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.

























Ethnicity and Nationalism

Anthropological Perspectives

London: Pluto Press

Third edition (2010)

Scroll down for chapter 2 (Us and them: Ethnic classification)




Preface to the third edition

It would be an exaggeration to claim that our entire way of thinking about ethnicity and nationalism has changed since the second edition of this book was completed in 2002, but it cannot be denied that research agendas have moved on and shifted somewhat in response to changing historical circumstances; new themes have been introduced, and some old themes have been rephrased, sometimes for the better. A few new topics in this edition, dealt with cursorily or not at all in the first two editions of this book, are cultural property rights, the role of genetics in the public understanding of identification, commercialisation of identity, and the significance of the internet. Arguments about globalisation, hybridisation and the need for a more inclusive concept of identity politics have been developed further, as have the sections about the relative degree of group cohesion, the role of culture in ethnic identification, the concept of race, and migration. Apart from these fairly major revisions, I have updated the text and made minor changes where necessary.

     As always, I am grateful to my students, colleagues and translators to languages other than English for their encouragement, but also for pointing out inconsistencies, debatable points, lacunae and incomplete arguments, and I have done my best to deal with relevant objections.


2  ethnic classification: Us and them



He came of good class, had a light olive complexion and hair with large waves (‘good’ hair, Miss Henery thought of it as; as a member of the West Indian coloured middle class, she conceived of human hair in terms of ‘good’ and ‘bad’ – sometimes ‘good’ and ‘hard’; ‘good’ hair is hair that is European in appearance; ‘bad’ or ‘hard’ hair is of the kinky, negroid type).

Edgar Mittelholzer (1979: 58)

The first fact of ethnicity is the application of systematic distinctions between insiders and outsiders; between Us and Them. If no such principle exists there can be no ethnicity, since ethnicity presupposes an institutionalised relationship between delineated categories whose members consider each other to be culturally distinctive. From this principle, it follows that two or several groups who regard themselves as being distinctive may tend to become more similar and simultaneously increasingly concerned with their distinctiveness if their mutual contact increases. Ethnicity is thus constituted through social contact. This chapter will present general aspects of these processes of contact. In later chapters, wider contexts for ethnic relations at the interpersonal level will be elucidated – from the formation of ethnic groups (chapter 3) and the creation of ethnic identities and ideologies (chapter 4), to the historical conditions for ethnicity (chapter 5), the relationship between ethnicity and the state, including debates about nationalism, indigenous peoples and multiculturalism (chapters 6, 7 and 8), and the implications of globalisation for identity politics (chapters 8 and 9). Although ethnicity is not wholly created by individual agents, it can simultaneously provide agents with meaning and with organisational channels for pursuing their culturally defined interests. It is very important to be aware of this duality.

The ecology of the city


Some of the earliest empirical research on complex polyethnic societies was undertaken by the group which has come to be known as the Chicago School, comprising urban sociologists as well as anthropologists (Park, 1950; cf. Hannerz, 1980). Among the main problems investigated by Robert Park and his associates in the 1920s and 1930s was how it could be that ethnic groups remained distinctive in American cities – and to what extent they did so through time. In other words, they were concerned with continuity and change in ethnic relations. We owe the widespread use of concepts of ‘acculturation’ and ‘the American melting-pot’ to the efforts of Park and his colleagues. By acculturation, they meant the adaptation of immigrants to their new cultural context. It could, but did not have to, eventually lead to total assimilation or loss of ethnic distinctiveness.
Park regarded the city as a kind of ecological system with its own internal dynamic, creating diverse opportunities and constraints for different individuals and groups. At the same time it contained several distinct ‘social worlds’ based on class, ‘race’ or national origin. These social worlds corresponded to distinctive physical neighbourhoods divided by unequal access to economic resources as well as ethnic differences. The combination of economic adaptation and ethnic identity thus created ‘natural areas’ such as Little Sicily and the ‘Black Belt’ in Chicago, more or less sharply distinguished from each other through their respective places in the division of labour and the cultural identities of their inhabitants. Economic, political and cultural resources were to a great extent pooled within each ethnic subsystem so that the individual could achieve many of his or her goals through ethnic networks. Mobility within the system as a whole could be achieved through acculturation – the adoption of the white, English-speaking majority’s values and ways of life – which in turn depended on the economic success of individuals or groups. ‘The typical “race relations” cycle,’ remarks Ulf Hannerz (1980: 44) in an assessment of the Chicago School, ‘would lead from isolation through competition, conflict, and accommodation to assimilation.’ Park generally assumed that ‘acculturation’ would eventually replace ethnic entrenchment, except in the case of the blacks. Another influential analysis of ‘acculturation’ from the same period is Bateson’s article about culture contact and schismogenesis, published in 1935 (Bateson 1972a*), where Bateson argues that contrary to what many expect, group differences may just as well be accentuated as reduced in situations of contact.

