural frames constitute two of the forms of discourse in any society. People learn to use both types of frames as they learn to competently participate in a culture. Cultural frames provide a 'common sense' pattern that systematically shapes the way we interpret images by drawing our attention to some aspects of these images while encouraging us to ignore others. When we frame images generated during a public policy dispute, we frame an object around which the issues revolve, rather than the issues themselves. Cultural frames are flexible. We can choose from a variety of frames to make sense of any particular information, and people ascribing to highly conflicting ideologies can find the same frames equally useful.

We should recognise the limitations of frame analysis. Neither the study of discrusive structural frames nor cultural frames by itself provides data which would enable the researcher to make prescriptive suggestions to social movements or other groups of people seeking to influence public policy. Cultural frame analysis enables scholars to study how people understand an issue, and to track the way in which this understanding changes over time. No direct connections tie cultural frames to language. Scholars cannot identify frames by counting the appearance of key words and phrases, or by specific argumentative structures. Instead, one must look for storylines about what is to be comprehended. Researchers have successfully identified a cultural frame when they can interchange the frame for the topic of the text without changing the meaning of the text. Since cultural frames lack fixed and quantifiable markers, frame analysis methodology may well only provide useful information to researchers studying cultures of which they are members.


1 Goffman's discussion of gender stereotyping in visual images closely parallels work by John Berger (1972), who observed that contemporary photographs and advertisements mirror the gender stereotyping in European oil paintings from the 19th century. Berger argues that these stereotypes arise from the male-centred attitudes which have dominated democratic thought from the rise of ancient Greece through the present century, though he also contends that the advent of capitalism accentuated the perniciousness of gender stereotypes which define men as social actors who can move in and out of the public gaze, and women as passive agents trapped permanently in the public gaze. Berger does not address the question of how male-centred values came to dominate democratic thought.

2 Snow and Benford outline ways of framing grievances which inhibit collective. They warn that movements should not frame a problem as hopeless, or as resolvable by advanced technology or another force over which most members of the public have no control. Additionally, they caution against framing a problem in a manner which leaves the action people might take unclear (Snow and Benford, 1988: pp. 203 - 4).

3 People must be able to identify a concrete target for their anger to develop an injustice frame. Nevertheless, people need not 'correctly' identify the cause of the problem to perceive it through the lens of injustice. As long as people can focus on a target, whether or not their anger is misplaced, they can develop an injustice frame (Gamson 1995: p. 91; Snow et al, 1986: p. 474).

4 Gamson stresses that 'this message comes through more equivocally on some issues than on others' (Gamson, 1995: p. 97).

5 In reviewing the literature on the psychology of language, Harley rejects the Chomskian notion that language and cognition function separately, the Piaget theory that language development depends on a person's passage through stages of cognitive development, and strong (but not all weak) versions of the Sapir-Worph hypothesis that language structures thought (Harley, 1995: see Chapter Ten in particular).

6 Van Dijk notes a special case of texts or texts fragments in which microstructures and macrostructures coincide, such as the command 'go home'. Van Dijk contends that such cases operate according to a 'zero rule', that is, the selection and deletion process does not operate as the whole message is selected as indicating the macrostructure.

7 Swidler attributes the term 'tool kit' to Hannerz (1969).

8 Cultural frames could also be said to resemble myths, depending on how one defines the latter concept. Roland Barthes defines a myth as follows:

It can be seen that to purport to discriminate among mythical objects according to their substance would be entirely illusory: since myth is a type of speech, everything can be a myth provided it is conveyed by a discourse. Myth is not defined by the object of its message, but by the way in which it utters this message: there are formal limits to myth, there are no `substantial' ones. Everything, then, can be a myth? Yes, I believe this, for the universe is infinitely fertile in suggestions (Barthes, 1972 translation: p. 109).

