ROGERS BRUBAKER
Ethnicity without groups
I. Common sensegroupism
Few SOCIAL science concepts would seem as basic, even indispensable, as that of group. In disciplinary terms, 'group' would appear to be a core concept for sociology, political science, anthropology, demography and social psychology. In substantive terms, it would seem to be fundamental to the study of political mobilization, cultural identity, economic interests, social class, status groups, collective action, kinship, gender, religion, ethnicity, race, multiculturalism, and minorities of every kind.
Yet despite this seeming centrality, the concept 'group' has remained curiously unscrutinized in recent years. There is, to be sure, a substantial social psychological literature addressing the concept (Hamilton et al. 1998, McGrath 1984), but this has had little resonance outside that sub-discipline. Elsewhere in the social sciences, the recent literature addressing the concept 'group' is sparse, especially by comparison with the immense literature on such concepts as class, identity, gender, ethnicity, or multiculturalism—topics in which the concept 'group' is implicated, yet seldom analyzed its own terms (1). 'Group' functions as a seemingly unproblematic, taken-for-granted concept, apparently in no need of particular scrutiny or explication. As a result, we tend to take for granted not only the concept 'group', but also 'groups'—the putative things-in-the-world to which the concept refers.
Earlier versions of this paper were presented to the conference 'Facing Ethnic Conflicts', Center for Development Research, University of Bonn, December 14, 2000; the Working Group on Ethnicity and Nationalism, UCLA, January 13, 2001; the Anthropology Colloquium, University of Chicago, February 26, 2001; and the Central European University, Budapest, March 20, 2001. Thanks to participants in these events for their comments and criticisms, and to Margit Feischmidt, Jon Fox,
Liana Grancea, David Laitin, Mara Loveman, Emanuel ScheglofT, Peter Stamatov, Peter Waldmann, and Andreas Wimmer for helpful written comments,
(r) Foundational discussions include Coo-ley 1962 [1909], chapter 3 and Homans 1950 in sociology; Nadel 1957, chapter 7 in anthropology; Bentley 1908, chapter 7 and Truman 1951 in political science. More recent discussions include Olson 1965, Tilly 1978 and Hechter 1987.
163
Rogers Brubaker, University of California(LosAngeles).
Arch. europ.socitjl ,XL1U, zu002), 163-189—0003-9756/00/0000-7413(07 50 per an + So. 10 per page'D 2002 A .E .S.
ROGERS BRUBAKER
ETHNICITY
My aim in this paper is not to enter into conceptual or definitional casuistry about the concept of group. It is rather to address one problematic consequence of this tendency to take groups for granted in the study of ethnicity, race and nationhood, and in the study of ethnic, racial and national conflict in particular. This is what I will call groupism: the tendency to take discrete, sharply differentiated, internally homogeneous and externally bounded groups as basic constituents of social life, chief protagonists of social conflicts, and fundamental units of social analysis (2). In the domain of ethnicity, nationalism and race, I mean by 'groupism' the tendency to treat ethnic groups, nations and races as substantial entities to which interests and agency can be attributed. I mean the tendency to reify such groups, speaking of Serbs, Croats, Muslims and Albanians in the former Yugoslavia, of Catholics and Protestants in Northern Ireland, of Jews and Palestinians in Israel and the occupied territories, of Turks and Kurds in Turkey, or of Blacks, Whites, Asians, Hispanics and Native Americans in the U.S. as if they were internally homogeneous, externally bounded groups, even unitary collective actors with common purposes. I mean the tendency to represent the social and cultural world as a multichrome mosaic of monochrome ethnic, racial or cultural blocs.
From the perspective of broader developments in social theory, the persisting strength of groupism in this sense is surprising. After all, several distinct traditions of social analysis have challenged the treatment of groups as real, substantial things-in-the-world. These include such sharply differing enterprises as ethnomethodology and conversation analysis, social network theory, cognitive theory, feminist theory, and individualist approaches such as rational choice and game theory. More generally, broadly structuralist approaches have yielded to a variety of more 'constructivisť theoretical stances, which tend—at the level of rhetoric, at least—to see groups as constructed, contingent, and fluctuating. And a diffuse postmodernist sensibility emphasizes the fragmentary, the ephemeral, and the erosion of fixed forms and clear boundaries. These developments are disparate, even contradictory in analytical style, methodological orientation and epistemological commitments. Network theory, with its methodological (and sometimes ontological) relationalism (Emirbayer and Goodwin 1994; Wellman
(2) In this very general sense, groupism extends well beyond the domain of ethnicity, race and nationalism to include accounts of putative groups based on gender, sexuality, age, class, abledness, religion, minority status, and any kind of 'culture', as well as putative
groups based on combinations of these categorical attributes. Yet while recognizing that it is a wider tendency in social analysis, I limit my discussion here to groupism in the study of ethnicity, race and nationalism.
164
1988) is opposed to rational choice theory, with its methodological (and sometimes ontological) individualism; both are sharply and similarly opposed, in analytical style and epistemological commitments, to postmodernist approaches. Yet these and other developments have converged in problematizing groupness and undermining axioms of stable group being.
Challenges to 'groupism', however, have been uneven. They have been striking—to take just one example—in the study of class, especially in the study of the working class, a term that is hard to use today without quotation marks or some other distancing device. Yet ethnic groups continue to be understood as entities and cast as actors. To be sure, constructivist approaches of one kind or another are now dominant in academic discussions of ethnicity. Yet everyday talk, policy analysis, media reports, and even much ostensibly constructivist academic writing routinely frame accounts of ethnic, racial and national conflict in groupist terms as the struggles 'of ethnic groups, races, and nations (3). Somehow, when we talk about ethnicity, and even more so when we talk about ethnic conflict, we almost automatically find ourselves talking about ethnic groups.
Now it might be asked: 'What's wrong with this?' After all, it seems to be mere common sense to treat ethnic struggles as the struggles of ethnic groups, and ethnic conflict as conflict between such groups. I agree that this is the—or at least a—common-sense view of the matter. But we cannot rely on common sense here. Ethnic common sense—the tendency to partition the social world into putatively deeply constituted, quasi-natural intrinsic kinds (Hirschfeld 1996)—is a key part of what we want to explain, not what we want to explain things with; it belongs to our empirical data, not to our analytical toolkit (4). Cognitive anthropologists and social psychologists have accumulated a good deal of evidence about common-sense ways of carving up the social world—about what Lawrence Hirschfeld (1996) has called 'folk sociologies'. The evidence suggests that some common sense social categories—and notably
(3) For useful critical analyses of media representations of ethnic violence, see the collection of essays in Allen and Seaton [990, as well as Seaton 1999.
(4) This is perhaps too sharply put. To the extent that such intrinsic-kind categories are indeed constitutive of common-sense understandings of the social world, to the extent that such categories are used as a resource for participants, and are demonstrably deployed or oriented to by participants in interaction, they
can also serve as a-resource for analysts. But as Emanuel Schegloff notes in another context, with respect to the category 'interruption', the fact that this is a vernacular, common-3ense category for participants 'does not make it a first-order category usable for professional analysis. Rather than being employed in professional analysis, it is better treated as a target category for professional analysis' (2000: 27). The same might well be said of common sense ethnic categories.
165
ROGERS BRUBAKER
ETHNICITY
common sense ethnic and racial categories—tend to be essentializing and naturalizing (Rothbart and Taylor 1992; Hirschfeld 1996; Gil-White 1999). They are the vehicles of what has been called a 'participants' primordialism' (Smith 1998: 158) or a 'psychological essential-ism' (Medin 1989). We obviously cannot ignore such common sense primordialism. But that does not mean we should simply replicate it in our scholarly analyses or policy assessments. As 'analysts of naturaliz-ers', we need not be 'analytic naturalizers' (Gil-White 1999: 803).
Instead, we need to break with vernacular categories and common-sense understandings. We need to break, for example, with the seemingly obvious and uncontroversial point that ethnic conflict involves conflict between ethnic groups. I want to suggest that ethnic conflict—or what might better be called ethnicized or ethnically framed conflict—need not, and should not, be understood as conflict between ethnic groups, just as racial or racially framed conflict need not be understood as conflict between races, or nationally framed conflict as conflict between nations.
