This website will be closed down during 2014.

Thomas Hylland Eriksen's new site is now open at

You are now being redirected to the new Engaging with the world site.

























The epistemological status of the concept of ethnicity

Thomas Hylland Eriksen


Conference paper, Amsterdam ("The Anthropology of Ethnicity"), December 1993. Published in Anthropological Notebooks (Ljubljana, Slovenia) in 1996.

Introduction: The problem

It is widely held, by social scientists as well as by lay people, that the members of human groups have an "innate" propensity to distinguish between insiders and outsiders, to delineate social boundaries and to develop stereotypes about "the other" in order to sustain and justify these boundaries. If this is indeed the case, ethnicity can be conceived of as being nearly as universal a characteristic of humanity as gender and age - unlike phenomena like nationhood and nationalism, which have been so conceptualised in the academic community as to concern the modern world only (Anderson 1991, Gellner 1983). Marx and Engels held, probably correctly, that sex, age and the insideroutsider distinction were universal criteria of differentiation. If, on the other hand, ethnicity as we conceptualise it can be shown to be a product of a particular kind of society, it can of course not be regarded as an ahistorical and universal phenomenon.

The question to be explored in this paper, at this stage little more than a sketch, consists in this ambiguity in the conceptualisation of ethnicity. Can it fruitfully be regarded as a relational property of any social system, or should the concept rather be confined to a specific kind of historical society, notably those defined as modern societies? In exploring this question, I shall draw on selected theoretical and meta-theoretical contributions to the analysis of ethnicity, beginning with Fredrik Barth's "Introduction" to his edited volume Ethnic Groups and Boundaries (Barth 1969), which to my mind represents the clearest defence of an ahistorical concept of ethnicity.

Defining concept or defined space?

While certain arguments in Barth's discussion may seem obsolete (such as the section on ecology) and others were perhaps less original than they may have seemed at the time (cf. e.g. the first chapter of Leach's Political Systems of Highland Burma, Leach 1954, or Mitchell's The Kalela Dance, Mitchell 1956), his article represented an important break with a formerly widespread, essentialist view of ethnicity, and it therefore deserves its prominent place in the history of anthropological ethnicity studies. Instead of highlighting the concern with boundaries as a replacement for "cultural differences", which is the single most influential idea in the article, I shall draw attention to the underlying epistemological assumptions of the model.

The concept of ethnicity developed by Barth in his "Introduction" could roughly be labelled a naturalist one. Although a main original contribution of his essay consisted in stressing that ethnic identities are created from within and not by virtue of "objective" cultural differences, thereby giving him the label "subjectivist" in some quarters, he also makes it clear, if implicitly, that ethnic phenomena are endemic to humanity and not to any particular kind of society. More specifically, Barth locates the emergence of ethnic distinctions to differentiation within a society and the concomitant development of divergent standards of evaluation and constraints on interaction (Barth 1969, pp. 17-18). Disentangling the concept of ethnicity from concepts of race and culture, the main epistemological contribution of Barth's article consisted, perhaps, in his refining and relativising the concept of society seen as a natural phenomenon of cultural humans, while not discarding it completely. He shows that societies may be poly-ethnic and thus contain delineated and distinctive groups, that the boundaries of societies may be not only relative but also "permeable" in the sense that people may permanently cross into another society (i.e. another ethnic group), and finally, that the members of an ethnic group need not share all the characteristics deemed as defining of the group (a polythetic "family resemblance" is sufficient).

On the other hand, the actual status differentiation within a society (notably ethnicity as an imperative status) is taken for granted in the greater part of the text and so is, by implication, the social structure. The systems of relationships entailed by ethnicity in various contexts are implicitly regarded by Barth as comparable, and in the final parts of his essay (pp. 29ff.), he goes on to discuss contextual variations and their implications for analysis. The actual boundary mechanism which defines ethnicity is, in other words, held constant and is implicitly assumed to be context-independent. Ethnicity thus becomes, in Barth's version, an important defining concept and thereby a formal comparative concept, an analytical bridgehead not confined to any particular kind of society or historical era. Barth's view is underpinned by the other contributions to Ethnic Groups and Boundaries (cf. especially Haaland's, Izikowitz' and Knutsson's contributions), which largely deal with interethnic relations in non-modern or non-industrial societies where ethnicity has yet to become a mobilising force in mass politics.