A main point in Park’s work is that every society is a more or less successful melting-pot where diverse populations are merged, acculturated and eventually assimilated, at different rates and in different ways, depending on their place in the economic and political systems.

The melting-pot metaphor

The American notion of the melting-pot has a long pedigree. It seems to have been used first in Crèvecoeur’s Letters from an American Farmer (1904/1788), where the author asked ‘What is the American, this new man?’ and answered ‘here individuals of all nations are melted into a new race of men, whose labours and posterity will one day cause great changes in the world’.   The philosopher Emerson spoke about ‘the smelting pot’ in the mid-19th century, and with Israel Zangwill’s immensely popular play ‘The Melting Pot’ from 1908, the term became a label of self-description for many new Americans.
The interethnic contexts investigated by the Chicago School were transient, recently constituted and perhaps atypical. In 1900, almost 80 per cent of Chicago’s population consisted of immigrants and their children; as late as 1930, about 35 per cent of the population were foreign-born. Following the ‘ethnic revival’ of the 1960s and 1970s, it has become commonplace to criticise the notion of the melting-pot for having been empirically wrong since it predicted the demise of ethnicity. As a matter of fact, the critics would maintain, the diverse ethnic groups never merged, and indeed the differences between them seem to have been accentuated after two generations or more of mutual adaptation.

This kind of development (cf. chapter 7) might have been surprising, but not incomprehensible to Park. He stressed that ethnicity, and ethnic conflict (or race prejudice), was an aspect of the relationship between groups and that it was caused by threats, real or imaginary, to an existing ‘ecological pattern’ of mutual adjustment. In other words, the social mobility – downwards or upwards – of any ethnic group would lead to tension in relation to the other groups.

Park was also aware of the fluid character of ethnic categorisations. As an individual moves between social contexts in the flux and transience of urban life, the relative importance of his or her ethnic membership changes. Thus an ‘individual may have many “selves” according to the groups to which he belongs and the extent to which each of these groups is isolated from the others’ (Park, 1955 [1921], quoted in Lal, 1986: 290).
Later scholars have criticised some of the premises from the Chicago school. Expanding his critique to include later representatives of the school, notably Robert Redfield, A. P. Cohen mentions three ‘myths’ of the Chicago school (A. P. Cohen, 1985: 28ff.): The myth of simplicity (the idea that rural societies were by default simpler than urban ones); the myth of egalitarianism (also assumed to be typical of rural societies), and the myth of inevitable conformity (in rural society). While they doubtless exaggerated the contrast between urban complexity and (assumed) rural simplicity, several insights of the Chicago School have proved to be of lasting value in the study of ethnicity: they showed that ethnic relations are fluid and negotiable; that their importance varies situationally; and that, for all their claims to primordiality and cultural roots, ethnic identities can be consciously manipulated and invested in economic competition in modern societies. The trend in American ethnicity studies (as well as in sociology more generally) known as symbolic interactionism (Gans, 1979) was initiated by the Chicago School. As will now be made clear, conclusions which were by and large compatible with those of the Chicago School also emerged, slightly later, from anthropological studies of ‘tribalism’ and interethnic relations in urbanising Southern Africa.

Communicating cultural difference

The intergroup contacts that constitute ethnicity may be caused by a variety of factors, among them population growth, the establishment of new communication technologies facilitating trade, inclusion of new groups in a capitalist system of production and exchange, political change incorporating new groups in a single political system, and/or migration.

In the 1930s, there was a growing demand for labour in the copper mines in the part of Northern Rhodesia (now Zambia) known as the Copperbelt. This spurred a stream of migration from rural areas to the mining towns, and the workers were settled in large barracks. There were several important changes to the social situation of these workers. They had until recently been subsistence farmers in rural villages; now they had become wage labourers in towns with a predominantly monetary economy. In most cases, their social organisation had formerly been based on kinship; now they were tied to the mining enterprise through individual labour contracts. Most of the workers lived alone in the barracks. If they were married, their families were left behind in the village, at least at the early stages. Finally, they were taking part in a social system of a much larger scale and greater complexity than formerly. Whereas the villagers were more or less self-sufficient and had only sporadic contacts with members of other ethnic groups, as town-dwellers they were in continuous interaction with a large number of individuals from ethnic groups other than their own. They shared housing, working places and leisure facilities with others. In some of the towns, dozens of ‘tribes’ were represented.