Participants, of course, regularly do represent ethnic, racial and national conflict in such groupist, even primordialist terms. They often cast ethnic groups, races or nations as the protagonists—the heroes and martyrs—of such struggles. But this is no warrant for analysts to do so. We must, of course, take vernacular categories and participants' understandings seriously, for they are partly constitutive of our objects of study. But we should not uncritically adopt categories of ethnopolitical practice as our categories of social analysis. Apart from the general unreliability of ethnic common sense as a guide for social analysis, we should remember that participants' accounts—especially those of specialists in ethnicity such as ethnopolitical entrepreneurs, who, unlike nonspecial-ists, may live 'off as well as 'for' ethnicity—often have what Pierre Bourdieu has called a performative character. By invoking groups, they seek to evoke them, summon them, call them into being. Their categories are for doing—designed to stir, summon, justify, mobilize, kindle and energize. By reifying groups, by treating them as substantial things-in-the-world, ethnopolitical entrepreneurs may, as Bourdieu notes, 'contribute to producing what they apparently describe or designate' (1991a: 220) (5).
Reification is a social process, not simply an intellectual bad habit. As a social process, it is central to the practice of politicized ethnicity. And
(5) Such performative, group-making prac-      zation and representation (Bourdieu 1991b: tices, of course, are not specific to ethnic 248-251). entrepreneurs, but generic to political mobili-
appropriately so. To criticize ethnopolitical entrepreneurs for reifying ethnic groups would be a kind of category mistake. Reifying groups is precisely what ethnopolitical entrepreneurs are in the business of doing. When they are successful, the political fiction of the unified group can be momentarily yet powerfully realized in practice. As analysts, we should certainly try to account for the ways in which—and conditions under which—this practice of reification, this powerful crystallization of group feeling, can work. This may be one of the most important tasks of the theory of ethnic conflict. But we should avoid unintentionally doubling or reinforcing the reification of ethnic groups in ethnopolitical practice with a reification of such groups in social analysis.
II. Beyondgroupism
How, then, are we to understand ethnic conflict, if not in common sense terms as conflict between ethnic groups? And how can we go beyond groupism? Here I sketch eight basic points and then, in the next section, draw out some implications of them. In the final section, I illustrate the argument by considering one empirical case.
Rethinking ethnicity
We need to rethink not only ethnic conflict, but also what we mean by ethnicity itself. This is not a matter of seeking agreement on a definition. The intricate and ever-recommencing definitional casuistry in studies of ethnicity, race and nationalism has done little to advance the discussion, and indeed can be viewed as a symptom of the non-cumulative nature of research in the field. It is rather a matter of critically scrutinizing our conceptual tools. Ethnicity, race and nation should be conceptualized not as substances or things or entities or organisms or collective individuals—as the imagery of discrete, concrete, tangible, bounded and enduring 'groups' encourages us to do—but rather in relational, pro-cessual, dynamic, eventful and disaggregated terms. This means thinking of ethnicity, race and nation not in terms of substantial groups or entities but in terms of practical categories, cultural idioms, cognitive schemas, discursive frames, organizational routines, institutional forms, political projects and contingent events. It means thinking of ethnicization, racialization and nationalization as political, social, cultural and psychological processes. And it means taking as a basic analytical category not the 'group' as an entity but groupness as a contextually fluctuating
167
ROGERS BRUBAKER
ETHNICITY
conceptual variable. Stated baldly in this fashion, these are of course mere slogans; I will try to fill them out a bit in what follows.
The reality of ethnicity
To rethink ethnicity, race and nationhood along these lines is in no way to dispute their reality, minimize their power or discount their significance; it is to construe their reality, power and significance in a different way. Understanding the reality of race, for example, does not require us to posit the existence of races. Racial idioms, ideologies, narratives, categories and systems of classification and racialized ways of seeing, thinking, talking and framing claims are real and consequential, especially when they are embedded in powerful organizations. But the reality of race—and even its overwhelming coercive power in some settings—does not depend on the existence of 'races'. Similarly, the reality of ethnicity and nationhood —and the overriding power of ethnic and national identifications in some settings—does not depend on the existence of ethnic groups or nations as substantial groups or entities.
Groupness as event
Shifting attention from groups to groupness and treating groupness as variable and contingent rather than fixed and given (6), allows us to take account of—and, potentially, to account for—phases of extraordinary cohesion and moments of intensely felt collective solidarity, without implicitly treating high levels of groupness as constant, enduring or definitionally present. It allows us to treat groupness as an event, as something that 'happens', as E. P. Thompson famously said about class. At the same time, it keeps us analytically attuned to the possibility that groupness may not happen, that high levels of groupness nay fail to crystallize, despite the group-making efforts of ethnopolitical entrepreneurs and even in situations of intense elite-level ethnopolitical conflict. Being analytically attuned to 'negative' instances in this way enlarges the domain of relevant cases and helps correct for the bias in the literature toward the study of striking instances of high groupness, successful mobilization or conspicuous violence—a bias that can engender an 'overethnicized' view of the social world, a distorted representation of whole world regions as 'seething cauldrons' of ethnic tension (Brubaker 1998) and an overestimation of the incidence of ethnic violence (Fearon
(6) For accounts (not focused specifically on one another, focus on variability in groupness
ethnicity) that treat groupness as variable, see across cases; my concern is primarily with
Tilly 1978: 6zff; Hechter 1987: 8; Hamilton et variability in groupness over time. al. 1998. These accounts, very different from
168
and Laitin 1996). Sensitivity to such negative instances can also direct potentially fruitful analytical attention toward the problem of explaining failed efforts at ethnopolitical mobilization.
Groups and categories
Much talk about ethnic, racial or national groups is obscured by the failure to distinguish between groups and categories. If by 'group' we mean a mutually interacting, mutually recognizing, mutually oriented, effectively communicating, bounded collectivity with a sense of solidarity, corporate identity and capacity for concerted action, or even if we adopt a less exigent understanding of 'group', it should be clear that a category is not a group (Sacks 1995, I: 41, 401; Handelman 1977; McKay and Lewins 1978; Jenkins 1997: 53$) (7). It is at best a potential basis for group-formation or 'groupness' (8).
By distinguishing consistently between categories and groups, we can problematize—rather than presume—the relation between them. We can ask about the degree of groupness associated with a particular category in a particular setting and about the political, social, cultural and psychological processes through which categories get invested with groupness (Petersen 1987). We can ask how people—and organizations—do things with categories. This includes limiting access to scarce resources or particular domains of activity by excluding categorically distinguished outsiders (Weber 1968 [1922]: 43/jf, 341#; Barth 1969; Brubaker 1992; Tilly 1998), but it also includes more mundane actions such as identifying or classifying oneself or others (Levine 1999) or simply 'doing being ethnic' in an ethnomethodological sense (Moer-man 1968). We can analyze the organizational and discursive careers of categories—the processes through which they become institutionalized and entrenched in administrative routines (Tilly 1998) and embedded in culturally powerful and symbolically resonant myths, memories and narratives (Armstrong 1982; Smith 1986). We can study the politics of
(7) Fredrik Barth's introductory essay to the collection Ethnic Groups and Boundaries (1969) was extraordinarily influential in directing attention to the workings of categories of self- and other-ascription. But Barth does not distinguish sharply or consistently between categories and groups and his central metaphor of 'boundary' carries with it connotations of boundedness, entitativity and groupness.
(8) This point was already made by Max Weber, albeit in somewhat different terms. As Weber argued—in a passage obscured in
the English translation—ethnic commonality, based on belief in common descent, is 'in itself mere (putative) commonality [(geglaubte) Gemeinsamkeit], not community [Gemeinschaft] [...] but only a factor facilitating communal action [Vergemeinschaftung]' (1964: 307; cf. 1968: 389). Ethnic commonality means more than mere category membership for Weber. It is—or rather involves—a category that is employed by members themselves. But this shows that even self-categorization does not create a 'group'.
169
ROGERS BRUBAKER
ETHNICITY
categories, both from above and from below. From above, we can focus on the ways in which categories are proposed, propagated, imposed, institutionalized, discursively articulated, organizationally entrenched and generally embedded in multifarious forms of 'governmentality' (Noiriel 1991; Slezkine 1994; Bmbaker 1994; Torpey 2000; Martin 2001). From below, we can study the 'micropolitics' of categories, the ways in which the categorized appropriate, internalize, subvert, evade or transform the categories that are imposed on them (Dominguez 1986). And drawing on advances in cognitive research, ethnomethodology and conversation analysis (9), we can study the sociocognitive and interactional processes through which categories are used by individuals to make sense of the social world; linked to stereotypical beliefs and expectations about category members (10); invested with emotional associations and evaluative judgments; deployed as resources in specific interactional contexts; and activated by situational triggers or cues. A focus on categories, in short, can illuminate the multifarious ways in which ethnicity, race and nationhood can exist and 'work' without the existence of ethnic groups as substantial entities. It can help us envision ethnicity without groups.