Barthes definition of myth differs from cultural frames in two respects. First, myth, for Barthes, centres around a specific, or at least partially specific object, person, or event (the properties of milk, the face of Garbo, the world of wrestling, or Romans in films). While myth, as a form of discourse, 'is not defined by the object of its message', some specific object nevertheless serves as the focus of any particular myth. While one type of cultural frame, the object analogy frame, entails a comparison between one thing and another, cultural frames do not necessarily revolve around discussion of a specific object, person or event ('no gain without cost', 'everyone deserves an equal chance'). Object analogy frames function metaphorically, while Barthes does define metaphor as a necessary part of myths. Secondly, the myths which Barthes describes have limited application to other objects, persons or events outside the myth story. People use cultural frames, on the other hand, to understand new information which they have not previously encountered, as well as to make sense of things with which they are already familiar. Barthe's understanding of myth, however, does share common properties with cultural frames. Barthes argues that everything, potentially, 'can be a myth' as 'the universe is infinitely fertile in suggestions'. I see no reason to reject the possibility that any concept or object could be included in a frame. Likewise, cultural frames, like myths, are 'a type of speech' (as Barthes defines this term) which 'is conveyed by a discourse'.

9 I deliberately leave the term philosophical perspective unspecified. While most readers would likely presume that this position could be identified as a pacifist philosophy, some recent anti-militarist writings from people such as Cynthia Enloe and Nira Yuval-Davis question the legitimacy of the binary opposition of peace and war implicit in the concept of pacifism.

10 Scholars representing a wide range of perspectives have discussed the ability of exceptional individuals to significantly transform societies. Weber ascribed such power to charismatic political leaders, while Trotsky contended that the Russian Revolution could not have occurred without Lenin, even if no other participant in that revolution had been indispensable.


This text is based on a paper originally given in the Applications of Frame Analysis to the Study of Social Movements Panel at the Fourth International Conference of Social Science Methodology held at the University of Essex in July 1996.


BARTHES, Roland (1972) Mythologies. New York : Hill and Wang.

BENFORD, Robert D. (1994) 'Review of Talking Politics, by William A. Gamson' American Journal of Sociology, vol. 99, no. 4, pp. 1103 - 4.

BERGER, John (1972) Ways of Seeing. London: BBC and Penguin Books.

DONATI, Paolo R. (1992) 'Political Discourse Analysis' in Mario Diani and Ron Eyerman (editors) Studying Collective Action. London: Sage.

DONATI, Paolo R. (1994) A Primer for Political Discourse Analysis (With an Application to Sustainable Development) paper prepared for Dr. Carlo Ruzza, University of Essex, for Sustainable Development Research project.

DUERST-LAHTI, Georgia and Rita Mae Kelly (1995) (editors) Gender, Power, Leadership, and Governance. Ann Arbor: University of Michigan Press.

FISHER, Kimberly (1997) 'The Press and the Framing of Military Gender and Sexuality Policies in Britain and the United States'. PhD thesis, supervised by Dr. Carlo Ruzza, Professor Ken Plummer, and Professor A.P.M. Coxon. Department of Sociology, University of Essex, Wivenhoe Park, Colchester CO4 3SQ, United Kingdom.

GAMSON, William A. (1988) 'Political Discourse and Collective Action' in Bert Klandermans, Hanspeter Kriesi, and Sidney Tarrow (editors) International Social Movement Research, Volume 1. London: JAI Press.

GAMSON, William A. (1992) Talking Politics. Boston: Cambridge University Press.

GAMSON, William A. (1995) 'Constructing Social Protest' in Hank Johnston and Bert Klandermans (editors) Social Movements and Culture. London: UCL Press.

GAMSON, William A., David CROTEAU, William HOYNES and Theordore SASSON (1992) 'Media Images and the Social Construction of Reality', Annual Review of Sociology, vol. 18, pp. 373 - 393.

GAMSON, William A. and David STUART (1992) 'Media Discourse as a Symbolic Contest: The Bomb in Political Cartoons', Sociological Forum, vol. 7, no. 1, pp. 55 - 69.

GAMSON, William A. and Gadi WOLFSFELD (1993) 'Movements and Media as Interacting Systems' in Russell J. Dalton (special editor) The Annals of the American Academy of Political and Social Science: "Citizens, Protest and Democracy", no. 528:, pp. 114 - 125.

GIDDENS, Anthony (1989) Sociology. Cambridge: Polity Press.

GOFFMAN, Erving (1974) Frame Analysis: An Essay on the Organization of Experience. London: Harper and Row.

GOFFMAN, Erving (1979) Gender Advertisements. London: Macmillan.

HANNERZ, Ulf (1969) Soulside: Inquiries into Ghetto Culture and Community. New York: Columbia University Press.

HARLEY, Trevor A. (1995) The Psychology of Language: From Data to Theory Hove: Erlbaum.