This process of urbanisation was investigated by anthropologists based at the Rhodes-Livingstone Institute in Lusaka, North Rhodesia, now the capital of Zambia. Among the most prominent of these were Godfrey Wilson, Max Gluckman, J. Clyde Mitchell and A. L. Epstein. This group is today known as the Manchester School because of its members’ later affiliation with the University of Manchester. Some of their studies, including Wilson’s Essay on the Economics of Detribalization (G. Wilson, 1941–2), focused almost exclusively on change, whereas others, such as Mitchell’s small monograph The Kalela Dance (Mitchell, 1956), looked into the relationship between social and cultural change and continuity. More recently, James Ferguson (1999) has produced an updated ethnography of the Copperbelt, tracing historical changes over the decades. Whereas Wilson described what he saw as a process of detribalisation, Mitchell emphasised that a form of retribalisation (what we would today call modern ethnicity) was taking place in the mining towns.

Although kin groups and ‘tribes’ were economically relatively unimportant in the towns, group membership was emphasised to the extent of being overcommunicated (Goffman, 1959) in public rituals as well as in casual interaction. This means that ethnicity was deliberately ‘shown off’. In other polyethnic situations ethnicity may rather be undercommunicated, which means that the actors tried to play it down and not to make it an important aspect of the definition of a situation.

Although people in the towns were not socially organised along tribal or ethnic lines, they grew strongly self-conscious of their ethnic identity under these circumstances of extensive contact with others. They developed standardised ways of behaving vis-à-vis each other, and oriented themselves socially according to ethnic ‘maps’ which would have been quite superfluous in a village setting, where most of one’s contacts were intraethnic. Many of the new social subsystems that developed in the urban environment, such as clubs and peer groups assembling in beer-halls, were based on ethnic membership.

Mitchell (1956) focuses on one such new institution, the kalela dance. It was performed every Sunday afternoon in Luanshya by male members of the Bisa category. They were dressed in a modern way, and the dance did not form part of the group’s traditional cultural repertoire. However, the kalela dance and accompanying songs were conspicuous and overt markers of group identity: most of the songs ridiculed the other groups and praised the homeland of the Bisa. Similar performances were carried out by other groups as well. In this way people’s social identities were established and emphasised in a striking way. In a village setting such rituals would have been unnecessary, both because the inhabitants knew each other and because villages were as a rule mono-ethnic.


In the Copperbelt, cultural differences were communicated in private situations as well. When two individuals met for the first time, the first information they would gather about one another would be their ethnic membership. When this fact was established they would know roughly how to behave towards each other, since there were standardised relationships between groups. Some groups had a ‘friendly’ relationship, some had a ‘hostile’ one, and yet others had ‘joking’ relationships. If one knew someone’s ethnic identity, one would know what kind of behaviour towards them would be appropriate. The members of each group had particular notions about the vices and virtues of the others, and these notions were articulated and dramatised in public rituals such as the kalela dance.

When such notions become part and parcel of the ‘cultural knowledge’ of a group and thus regularly and more or less predictably guide their relationships with others, we may describe them as ethnic stereotypes. Mitchell explains:

Town-dwellers display their ethnic origin by the language they speak and their way of life generally. This enables members of other tribal groups immediately to fit their neighbours and acquaintances into categories which determine the mode of behaviour towards them. For Africans in the Copperbelt ‘tribe’ is the primary category of social interaction, i.e. the first significant characteristic to which any African reacts in another. (Mitchell, 1956: 32)

Stereotypes are often mentioned in connection with racism and discrimination, so that, for example, white Americans may justify discrimination against blacks by referring to the latter’s ‘lazy and erratic ways’. Stereotypes tend to be more or less pejorative, although this is not necessarily the case. Many Europeans have positive stereotypes of ‘primitive peoples’, arguing that their quality of life is higher than their own. Used analytically in social anthropology, the concept of stereotyping refers to the creation and consistent application of standardised notions of the cultural distinctiveness of a group. Stereotypes are held by dominated groups as well as by dominating ones, and they are widespread in societies with siginficant power differences as well as in societies where there is a rough power equilibrium between ethnic groups. In most polyethnic societies, ethnic stereotypes exist, although there always exist individuals who do not hold such stereotypes – as well as individuals who are acknowledged to be ‘exceptions’.

In the polyethnic Indian Ocean island society of Mauritius, the entire population of slightly over one million consists of the descendants of immigrants who have arrived in successive waves since the French colonisation in 1715. The most important ethnic categories are Hindus and Muslims (of Indian descent), Creoles (of largely African and Malagasy descent), Coloureds (of ‘mixed’ descent), Sino-Mauritians (of Chinese descent) and Franco-Mauritians (of French and British descent). The groups tend to have mutual stereotypes of each other and of themselves (Eriksen, 1988; 1998; see also Boswell, 2006). The most important of these stereotypes are summarised in Figure 2.1.