Group-making as project
If we treat groupness as a variable and distinguish between groups and categories, we can attend to the dynamics of group-making as a social, cultural and political project, aimed at transforming categories
(9) Ethnomethodology and conversation analysis have not focused on the use of ethnic categories as such, but Sacks, Schegloff and others have addressed the problem of situated categorization in general, notably the question of the procedures through which participants in interaction, in deploying categories, choose among alternative 3et3 of categories (since there is always more than one set of categories in terms of which any person can be correctly described). The import of this problem has been formulated as follows by Schegloff (2000: 30-31): 'And given the centrality of [...] categories in organizing vernacular cultural 'knowledge', this equivocality can be profoundly consequential, for which category is employed will carry with it the invocation of common-sense knowledge about that category of person and bring it to bear on the person referred to on some occasion, rather than bringing to bear the knowledge implicated with another category, of which the person
170
being referred to is equally a member'. For Sacks on categories, see 1995: I, 40-48, 333-340, 396-403, 578-596; II, 184-187.
(to) The language of 'stereotypes' is, of course, that of cognitive social psychology (for a review of work in this tradition, see Hamilton and Sherman 1994). Bur the general eth.no-methodological emphasis on the crucial importance of the rich though tacit background knowledge that participants bring to interaction and—more specifically—-Harvey Sacks' discussion of the 'inference-rich' categories in terms of which much everyday social knowledge is stored (1995: I, 4off et passim; cf. Schegloff 2000: 29/f) and of the way in which the knowledge thus organized is 'protected against induction' (ibid., 336//), suggest a domain of potentially converging concern between cognitive work on the one hand and ethnomethodologica! and conversation-analytic work on the other—however different their analytic stances and methodologies.
into groups or increasing levels of groupness (Bourdieu 1991a, 1991b). Sometimes this is done in quite a cynical fashion. Ethnic and other insurgencies, for example, often adopt what is called in French a. politique du pire, a politics of seeking the worst outcome in the short run so as to bolster their legitimacy or improve their prospects in the longer run. When the small, ill-equipped, ragtag Kosovo Liberation Army stepped up its attacks on Serb policemen and other targets in early 1998, for example, this was done as a deliberate—and successful—strategy of provoking massive regime reprisals. As in many such situations, the brunt of the reprisals was borne by civilians. The cycle of attacks and counterattacks sharply increased groupness among both Kosovo Albanians and Kosovo Serbs, generated greater support for the KLA among both Kosovo and diaspora Albanians and bolstered KLA recruitment and funding. This enabled the KLA to mount a more serious challenge to the regime, which in turn generated more brutal regime reprisals and so on. In this sense, group crystallization and polarization were the result of violence, not the cause (Brubaker 1999).
Of course, this group-making strategy employed in the late 1990s did not start from scratch. It had already begun with relatively high levels of groupness, a legacy of earlier phases of conflict. The propitious 'raw materials' the KLA had to work with no doubt help explain the success of its strategy. Not all group-making projects succeed and those that do succeed (more or less) do so in part as a result of the cultural and psychological materials they have to work with. These materials include not only, or especially, 'deep', longue-duree cultural structures such as the mythomoteurs highlighted by Armstrong (1982) and Smith (1986), but also the moderately durable ways of thinking and feeling that represent 'middle-range' legacies of historical experience and political action. Yet while such raw materials—themselves the product and precipitate of past struggles and predicaments—constrain and condition the possibilities for group-making in the present, there remains considerable scope for deliberate group-making strategies. Certain dramatic events, in particular, can serve to galvanize and crystallize a potential group, or to ratchet up pre-existing levels of groupness. This is why deliberate violence, undertaken as a strategy of provocation, often by a very small number of persons, can sometimes be an exceptionally effective strategy of group-making.
Groups and organizations
Although participants' rhetoric and common sense accounts treat ethnic groups as the protagonists of ethnic conflict, in fact the chief
171
ROGERS BRUBAKER
ETHNICITY
protagonists of most ethnic conflict—and a fortiori of most ethnic violence—are not ethnic groups as such but various kinds of organizations, broadly understood and their empowered and authorized incumbents. These include states (or more broadly autonomous polities) and their organizational components such as particular ministries, offices, law enforcement agencies and armed forces units; they include terrorist groups, paramilitary organizations, armed bands and loosely structured gangs; and they include political parties, ethnic associations, social movement organizations, churches, newspapers, radio and television stations and so on. Some of these organizations may represent themselves, or may be seen by others, as organizations of and for particular ethnic groups (i i). But even when this is the case, organizations cannot be equated with ethnic groups. It is because and insofar as they are organizations and possess certain material and organizational resources, that they (or more precisely their incumbents) are capable of organized action and thereby of acting as more or less coherent protagonists in ethnic conflict (12). Although common sense and participants' rhetoric attribute discrete existence, boundedness, coherence, identity, interest and agency to ethnic groups, these attributes are in fact characteristic of organizations. The IRA, KLA and PKK claim to speak and act in the name of the (Catholic) Irish, the Kosovo Albanians and the Kurds; but surely analysts must differentiate between such organizations and the putatively homogeneous and bounded groups in whose name they claim to act. The point applies not only to military, paramilitary and terrorist organizations, of course, but to all organizations that claim to speak and act in the name of ethnic, racial or national groups (Heisler 1991).
A fuller and more rounded treatment of this theme, to be sure, would require several qualifications that I can only gesture at here. Conflict and violence vary in the degree to which, as well as the manner in which, organizations are involved. What Donald Horowitz (2001) has called the deadly ethnic riot, for example, differs sharply from organized ethnic insurgencies or terrorist campaigns. Although organizations (sometimes ephemeral ones) may play an important role in preparing, provoking and permitting such riots, much of the actual violence is committed by broader sets of participants acting in relatively spontaneous fashion and in starkly polarized situations characterized by high levels of groupness.
(11) One should remember, though, that tion perspective on social movements, eclipsed organizations often compete with one another in recent years by identity-oriented new social for the monopolization of the right to re- movement theory, has much to offer students present the same (putative) group. of ethnicity. For an integrated statement, see
(12) In this respect the resource mobiliza- McCarthy and Zald 1977.
Moreover, even where organizations are the core protagonists, they may depend on a penumbra of ancillary or supportive action on the part of sympathetic non-members. The 'representativeness' of organizations—the degree to which an organization can justifiably claim to represent the will, express the interests and enjoy the active or passive support of its constituents—is enormously variable, not only between organizations, but also over time and across domains. In addition, while organizations are ordinarily the protagonists of conflict and violence, they are not always the objects or targets of conflict and violence. Entire population categories—or putative groups—can be the objects of organized action, much more easily than they can be the subjects or undertakers of such action. Finally, even apart from situations of violence, ethnic conflict may be at least partly amorphous, carried out not by organizations as such but spontaneously by individuals through such everyday actions as shunning, insults, demands for deference or conformity, or withholdings of routine interactional tokens of acknowledgment or respect (Bailey 1997). Still, despite these qualifications, it is clear that organizations, not ethnic groups as such, are the chief protagonists of ethnic conflict and ethnic violence and that the relationship between organizations and the groups they claim to represent is often deeply ambiguous.
Framing and coding (13)
If the protagonists of ethnic conflict cannot, in general, be considered ethnic groups, then what makes such conflict count as ethnic conflict? And what makes violence count as ethnic violence? Similar questions can be asked about racial and national conflict and violence. The answer cannot be found in the intrinsic properties of behavior. The 'ethnic' quality of 'ethnic violence', for example, is not intrinsic to violent conduct itself; it is attributed to instances of violent behavior by perpetrators, victims, politicians, officials, journalists, researchers, relief workers or others. Such acts of framing and narrative encoding do not simply interpret the violence; they constitute it as ethnic.
Framing may be a key mechanism through which groupness is constructed. The metaphor of framing was popularized by Goffman (1974), drawing on Bateson 1985 [1955]. The notion has been elaborated chiefly in the social movement literature (Snow et al. 1986; Snow and Benford 1988; Gamson and Modigliani 1989; Gamson 1992; uniting rational choice and framing approaches, Esser 1999). When ethnic fram-
(13) These paragraphs draw on Brubaker and Laitin 1998.
