The audience is one of the central elements of media studies. Yet, the term connotes many different, sometimes contradictory things. This article surveys the field of audience studies, noting common concerns in otherwise disparate bodies of research and theory. It identifies three basic models of the audience: audience-as-mass, audience-as-outcome, and audience-as-agent. It concludes with a number of recommendations for future work on the audience.

For over a century, media audiences have been a concern of entrepreneurs and social critics alike. Audiences are the raison d'etre for mass media organizations. In fact, it is hard to imagine any form of media studies that is not, on some level, about audiences. Under such circumstances one might assume that the concept of an audience was well understood and universally applied. As McQuail (1997) has noted, "It is a term that is understood by media practitioners and theorists ... and it has entered into everyday usage, recognized by media users as an unambiguous description of themselves" (p. 1). But this easy use of the term belies many different connotations. As we approach the next century, the idea of an audience is less settled than at any time in the past.

The tumult results from a combination of rapid technological changes reshaping the media landscape and shifting intellectual paradigms within universities. New media, the Internet, and a host of other developments have threatened to "upset the apple cart" of established practices in industry research and marketing. In academe, scholars of all stripes have argued for a reevaluation of the concept of audience. Some have even gone so far as to argue that audiences exist only as "discursive constructs." All these developments raise questions about the meaning and definition of audiences, and in turn, the proper scope of audience studies. Against this backdrop, it is useful to take stock of where we have been and reconsider where we should to be going.

Audience-as-Mass

Conceiving of the audience as a mass is the most common way to think about the media audience. Under this model, the audience is seen as a large collection of people scattered across time and space who act autonomously and have little or no immediate knowledge of one another. They are defined as an entity by their common exposure to media. This is similar in concept to Blumer's (1946) definition of the "mass" in social theory. Certainly if we include research done by and for media industries, and the many social institutions that have a stake in the audience, a case can be made that this is the "dominant model" of the audience (Webster & Phalen, 1997). As Bogart (1996) has observed, "The bulk of communication research is commercial research and is addressed to the question of measuring audiences, rather than to study of the process through which audiences reject or ingest the information presented to them" (p. 138).

The central question of this model is "What media do people consume?" Other concerns, such as how individuals make sense of the material they encounter, are of secondary importance. Under this model it is usually sufficient to know what media offerings have captured people's attention, how and why the audience moves from one encounter to the next, what they are likely to consume tomorrow, and over the long haul, how media must adapt if they are to maintain an audience. The usual way to answer these questions is by aggregating large numbers of people to reveal predictable patterns of response. In the process, a theoretical preference is given to macro-level determinants of exposure rather than dwelling on individual cases. The body of the audience--the mass--is of primary importance.

This way of thinking about the audience has its roots in the Industrial Revolution. It became necessary because modern audiences, unlike congregations, were dispersed and essentially invisible. The growth of a reading public, mass marketing, and the rise of "statistical thinking" Porter, 1986) set the stage for the idea of a mass audience. It was not until the 1930s, though, that the concept of a mass audience and its related research practices crystallized (Webster & Phalen, 1997). In that decade, we saw for the first time the widespread use of survey research to measure actual instances of exposure. We also felt the influence of Paul Lazarsfeld and what some have described as a "marketing" approach to communication studies (Chaffee & Hochheimer, 1985; Rogers, 1994). Today, the mass audience concept underlies most work in media marketing and audience segmentation.

There are many examples of this model in the theory and practice of audience studies. The whole structure of audience ratings research would seem to qualify (Beville, 1988; Webster & Lichty, 1991). The business of turning the audience into a commodity that can be bought and sold depends on invoking a kind of audience-as-mass mentality. This is true whether we consider mainstream economic analysis (e.g., Owen & Wildman, 1992) or the work of "political economists" (e.g., Mosco, 1996; Smythe, 1977). Formal attempts to model mass audience behavior occur in marketing, media planning, and programming (Barwise & Ehrenberg, 1988; McPhee, 1963; Rust, 1986). The model also has some applicability in studies of popular culture and media events, as these seem to achieve much of their cultural significance by attracting vast audiences (Dayan & Katz, 1992). The intersection of audience-as-mass and other models locates many more important bodies of research and theory, as described in the sections that follow.