Creoles            Lazy, merry, careless
Hindus            Stingy, dishonest, hardworking
Muslims            Religious fanatics, non-minglers
Sino-Mauritians             Greedy, industrious
Franco-Mauritians            Snobbish, decadent, undemocratic
Coloureds            Clever, conceited, too ambitious


Creoles            Funloving, compassionate, friendly
Hindus            Sensible, care for family
Muslims            Members of a proud, expanding culture
Sino-Mauritians            Clever, industrious
Franco-Mauritians            ‘True Mauritian’, dignified
Coloureds            ‘True Mauritian’, intelligent

Figure 2.1: Mauritian ethnic stereotypes Source: Eriksen, 1988.

Here, we should keep in mind that actual interethnic relations may very well diverge from stereotypes as they are presented in casual conversations; that there may be a discrepancy between what people say and what they do. In a famous study of the relationship between attitudes and actions in the US, La Piere (1934) toured the American West Coast with a Chinese couple and visited a large number of restaurants and hotels with them. They were refused service only once. He then sent out a questionnaire to the owners of the establishments, asking them whether or not they would accept ‘members of the Chinese race’ as guests. The vast majority affirmed that they would not.

Stereotypes need not refer to a social reality, and they do not necessarily give accurate hints of what people actually do. Therefore, we must reflect on the causes and uses of stereotypes.
First of all, in Mauritius as well as in the Copperbelt, stereotypes help the individual to create order in an otherwise excruciatingly complicated social universe. They make it possible to divide the social world into kinds of people, and they provide simple criteria for such a classification. They give the individual the impression that he or she understands society.

Second, stereotypes can justify privileges and differences in access to a society’s resources. Conversely, negative stereotypes directed towards a ruling group may alleviate feelings of powerlessness and resignation: they can be seen as the symbolic revenge of the downtrodden.

Third, stereotypes are crucial in defining the boundaries of one’s own group. They inform the individual of the virtues of his or her own group and the vices of the others, and they thereby serve to justify thinking that ‘I am an X and not a Y.’ In the vast majority of cases stereotypes imply, in some way or other, the superiority of one’s own group. However, there are also minorities who have largely negative stereotypes of themselves and positive ones of the dominating group.

Stereotypes can sometimes function as self-fulfilling prophecies. A dominating group can stunt the intellectual development of a dominated group by systematically telling them that they are inferior. There are, of course, many stereotypes which have little or no truth, such as the ideas traditionally held by many African peoples and others to the effect that their neighbours are cannibals (Arens, 1978).

Finally, stereotypes can be morally ambiguous and contested by different parties. In Mauritius, it is often said that ‘If a Creole has ten rupees, he will spend fifteen; but if a Hindu has ten rupees, he spends seven and invests the rest.’ This saying is sometimes invoked by Creoles as well as Hindus as proof of their own community’s moral superiority.

The moral character of stereotyping is not the main point here. Rather, it should be emphasised that stereotypes contribute to defining one’s own group in relation to others by providing a tidy ‘map’ of the social world, and that they can be invoked in attempts to justify systematic inequalities in access to resources.

Folk taxonomies and social distance

As noted above, informal groupings in the Copperbelt tended to be based on ethnic membership. For example, a vast majority of town-dwellers chose drinking companions from their own ‘tribe’ or ethnic category. In the barracks, they preferred to have room mates from their own group. However, if this was not possible they would rather share their room with people whom they perceived as close than with people they perceived as distant (Mitchell, 1974). Perceptions of distance, Mitchell notes, combined cultural and geographic criteria so that, for example, matrilineal peoples from the north would rank other matrilineal peoples from the north as those closest to themselves. In a large survey of townspeople (which was probably male-biased), Mitchell and his assistants used the following scale of ‘stages of social distance or social nearness’:

1. Would admit him to near kinship by marriage.
2. Would share a meal with him.
3. Would work together with him.
4. Would allow to live nearby in my village.
5. Would allow to settle in my tribal area.
6. Would allow as a visitor only in my tribal area.
7. Would exclude from my tribal area.
(Mitchell, 1956: 23)

On the basis of such perceptions of social distance, the town-dwellers developed – and reconfirmed, through interaction – a system of social classification where one did not just distinguish between Us and Them, but where there were various degrees of group inclusion and exclusion. In other words, there were different Us and Them groups. Depending on the situation, different levels of group membership could be activated. For instance, in local politics an individual would behave as a member of a larger group than he or she would concerning questions of marriage.