172
173
ROGERS BRUBAKER
ETHNICITY
ing is successful, we may 'see' conflict and violence not only in ethnic, but in groupist terms. Although such imputed groupness is the product of prevailing interpretive frames, not necessarily a measure of the groupness felt and experienced by the participants in an event, a compelling ex post interpretive framing or encoding may exercise a powerful feedback effect, shaping subsequent experience and increasing levels of groupness. A great deal is at stake, then, in struggles over the interpretive framing and narrative encoding of conflict and violence.
Interpretive framing, of course, is often contested. Violence—and more generally, conflict—is regularly accompanied by social struggles to label, interpret and explain it. Such 'metaconflicts' or 'conflict[s] over the nature of the conflict*, as Donald Horowitz has called them (1991: 2), do not simply shadow conflicts from the outside, but are integral and consequential parts of the conflicts. To impose a label or prevailing interpretive frame—to cause an event to be seen as a 'pogrom' or a 'riot' or a 'rebellion'—is no mere matter of external interpretation, but a constitutive act of social definition that can have important consequences (Brass 1996b). Social struggles over the proper coding and interpretation of conflict and violence are therefore important subjects of study in their own right (Brass 1996a, 1997, Abelmann and Lie 1995).
Coding and framing practices are heavily influenced by prevailing interpretive frames. Today, ethnic and national frames are accessible and legitimate, suggesting themselves to actors and analysts alike. This generates a 'coding bias' in the ethnic direction. And this, in turn, may lead us to overestimate the incidence of ethnic conflict and violence by unjustifiably seeing ethnicity everywhere at work (Bowen 1996). Actors may take advantage of this coding bias and of the generalized legitimacy of ethnic and national frames, by strategically using ethnic framing to mask the pursuit of clan, clique or class interests. The point here is not to suggest that clans, cliques or classes are somehow more real than ethnic groups, but simply to note the existence of structural and cultural incentives for strategic framing.
Ethnicity as cognition (14)
These observations about the constitutive significance of coding and framing suggest a final point about the cognitive dimension of ethnicity. Ethnicity, race and nationhood exist only in and through our perceptions, interpretations, representations, categorizations and identifications. They are not things in the world, but perspectives on the
(14) These paragraphs draw on Brubaker et al. 2001.
world (15). These include ethnicized ways of seeing (and ignoring), of construing (and misconstruing), of inferring (and misinferring), of remembering (and forgetting). They include ethnically oriented frames, schemas and narratives and the situational cues that activate them, such as the ubiquitous televised images that have played such an important role in the latest intifada. They include systems of classification, categorization and identification, formal and informal. And they include the tacit, taken-for-granted background knowledge, embodied in persons and embedded in institutionalized routines and practices, through which people recognize and experience objects, places, persons, actions or-situations as ethnically, racially or nationally marked or meaningful.
Cognitive perspectives, broadly understood (16), can help advance constructivist research on ethnicity, race and nationhood, which has stalled in recent years as it has grown complacent with success. Instead of simply asserting that ethnicity, race and nationhood are constructed, they can help specify how they are constructed. They can help specify how—and when—people identify themselves, perceive others, experience the world and interpret their predicaments in racial, ethnic or national rather than other terms. They can help specify how 'groupness' can 'crystallize' in some situations while remaining latent and merely potential in others. And they can help link macro-level outcomes with micro-level processes.
III. Implications
At this point a critic might interject: 'What is the point of all this? Even if we can study 'ethnicity without groups', why should we? Concepts invariably simplify the world; that the concept of discrete and bounded ethnic groups does so, suggesting something more substantial and clear-cut than really exists, cannot be held against it. The concept of
(15) As Emanuel Schegloff reminded me in a different context, this formulation is potentially misleading, since perspectives on the world—as every Sociology 1 student is taught—are themselves in the world and every bit as 'real' and consequential as other sorts of things.
(16) Cognitive perspectives, in this broad sense, include not only those developed in cognitive psychology and cognitive anthropology but also those developed in the post- (and
anti-) Parsonian 'cognitive turn' (DiMaggio and Powell 1991) in sociological and (more broadly) social theory, especially in response to the influence of phenomenological and ethno-methodological work (Schutz 1962; Garfinkei 1967; Heritage t984). Cognitive perspectives are central to the influential syntheses of Bourdieu and Giddens and—in a very different form—to the enterprise of conversation analysis.
174
175
ROGERS BRUBAKER
ETHNICITY
ethnic group may be a blunt instrument, but it is good enough as a first approximation. This talk about groupness and framing and practical categories and cognitive schemas is all well and good, but meanwhile the killing goes on. Does the critique matter in the real world, or—if at all—only in the ivory tower? What practical difference does it make?
I believe the critique of groupism does have implications, albeit rather general ones, for the ways in which researchers, journalists, policymakers, NGOs and others come to terms, analytically and practically, with what we ordinarily—though perhaps too readily—call ethnic conflict and ethnic violence. Here I would like to enumerate five of these, before going on in the final section to discuss an empirical case.
First, sensitivity to framing dynamics, to the generalized coding bias in favor of ethnicity and to the sometimes strategic or even cynical use of ethnic framing to mask the pursuit of clan, clique or class interests can alert us to the risk of over-ethnicized or overly groupist interpretations of (and interventions in) situations of conflict and violence (Bowen 1996). One need not subscribe to a reductionist 'elite manipulation' view of politicized ethnicity (Brubaker 1998) to acknowledge that the 'spin' put on conflicts by participants may conceal as much as it reveals and that the representation of conflicts as conflicts between ethnic or national groups may obscure the interests at stake and the dynamics involved. What is represented as ethnic conflict or ethnic war—such as the violence in the former Yugoslavia, may have as much or more to do with thuggery, warlordship, opportunistic looting and black-market profiteering than with ethnicity (Mueller 2000; cf. Collier 1999).
Second, recognition of the centrality of organizations in ethnic conflict and ethnic violence, of the often equivocal character of their leaders' claims to speak and act in the name of ethnic groups and of the performative nature of ethnopolitical rhetoric, enlisted in the service of group-making projects, can remind us not to mistake groupist rhetoric for real groupness, the putative groups of ethnopolitical rhetoric for substantial things-in-the-world.
Third, awareness of the interest that ethnic and nationalist leaders may have in living off politics, as well as for politics, to borrow the classic distinction of Max Weber (1946: 84), and awareness of the possible divergence between the interests of leaders and those of their putative constituents, can keep us from accepting at face value leaders' claims about the beliefs, desires and interests of their constituents.
Fourth, sensitivity to the variable and contingent, waxing and waning nature of groupness and to the fact that high levels of groupness may be more the result of conflict (especially violent conflict) than its under-
lying cause, can focus our analytical attention and policy interventions on the processes through which groupness tends to develop and crystallize and those through which it may subside. Some attention has been given recently to the former, including tipping and cascade mechanisms (Laitin 1995, Kuran 1998) and mechanisms governing the activation and diffusion of schemas and the 'epidemiology of representations' (Sperber 1985). But declining curves of groupness have not been studied systematically, although they are just as important, theoretically and practically. Once ratcheted up to a high level, groupness does not remain there out of inertia. If not sustained at high levels through specific social and cognitive mechanisms, it will tend to decline, as everyday interests reassert themselves, through a process of what Weber (in a different but apposite context [1968 (1922): 246-254]) called 'routinization' (Verall-taeglichungj literally 'towards everydayness').
Lastly, a disaggregating, non-groupist approach can bring into analytical and policy focus the critical importance of intra-ethnic mechanisms in generating and sustaining putatively interethnic conflict (Brubaker and Laitin 1998: 433). These include in-group 'policing,' monitoring, or sanctioning processes (Laitin 1995); the 'ethnic outbidding' through which electoral competition can foster extreme ethnici-zation (Rothschild 1981; Horowitz 1985); the calculated instigation or provocation of conflict with outsiders by vulnerable incumbents seeking to deflect in-group challenges to their positions; and in-group processes bearing on the dynamics of recruitment into gangs, militias, terrorist groups or guerrilla armies, including honoring, shaming and shunning practices, rituals of manhood, intergenerational tensions and the promising and provision of material and symbolic rewards for martyrs.
IV. Ethnicity at work in a Transylvanian town
At this point, I would like to add some flesh to the bare-bones analytical argument sketched above. It is tempting to comment on the United States. It would be easy to score rhetorical points by emphasizing that the 'groups' taken to constitute the canonical 'ethnoracial pentagon' (Hollinger 1995)—African Americans, Asian Americans, Whites, Native Americans and Latinos—are (with the partial exception of African Americans) not groups at all but categories, backed by political entrepreneurs and entrenched in governmental and other organizational routines of social counting and accounting (Office of Manage-
176
177
ROGERS BRUBAKER ETHNICITY
ment and Budget 1994). It wou'd De easY to highlight the enormous cultural heterogeneity within these and other putative 'groups,' and the minimal degree of groupness associated with many ethnic categories in the US (Gans 1979; Heisler 1991).