Unfortunately, the word "mass" carries a good deal of excess baggage. For many it connotes passivity, susceptibility to influence, fickleness, or indiscriminate tastes (Williams, 1983). None of these seem essential to defining the term as it is used here. But even if its meaning is pared back to reflect actual research practice, it is still the subject of considerable criticism. It is faulted for being too behaviorist in orientation, too insensitive to the social context of media use, and at worst, a tool for the repression and colonization of the audience by institutional interests (e.g., Ang, 1991; Gandy, 1993). All of these criticisms have some merit and illustrate the limitations of any one approach to studying audiences.

Audience-as-Outcome

If we think of audience studies as a purely academic enterprise, one might conclude that a kind of "audience-as-outcome" model has dominated the field of media studies from the beginning. This way of thinking about the audience sees people as being acted upon by media. Typically, it reflects a concern about the power of media to produce detrimental effects on individuals, and by implication on society as a whole. Less frequently, it speaks to the pro-social potential of the media. When the action is taken for the audience's "own good" it has been referred to as an audience-as-public model. When the action is taken for some corporate purpose it has been labeled an audience-as-market model (Ang, 1991). Whether for good or ill, all such models position individuals in a fairly reactive role. In all cases, the defining concern of the model is expressed in the off-quoted question, "What do media do to people?"

Though social commentators have long worried about the ability of media to lead people astray, the formal academic study of media effects probably began in the late 1920s and early 30s with propaganda analysis and the Payne Fund studies of cinema (Cmiel, 1996; Lowery & DeFleur, 1987). For many social scientists, it continues to be of central importance (Bryant & Zillmann, 1994). Even among those who question this way of thinking about the audience, there is widespread recognition that the study of effects helped legitimize communication as an academic discipline (Jenzen & Rosengren, 1990).

As one might imagine, there is a large quasi-scientific literature that takes an audience-as-outcome perspective. Decades of experimental research on persuasion and attitude change or the effects of violence and pornography are illustrative (Bryant & Zillmann, 1994). The work of Carl Hovland, one of the "founding fathers" of communication studies, has been characterized in the following way:

The audience of persuasive communication was conceptualized by Hovland as a passive receiver of the information contained in the message. In conjecturing how some variable (e.g., source credibility, level of fear appeal, or time passage) affected the attitude-change impact of the message, Hovland would think in terms of how it affected the mediating information-encoding steps such as attention, comprehension, and agreement. That he used this passive audience concept is not surprising considering Hovland's background in the 1933-1942 period when his interest focused on memorization and rote learning. (McGuire, 1996, p. 55)

This way of thinking implies a "transmission" model of communication (Carey, 1989) and, according to McGuire, remains prevalent in social psychology even today. A similar mind-set underlies many government policies, from regulating indecency to "equal time" for political candidates (Webster & Phalen, 1994). There are also strains of technological determinism that focus on the properties of a medium rather than specific items of content as a cause of various effects (Kubey & Csikszentmihalyi, 1990; McLuhan, 1964; Meyrowitz, 1985).

But the audience-as-outcome model is much broader than the social science of effects and its various applications in industry or government. Some forms of literary criticism and film theory adopt a similar posture in regard to audiences. "Screen theory," which rose to prominence during the 1970s, is a case in point. According to Moores (1993), "'Screen theory' was an application of ... Lacanian insights to an analysis of film as discourse. The aim was to uncover the symbolic mechanisms through which cinematic texts confer subjectivity upon readers, sewing them into the film narrative through the production of subject positions" (p. 13). In a nutshell, screen theory and related modes of critical thinking argue that viewers became unwitting accomplices in the production of meaning as orchestrated by the text. Because all viewers participated in this so-called "subject effect" it had the potential to produce widespread ideological consequences, though that was rarely demonstrated empirically.

