INTRODUCTION

The attribution of classic status to any sociological theory is never settled, never final. In the English-speaking world in the early decades of this century, for example, the theories of Durkheim were readily dismissed for supposedly exemplifying the" group mind fallacy." This interpretation was a familiar one at least until the substantial reassessments of Parsons (1937) and Alpert (1939). Thus classic status is often hard-won and rarely secure. Moreover the qualifying criteria, in a multiparadigmatic discipline, are enormously variable. Some commonalities, however, can be suggested. According to Murray S. Davis (1986) what makes a sociological theory "classic" is that it seductively denies aspects of its audience's commonplace assumptions about the world (for a nihilist version of this definition see Travers 1989). Davis identifies features of the rhetoric of Marx, Weber, Durkheim, Simmel and Freud which serve to subvert audience assumptions and thus earn them classic status 1 but he emphasises that this rhetoric may not endure and that other (and no less valid) rhetorics to justify classic status may be extracted from the analysis of more contemporary theories. In this paper I wish take up this latter suggestion and consider those aspects of Goffman's Gender Advertisements(henceforth GA) that might qualify it as a visual sociology classic 2. This affords the opportunity to review the book and its fate two decades on from its publication.

GA is perhaps Goffman's most singular work. Within the context of Goffman's oeuvre, the 11 books and numerous papers which Goffman published between 1951 and his premature death in 1982, it is distinctive in that it is the only place where he addresses a single body of data: a collection of some 500 advertising and news photographs. Within the context of the discipline of sociology, it represents a rare and exemplary instance of an empirical study which treats photographic materials as data, worthy of analysis in their own right, and not merely a handy illustrative resource intended only to vivify the serious business of analysis accomplished by the written text (Ball & Smith, 1992). Album-sized, GA simply looks different from any other book readily describable as a work of sociology. Exactly two-thirds (56) of its 84 pages consist of arrays of photographs. The pictures are arranged in sets and each set is accompanied by a brief commentary. The pictures are set out in columns and meant to be looked at in the same manner in which newspaper columns are read. It is also significant that GA is alone among Goffman's books published in his own lifetime that is prefaced by a commentary authored by another writer 3. The US edition features an "Introduction" by Vivian Gornick ("What Erving Goffman shares with contemporary feminists is the felt conviction that beneath the surface of ordinary social behaviour innumerable small murders of the mind and spirit take place daily" [Gornick, 1979: ix]; see also West 1994) whilst the early British imprints contain a "Foreword" by Richard Hoggart ("this brilliant, suggestive book" [Hoggart 1979: viii]). The provision of such prefatory material seems to signal a publishing" event."

But coffee table sociology it most assuredly is not. The first two written sections of the book contain as dense (in both senses) and as extensive an exposition of EG's analytic preoccupations and methodological reasoning as is to be found in any of his writings. Chapter One establishes the principal features of his" special concern", namely gender display, the culturally conventional portrayals of sex-class membership ordinarily available and noticeable to society members at a glance. Chapter Two offers a lengthy disquisition on the varying senses in which pictures can be said to" really" depict their referent 4. The book versions of GA were published simultaneously in the UK and USA in 1979 and differ little from the original journal version of 1976 5. It is worth remarking that the paper quality is poorer in the book, which compromises the reproduction of some of the darker images. The glossy paper of the 1976 journal publication reproduces black and white tones more effectively -- important when one remembers that many of the advertising images in GA will have appeared in colour in their original textual sites.

The publication of the book in the UK occasioned some controversy. The first imprint of the UK edition used a cover photograph featuring two female models posed in a manner contrived to be alluring to the male gaze. It provoked one British reviewer to speak of the" offensively misleading cover of Gender Advertisements"(Kuhn, 1980: 316). Another (Hunt 1980) observed that the photograph was a glaring example of" the use of women as sex-objects to promote the sale" of the book. She continued:" What are we to make of it? Has Goffman or his publisher picked up some useful hints in this study of the advertiser's trade?" (Hunt, 1980: 443). Goffman apparently did recognise that the picture exploited the very matter the text was meant to criticise. He insisted that GA was concerned with analysing the merchandising of culture, not aiding and abetting the process. Of course, a book cannot be judged by its cover, but a raw nerve in the politics of representation had evidently been touched and subsequent printings of the UK edition used the more innocuous US jacket.