The view of ethnicity presented in Abner Cohen's "essay on the anthropology of power and symbolism" (Cohen 1974a), as well as in his important "Introduction" to his edited ASA monograph on Urban Ethnicity (Cohen 1974b), differs from that of Barth in this regard. Wheras Barth could be represented as a moderate realist, Cohen makes it clear that he is a committed nominalist: to him, ethnicity is neither more nor less than a useful heuristic concept tailored to make sense of particular, historically delineated processes such as urbanisation in Africa. Cohen thus identifies ethnicity with the processes whereby "some interest groups exploit parts of their traditional culture in order to articulate informal organisational functions that are used in the struggle of these groups for power" (1974a, p. 91). A few pages on, he elaborates the notion by adding, among other things, that ethnicity "involves a dynamic rearrangement of relations and of customs and is not the result of cultural conservatism or continuity" (1974a, p. 97). In Cohen's analysis, ethnicity appears as neither more nor less than a form of corporate traditionalism, and is as such confined to modern circumstances - presumably presupposing institutional differentiation, literacy and the state.

However, like Barth, Cohen also depends on ahistorical, formal, defining concepts. Unlike Barth, who developed his boundary model partly to evade the pitfalls of structural-functionalism (cf. Barth 1966), Cohen embraces structural-functionalist explanation, using it explicitly as a general comparative frame into which he puts, under certain historical and political circumstances, the empirical phenomena classified as ethnicity. This logic is also evident in Cohen's ethnographic analyses of Hausa in Ibadan and Creoles in Freetown. In other words: Whereas ethnicity appears as a natural social phenomenon in Barth, it is relegated to the status of an historical contingent phenomenon in Cohen - in other words, as a part of the defined space; as a part of the society under scrutiny. A question to which we shall have to return is, obviously, whether the two authors have the same phenomenon in mind when talking about ethnicity.

Constructivism and historical accounts

Arnold Epstein's Ethos and Identity (Epstein 1978) marked a decisive shift in focus in the social anthropological study of ethnicity. Whereas Barth and Cohen implicitly agree that ethnicity is best seen as a kind of politics, Epstein calls attention to the identity dimension of ethnicity rather than the political dimension, and draws heavily on social psychology, notably Erik H. Erikson's work, in arguing that "we need to supplement conventional sociological perspectives by paying greater attention to the nature of ethnic identity" (Epstein 1978, p. 5). However, contrary to what one might fear from Epstein's programmatic statement at the outset of his three studies, he does not leave "conventional sociological perspectives" out. In a manner which resembles, probably not accidentally, Victor Turner's analysis of the work of ritual symbols, Epstein combines a cognitivist concern with symbolic meaning and a foundation in sociological analysis in what is one of the most profound studies of ethnic identity to date. It is also clear that Epstein, like Cohen, considers ethnicity largely as the creation of modernity. His cases, from the Copperbelt, Melanesia and contemporary Jewish diaspora, indicate that the formation of ethnic identities, and that socially organised and orchestrated communication of cultural distinctiveness which constitutes ethnicity, belong to situations of rapid and uncontrollable social change - in a word, the impact of the modern world. Epstein's social psychological approach could be seen as complementary to Cohen's sociological view, in that it looks into the non-utilitarian and nonfunctional aspects of individual meaning creation in the process of ethnogenesis, but deals with essentially the same kind of social situation.