But rather than take this tack, I will try to address a harder case, drawn from a region historically characterized by much higher degrees of ethnic and national groupness. I want to consider briefly how ethnicity works in an East Central European context characterized by continuous and often intense elite-level ethnonational conflict since the fall of communism (and, of course, by a much longer history of ethnonational tension). Here too, I want to suggest, we can fruitfully analyze ethnicity without groups.
The setting, familiar to me from field research conducted in the second half of the 1990s, is the city of Cluj, the main administrative, economic and cultural center of the Transylvanian region of Romania. Of the approximately 330,000 residents, a substantial minority— somewhere between 14 and 23 percent—identify themselves as Hungarian by ethnocultural nationality (17). The city, as I indicated, has been the site of protracted and seemingly intractable ethnonational conflict since the collapse of the Ceausescu regime in December 1989. But this is not, I will argue, best understood as a conflict between ethnic or national groups. To think of it as a conflict between groups is to conflate categories ('Hungarian' and 'Romanian') with groups ('the Hungarians', 'the Romanians'); to obscure the generally low, though fluctuating, degree of groupness in this setting; to mistake the putative groups invoked by ethnonational rhetoric for substantial things-in-the-world; to accept, at least tacitly, the claims of nationalist organizations to speak for the 'groups' they claim to represent; and to neglect the everyday contexts in which ethnic and national categories take on meaning
(17) In the US and much of northern and western Europe, 'nationality' ordinarily means 'citizenship', that is, membership of the state; and 'nation' and 'state' are often used interchangeably. In central and eastern Europe, by contrast, 'nation' and 'nationality' do not refer in the first instance to the state, but ordinarily invoke an ethnocultural frame of reference independent of—and often cutting across the boundaries of—statehood and citizenship. To identify oneself as Hungarian by nationality in Transylvania is to invoke a state-transcending Hungarian ethnocultural 'nation'. In the text, following the usage in this setting, I use 'ethnic' and 'national' interchangeably.
At the last census, conducted in 1992, 23 % of the population of Cluj identified as Hungarian. More recent statistics, however, suggest a smaller population identifying as Hungarian, at least among younger age cohorts. Of persons getting married in 1999, 14.4% identified their nationality as Hungarian. Of primary, middle school and secondary school students in 1999-2000, 15.1%, 14.3% and 14.8%, respectively, identified as Hungarian. Contextual differences in identification may account for part of the difference, as might different age structures of Romanian and Hungarian populations and differential emigration rates during the 1990s.
and the processes through which ethnicity actually 'works' in everyday life.
Here, as elsewhere, the protagonists of the conflict have been organizations, not groups. The conflict has pitted the town's three-term mayor—the flamboyant Romanian nationalist Gheorghe Funar—and the statewide Romanian nationalist parties against the Cluj-based Democratic Association of Hungarians of Romania (DAHR), at once a statewide political party with its electoral base in Transylvania and an organization claiming to represent and further the interests of the Hungarian minority in Romania. Rhetoric has been heated on both sides. Mayor Funar has accused Hungary of harboring irredentist designs on Transylvania (i 8); he has called the DAHR a 'terrorist organization'; and he has accused Transylvanian Hungarians of secretly collecting weapons, forming paramilitary detachments and planning an attack on Romanians. Funar has ordered bilingual signs removed from the few buildings that had them; banned proposed celebrations of the Hungarian national holiday; called for the suspending of Hungarian language broadcasts on Romanian state television; called for punishment of citizens for displaying the Hungarian flag or singing the Hungarian anthem; and proposed to rename after Romanian personages the few Cluj streets that bear the names of Hungarians.
The DAHR, for its part, is committed to a number of goals that outrage Romanian nationalists (19). It characterizes Hungarians in Romania as an 'indigenous community' entitled to an equal partnership with the Romanian nation as a constituent element of the Transylvanian state—thereby directly challenging the prevailing (and constitutionally enshrined) Romanian understanding of the state as a unitary nation-state like France. At the same time, it characterizes Transylvanian Hungarians as an 'organic part of the Hungarian nation' and as such claims the right to cultivate relations with the 'mother country' across the border, which leads Romanian nationalists to call into question their loyalty to the Romanian state. It demands collective rights for Hungarians as a national minority and it demands autonomy, including territorial autonomy, for areas in which Hungarians live as a local majority, thereby raising the specter of separatism in the minds of Romanian nationalists. It demands that Hungarians have their own institutional system in the domain of education and culture—yet that this institutional system
(18) Transylvania had belonged to Hungary for half a century before the First World War and again for four years during the Second World War.
(19) The DAHR program can be found in English at http: //www.rmdsz.ro/angol/ aboutus/prog.htm.
178
'79
ROGERS BRUBAKER
ETHNICITY
should be financed by the Romanian state. It demands the right to public, state-funded education in Hungarian at every level and in every branch of the educational system, including vocational education. It demands the right to take entrance exams to every school and university in Hungarian, even if the school or department to which the student is applying carries out instruction in Romanian. And it demands the re-establishment of an independent Hungarian university in Cluj and the establishment of publicly funded but independent Hungarian language radio and TV studios.
Like ethnic and nationalist organizations everywhere, the DAHR claims to speak for the Hungarian minority in Romanian, often characterizing it as a singular entity, 'the Hungariandom of Romania' (a romdniai magyarsdg). But no such entity exists (20). The many Cluj residents who self-identify as Hungarian are often sharply critical of the DAHR and there is no evidence that the demands of the DAHR are the demands of 'the Hungarians'. On the question of a Hungarian university—the most contentious political issue of the last few years—a survey conducted by a Hungarian sociologist found that a plurality of Hungarian university students in Cluj preferred an autonomous system of Hungarian-language education within the existing university to the DAHR goal of re-establishment of a separate Hungarian university (Magyari-Nandor and Peter 1997). Most Hungarians, like most Romanians, are largely indifferent to politics and preoccupied with problems of everyday life—problems that are not interpreted in ethnic terms. Although survey data and election results suggest that they appear to vote en bloc for the DAHR, most Hungarians are familiar only in a vague way with the DAHR program. Similarly, there is no evidence that Mayor Funar's anti-Hungarian views are widely shared by the town's Romanian residents. When Funar is praised, it is typically as a 'good housekeeper' (bun gospodar); he is given credit for sprucing up the town's appearance and for providing comparatively good municipal services. Almost everyone—Romanian and Hungarian alike—talks about ethnic conflict as something that 'comes from above' and is stirred up by politicians pursuing their own interests. The near-universal refrain is that ethnicity is 'not a problem'. To be sure, a similar idiom—or perhaps ideology—of everyday interethnic harmony can be
(20) Of course this point holds not only, or especially, of the Hungarian minority, or of minorities generally. In Romania as elsewhere, those who claim to speak for dominant nations—nations that are closely identified with the states that bear their names, referred
l8o
to in German as Staatsvölker or 'state peoples'—also routinely reify those 'nations' and characterize them as singular entities with a common will and common interests, where in fact no such entity exists.
found in many other settings, including some deeply divided, violence-plagued ones. So the idiom cannot be taken as evidence of the irrelevance of ethnicity. The point here is simply to underscore the gap between nationalist organizations and the putative 'groups' in whose names they claim to speak.
Despite the continuous elite-level ethnopolitical conflict in Cluj since the fall of Ceausescu, 'groupness' has generally remained low. At no time did Hungarians and Romanians crystallize as distinct, solidary, bounded groups; in this sense groupness failed to 'happen'. The contrast with Targu Mures, another Transylvanian city where groups did crystallize in 1990, is instructive. In Targu Mures, ethnically framed conflict over the control of a high school and over the control of local government in the immediate aftermath of the fall of Ceausescu intensified and broadened into a generalized conflict over the 'ownership' and control of the ethnodemographically evenly divided city. The conflict culminated in mass assemblies and two days of street fighting that left at least six dead and 200 injured. In the days leading up to the violent denouement, categories had become palpable, sharply bounded groups, united by intensely felt collective solidarity and animated by a single overriding distinction between 'us' and 'them'. The violence itself reinforced this sense of groupness, which then subsided gradually as life returned to normal and no further Hungarian-Romanian violence occurred, here or elsewhere in Transylvania.