Obviously one can advance the audience-as-outcome model while invoking a wide range of theories and methods. This makes any uniform critique quite difficult. Some feel uncomfortable with the "administrative" applications of much of the work. Others find the potential for manipulation disquieting. Perhaps the most fundamental criticism of the model is that it positions audience members in a passive role, unaware of how the media act upon them. Not only is this a rather pessimistic view of individuals, it flies in the face of recent trends in communication theory. In fact the third and final model can be seen, in part, as a reaction against the audience-as-outcome.

Audience-as-Agent

Since the early 1970s, there has been growing interest within the academy in an alternative conception of the audience. Rather than seeing people as acted upon by media, people are conceived of as free agents choosing what media they will consume, bringing their own interpretive skills to the texts they encounter, making their own meanings, and generally using media to suit themselves. Under this model, the audience acts upon the media, not vice versa. This way of thinking about the audience takes as its central concern the question "What do people do with media?" One should hasten to add that this positioning of the audience does not mean that people are free of outside influences. Proponents of this model typically recognize that audience actions are somehow determined by their social and cultural milieu. In fact, in many applications of this model audience actions are deemed to be uninterpretable without reference to some broader structure. But in the interface between audience and media, it is the individual viewer who has the upper hand. Controlling this engagement is one way in which audiences are thought to exercise power.

Like the previous model, the audience-as-agent model is an amalgam of theoretical perspectives. Social scientists who have adopted this approach share a history with effects researchers. Others who argue for the application of an "interpretive paradigm" find their roots in hermeneutics, phenomenology, and the work of several 19th and 20th century philosophers (Lindlof, 1995). All told, the audience-as-agent model has attracted many adherents in communication studies. In fact, a case can be made that it now represents a kind of majority opinion within the academy. As Bryant and Street (1988) observed:

The notion of the "active communicator" is rapidly achieving preeminent status in the communication discipline. In the mass and interpersonal literatures alike, we read statement after statement claiming that today's message receivers have abundant message options and actively select from and act on these messages. (p. 162)

Studies of "selective processes," which are often associated with the audience-as-outcome model, actually credit audience members with a considerable capacity to make their own meanings (Schramm, 1955; Zillmann & Bryant, 1985). More recently, proponents of "uses and gratifications" research have posited individual needs as the driving force behind people's consumption of the media (Katz, Blumler, & Gurevitch, 1974; Rosengren, Wenner, & Palmgreen, 1985). Quantitative methods of one sort or another have been standard operating procedure in such investigations. By the early 1980s, qualitative modes of investigation also became popular. Most noteworthy are the audience ethnographies of Lull (1980), Morley (1980), and other proponents of "British cultural studies" (Moores, 1993). Beyond the immediate parameters of mass media, one also finds academic work on reader response theory and interpretive communities (Fish, 1980; Radway, 1991) as more or less in step with an audience-as-agent model.

Criticisms of this model run the gamut. Work of a more traditional quantitative sort is faulted for being too mentalistic, too administrative in its orientation, and too susceptible to the pitfalls of logical positivism. Qualitative studies of the audience are often ungeneralizable since each person's construction of the text and its meaning is seen as unique and "incommensurable" (Lindlof, 1987). Further, some advocates of cultural studies abandon empiricism altogether, preferring instead to cite personal experience or the reactions of some imputed reader. These features of interpretive studies make them of limited value in administrative application--a source of pride for some of their proponents (Ang, 1991). No matter the method of investigation, all approaches are occasionally guilty of making outlandish claims of audience power, leading to charges of "vulgar gratificationism" (Blumler, Gurevitch, & Katz, 1985).

Mixed Models

A number of important traditions in audience studies do not fall neatly within the three models outlined thus far. Rather, they exist on the margins where those models overlap. By definition, they offer more complex conceptions of the audience, as they embody attributes of two or more basic types. Each area of intersection merits some comment.