I have explored the functioning of ethnic classification in Mauritius, which officially has four ethnic ‘communities’; that is to say, the Constitution of Mauritius acknowledges the existence of four communities: Hindus (52 per cent), Muslims (16 per cent), Sino-Mauritians or Chinese (3 per cent) and ‘general population’ (29 per cent). The general population is a residual category which encompasses people of African, European and mixed descent. Nearly all of them are Catholics, but they do not consider themselves, nor are considered by others, an ethnic group. They rarely intermarry and do not vote together at elections. Moreover, it transpires that the ‘Hindus’ cannot be considered an ethnic group either, especially since this category includes both Biharis from north India (the most numerous segment) and a fair proportion of Tamils and Telugus, who do not identify themselves as members of the same ethnic group as the northerners, and who have periodically formed their own political parties.

It is impossible to tell straightforwardly how many ethnic groups exist in Mauritius. Cultural differences are communicated in a variety of situations, but they do not always refer to the same social distinctions. A Mauritian Hindu, for example, can be morally and socially compelled to marry at the caste level, but will usually vote for the party representing all (northern) Hindus. Further, distinctions are made between groups whose existence is ignored by other Mauritians, such as when Creoles distinguish between Rodriguais (from Mauritius’ island dependency Rodrigues) and Mauritian Creoles. Similarly, Mauritian Tamils would distinguish between urban and rural Tamils, sometimes to the extent of discouraging intermarriage, but such a distinction is not widely known outside of the Tamil ‘community’. To the question of ‘how many ethnic groups exist in Mauritius?’, therefore, we must reply that this depends on the situation.

As a general rule, ethnic folk taxonomies are at their most detailed closest to the actor. To a white Franco-Mauritian, it is of little consequence that Shi’ite and Sunni Muslims do not intermarry, or that there is little political loyalty between Marathis and Biharis (both of them Hindus). To the agents themselves, such distinctions may be of great importance in practical matters as well as in matters relating to identity and definition of self in relation to others.

Contrasting and matching

Many studies of ethnicity have stressed the relative distinctiveness of ethnic groups. Very often it is taken for granted that the groups in a polyethnic social system remain apart and different in most regards, and a great number of studies focus on the ways in which the groups manage to remain discrete (see chapter 3). However, since ethnicity is an aspect of relationship, one may equally well stress the mutual contact and the integrative aspect. To some extent this was emphasised in Fredrik Barth’s early study of ethnic ‘niches’ in Swat, where the biological metaphor of symbiosis was used to describe group relations (Barth, 1956), and it was a central point in the Chicago School (for example Wirth, 1956 [1928]) that the degree of isolation varied in interethnic relationships. Barth showed how the three ethnic groups of Swat valley (in north Pakistan), the Pathans, the Kohistanis and the Gujars, had adapted economically not only to the natural environment but also to the human aspect of their environment; that is to say, to each other. They had gradually developed mutual interdependencies through trade, exchanging necessities and services each of them had specialised in providing. The transhumant mountain Gujars, for example, depended on the lowland Pathans for fodder, while the Pathans bought dairy products from the Gujars.

In Harald Eidheim’s (1971) studies of the Sami in northern Norway, processes of interethnic accommodation are described in great detail at the level of interaction. Eidheim shows how negative stereotyping can be interrelated with a shared cultural repertoire – indeed, that both aspects are probably necessary components of a stable system of interethnic relations.

Group membership and loyalties are confirmed and strengthened through stereotyping and the articulation of conflict or competition between Sami and Norwegians. This mutual demarcation process can be called contrasting, or in Eidheim's terms, ‘dichotomisation’.

For interethnic interaction to take place at all, however, there must be some mutual recognition inherent in the process of communicating cultural differences. Otherwise, the ethnic identity of at least one of the parties will necessarily be neglected and undercommunicated in a situation of interaction. Such an acknowledgement of differences can be labelled matching (Eidheim uses the term ‘complementarisation’). Here, the cultural differences communicated through ethnicity are considered a fact and frequently an asset. Whereas contrasting essentially expresses an Us–Them kind of relationship, matching can be described as a We–You kind of process. When one enters an interethnic relationship, it is necessary to establish a field of complementarity. This could be a shared language within which interaction can take place.

In relation to power, matching can lead to two opposite results. Indigenous and other minority movements which seek recognition by the majority may try to establish an ideology of complementarity in order to be able to negotiate on an equal footing with the majority. On the other hand, dominant groups may also speak of complementarity in order to justify exploitation of and discrimination against minorities. This may be particularly relevant in societies with an ethnic division of labour, where, for example, particular ethnic groups carry out most of the underpaid manual work. In such situations, dominant groups may emphasise that it is the ‘nature’ of the members of group X to do manual work; that they are ‘unsuitable’ – by nature or by culture – to carry out prestigious jobs. The former apartheid system of South Africa exemplifies this hierarchical kind of complementarity, as did race relations in the USA before the civil rights movement. The parallel with gender studies is obnvious here; male-centred (or androcentric) ideologies of gender tend to justify the subjugation of women by referring to ideals of complementarity.