No such crystallization occurred in Cluj. There were, to be sure, a few moments of moderately heightened groupness. One such moment—among Hungarians—occurred when Mayor Funar ordered a new plaque installed on the base of a monumental equestrian statue of Matthias Corvinus, celebrated king of Hungary during the late fifteenth century, in the town's main square. The statue, erected at the turn of the last century at a moment of and as a monument to, triumphant Hungarian nationalism, is perceived by many Hungarians as 'their own' and the new plaque deliberately affronted Hungarian national sensibilities by emphasizing the (partly) Romanian origin of Matthias Corvinus and representing him—contrary to the triumphalist image projected by the statue—as having been defeated in battle by 'his own nation', Moldavia (Feischmidt 2001). Another moment occurred when archeological excavations were begun in front of the statue, again in a manner calculated to affront Hungarian national sensibilities by highlighting the earlier Roman—and by extension, Romanian—presence on the site. A third moment occurred in March 1998, when Mayor Funar tried to bar Hungarians from carrying out their annual March 15 celebration com-
181
ROGERS BRUBAKER
ETHNICITY
memorating the revolution of 1848, this year's celebration, in the ses-quicentennial year, having special significance (Brubaker and Feisch-midt 2002) (21). A final moment occurred in June 1999 at the time of a much-hyped soccer match in Bucharest between the national teams of Romania and Hungary. In Cluj, the match was televised on a huge outdoor screen in the main square and some fans chanted lAfara, afara, cu Ungurii din fatal' (out, out, the Hungarians out of the country!) and vandalized cars with Hungarian license plates (Adevarul de Cluj 1999).
In each of these cases, groupness—^especially among Hungarians, though in the final case among Romanians as well—was heightened, but only to a modest degree and only for a passing moment. The first event occasioned a substantial but isolated Hungarian protest, the second a smaller protest, the third some concern that the commemoration might be broken up (in the event it proceeded without serious incident) and the last some moments of concern for those who happened to be in the town center during and immediately after the soccer match. But even at these maximally group-like moments, there was no overriding sense of bounded and solidary groupness for those not immediately involved in the events (22). In short, when one shifts one's focus from presupposed groups to variable groupness and treats high levels of groupness as a contingent event, a crystallization, something that happens, then what is striking about Cluj in the 1990s is that groupness remained low and groups failed to happen or to crystallize.
(21) To Romanian nationalists, Hungarians' commemoration of 1848 is illegitimate, for it celebrates a regime that was as much nationalist as revolutionary, aspiring to—and briefly securing—unitary control over Transylvania. Romanian nationalist mythology commemorates not the revolution, but the guerilla struggle against the Hungarian revolutionary regime, led by Avram Iancu, to whom a colossal monument was erected under Funar's sponsorship in 1993.
(22) Even for those who were involved in the events, one should be cautious about inferring an overriding sense of groupness. I was in Cluj in the summer of 1994, when excavations in the main 'Hungarian' square were about to begin. I was staying with the family of a leading figure of the DAHR, albeit one of the more liberal figures. At one point, he proposed: 'Menjiink asni? [Shall we go dig?]' At a moment of overriding groupness, such a joke would be unthinkable; here, the nationalist
projects of Mayor Funar were—at least for some—a joking matter. One further incident is worth mentioning in this connection. In 1997, a long-closed Hungarian consulate re-opened in Cluj, reflecting a warming of relations between Budapest and the newly elected pro-western government in Bucharest. Funar protested—in vain—against its opening and when it opened, tried to fine it for flying the Hungarian flag. A few weeks after its opening, five men pulled up in a pickup truck, placed an extendable ladder against the side of the building, and—removed the flag, in broad daylight, as a small crowd looked on. The next day, they were apprehended by the police; Funar characterized them as 'Romanian heroes'. FIsewhere, this sort of incident—which could easily be construed as involving the desecration of a sacred national symbol—has been enough to trigger a riot. Here, nobody paid much attention; the incident was coded as farce, not as sacred drama.
To note the relatively low degree of groupness in Cluj and the gap between organizations and the putative groups they claim to represent, is not to suggest that ethnicity is somehow not 'real' in this setting, or that it is purely an elite phenomenon. Yet to understand how ethnicity works, it may help to begin not with 'the Romanians' and 'the Hungarians' as groups, but with 'Romanian* and 'Hungarian' as categories. Doing so suggests a different set of questions than those that come to mind when we begin with 'groups'. Starting with groups, one is led to ask what groups want, demand or aspire towards; how they think of themselves and others; and how they act in relation to other groups. One is led almost automatically by the substantiahst language to attribute identity, agency, interests and will to groups. Starting with categories, by contrast, invites us to focus on processes and relations rather than substances. It invites us to specify how people and organizations do things with and to ethnic and national categories; how such categories are used to channel and organize processes and relations; and how categories get institutionalized and with what consequences. It invites us to ask how, why and in what contexts ethnic categories are used—or not used—to make sense of problems and predicaments, to articulate affinities and affiliations, to identify commonalities and connections, and to frame stories and self-understandings.
Consider here just two of the many ways of pursuing a category-centered rather than a group-centered approach to ethnicity in Cluj. First, a good deal of common-sense cultural knowledge about the social world and one's place in it, here as in other settings, is organized around ethnonational categories (23). This includes knowledge of one's own and others' ethnocultural nationality and the ability to assign unknown others to ethnonational categories on the basis of cues such as language, accent, name, sometimes dress, hair style, and even phenotype. It includes knowledge of what incumbents of such categories are like (24), how they typically behave and how ethnonational category membership matters in various spheres of life. Such common-sense category-based
(23) On categories as 'repositories] for common sense knowledge' generally (Sche-gloff 2000: 29), see Sacks 1995, I, 40-48, 333-340. For cognitive perspectives on social categories as structures of knowledge, with special regard to ethnic, racial and other 'natural kind", like categories, see Rothbart and Taylor 1992; Hamilton and Sherman 1994; Hirsch-feld 1996.
(24) Even when such common-sense, category-based stereotypical knowledge is overridden, the very manner of overriding may
testify' to the existence (and the content) of the category-based knowledge that is being overridden. On the general phenomenon of 'modifiers' that work by asserting that what is generally known about members of a category is not applicable to some particular member, see Sacks (1995), I, 44-45- Among Hungarians—even liberal, cosmopolitan Hungarians—I have on several occasions heard someone referred to as 'Roman, de rendes' (Romanian, but quite all right) or something to that effect.
183
ROGERS BRUBAKER
ETHNICITY
knowledge shapes everyday interaction, figures in stories people tell about themselves and others and provides ready-made explanations for certain events or states of affairs. For Hungarians, for example, categorizing an unknown person as Hungarian or Romanian may govern how one interacts with him or her, determining not only the language but also the manner in which one will speak, a more personal and confidential (bizalmas) style often being employed with fellow Hungarians. Or for Romanians, categorizing two persons speaking Hungarian in a mixed-language setting as Hungarian (rather than, for example, as friends who happen to be speaking Hungarian) provides a ready-made explanation for their conduct, it being common-sense knowledge about Hungarians that they will form a bisericufa (clique, literally: small church) with others of their kind, excluding co-present Romanians, whenever they have the chance. Or again for Hungarians, categorically organized common-sense knowledge provides a ready-made framework for perceiving differential educational and economic opportunities as structured along ethnic lines, explaining such differentials in terms of what they know about the bearing of ethnic nationality on grading, admissions, hiring, promotion and firing decisions and justifying the commonly voiced opinion that 'we [Hungarians] have to work twice as hard' to get ahead (Feischmidt 2001; Fox 2001). These and many other examples suggest that ethnicity is, in important part, a cognitive phenomenon, a way of seeing and interpreting the world and that, as such, it works in and through categories and category-based common sense knowledge.
Ethnic categories shape institutional as well as informal cognition and recognition. They not only structure perception and interpretation in the ebb and flow of everyday interaction but channel conduct through official classifications and organizational routines. Thus ethnic (and other) categories may be used to allocate rights, regulate actions, distribute benefits and burdens, construct category-specific institutions, identify particular persons as bearers of categorical attributes, 'cultivate' populations or, at the extreme, 'eradicate' unwanted 'elements' (25).
In Cluj—as in Romania generally—ethnic categories are not institutionalized in dramatic ways. Yet there is one important set of institutions built, in part, around ethnic categories. This is the school system (26).
(25) On 'population politics' and the metaphor of the gardening state, see Holquist 1997: 131; Bauman 2000; Weiner 2001. Genocide, as Bauman observes, 'differs from other murders in having a category for its object' (2000: 227, italics in original).