The combination of audience-as-mass and audience-as-outcome is rich with both applied and theoretical constructions of the audience. The various manifestations of mass marketing (e.g., advertising, public relations, public opinion research, etc.) can be found at this juncture. All conceptualize audiences as atomized masses, but they do so as a step along the way to understanding and managing various outcomes (i.e., to know what media do to people, one must first know what media they consume). Mass society theory, often attributed to the "Frankfurt School," bears significant similarities. The audience is seen as a mass of isolated individuals who are inherently susceptible to manipulation, though they may fail to recognize the hegemonic processes at work (Adorno & Horkheimer, 1977; Beniger, 1987). More recent theories like "cultivation analysis" (Gerbner & Gross, 1976) and the "spiral of silence" (Noelle-Neumann, 1993) employ a similar framework. As Katz (1987) noted:

These come, respectively from the Left and the Right of the political spectrum (which means that a case can be made that the Right can also have a critical theory). Both studies assume a classical mass society in which the individual is atomized, locked into his home for fear of going out (Gerbner), or locked into silence for fear of being ostracized (Noelle-Neumann). (p. S32)

The mass is not just a kind of aggregated extension of the audience-as-outcome model. It interacts equally well with the audience-as-agent model. Examples of this intersection include a range of work on markets and audience preferences. In economics we find a half century of theorizing about "models of program choice" (Owen & Wildman, 1992; Steiner, 1952). More pragmatic work in market research is cut from the same cloth (Frank & Greenberg, 1980; Rust, Kamakura, & Alpert, 1992). Both assume that individual preferences drive program choices, and that in the aggregate these choices form markets or audience segments to which industry will respond. This particular line of theorizing has had a tangible impact on American communications policy (Webster & Phalen, 1994). Other academic work less explicitly tied to economics employs a similar concept of audience (Cantor, 1994; Peterson, 1994).

The juncture between mass and agent also locates critical studies on the sociology of taste. Rather than focusing on the individual reader, this work is particularly sensitive to collectivities formed around patterns of taste and media consumption. Gans (1974) has described these notions:

Taste cultures are not cohesive value systems, and taste publics are not organized groups; the former are aggregates of similar values and usually but not always similar content, and the latter are aggregates of people with usually but not always similar values making similar choices from the available offerings of culture. Moreover, they are analytical aggregates which are constructed by the social researcher, rather than real aggregates which perceive themselves as such. (p. 69-70)

Like work in neoclassical economics and marketing, this sort of enterprise ascribes considerable importance to choice making en masse. Unlike more administrative perspectives, however, it is much more sensitive to the social and cultural determinants of taste (Bourdieu, 1984; Lewis, 1981), and is less likely to be satisfied with the simplifying assumption that peoples' media preferences exist a priori (Gandy, 1992).

A particularly rich vein of work crops up at the juncture of outcome and agent. In their most extreme versions, these basic models portray the audience in absolutely contradictory terms. Audience-as-outcome sees media acting upon individuals in myriad ways. Hence people are labeled "passive." Audience-as-agent views people acting upon media. Hence they are "active." A number of thoughtful, long-standing bodies of work reject this either/or dichotomy. Instead, the audience is most often located between the extremes, varying from time to time or person to person (Blumler, 1979). The best known of these approaches is the limited effects perspective, sometimes called the "dominant paradigm" of media sociology (Gitlin, 1978). It was deliberately posed as an alternative to mass society theory. Instead of seeing an audience of free floating atoms, people were conceptualized as members of social networks. Instead of being vulnerable to direct media manipulation, people actively select and reinterpret the messages they encounter (Katz & Lazarsfeld, 1955; Klapper, 1960). The limited effects perspective never denied the possibility people could be manipulated, but it did credit them with offering considerable resistance.