GENDER DIFFERENTIATION AND THE INTERACTION ORDER

In GA the sociologist of the interaction order -- that domain of social life generated by the copresence of persons (Goffman, 1983; Rawls, 1987) --investigates interactional manifestations of gender difference. The book's focus on gender displays extends and particularises the earlier and controversial" institutional reflexivity" 6 theory of gender differentiation (Goffman, 1977; Wedel, 1978). Briefly, Goffman's claim is that the differential treatment of males and females is often justified by folk beliefs which presume some essential biological differences between the sexes. But for Goffman biology cannot determine social practices which have to be treated as sui generis for sociological purposes. Therefore many social practices, frequently presented and excused as natural consequences of the differences between the sexes, are actually the means through which those self-same differences are honoured and produced. Biology is not an external constraint upon social organisation. Gender differentiation, at least in modern industrial societies, is produced and reproduced in interaction. Further, these interactional practices hold definite implications for the presumed human natures of gendered persons. So practices which ostensibly reflect consequential biologically-based differences in our human natures, everything from the engendering of pronouns and first names in European languages to the segregation of toilet facilities in public places, come in Goffman's view to constitute the differences between the presumed natures of the sexes. Goffman's institutional reflexivity theory is thus primarily concerned with gender differentiation rather than stratification, approached from a social constructionist standpoint.

The critique of common sense biological thinking about gender (as well as popular ethology) is taken further in GA. Gender displays are most emphatically not to be regarded as residues or remnants of the evolutionary development of the human species, nor are they" natural expressions" of our supposedly "essential" nature as men and women. Instead, Goffman contends that" there is only a schedule for the portrayal of gender ... only evidence of the practice between the sexes of choreographing behaviourally a portrait of relationship" (1979: 8; see also Clough, 1992: 102-107). Persons as gendered agents enact an appropriate schedule of gender displays. Nor are the displays to be treated simply as part of the froth of social existence: in the hierarchical relations between the sexes they are" the shadow and the substance" (Goffman, 1979: 6) of gendered social life. Gender displays serve to affirm basic social arrangements (keeping women in their place) and they present ultimate conceptions of the nature of persons (our" essential" gender identity). These displays are suffused with a behavioural vocabulary typical of parent-child relationships. The" orientation license", "protective intercession"," benign control" and "non-person treatment" which parents ideally extend to children also serves as a model which characterises the socially situated treatment of adult women by men. Thus," ritually speaking, females are equivalent to subordinate males and both are equivalent to children" (1979: 5).

The largest and in many ways most significant part of GA is devoted to a" pictorial pattern analysis" (1979: 25) of the presentation of gender (and femininity in particular) in advertisements. Goffman undertakes to describe some principles of gender display in contemporary Anglo-American society: relative size, the feminine touch, function ranking, the family, the ritualization of subordination, licensed withdrawal. The use of collections of photographs has the considerable advantage of allowing subtle features of gender displays to be exhibited, not merely described. The persuasive force of this analytical strategy is considered further below. It depends on the way empirical materials function as illustrations of an analytic theme. At first sight the pictures in GA appear to have a broadly equivalent function to transcripts in conversation analysis. They seem, like transcripts, to allow readers the opportunity to assess the adequacy of the interpretations presented by Goffman, to see how far his reading of the pictures works for us. But in fact the procedure is much more one-sided.

PICTORIAL PATTERN ANALYSIS: HOW IT WORKS

A unique feature of GA is the format of the long pictorial section. Arrays of numbered pictures are accompanied by an understated interpretive commentary which gives each page a distinctive look. Goffman's procedure is to first present us with his written observations about a particular gender display. These observations are then followed by a series of advertising images which" illustrate" the themes earlier articulated in words. The pictures are" arranged to be 'read' from top to bottom, column to column, across the page" (Goffman, 1979:26). Sometimes the series is concluded with exceptions ("sex role reversals") which presumably prove the rule. These exceptions are identified by black edging surrounding the picture 7.

The reader thus has to engage in a kind of search procedure, visually scrutinising each series for evidence of the gender display Goffman has just described in words. The reader scans the series of pictures looking for a family resemblance in the collection and, to the extent that the reader finds the resemblance Goffman has indicated, the written description is corroborated visually. The process involves something more than just giving empirical reference to the written observation. As readers we seem to employ what Garfinkel (1967) calls the documentary method of interpretation. Making sense is a two-tiered process. The upper tier consists of surface particulars, the lower tier the presumed underlying structure or pattern indexed and developed by surface particulars. In the case of GA, surface particulars are provided by the words and pictures of Goffman's text. The underlying pattern is the sense we arrive at about" mock assault games" or" body clowning" from reading Goffman's words and looking at his collections of pictures in that way. (See pictures 224-243 [Goffman, 1979:52-53] for" mock assault games" and pictures 207-216 [Goffman, 1979:50] for" body clowning" 8 ). Our understanding of the features of any given gender display is thus built up through this to-and-fro process of mutual elaboration of surface particulars and underlying pattern.