JOHNSTON, Hank (1995) 'A Methodology for Frame Analysis: From Discourse to Cognitive Schemata' in Hank Johnston and Bert Klandermans (editors) Social Movements and Culture. London: UCL Press.

JUDD, Charles M., Eliot R. SMITH and Lousie H. KIDDER (1991) Research Methods in Social Relations: International Edition: Sixth Edition London: Harcourt Brace Jovanovich.

LAKOFF, George (1987) Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press.

LAKOFF, George, and Mark JOHNSON (1980) Metaphors We Live By. Chicago: University of Chicago Press.

LEVI-STRAUSS, Claude (1985) (translated by Joachim Neugroschel and Phoene Hoss) The View from Afar. Oxford: Blackwell.

MINISTER, Kristin (1991) 'A Feminist Frame for the Oral History Interview' in Sherna Berger Gluck and Daphne Patai (editors) Women's Words: The Feminist Practice of Oral History. London: Routledge.

MOSCOVICI, Serge (1971 (printed 1972)) 'Preface' in Serge Moscovici (editor) The Psychosociology of Language. Chicago: Markham Publishing.

MOSCOVICI, Serge (1984) 'The Phenomenon of Social Representations' in Robert M. Farr and Serge Moscovici (editors) Social Representations. London: Cambridge University Press.

REIN, Martin (1983) From Policy to Practice Armor, New York: ME Sharpe.

RUZZA, Carlo E. and Oliver SCHMIDTKE (1993) 'Roots of Success of the Lega Lombarda', West European Politics, vol. 16, no. 2, pp. 1 - 23.

SAKACH, Jamie (22 February, 1993) 'Clinton's Martial Plan', Newsweek, no. 121, p. 12.

SCHLESINGER, Philip (1991) Media, State and Nation: Political Violence and Collective Identities. London: Sage Publications.

SNOW, David A. and Robert D. BENFORD (1988) 'Ideology, Frame Resonance, and Participant Mobilization' in Bert Klandermans, Hanspeter Kriesi, and Sidney Tarrow (editors) International Social Movement Research: Volume 1. London: JAI Press.

SNOW, David A. and Robert D. BENFORD (1992) 'Master Frames and Cycles of Protest' in A. D. Morris and C. McClurg-Mueller (editors) Frontiers in Social Movement Theory. New Haven, CN: Yale University Press.

SNOW, David A., E. Burke ROCHFORD, Jr., Steven K. WORDEN and Robert D. BENFORD (1986) 'Frame Alignment Processes, Micromobilization and Movement Participation', American Sociological Review, vol. 51, no. 4, pp. 464 - 81.

STEVENSON, D. E. (20 July, 1946) 'The Four Graces', Woman's Weekly, no. 1811, pp. 65 - 9; 82 - 3.

SWANSON, Guy E. (July 1975) 'Erving Goffman - Frame Analysis' The Annals of the American Academy of Political and Social Science: Adjusting to Scarcity, no. 420, pp. 218 - 20.

SWIDLER, Ann (1986) 'Culture in Action', American Sociological Review, vol. 51, no. 2, pp. 273 - 86.

SWIDLER, Ann (1995) 'Cultural Power and Social Movements' in Hank Johnston and Bert Klandermans (editors) Social Movements and Culture. London: UCL Press.

TRIANDAFYLLIDOU, Anna (1995) 'Second Project Report: A Frame Analysis of Institutional Discourse', report produced for the Sustainable Development Research Project, directed by the Centre for European Social Research, Cork, Ireland, submitted to Dr. Carlo Ruzza, Professor Nigel Gilbert and Dr. Patt O'Mahoney.

van DIJK, Teun A. (1977) Text and Context Explorations in the Sematics and Pragmatics of Discourse. London: Longman.

van DIJK, Teun A. (1980) Macrostructures: An Interdisciplinary Study of Global Structures in Discourse, Interaction and Cognition. Hillsdale, NJ: Lawrence Erlbaum.

van DIJK, Teun A. (1985) Handbook of Discourse Analysis. New York: Academic Press.

WASBURN, Philo C. (1994) 'Review of William A. Gamson Talking Politics', Social Forces, vol. 72, no. 3, pp. 923 - 4.

WEST, Candace (1996) 'Goffman in Feminist Perspective', Sociological Perspectives, vol. 39. no. 3, pp. 353 - 369.

Copyright Sociological Research Online, 1997