Studies of a more interpretive ilk also locate the audience somewhere between outcome and agent. An early example of this kind of thinking is symbolic interactionism, commonly associated with the "Chicago School" of sociology (Blumer, 1969; Mead, 1934). Here, media are seen to contain "significant symbols" that carry meaning. Yet these meanings are not self-evident: they must be learned. So the individual has a role to play in deciphering media. Each person is at once actor and acted upon. Some of the more balanced examples of cultural studies also put audience members at the juncture of outcome and agency. Stuart Hall's (1980) widely cited piece on audience decodings recognizes that while texts may have certain encoded meanings, individuals are nonetheless capable of "negotiating" those meanings. More recent work on reception analysis and social semiotics locates the audience in a similar way (Jensen, 1995).

The last margin to consider is the overlap of all three circles. By definition, this space locates very complicated notions of audience. Perhaps this is simply a trap of the Venn diagram--an empty space that should remain unfilled. I think, however, there are a few constructions of audience that belong here. None is unproblematic. Each is, in its own way, a challenge to students of the audience.

The "marketplace of ideas" is one of the most popular and potent metaphors in communications law and policy. Rooted in "First Amendment theory" and the writing of 18th century philosophers such as John Stuart Mill, it implies an audience interested in issues of public importance and inclined to debate them in some public forum. This mechanism is thought to promote diversity of expression, social cohesion, and the discovery of "truth." The "public sphere" is a similar construct (Habermas, 1991). It too implies an audience engaged in rational debate over issues of common concern, finding historical precedents in the salons of 18th century France, the coffeehouses and social clubs of 19th century Western Europe, and the town meetings of New England (Calhoun, 1992).

In contemporary application to mass media, the marketplace of ideas has been contested by analysts espousing either a "social value" or a "market economics" perspective (Entman & Wildman, 1992). These schools of thought generally comport with the audience-as-outcome and audience-as-agent models, respectively. The former sees the audience as being acted upon in ways that are potentially detrimental to society, justifying government intervention. The latter credits that audience with being able to act in its own self interest if only it is left free to choose an unrestricted diet of media offerings (Webster & Phalen, 1994).

But it is not enough to locate the marketplace or the public sphere as wavering between the poles of agent and outcome. To be relevant in a modern age, it must somehow reckon with the mass audience. As Habermas (1974) himself noted:

Citizens behave as a public body when they confer in an unrestricted fashion--that is, with the guarantee of freedom of assembly and association and the freedom to express and publish their opinions--about matters of general interest. In a large public body this kind of communication requires specific means for transmitting information and influencing those who receive it. Today newspapers and magazines, radio and television are the media of the public sphere. (p. 49)

Despite recognizing the importance of mass media, the constructs of public sphere and marketplace are troublesome because they are rooted in models of interpersonal communication. As Garnham noted, "The initial theory and subsequent related ideologies are based on face-to-face communication in a single physical space" (1992,p. 365). In such a world there is equality of access. Proponents of different ideas can thrash them out before an assembly of engaged listeners. This is hardly a model of the media audience. While there have been attempts to update the marketplace metaphor, just what role the mass audience is to play in the workings of the public sphere remains unclear (Entman & Wildman, 1992; Peters, 1995).

Finally, I am tempted to locate various "postmodern" notions of audience in the three-way space of outcome, agent, and mass. There is admittedly some danger in even attempting to pigeonhole such an elusive body of work--one that characterizes itself through its contradictions. On one hand, audience members are portrayed as skilled readers of culture and makers of meaning beyond ideological manipulation. On the other hand, they are consumed by the modern media systems of which they are a part. "The rapidly expanding number, diversity, and pace of these communications overwhelms the individual's ability to interpret their meaning rationally" (Harms & Dickens, 1996, p. 216). At least some postmodern theorists have also taken to conceiving of the audience as mass, in a surprisingly literal way. The audience is seen as a social center of gravity, a "black hole," operating by its own set of rules (Baudrillard, 1983). Interestingly, this concept of the mass audience is not worlds apart from more conventional notions of the mass as a dynamic entity in its own right. As Webster & Phalen (1997) have noted, "In a very real sense, the mass takes on a life of its own, at some distance from the individuals who form it" (p. 23). Consequently, postmodernism invokes elements of all three of the basic models.