The procedure is a powerful persuasive device which makes it difficult for the reader to resist the interpretation which Goffman seeks to obtain from the arrays. In an examination of the use of ethnographic and other case materials in Goffman's earlier writings, Watson (1987, 1989) argues that he provides an "instructed reading" which very effectively transforms the reader's endogenous understandings of these illustrations to secure his analytic points. A kind of gestalt switch is thereby achieved in which the reader's understanding of an illustration is altered in the direction indicated by Goffman's analytic schema. A broadly similar procedure appears to be at work in GA; we could say that the reader engages in" instructed viewing." The difference is that the illustrations are presented visually rather than verbally and it is the visual status of the illustrations which makes the procedure even more effective. Goffman is so successful in exploiting the connotative penumbra surrounding the visual data precisely because the laconic text and concatenations (cf. Barthes, 1977) of pictures actively require readers' and lookers' work for the analysis to emerge. For Watson, like other ethnomethodologists (e.g. Schegloff, 1988), the overlooking of endogenous understandings represents a regrettable loss of a significant topic for sociological inquiry. However for Goffman it is precisely the development of a new perception, generated by the alignment of the written text and the concatenation of pictures, that yields a fresh understanding of the underlying, taken-for-granted features of gender codes.

A CANDIDATE CLASSIC?

Citation counts alone cannot adequately convey the classic potential of GA 9 (see also Winkin 1990). Furthermore, a full assessment of GA would need to address the larger issue of the overall worth of Goffman's sociology, and it has to be recognised that some are sceptical 10. If GA has any claim on classic status then an estimation of its fertility and novelty must be essayed. At least three questions need to be addressed in order to provide such an assessment:

1. can the ideas of GA be incorporated into systematic investigations of communication content?
2. can the analytic framework of GA itself be extended?
3. what are the prospects for developing Goffman's work on gender display beyond the non-vocal realm?

CONTENT ANALYSIS AND GA

Content analysis is the foremost systematic method for the investigation of communicative data in the human sciences (Ball & Smith, 1992: 20-31). One very simple use of GA by content analysts has been to sensitize researchers (and even beginning students; Jones, 1991) to everyday forms of gender dominance and subordination, as in Provenzo's (1991) analysis of video game characters. However a major difficulty in adopting Goffman's gender display categories is that they do not meet two of the essential coding requirements of standard forms of content analysis: that categories are mutually exclusive, and that the system of categories is designed to be exhaustive of all the content under investigation. It seems that a given advertisement can readily support two or more of the genderisms Goffman identifies. Take the example of picture 343 in GA(1979:64): Goffman classifies it as an example of licensed withdrawal via the withdrawal of gaze but it could just as readily exemplify the bashful knee bend, a type of ritualization of subordination (Moore, 1990:12). One solution is for the content analyst is to seek to identify the" dominant" (Moore, 1990:11) gender display to satisfy the mutually exclusive categorization requirement. Another is for the researcher to count as many genderisms as are readily identifiable in a given corpus of documents (Belknap and Leonard, 1991). The latter procedure breaches the canons of conventional content analysis (Berelson, 1952) but it does give a rough indication of the prevalence six main genderisms identified by Goffman. Moreover, adopting this procedure leads Belknap and Leonard (1991: 115) to the counterintuitive finding that Goffman's genderisms are nowadays more prevalent in advertisements in "modern" magazines (Ms, Gentlemen's Quarterly, Rolling Stone) than in" traditional" ones (Good Housekeeping, Sports Illustrated, Time). Berelson would not have endorsed this procedure, and Goffman (who was taught content analysis by Berelson) is wise enough to concede the point (Goffman 1979: 24-25).