A fourth main position could be described as the social constructivist view. Drawing inspiration from all three "classic" perspectives outlined, but defending a reflexive position, representatives of this view more explicitly and frequently more viciously dissociate ethnicity from "race" and "culture", often focusing on the ways in which ethnic identities and boundaries are historically arbitrary and the constructs of members of an élite looking for political power and/or material gain - or the construct of a dominating group seeking to intimidate dominated groups by imposing ethnic labels on them. Each in their way, Eugeen Roosens (1989) and the editors of the ASA monograph History and Ethnicity (Chapman et al. 1989) represent such a strategy, which stresses the importance of the "native's point of view" in the development of ethnic identities. Their views are perhaps truly "subjectivist" (unlike Barth's, which combines subjective and objective factors) since they regard ethnic groups as possible, but not necessary products of creative endeavours under particular historical circumstances. In their view, culturalist explanations of ethnicity are as invalid as racist explanations of social race, since ethnic identity formation involves the more or less haphazard appropriation of and overcommunication of alleged cultural traits. The degree to which societal factors are granted explanatory power within this exploratory matrix varies; Benedict Anderson, for example, could clearly be seen as a constructivist, although he insists on the necessity of objective, enabling technological forces for ethnic (or national) identities to appear.

The final approach to ethnicity studies to be mentioned here, could be described as the historical one. This view has emerged as a component of the general increased interest in historical analysis in anthropology, which began when the Marxist and so-called neo-Marxist currents were in fashion and which has continued up to this day. John and Jean Comaroff, in Ethnography and the Historical Imagination (Comaroff & Comaroff 1992), make a clear statement of related positions when they state, in their engaging comparison of ethnicity and totemism, that "contrary to the tendency to view it as a function of primordial ties, ethnicity always has its genesis in specific historical forces, forces which are simultaneously structural and cultural" (p. 50). The modern world renders values comparable through monetarisation; it also renders cultural differences comparable and thereby stimulates the emergence of ethnic groups - in Comaroff & Comaroff's words, "ethnicity has its origins in the asymmetrical incorporation of structurally dissimilar groupings into a single political economy" (p. 54). This kind of view is supported by several other scholars, for instance in John Peel's important work from Nigeria (e.g. Peel 1989), where he argues, contra Abner Cohen, that although Yoruba ethnicity emerged historically in response to social changes imposed by colonialism, it did emerge under specific circumstances which included important cultural dimensions - and that Yoruba ethnicity could not be accounted for satisfactorily without a consideration of cultural and historical factors. This argument could well be directed against extreme social constructivist positions as well as against Cohen's structural-functionalist explanation: Peel shows that although ethnic identities may appear as inventions, they are certainly not arbitrary inventions, and historical causation severely limits not only the range of options for the intentional construction of identities, but also their form.

The map–territory problem

The discussion engaged in so far may seem to resemble former metatheoretical discussions of the ethnicity concept; notably the old subjectivist–objectivist and primordialist–instrumentalist debates. However, these distinctions no longer seem to reflect the main theoretical differences. The positions outlined could perhaps be described as naturalist (Barth), structural-functionalist (Cohen), mentalist (Epstein), constructivist (Roosens, Chapman et al.), and historicist (Comaroff, Peel), and could apparently be arranged on a continuum from extreme subjectivism (the constructivist positions) to a moderate objectivism (Comaroff). However, to be fair, all positions grant that ethnicity is contingent on a combination of subjective and objective factors although the stress on either dimension varies.

Regarding the issue of culture as a determining force in ethnicity vs. culture as reified ideological fantasy, the most primordialistically inclined views are perhaps those of Epstein and Barth, but all authors dealt with have distanced themselves from the Weberian idea that ethnic identities are "natural" and could therefore be treated as independent variables. When Barth is labelled a "naturalist", this is merely intended to call attention to his formal, empirically empty model of ethnicity - his ethnicity concept is a defining one, but not an agent of causation.