Exploit the interdisciplinary potential of audience studies

To observe that an interest in audiences occurs across disciplines is not to observe that individual studies of the audience are interdisciplinary in character. More typically, each has a specific lineage that turns a blind eye to related work in other disciplines. As noted at the beginning of this piece, even in wide-ranging surveys of the field, the custom is to organize the material by reviewing "antagonistic" traditions (e.g., Lazarsfeld, 1941; Schroder, 1987), historically or theoretically discrete disciplines (e.g., Allor, 1988; Jensen & Rosengren, 1990), or multiple paradigms (e.g., Katz, 1987; Pietila, 1994). This has the effect of highlighting differences of epistemology, ideology, and nomenclature. We should, of course, be cognizant of these differences, but we could do a better job of identifying common concerns and exploiting the insights they offer.

That has been the logic of this article. How do different bodies of work position the audience in relation to the media text? What fundamental question is being asked of that relationship? This way of thinking tends to cut across heretofore distinct bodies of work. In the process are created some admittedly strange bedfellows. Effects research is grouped with film theory, selective processes are categorized with reader response, and economic models of program choice are classified with taste publics. The purpose is not to minimize their differences, but to suggest that audience studies with very different pedigrees have all been grappling, in their own way, with the same limited set of questions. Too often we ignore disparate disciplines as a source of insight and inspiration because they seem too foreign or old fashioned. Rather, audience studies have a penchant for proclaiming new paradigms. In the process, they tend to trivialize, misrepresent, or simply ignore what has gone before. The result is that students of the audience have a distressing tendency to "reinvent the wheel." Perhaps an interdisciplinary scheme of organization could help.

Employ an enlightened empiricism when studying audiences

Raymond Williams (1983) wrote, "There are in fact no masses; only ways of seeing people as masses" p. 300). The same might be said of audiences. They are at once real and imagined. All audience studies reduce people to a theoretical abstraction of one sort or another. In conventional audience research, people are typically sorted into a small number of categories (e.g., age, gender, income) and treated as if one person is the functional equivalent of everyone else in the category. Surveys of public opinion introduce additional variables, but take much the same approach. In gratificationist work, categories are also established to reflect theorized motivations and uses. These methods allow for a kind of "headcounting" and generalizability supported by inferential statistics.

Audience ethnography is sometimes touted as offering a richer, more authentic view of the audience--a way to understand the "lived reality behind the ratings" (Jensen, 1987, p. 25). But ethnography has its own limitations. As Anderson (1996) has noted, if it implies generalizable results it is "another way of expressing categorical aggregates, the difference being solely in the method of collecting information about them" (p. 84). If it makes no claim to generalizabilty, it asks us to consider a few idiosyncratic, ever changing "micro-situations." In a world of infinite variability and incommensurable meanings, it only moves us closer to the "truth" one person at a time. But it moves us nonetheless.

In some manifestations of film theory and cultural studies, however, the empirical referent is lost altogether. The "spectator" has become an idealized reader of texts. There is no pretense that the audience is anything other than an imagined role player. The fabricated quality of the audience has caused some to argue that the audience "exists nowhere; it inhabits no real space, only positions within analytical discourses" (Allor, 1988, p. 228). While this posture may set the stage for interesting theorizing, it goes too far. It runs the risk of having us dispense with empiricism altogether. As Lull (1988) observed in response to Allor:

What is presented in much cultural studies writing about audiences is actually the writer's position, his or her relation to media content, to the family, and the social environment. The logic of the theoretical argument and the style of presentation are given more weight than descriptions and grounded interpretations of what audiences really think and do. (p. 240)

Just because audience studies inevitably present an incomplete picture of the audience does not mean that audiences are unreal. When it suits them, analysts from every point on the methodological compass make reference to an "actual audience." Most would agree that there are real human beings out there who constitute the audience. Studying the actual audience, and the actual institutions that serve it, can and should offer a check against unfettered theorizing. What is needed is a kind of enlightened empiricism--one that makes room for a number of methods, each compensating for the limitations of the other--one that compels analysts to go into the real world, recognizing that audiences are never completely knowable.