An important point demonstrated by the preceding discussion is that interethnic relations are not necessarily conflictual. Although there are frequently discrepancies of power (in Swat, the Pathans are clearly the dominant group), interethnic systems of communication and/or exchange may well be based on cooperation and mutual acknowledgement. Indeed, if there is little complementarisation in interethnic relations, there will usually be a tendency towards identity shift or assimilation among members of the weaker group. To sum up: ethnicity entails the establishment of both Us–Them differences (contrasting) and a shared field for interethnic discourse and interaction (matching).

Ethnic stigma

Although it has scarcely been accorded a central place in the anthropological study of ethnicity (as opposed to the sociological tradition of studying ‘race relations’), it is a fact that many interethnic relations are highly asymmetrical regarding access to political power and economic resources. It therefore seems appropriate at this point to present an interethnic relationship which has for centuries been marked by clearly hierarchical aspects.

Unlike the transhumant Sami of the mountain tundra of northern Scandinavia, the Sami of the Norwegian Arctic coast are not reindeer herders. Like the Norwegians who live in the same area, they obtain their livelihood from fishing and marginal agriculture. The two populations have been in contact for many centuries. They occupy the same economic niche, they live in the same kinds of houses, wear the same kind of clothing and practise the same Protestant religion. Upon arriving in one of these mixed communities, Eidheim (1969; 1971) looked in vain for cultural traits distinguishing Sami from Norwegians. During the first months of his fieldwork, the locals took great pains to show off their Norwegianness. They always spoke the local Norwegian dialect. The housewives had what to Eidheim seemed a craze for cleanliness. (Uncleanliness was considered a typical Sami vice by ethnic Norwegians.) On the face of it, there were no Sami in the community. However, although ‘there is a conspicuous lack of “contrasting cultural traits” between … [Sami] and Norwegians, … these ethnic labels are attached to communities as well as to families and individual persons, and are in daily use’ (Eidheim, 1971: 51).

Gradually some of Eidheim’s informants took him into their confidence, realising that he, a southerner and an unusual one at that, had no stake in the local interethnic system. As he grew to know them better, it turned out that many of the locals habitually spoke Sami (a Finno-Ugric language unrelated to Norwegian) at home. Indeed, a majority of the fjord population were Sami. However, it was impossible to engage people in conversations about ethnicity in public. In such situations, at the shop or at the quay for example, people would always act emphatically Norwegian. They would certainly speak Norwegian in such situations.

In this part of the country the Sami have traditionally been the weaker party in a patron–client relationship, and they had for centuries been considered primitive, backward, stupid and dirty by the dominant Norwegians. Therefore, Sami ethnic identity was consistently undercommunicated in public situations. Conversely, their command of modern Norwegian culture was strongly overcommunicated; they presented themselves as Norwegians to others. Sami identity became a kind of secret. Still, everybody in the community knew who was ‘really’ a Sami and who was not. Thus a total identity change was nearly impossible in the short run (say, within an individual’s lifetime), even if there were few ‘objective cultural differences’ between Sami and Norwegians. Since it was connected with undesirable and presumably immutable personality traits, Sami identity could be described as a stigmatised identity. Being recognised as a Sami entailed that one was considered inferior to Norwegians, and this, of course, was the main reason why Sami identity was being undercommunicated. Moreover, many Sami themselves shared the dominant, pejorative view of Sami culture, and refused to teach their children Sami. This kind of self-contempt is characteristic of powerless groups in polyethnic contexts.

Since the 1950s, the mountain Sami have gone through a process of ethnic incorporation: they have organised themselves politically on an ethnic basis. This coastal Sami population has rather moved towards assimilation, gradually losing their markers of distinctiveness and merging into the majority population. Eventually, it seemed at the time of Eidheim’s fieldwork, the descendants of these Sami would become Norwegian, just like the inhabitants of many small fishing communities on this coast, which were formerly Sami but which are now – after generations of cultural ‘Norwegianisation’ – considered Norwegian. This kind of process is very common among discriminated minorities, but it presupposes that there is a real, practical possibility of removing the stigma imposed by the dominant population. If, for example, the Sami had been physically very different from the Norwegians, the process of assimilation would probably have been more difficult.

It should be noted, however, that many coastal Sami have remained ‘split’ between Norwegian and Sami identities in a sometimes problematic way (Hovland, 1996). Aware of their Sami ancestry and of the fact that their grandparents (and sometimes parents) had a way of life that was very distinctive from the Norwegian one, many feel attached to their Sami identity despite its low public status. In other Sami areas there has actually been a strong Sami revitalisation movement in recent years, proclaiming the virtues of Sami identity in a manner reminiscent of the ‘Black is beautiful’ movement in the United States.