The debates on subjective vs. objective and primordial vs. instrumental thus seem to have been transcended. The most pressing issue regarding the epistemological status of the concept of ethnicity, in other words, seems to be the relationship between ethnicity as a property of intergroup relationships tout court, and ethnicity as the product of a particular kind of historical situation.

In order to disentangle different aspects of this question, it may be useful to look into the actual conceptualisations. In other words: Do different analysts speak of the same thing when they say "ethnicity"?

Levels of inquiry

The most fundamental fact of ethnicity, as investigated by anthropologists, is the application of a systematic distinction between "we" and "the others". A mass of anthropological work from the most diverse societies indicate that such distinctions are indeed universal. Studies such as Middleton's of Lugbara witchcraft or Leach's of Kachin society, indicate that the we–they distinction is a perennial feature of human groups. The moment they come into contact with other groups, it seems, ethnicity appears. (With hindsight, we may, perhaps, add: Ethnicity then appears at least in the eye of the beholder.) Further, as Hirschfeld (1988) has argued, even very young children seem to be able distinguish more or less spontaneously between "kinds" of people - in other words, humans seem to be genetically predisposed for this kind of distinction. Epstein's view of ethnicity resonates with this perspective on humanity. He regards the search for, and psychological need for, a sound, secure and more or less bounded social identity as fundamental, and connects this need - under particular historical circumstances - to the formation of ethnic identities.

Are social identities, seen as contrastive labels and mutually exclusive classifications, necessarily and always to be considered as ethnic ones? Do any other features of relationship need to be present in order for a particular configuration to be labelled "ethnic"? This seems to me to be the heart of the matter. The Comaroffs' reply is "yes", and they classify e.g. the Nuer–Dinka relationship as a totemic one, not an interethnic one, since it is the expression of a "different kind of consciousness" (1992, p. 55). Ethnicity occurs when perceived cultural differences make a social difference; that much is granted. A useful comparative definition nevertheless requires greater accuracy, and it might here be noted that all of the analysts cited arrive at heuristically useful, although different definitions.

The next, inevitable question must be: Which conceptualisation of ethnicity is the most useful one in anthropological comparison? The empirically empty, formal concept advocated by Barth gives us the option of very wide-ranging comparisons of ethnic phenomena, as it theoretically includes Yanomamö–Cholo, Norwegian–Sami, Fur–Baggara, Black British–White British relationships as well as many other relationships which have little in common apart from the ongoing social reproduction of ethnic boundaries. At least in the case of the Fur–Baggara relationship, modernisation did not seem to have influenced it to any significant degree at the time of Haaland's (1969) fieldwork (notwithstanding O'Brien's, 1986, historical analysis of Sudanese ethnogenesis).

The outcome of comparisons at this level of generality is limited to highly abstract findings such as: Ethnicity implies both dichotomisation and complementarisation - both contrasting and comparison (Eidheim 1969); individuals may cross ethnic boundaries without disturbing their basic functioning (Haaland 1969); the relationship between culture and ethnicity is not a one-to-one relationship (Blom 1969); and at an even more general level: ethnicity entails making differences comparable. This kind of comparison, while it runs the risk of comparing apples and pears, clearly has its value since it enables us to ask further, more specific questions to our material. For example, the various interethnic situations may be compared with regard to openness vs. closure, dominance vs. equity, and the correlation between ethnicity and the division of labour.

What this conceptualisation of ethnicity does not promise, is a better understanding of the emergence of particular ethnic identities and particular interethnic relationships. That, however, can be achieved from research on particular societies; the point is that an all-encompassing, formal concept of ethnicity may serve as a starting-point or a bridgehead for the investigation of the unique features of a particular society (cf. Eriksen 1992:chaps. 1–2).

At this point, it could be objected that such a maneuvre entails a brutal and deeply positivistic comparison of contexts which cannot meaningfully be compared - for example individualistic and holistic societies, where "ethnicity" cannot exist in the same way. This would probably have been Dumont's argument, had he engaged in this discussion: elsewhere (e.g. Dumont 1983), he has criticised post-Tönniesian European social scientists for presuming that the European logic of action and individualist metaphysics were universal.