Consider audiences as collectivities as well as individuals

The audience is most often conceptualized as "a large, loosely connected mass on the receiving end of the media" (Webster & Phalen, 1997, p.1). Certainly, this is true if we include the day-to-day practices of media institutions. It is surprising, therefore, to discover how little academic theorists have to say about the audience as a larger body of people. Instead, attendance to "mass communication" is reduced to a matter of individual response--it is about the experimental subject, the lone reader, or the individual operating within the intimate bounds of family and group membership. What happens beyond the micro-level of analysis is often lost in the shuffle. As Blumler (1996) has observed, "Most empirical research into mass media communicators' audience images has been microscopic--more productive of closeups than panoramas" (p. 99).

This bias toward the individual is hardly unique to audience studies. Giddens (1987) has noted, "The human agent is treated as the prime focus of social analysis. That is to say, the main concern of the social sciences is held to be the purposeful, reasoning actor" (p. 59). Similarly, McPhee (1963) complained some years ago that "most of our modern social research is irretrievably microscopic; that is it is only about individuals in the mass" (p. 4). Unfortunately, it is not sufficient to conceptualize the mass and its consequences as the simple summation of individual responses. As noted above, audiences have characteristics that are invisible at lower levels of analysis. How are these collectivities formed? Do audiences result from some underlying social structure or are they manufactured by the media? In an age when technology seems to empower ever greater freedom of choice, these are particularly pressing questions.

There is much more to be learned by considering the audience as a social force in its own right. The process of aggregation seems to empower audiences. As Ettema & Whitney (1994) have argued, "Actual receivers are not powerless but ... they wield influence within the institution only when they have been constituted as some effective audience such as an identifiable and desirable market segment" (p. 11). This is critical to understanding the larger operation of media economics, popular culture, and mass marketing (e.g., Cantor, 1994; Owen & Wildman, 1992; Turow, 1997).

In keeping with the prescription for using enlightened empiricism we should also recognize that not all audience collectivities can or should be measured by conventional survey research. Anderson (1996) has noted that "encoded" audiences, embedded in industry lore, shape the conduct of media institutions. Webster & Phalen (1997) have argued that "presumed audiences" may explain many of the social effects of mass media. Similarly "imagined communities," formed in the first instance by print media, are thought to underlie the modern nation state (Anderson, 1991). All these constructions of the audience have real consequences and constitute additional forms of audience power. Though they transcend the usual parameters of the audience-as-mass, they seem ripe for further investigation.

Abandon the active/passive continuum in audience studies

No review of audience studies could be complete without addressing the subject of audience activity. Volumes of material have been written on the subject (e.g., Biocca, 1988; Hearn, 1989; Levy & Windahl, 1985; Rubin, 1993). Usually, the "active audience" is pitted against the conceptual alternative of a "passive audience." As we have seen, this polarity defines differences between the audience models of outcome and agent. Gratificationists have been the most outspoken in championing the concept of audience activity. As Levy and Windahl (1985) have noted:

As commonly understood by gratifications researchers, the term 'audience activity' postulates a voluntaristic and selective orientation by audiences toward the communication process. In brief, it suggests that media use is motivated by needs and goals that are defined by audience members themselves, and that active participation in the communication process may facilitate, limit, or otherwise influence the gratifications and effects associated with exposure. (p. 110)

But audience activity is a slippery concept. Nowhere is it more troublesome than in the assertion that people are aware of why they do what they do, and that they can provide reliable reports of the meanings and gratifications of media use (Katz, Blumler, & Gurevitch, 1974). This assumption of self-awareness allows too little room for the operation of the unconscious. Surely selective processes operate without an individual's full awareness. Zillmann and Bryant (1985) have pointed out, "People usually pay little attention to why they choose what they choose when they choose it" (p. 163). Similarly, the meanings readers ascribe to a text undoubtedly involve learned associations of which they are not fully cognizant. Moreover, researchers now seem inclined to define "cognitive effort" as a form of activity. Hawkins and Pingree (1986) have noted, "Recently, researchers have made some serious attempts to measure amount of cognitive effort, using operationalizations centering around attention, brain waves, reported amounts of thinking while viewing and the results of cognitive effort" (p. 239). If we allow for the possibility that "activity" might legitimately include mental gyrations of which the individual is not consciously aware, one is left to wonder what "passivity" could possibly mean. Biocca (1988) has pointed out that "In some extreme formulations of the active-passive audience dichotomy, only a corpse propped in front of a television set could be registered as a member of the much scorned 'passive audience'" (p. 75).