(26) Traditional churches, too, are built
around ethnic categories, with two 'Hungarian' churches (Roman Catholic and Calvinist) and two 'Romanian' churches (Orthodox and Greek-Catholic or.Uniate). With aging congregations, dwindling influence and increased competition from less ethnically marked neo-Protestant denominations, the traditional
184
In Cluj, as in other Transylvanian cities, there is a separate Hungarian-language school system paralleling the mainstream system and running from preschool through high school. These are not private schools, but part of the state school system. Not all persons identifying themselves as Hungarian attend Hungarian schools, but most do (85-90 percent in grades 1-4, smaller proportions, though still substantial majorities, in later grades) (27). In Cluj, moreover, there are also parallel tracks at the university level in many fields of study.
Categories need ecological niches in which to survive and flourish; the parallel school system provides such a niche for 'Hungarian' as an ethnonational category. It is a strategically positioned niche. Hungarian schools not only provide a legitimate institutional home and a protected public space for the category. They also generate the social structural foundations for a small Hungarian world within the larger Romanian one (Feischmidt 2001). Since the schools shape opportunity structures and contact probabilities and thereby influence friendship patterns (and, at the high school and university level, marriage patterns as well), this world is to a certain extent a self-reproducing one. Note that the (partial) reproduction of this social world—this interlocking set of social relationships linking school, friendship circles and family—does not require strong nationalist commitments or group loyalties. Ethnic networks can be reproduced without high degrees of groupness, largely through the logic of contact probabilities and opportunity structures and the resulting moderately high degrees of ethnic endogamy (28).
This brief case study has sought to suggest that even in a setting of intense elite-level ethnic conflict and (by comparison to the United States) deeply rooted and stable ethnic identifications, one can analyze the workings of ethnicity without employing the language of bounded groups.
V. Conclusion
What are we studying when we study ethnicitv and ethnic conflict? This paper has suggested that we need not frame our analyses in terms
churches are less significant than schools as institutional loci of ethnic categories.
(27) Data are drawn from figures provided by the School Inspectorate of Cluj County.
(28) Of the Hungarians who married in Cluj in i999> nearly 75 percent married other Hungarians, while about 25 percent married
Romanians. This suggests a moderately high degree of ethnic endogamy, but only moderately high, for about 40 percent of all marriages involving Hungarians were mixed marriages. Data were compiled from forms filled out by couples, consulted at the Cluj branch of the National Commission for Statistics.
185
ROGERS BRUBAKER
ETHNICITY
of ethnic groups and that it may be more productive to focus on practical categories, cultural idioms, cognitive schemas, commonsense knowledge, organizational routines and resources, discursive frames, institutionalized forms, political projects, contingent events and variable groupness. It should be noted in closing, however, that by framing our inquiry in this way and by bringing to bear a set of analytical perspectives not ordinarily associated with the study of ethnicity—cognitive theory, ethnomethodology, conversation analysis, network analysis, organizational analysis and institutional theory, for example—we may end up not studying ethnicity at all. It may be that 'ethnicity' is simply a convenient—though in certain respects misleading—rubric under which to group phenomena that, on the one hand, are highly disparate and, on the other, have a great deal in common with phenomena that are not ordinarily subsumed under the rubric of ethnicity (29). In other words, by raising questions about the unit of analysis—the ethnic group—we may end up questioning the domain of analysis: ethnicity itself. But that is an argument for another occasion.
REFERENCES
Abelmann, Nancy and John Lie, 1995, Blue Dreams: Korean Americans and the Los Angeles Riots (Cambridge: Harvard University Press).
Adevarul de Cluj, 1999, Confruntarea dintre Romania si Ungaria a continuat si dupa meci [The confrontation between Romania and Hungary continued also after the match], June 7,1999.
Allen, Tim and Jean Seaton (eds), 1999. The Media of Conflict: War Reporting and Representations of Ethnic Violence (London: Zed Books).
Armstrong, John A., 1982, Nations Before Nationalism (Chapel Hill: University of North Carolina Press).
Baii^ey, Benjamin, 1997, Communication of Respect in Interethnic Service Encounters, Language in Society 26, 327-356.
Barth, Fredrik, 1969, Introduction, in Fredrik Barth (ed.), Ethnic Groups and Boundaries:
(29) As Weber put it nearly a century ago (1964 [1922]: 313; cf. 1968 [1922]: 394-5). a precise   and   differentiated   analysis would
The Social Organization of Culture Difference (London: George Allen & Unwin), 9-38.
Bateson, Gregory, 1985 [1955], A Theory of Play and Fantasy, in Robert E. Innis (ed.), Semiotics: An Introductory Anthology, 131-144.
Bauman, Zygmunt, 2000 [1989], Modernity and the Holocaust (Ithaca: Cornell University Press).
Bentley, Arthur F., 1908, The Process of Government: A Study of Social Pressures (Chicago: University of Chicago Press).
BouRDlEU, Pierre, 1991a, Identity and Representation: Elements for a Critical Reflection on the Idea of Region, in Pierre Bourdieu, Language and Symbolic Power (Cambridge: Harvard University Press), 220-228.
—, 1991b, Social Space and the Genesis of 'Classes', in Pierre Bourdieu, Language and
'surely throw out the umbrella term "ethnic" altogether', for it is 'entirely unusable' for any 'truly rigorous investigation'.
Symbolic Power (Cambridge: Harvard University Press), 229-251. Bo wen, John R., 1996, The Myth of Global Ethnic Conflict, Journal of Democracy 7(4), 3-14-
Brass, Paul R. (ed.), 1996a, Riots and Pogroms (Washington Square: New York University Press).
—, [996b, Introduction: Discourse of Ethnicity, Communalism and Violence, in Paul R. Brass (ed.), Riots and Pogroms (Washington Square: New York University Press), 1-55.
—, 1997, Theft of an Idol: Text and Context in the Representation of Collective Violence (Princeton: Princeton University Press).
Brubaker, Rogers, 1992, Citizenship and Nationhood in France and Germany (Cambridge, MA: Harvard University Press).
—, 1994, Nationhood and the National Question in the Soviet Union and Post-Soviet Eurasia, Theory and Society 23t 1), 47-78.
—, 1998, Myths and Misconceptions in the Study of Nationalism, in John Hall (ed.), The State of the Nation: Ernest Getlner and the Theory of Na tionalism (New Yo rk: Cambridge University Press), 272-306.
—, 1999, A Shameful Debacle, UCLA Magazine, Summer 1999,15-16.
Brubaker, Rogers and Margit Feischmidt, 2002, 1848 in 1998: The Politics of Commemoration in Hungary, Romania, and Slovakia, forthcoming in Comparative Studies in Society and History.
Brubaker, Rogersand David D. Laitin, 1998, Ethnic and Nationalist Violence, Annual Review of Sociology 24,423-452.
Brubaker, Rogers, Mara Loveman and Peter Stamatov, 2001, 'Beyond Social Construction: The Case for a Cognitive Approach to Race, Ethnicity and Nationalism'. Unpublished manuscript.
Collier, Paul, 1999, Doing Well Out of War. http: lltoviw.worldbank.org/researchlconflictl papers jeconagenda .htm.
Cooley, Charles H., 1962 [1909], Social Organization (New York: Schocken Books).
DiMaggio, Paul J. and Walter W. PoWEu.,1991, Introduction, in Walter W. Powell and Paul J. DiMaggio (eds), The New Institutionalism in Organizational (Chicago: University of Chicago Press), 1-38.
DoMiNcuEZ, Virginia R., 1986, White by Definition: Social Classification in Creole Louisiana (New Brunswick: Rutgers University Press).
186
Emirbayer, Mustafa and Jeff Goodwin, 1994, Network Analysis, Culture and the Problem of Agency, American Journal of Sociology 99(6), 1411-1454.
Esser, Hartmut, 1999, Die Situationslogik ethnischer Konflikte: Auch eine Anmerkung zum Beitrag 'Ethnische Mobilisierung und die Logik von Identitätskämpfen' von Klaus Eder und Oliver Schmidtke, Zeitschrift für Soziologie 28(4), 245-262.
Fearon, James and David D. Laitin, 1996, Explaining Interethnic Cooperation, American Political Science Review 90(4), 715-735.
Feischmidt, Margit, 2001, Zwischen Abgrenzung und Vermischung: Ethnizität in der siebenbürgischen Stadt Cluj (K0I0-zsvär, Klausenburg). Ph. D dissertation (Humboldt University, Berlin).
Fox, Jon E., 2001, Nationness and Everyday Life: Romanian and Hungarian University Students in Transylvania. Paper presented to the Center for Comparative Social Analysis Workshop, Department of Sociology, UCLA.