Modern reflexivity

In line with this reasoning, it has been argued that not only should ethnicity be seen largely as a construct, but it is indeed, in many cases, the construct of the analyst (Fardon 1987, Ardener 1989). Earlier, Aidan Southall (1976) has argued that the "Nuer" and "Dinka", seen as ethnic groups, were more or less the constructs of Evans-Pritchard and Lienhardt, respectively: neither people could conceivably define themselves as members of "peoples". Fardon and Ardener argue what is essentially the same point with respect to two West African peoples, the Chamba and Kole, respectively. Surely, Fardon concedes, the Chamba invoked "us–them" distinctions in precolonial times, but these were more fluid, more situational and overlapping, and on the whole less clear-cut than ethnic distinctions, where a definite boundary is being drawn.

The history of nationalism may give clues as to the origin of the current anthropological concept of the ethnic group. As a matter of fact, the concept of "tribe", as well as that of "ethnic group", is conceptually close kin to the European concept of the nation as culturally homogenous, united and sovereign. The very concept of the ethnic group thus appears as a child of nationalism - in which case it may be of limited use in the study of non-modern societies.

However, in Fardon's article at least, it turns out that the Chamba have eventually come to regard themselves as an ethnic group in a way roughly reminiscent of the anthropological conceptualisation of ethnic groups. As a matter of fact, they have partly picked up the anthropological usage, partly been influenced by social processes transforming their society into one where ethnic groups (in this sense) may come about - in a word, literate capitalist state societies with a formal educational system. In an epistemologically and methodologically confusing way, thus, so-called informants have in recent years appropriated the concepts and analyses of anthropologists and have thereby turned these concepts into empirical material or "data". The situation may be one of parameter collapse (Ardener's apt term) where our concepts of culture and ethnicity, formerly defining concepts par excellence, collapse into the defined space - rather than serving as conceptual footholds, they become part of the social reality which needs to be accounted for.

This, evidently, is the kind of ethnicity Roosens talks of in his analyses of the Huron indians and the Luba of Kasai (Roosens 1989). The Hurons, in particular, have self-consciously fashioned "a culture" and a social identity which fits perfectly with the contemporary view of minorities and minority politics, which enables them to pursue political goals within the context of the Canadian state - and which owes a lot to anthropological concepts and models. This option was not present to Haaland's Fur, although it seems clear that they held ideological notions about their cultural distinctiveness vis-à-vis the Baggara.

Ethnicity and diffusion

Studies such as Roosens' Creating Ethnicity indicate the fruitfulness of an historically bound ethnicity concept. In this book, which presents a wide range of interethnic situations, the author has tailored an ethnicity concept encompassing the sociopsychological and reflexive dimensions of ethnicity as well as the political ones. Simultaneously, Roosens argues that "among human beings, whatever their cultural tradition, a number of material goods and values, whose production originated in Western society, are highly desirable". In Roosens' analysis, ranging from the Hurons of Quebec to minorities in Belgium, ethnicity, while it has not necessarily been directly diffused, appears as a kind of response to the diffusion of certain vital dimensions of modernity.

A more explicitly diffusionist view is taken by Peter Worsley (1984), who remarks that it is surely not without academic interest that Tamils in Sri Lanka, before the separatist Tamil Eelam movement emerged, must have watched newsreels from the West Bank informing them about the Palestinian struggle.