While there may be carefully delineated ways in which the terms "active" and "passive" can be of use, we should abandon them as the defining polarity of audience studies. It focuses too much on the micro-level questions of how the individual interacts with media texts. It undervalues the role of habit in audience behavior (e.g. Rosenstein & Grant, 1997). It focuses too little attention on the audience-as-mass by inviting simple-minded generalizations. For example, ever since the heyday of mass society theory, most academics have equated the mass with passivity. But, as Webster and Phalen (1997) have pointed out, "There is nothing in the concept of the mass as it was first articulated by Blumer (1946), or in contemporary advertising practice that requires us to position the audience as a bunch of passive dolts" (p. 117).

Abandoning the active-passive continuum would free us from a needlessly valueladen and unproductive way of thinking. This will not be easy. We have all adopted the vocabulary. It is in common use among the general public, and is deeply embedded in popular and scholarly literatures on the audience. Perhaps, at least for academics, an alternative framework would help us climb off this slippery slope.

Think of audience studies in terms of agency and structure

The alternative I have in mind is to think of audiences in terms of agency and structure. These concepts are familiar to students of sociology (Giddens, 1979) and are beginning to find a place in communication studies (e.g., Jensen, 1995; Mosco, 1996). While structure is sometimes cast as an external constraint on individual action, structure and agency can be mutually constituted. That is, while "human agents" are capable of acting freely, they are typically located within institutional and social structures that both shape and are shaped by individual actions. As Giddens (1987) has put it:

Structure is the very medium of the 'human' element of human agency. At the same time, agency is the medium of structure, which individuals routinely reproduce in the course of their activities ... Human beings normally know not only what they are doing at any moment, but why they are doing it. That is to say, it is characteristic of human agents that they routinely appraise what they do as a means of doing it, and that they are able discursively to give both an account of what they do and of their reasons for what they do ... But it does not follow that they know all there is to know about the consequences of what they do, for the activities of others or for their own activities in the future. Nor do they know all there is to know all about the conditions of their action, that is, the circumstances that are causally involved with its production. (p. 220-21)

This seems a profitable way to conceive of audiences. It incorporates much of the rhetoric of the active/passive continuum within the concept of agency. It recognizes that human beings are typically self-aware, but that they may not fully understand the causes and consequences of their actions. Structures, whether conceived of as group memberships, jobs, social classes, markets, or technological infrastructures, should be considered in understanding each person's use of media, as well as the formation and conduct of larger collectivities. For instance, Webster and Phalen (1997) have argued that the interplay of availability and media structures (e.g., channels, programming practices, etc.) offers the most telling way to understand mass audience behavior. If, for some reason, individual agents were to "act otherwise," institutional structures would adapt, supporting and promoting new patterns. Hence there is a reciprocity or, as Giddens (1987) puts it, a "duality of structure" (p. 220).

None of this overturns the fundamental models of audience with which we began. Indeed, one can find elements of this theoretical perspective in each. Being more explicit about agency and structure, however, might offer more sophisticated answers for the questions that have been raised. People can still be thought of as acting upon media. We can still ask "What media do people consume?" Likewise, we can ask "What do media do to people?" We recognize that in any of these matters people may not be mindful of all that they do, or all that is done to them. Similarly, audience members can be seen not only as individual agents but also as participants in the larger dynamics of social and institutional structure. All contribute to the creation and evolution of the audience.

DIAGRAM: Figure 1