Negotiating identity

An important insight from the Copperbelt studies, foreshadowed in Robert Park’s ‘urban ecology’, was that ethnicity and social identities in general are relative and to some extent situational. As Mitchell writes, an individual can behave as a ‘tribal’ in some situations and as a ‘town-dweller’ in others (Mitchell, 1966). This fact should remind us that even in typical polyethnic societies where cultural differences are pervasive, there are many situations where ethnicity does not matter. This holds good not only in intraethnic relationships, but also in interethnic ones. Mauritian Hindus and Creoles often meet without implicitly or explicitly referring to their respective ethnic identities, for instance where the situation is defined through their statuses as colleagues or business partners.

The material from the Copperbelt and Mauritius also indicates that the compass of the ‘We’ category may expand and contract according to the situation. At general elections in Mauritius an individual may identify him or herself with the Hindu community at large; when looking for a job the extended kin group may be the relevant category, and when abroad he or she may actually take on an identity as simply Mauritian, even to the extent of feeling closer to Christian and Muslim Mauritians than to Hindus from India (Eriksen, 1992a: chapter 9; 1998). Similarly, Scandinavian identity is at its strongest when a Scandinavian encounters people from the neighbouring Scandinavian countries abroad. In most other situations that particular identity is not activated; it does not seem relevant in the definition of social situations. In other words, individuals have many statuses and many possible identities, and it is an empirical question when and how ethnic identities become the most relevant ones.

This fluidity and relativity of identity can sometimes be studied in interaction as negotiation of identity. The Kalela Dance exemplifies such a negotiation, where the agents disagree about the definition of their relationship. Mitchell describes the situation in this way:

A man and three women are drinking beer together in a beer-hall. One of the women belongs to the Lozi tribe. The man is a Ngoni, while the two other women are Ndebele. Suddenly the Lozi woman snatches a coin from him, says, ‘A foreigner has lost his money,’ and buys herself a cup of beer. The man asks why she took the money and demands that she give it back. She replies that there is a joking relationship between their tribes and that she was therefore entitled to take the money. The man denies that such a relationship exists. It then turns out that there is a joking relationship between the Lozi and the Ndebele, and that the woman identifies the man as being ‘more or less’ a Ndebele. The Ngoni and Ndebele tribes are linguistically and geographically close. The man insists that he is not a Ndebele but a Ngoni, but the woman does not pay him back. (Mitchell, 1956: 39-40)

In this situation, the Lozi woman insisted that a Ngoni was for practical purposes ‘the same’ as a Ndebele and could therefore be dealt with in the standardised way, whereas the man insisted that he was certainly not Ndebele. He challenged the validity of her taxonomic extension including the Ngoni in the same general category as the Ndebele. Similarly, London Brahmins might feel offended if they were to be treated, by native English people, in the same standardised way as black Londoners of Jamaican origin. In such a situation the Brahmins would be challenging the English taxonomic category of ‘immigrant’ or ‘minority’, insisting that there were socially relevant differences between kinds of immigrants.

In other interethnic situations where identity is negotiated, the issue may rather be whether or not to make ethnic identity relevant. Although it may be difficult to neglect the ethnic dimension entirely in such situations, it can often effectively be over- or undercommunicated. Notably, members of stigmatised and powerless ethnic categories such as the coastal Sami would usually be prone to play down the importance of ethnicity in interaction with the dominant Norwegians – or they might try, in a negotiating approach, to present themselves as carriers of a Norwegian identity.

The point here is that ethnicity can be a fluid and ambiguous aspect of social life, and can to a considerable degree be manipulated by the agents themselves. Of course, ethnic identities cannot be manipulated indefinitely, and one cannot ascribe any identity to somebody by claiming, say, that an Irish person is ‘really’ a Jamaican. Ethnicity can be of varying importance in social situations, and it is often up to the agents themselves to decide upon its significance.

Ethnicity from the individual’s point of view

When does ethnicity matter? It has already been stated that ethnicity occurs in social contexts where cultural differences ‘make a difference’. But what kind of difference? This is a very complex question which we can only begin to explore here.

In the mining towns of the Copperbelt in the 1940s and 1950s, ethnicity played a small but not insignificant role in the allocation of jobs. Although workers were hired by the mining companies, people could use their ethnic networks as sources of information and recommendations when looking for work. Ethnic distinctions still had a part to play in matters pertaining to marriage. Mitchell (1956) and Epstein (1958; 1978; 1992) also report the modest emergence of what we would today call ethnic politics, although ethnicity or ‘tribalism’ remained ‘essentially a category of interaction in casual social intercourse’ and did ‘not form the basis for the organization of corporate groups’ (Mitchell, 1956: 42). However, groups speaking the same language would, for example, protest that church services were conducted in a language unrelated to their own, and thus ethnic identity could function politically in certain contexts.