This perspective is, if adopted in comparative studies of ethnicity, not without epistemological consequences. If we are to focus on ethnic self-consciousness as it is being spread by mass media or migration, or as it is being developed through similar historical processes in different societies, then the classical anthropological form of comparison through "quasi-experiments" relating discrete and presumably isolated societies to each other, must be discarded. In my view, it is clear that this method will have to go. In this seamless world, it is no longer feasible to keep variables constant in the anthropological quasi-experiment of comparison: there is too much distortion for this style of comparison to be viable. Instead, a sensible strategy for comparison must trace out interconnections between societies and account for their local expressions. These interconnections include "objective" as well as "subjective" aspects, and ethnicity, whether studied as contrasting identities or as political organisation, must in today's world be viewed in relation to globalisation processes. This does not imply that ethnicity is "ontologically" merely a by-product of capitalism, the state and modern mass media, but that its expressions are at least today contingent on such parameters.

Such an historically bound concept of ethnicity enables us to compare modernities within a larger conceptual framework, which I cannot present here (but cf. Eriksen 1993:ch. 8). A few points can nevertheless be suggested, however briefly.

Edvard Hviding, a Melanesianist, has argued (1993) that the general tendency towards cognatic kinship among Solomon Islanders has in recent years been challenged by a concern with unilinear kinship, especially of the patrilineal kind. Hviding explains this as an expression of political interests in achieving territorial rights and tight corporate organisation at the clan level. This development is very similar to the development described by Fardon (1987), in his study of Chamba ethnogenesis: it is contingent on modernity, colonialism, literacy, private property, individualism and several other parameters which, after their introduction, makes discrete contexts comparable along new lines. In other words, an historical location of ethnicity as we know it may include contextual variables as well as the naked facts of ethnic distinctions, and makes it possible to compare empirical features of societies, not simply formal characteristics.

Concluding remarks
Instead of framing the question of the nature of ethnicity somewhere within or around the familiar instrumentalist/primordialist dichotomy, I have chosen to discuss it as a concept. For ethnicity is first and foremost a concept and not a natural phenomenon. As a concept it exists at (at least) two levels, that of the analyst and that of the native. Is it possible - or even desirable - to keep the two apart? If we do, it will retain its wide-ranging comparative potential; if we do not, it will enable us to describe local contexts in a more experience-near fashion than otherwise.

Ethnicity can be seen as a universal social phenomenon, and it can be seen as a modern cultural construct. It can be conceptualised as a peculiar kind of informal political organisation (Cohen), as an aspect of personal identity involving contrastive, mutually exclusive labelling (Epstein), as the reflexive appropriation of a "cultural estate", history and concomitant political rights (Roosens), as a product of colonialism and capitalism (Comaroff, Fardon), or as a functional boundary mechanism separating endogamous groups (Barth). If say, the reflexive, traditionalist self-identity aspect forms the focus of the analysis, a narrow, historically bounded concept of ethnicity is called for. If, on the contrary, the research aims at mapping out say, aspects of basic processes of interaction or of social identity formation, then a more encompassing and formal conceptualisation is needed. The question should not, therefore, be framed as "what is ethnicity", but rather as "how can we most fruitfully conceptualise ethnicity?". I suspect that most controversies over definitions stem from an inadequate distinction in this regard: from an implicit (or explicit) concern with "essences" and a positivist innocence with regards to the ontological status of our concepts. Most of them, in the social disciplines at least, have a fairly short lifespan, and there is no reason to believe that the presently – still – useful concept of ethnicity will still be with us in a few years. As I have insinuated, the boundary between ethnicity as folk concept and as analytical tool is currently under stress - natives have their own, anthropologically informed theories of ethnicity - and it seems that the concept, hitherto a defining concept, is about to collapse, as Ardener would have put it, into a defined space. Perhaps it is time to prepare ourselves to replace our ethnicity concepts with terms like traditionalism, culturalism, politicised culture and informal political organisation in an increasing number of cases. Such concepts will create a new defined space, possibly one more beneficial to research. And possibly not.