An important step forward is taken by Mooney, Brabant and Moran's (1993) content analysis of birthday cards. Apparently insignificant, birthday cards serve as personally consequential" ceremonial tokens" which are differentiated along age and gender lines. Mooney, Brabant and Moran recognise that Goffman's gender display categories themselves require operationalization if they are to be effectively used in content analysis, and the study inventively devises indicators which draw upon both the visual imagery and verbal content of birthday cards. In so proceeding the study finds much evidence of three of Goffman's categories, the feminine touch, the ritualization of subordination and licensed withdrawal, leading the authors to conclude that" birthday cards convey messages of the subordination and devaluation of women and children in ceremonial activities" (Mooney, Brabant and Moran, 1993:626). The understated methodological message of this study is that the Goffman's gender display categories are primarily analytic categories which may need to be further operationalized if they are to retain their cogency.

These difficulties underline a general feature of Goffman's work: it is stronger on conceptual inspiration than methodological guidance. This can be seen clearly in Alexander's (1994) careful plotting of changes in the portrayal of children in twentieth century US magazine advertisements. Alexander's acknowledged point of departure is Goffman's suggestion about historical shifts in conventions for depicting dominance in family groups. Her analysis of these changes, however, employs the established methods of content analysis and categories unconnected to Goffman's. Dines-Levy (1990) takes a broadly similar line. Dines-Levy finds Goffman's thinking about the gendered properties of the" glimpsed world" helpful in the construction of an analytic framework for interpreting cartoon imagery, but the content analysis of cartoons owes little to Goffman. For these researchers, unconcerned with Goffman's own circumscribed preoccupation with the interaction order, his analysis of gender display is a place to begin inquiries which then extend beyond the confines of face-to-face interaction (Tseelon, 1995).

EXTENDING THE ANALYTIC FRAMEWORK

As conceptual innovation is so central to Goffman's sociology of the interaction order, another means of developing the analysis initiated by GA is to identify new forms of gender display. Chadwick (1988: 62-63) has a stab at this with his" woman as enigma" proposal: in a head and shoulders picture the female is posed fully frontal or slightly profiled, gaze aligned to the viewer, and lips closed or slightly parted. This genderism is expressed primarily through the eyes which wear a" mysterious" or "inviting" look, suggesting depths behind the outward visage. It is a conventionalised expression, Chadwick suggests, which is infrequently enacted by male models. Just how durable and widespread this gender display is awaits further investigation, as does the exploration of advertising representations of masculinity which would rectify the bias towards feminine imagery in Goffman's study. Few researchers seem to have taken up these challenges in the terms outlined by Goffman; he may well be regarded as a hard act to follow.

DEVELOPING GOFFMAN'S WORK ON GENDER BEYOND THE NON-VOCAL REALM

Some of the theoretical ideas of GA have been applied with profit outside the domain of non-vocal conduct. The "parent-child complex" has been used by West and Zimmerman (1977) to shed light on the gender patterning of interruptions in conversation. Hochschild (1990) echoes Wedel's (1978) earlier critique of Goffman's conception of feminity as archaic and monolithic, but uses that critique productively to identify two gender codes ("traditional" and" modern") in women's advice manuals. And despite the predominately empirical direction of Goffman's sociological interests, GA has also served as a resource for further theoretical disquisition. In the Derridean gloss supplied by Clough (1992: 107), gender displays are suffused with" an oedipal logic of realist narrativity" that is consonant with patriarchal capitalism's disavowal of desire and which portends the end of ethnography as we have come to know it.

In a number of works Gardner (1980; 1988; 1989; 1990; 1995) has explored the gendered character of experience of public places using an analytic framework originating in Goffman's sociology but which critiques, revises and develops it in significant ways. This body of work provides vivid illustration of Hochschild's (1990:278) observation that" in his analysis of gender Goffman did not use all of 'Goffman'" but it also leaves Goffman exposed to the charge that he failed to examine the interactionally real features of women's disadvantage -- that the substance escaped while he analyzed the shadow.

These studies indicate the considerable fertility of Goffman's thinking on gender, even from those who may regret that a more overtly feminist message was not drawn from data that would support it. But GA is also a novel work of contemporary sociology. Its principal claim to classic status resides in the way the pictorial pattern analysis induces the reader to see the world as Goffman describes it. The singular contribution of GA to a sociological understanding of gender difference centres on the effect produced by the interaction between text as caption and artfully sequenced arrays of photographs which co-opt the reader in a vivid and compelling way. This is a unique achievement, so far without substantial sociological issue, but on the basis of Goffman's increasing salience as a social theorist (Burns, 1992; Manning, 1992; Smith, forthcoming; Smith [ed] forthcoming) it is likely to inspire sociologists to become more adept at analysing the vast repositories of visual data in postmodern culture.