Gamson, William A., 1992, Talking Politics (New York: Cambridge University Press).
Gamson, William A. and Andre Modigliani,1989, Media Discourse and Public Opinion on Nuclear Power: A Constructionist Approach, American Journal of Sociology 95.1-37-
Gans, Herbert J., 1979, Symbolic Ethnicity: The Future of Ethnic Groups and Cultures in America, Ethnic and Racial Studies 2,1-20.
Garfinkel, Harold, 1967, Studies in Ethnomethodology (Englewood Cliffs, NJ.: Prentice-Hall, Inc.)
Gil-White, Francisco, 1999, How Thick Is Blood? The Plot Thickens...: If Ethnic Actors Are Primordialists, What Remains of the Circumstantialist/Primordialist Controversy? Ethnic and Racial Studies 22(5), 789-820.
Goffman, Erving, 1974, Frame Analysis (San Francisco: Harper Colophon Books).
Hamilton, David L. and Jeffrey W. Sherman, 1994, Stereotypes, in Robert S. Wyer and Thomas K. Srull (eds), Handbook of Social Cognition, 2nd ed. (Hillsdale, NJ: L. Erlbaum Associates), 1-68.
Hamilton, David L., Steven J. Sherman and Brian Lickel, 1998, Perceiving Social Groups: The Importance of the Entitativity Continuum, in Constantine Sedikides, John Schopler  and  Chester  A.  Insko (eds),
187
ROGERS BRUBAKER
ETHNICITY
Intergroup Cognition and Intergroup Behavior (Mahwah, NJ: Lawrence Erlbaum Associates), 47-74.
Handelman, Don, 1977, The Organization of Ethnicity, Ethnic Groups 1, 187-200.
hechter, Michael, 1987, Principles of Group Solidarity (Berkeley: University of California Press).
Heisler, Martin, 1991, Ethnicity and Ethnic Relations in the Modern West, in Joseph Montville (ed.), Conflict and Peacemaking in Multiethnic (Lexington: Lexington Books), 21-52.
Heritage, John, 1984, Garfinket and Ethno-methodology (Cambridge: Polity Press).
Hirschfeld, Lawrence A., 1996, Race in the Making: Cognition, Culture and the Child's Construction of Human Kinds (Cambridge, MA: MIT Press).
Holunoer, David A., 1995, Postethnic America: Beyond Multiculturalism (New York: Basic Books).
Holquist, Peter, 1997, Conduct Merciless 'Mass Terror': Decossackization on the Don, 1919, Cahiers du Monde russe 38(1-2), 127-162.
Homans, George C, 1950, 7%« Human Group (New York: Harcourt, Brace & World, Inc.)
Horowitz, Donald L., 1985, Ethnic Groups in Conflict (Berkeley, CA: University of California Press).
—, 1991, A Democratic South Africa?: Constitutional Engineering in a Divided Society (Berkeley: University of California Press).
—, 2001, The Deadly Ethnic Riot (Berkeley, CA: University of California Press).
Jenkins, Richard, 1997, Rethinking Ethnicity (London: Sage).
Kuran, Timur, 1998, Ethnic Norms and their Transformation through Reputation-a] Cascades, Journal of Legal Studies 27, 623-659.
Laitin, David D., 1995, National Revivals and Violence, Archives europeennes de sociologie xxxvi, 3-43.
Levine, Hal B., 1999, Reconstructing Ethnicity, Journal of the Royal Anthropological Institute (New Series) 5, 165-180.
Magyari-NAndor, Laszto and Laszlo Peter, 1997, Az egytemrol magyar diakszemmel [Hungarian students' views on the university], Korunk 1997/4, 11 a-119.
Martin, Terry, 2001, The Affirmative Action Empire (Ithaca: Cornell University Press).
McCarthy, John D. and Mayer N. Zald, 1977, Resource  Mobilization and Social
Movements: A Partial Theory, American Journal of Sociology 82 (6), 1212-1241. McGrath, Joseph E., 1984, Groups: Interaction and Performance (New Jersey: Prentice-Hall).
McKay, James and Frank Lewins,i978, Ethnicity and the Ethnic Group: A Conceptual Analysis and Reformulation, Ethnic and Racial Studies 1 (4), 412-427.
Medin, Douglas L., 1989, Concepts and Conceptual Structure, The American Psychologist^, 1469-1481.
Moerman, Michael, 1968, Being Lue: Uses and Abuses of Ethnic Identification, in June Helm (ed.), Essays on the Problem of Tribe (Washington: University of Washington Press).
Mueller, John, 2000, The Banality of 'Ethnic War', International Security 25,42-70.
Nadel, S. E, 1957, A Theory of Social Structure (London: Cohen & West).
Noiriel, Gerard, 1991,La Tyrannic du national: le droit d'asile en Europe ijgj-iggj (Paris: Calmann-Levy).
Office of Management and Budget,  1994, Standards   for  the  classification  of federal data on race and ethnicity, http://wtviv.whitehouse.gov/omb/fedreg/ notice-i 5 .html.
Olson, Mancur, 1965, The Logic of Collective Action: Public Goods and the Theory of Groups (Cambridge: Harvard University Press),
Petersen, William, 1987, Politics and the Measurement of Ethnicity, in William Alonso and Paul Starr (eds), The Politics of Numbers (New York: Russell Sage Foundation), 187-233.
Rothbakt, Myron and Marjorie Taylor, 1992, Category Labels and Social Reality: Do We View Social Categories As Natural Kinds? in Gun R. Semin and Klaus Fiedler (eds), Language, Interaction and Social Cognition (London: Sage Publications).
Rothschild, Joseph, 1981, Ethnopolitics: A Conceptual Framework (New York: Columbia University Press).
Sacks, Harvey, 1995, Lectures on Conversation (Oxford, Blackwell Publishers).
Schegloff, Emanuel A., 2000, Accounts of Conduct in Interaction: Interruption, Overlap and Turn-Taking, in J. H. Turner (ed.), Handbook of Sociological Theory (New York: Plenum). Forthcoming.
Schutz, Alfred, 1962, Collected Papers I: The Problem of Social Reality, Maurice Natan-son (ed.) (The Hague: Marinus Nijhoff).
Seaton, Jean, 1999, Why Do We Think the Serbs Do It? The New 'Ethnic' Wars and the Media, Political Quarterly 70(3), 2S4-270.
Slezkine, Yuri, 1994, The USSR As.a Communal Apartment, or How a Socialist State Promoted Ethnic Particularlism, Slavic Review 53(4), 414-452.
Smith, Anthony D., 1986, The Ethnic Origins of Nations (Oxford: Basil Blackwell).
—, 1998, Nationalism and Modernism: A Critical Survey of Recent Theories of Nations and Nationalism (London: Routledge).
Snow, David A. and Robert D. Ben-ford, 1988, Ideology, Frame Resonance and Participant Mobilization, International Social Movement Research 1, 197-217.
Snow, David A., E. B. Rochford Jr., Steven K. Worden and Robert D. Benford, 1986, Frame Alignment Processes, Micromobili-zation and Movement Participation, American Sociological Review 5 1, 464-481.
Sperber, Dan, 1985, Anthropology and Psychology: Towards an Epidemiology of Representations, Man 20, 73-89.
Tilly, Charles, 1978, From Mobilization to Revolution (Reading, Mass.: Addison-Wesley Publishing Company).
—, 1998, Durable Inequality (Berkeley: University of California Press).
Torpey, John, 2000, Tfie Invention of the Passport: Surveillance, Citizenship and the State (Cambridge: Cambridge University Press).
Truman, David B., 1951, The Governmental Process: Political Interests and Public Opinion. 2nd ed. (New York: Alfred A. Knopf).
Weber, Max, 1946, From Max Weber: Essays in Sociology, H. H. Gerth and Wright Mills (eds) (New York: Oxford University Press).
—, 1964, Wirtschaft und Gesellschaf. 4th ed. (Köln: Kiepenheurer & Witsch).
-—, 1968 [1922], Economy and Society, Guen-ther Roth and Claus Wittich (eds) (Berkeley: University of California Press).
Weiner, Amir, 2001, Making Sense of War: The Second World War and the Fate of the Bolshevik Revolution (Princeton: Pnnceton University Press).
Wellman, Barry, 1988, Structural Analysis: From Method and Metaphor to Theory and Substance, in Barry Wellman and S- D. Berkowitz (eds), Social Structures: A Network Approach (Cambridge: Cambridge University Press), 19-61.