Anderson, Benedict (1991 [1983]) Imagined Communities. Reflections on the Origins and Spread of Nationalism, 2nd edition. London: Verso.
Ardener, Edwin (1989) The Voice of Prophecy and other essays, ed. Malcolm Chapman. Oxford: Blackwell.
Barth, Fredrik (1966) Models of Social Organization. London: RAI Occasional Papers, no. 23.
- (1969) Introduction. In Fredrik Barth, ed., Ethnic Groups and Boundaries. The Social Organization of Culture Difference, pp. 9-38. Oslo: Scandinavian University Press.
Blom, Jan-Petter (1969) Ethnic and cultural differentiation. In Fredrik Barth, ed., Ethnic Groups and Boundaries. The Social Organization of Culture Difference, pp. 75-85. Oslo: Scandinavian University Press.
Chapman, Malcolm, Maryon McDonald & Elizabeth Tonkin (1989) Introduction - history and social anthropology. In Elizabeth Tonkin, Maryon McDonald and Malcolm Chapman, eds., History and Ethnicity, pp. 1-21. London: Routledge.
Cohen, Abner (1974a) Two-Dimensional Man. London: Tavistock.
- (1974b) Introduction: The lesson of ethnicity. In Abner Cohen, ed., Urban Ethnicity, pp. ix-xxii. London: Tavistock.
Comaroff, John and Jean Comaroff (1992) Ethnography and the Historical Imagination. Boulder: Westview.
Dumont, Louis (1983) Essais sur l'individualisme. Paris: Seuil.
Eidheim, Harald (1969) When ethnic identity is a social stigma, in Fredrik Barth, ed., Ethnic Groups and Boundaries. The Social Organization of Culture Difference, pp. 39-57. Oslo: Scandinavian University Press.
Epstein, A.L. (1978) Ethos and Identity: Three Studies in Ethnicity. London: Tavistock
Eriksen, Thomas Hylland (1992) Us and Them in Modern Societies: Ethnicity and Nationalism in Trinidad, Mauritius and Beyond. Oslo: Scandinavian University Press.
- (1993) Ethnicity and Nationalism: Anthropological Perspectives. London: Pluto.
Fardon, Richard (1987) 'African ethnogenesis': Limits to the comparability of ethnic phenomena. In Ladislav Holy, ed., Comparative Anthropology, pp. 168-188. Oxford: Blackwell.
Gellner, Ernest (1983) Nations and Nationalism. Oxford: Blackwell.
Haaland, Gunnar (1969) Economic determinants in ethnic processes, in Fredrik Barth, ed., Ethnic Groups and Boundaries. The Social Organization of Culture Difference, pp. 58-74. Oslo: Scandinavian University Press.
Hirschfeld, Lawrence (1988) On acquiring social categories: cognitive development and anthropological wisdom. Man, vol. 23:611-38.
Hviding, Edvard (1993) Personal communication.
Izikowitz, Karl G. (1969) Neighbours in Laos, in Fredrik Barth, ed., Ethnic Groups and Boundaries. The Social Organization of Culture Difference, pp. 135-48. Oslo: Scandinavian University Press.
Knutsson, Karl Eric (1969) Dichotomization and Integration, in Fredrik Barth, ed., Ethnic Groups and Boundaries. The Social Organization of Culture Difference, pp. 86-100. Oslo: Scandinavian University Press.
Leach, Edmund R. (1954) Political Systems of Highland Burma. London: Athlone.
Mitchell, J. Clyde (1956) The Kalela Dance. Manchester: Manchester University Press, Rhodes-Livingstone Papers, no. 27.
O'Brien, Jay (1986) Toward a reconstitution of ethnicity: Capitalist expansion and cultural dynamics in Sudan. American Anthropologist, vol. 88, pp. 898-906.
Roosens, Eugeen E. (1989) Creating Ethnicity. London: SAGE.
Southall, Aidan (1976) Nuer and Dinka are people: ecology, ethnicity and logical possibility. Man, vol. 11(4), pp. 463-91.
Worsley, Peter (1984) The Three Worlds: Culture & World Development. London: Weidenfeld & Nicholson.

©Thomas Hylland Eriksen 1993/1996