In Mauritius, which has a longer history as a plural society than the Copperbelt, ethnic membership can be important to individuals in a number of ways. Jobs have traditionally been allocated on an ethnic basis, usually through personal acquaintances or kinship. In many cases, religious associations and cults are also tightly linked with ethnic membership. Politics is thoroughly ‘ethnified’, and Mauritians tend to vote for parties which ostensibly represent the interests of their ‘community’. Youth clubs tend to be ethnic or religious in character, and this is often where Mauritians make friends and meet prospective wives or husbands. Most families have traditionally insisted that their children marry within the ‘community’. This means that in Mauritius, ethnic membership can provide people with their livelihoods, their spouses, their friends and their religion.  In addition, ethnic identity offers a sense of continuity with the past and personal dignity. This aspect of ethnicity will be looked at more closely from chapter 4 onwards.

For ethnic membership to have a personal importance, it must provide the individual with something he or she considers valuable. However, we must make one important reservation: in some cases, ethnic identities are imposed from the outside, by dominant groups, on those who do not themselves want membership in the group to which they are assigned.

For many years, sociology and social anthropology contended that modernisation would eventually level out and remove ethnic distinctions. The general argument was that it would no longer be profitable to pay allegiance to ethnic groups in modern, individualistic and bureaucratic societies, and that the processes of modernisation would also remove the cultural differences between groups. This was Max Weber’s view. Godfrey Wilson spoke of ‘detribalisation’, and in a later study of urbanisation in South Africa, Philip Mayer (1961) argued that ‘trade unions transcend tribes’, arguing along the same lines as Park, who described what he saw as melting-pot processes (see Bank 2009 for an update of the South African ethnography).
Do trade unions transcend tribes? Ethnicity has not only proved resilient in situations of change; it has also often emerged in forceful ways during the very processes of change which many believed would do away with it. On the other hand, there is no doubt that the meaning as well as the organisational form of ethnicity changes with other aspects of society. In order to find out what actually happens to ethnicity in the context of social change, we must therefore pose the question in more accurate terms than merely asking whether it disappears or stays the same. Its relative social importance is highly variable, for one thing, and in this respect, Steve Fenton (1999) has proposed a useful distinction between hot and cold ethnicity, which refers to its varying degrees of social importance and emotional intensity. Obviously, the situation in Rwanda during the massacres of 1994, where survival was contingent on ethnic identity, has little in common with the situation in South Dakota, where farmers of Scandinavian ancestry occasionally celebrate their culture of origin.

Criteria for ethnicity

Before we turn to look at ethnic group dynamics and processes of ethnic incorporation, we must enquire as to the substance of ethnic membership and classification. In other words, what is the stuff of ethnicity? How is it that some categories of people can be labelled ethnic while others cannot? Why is it that social classes, or the inhabitants of Somerset, or for that matter the members of a science-fiction club, are not considered ethnic groups, while the Sami, the Bisa and the Mauritian Creoles are? For a long time it was common to equate ‘ethnic groups’ with ‘cultural groups’; any category of people who had ‘a shared culture’ was considered an ethnic group. As we have seen, this position has become difficult to justify. As Moerman discovered during fieldwork in Thailand (Moerman, 1965), the sharing of cultural traits frequently crosses group boundaries and, moreover, people do not always share all their relevant ‘cultural traits’ with the people who belong to their ethnic group. One may have the same language as some people, the same religion as some of those as well as of some others, and the same economic strategy as an altogether different category of people. In other words, cultural boundaries are not clear-cut, nor do they necessarily correspond with ethnic boundaries. As Eugeen Roosens remarks: ‘There is more chance that the Flemish in Brussels, who always have to speak French, will become more “consciously” Flemish than their ethnic brothers and sisters in the rather isolated rural areas of West Flanders or Limburg’ (Roosens, 1989: 12). With this observation, we are also reminded of the fact that ethnicity is an aspect of relationship, not a cultural property of a group. If a setting is wholly mono-ethnic, there is effectively no ethnicity, since there is nobody there to communicate cultural difference to.

It is also clear that the criteria which constitute ethnicity vary. It will simply not do to state that an ethnic group is marked by shared culture, or even to point at specific ‘shared traits’ such as shared religion, language and/or customs. The Mauritian case brings this out clearly. Of the four ethnic groups which legally exist in Mauritius, two are defined in relation to religion (Hindus and Muslims), one in relation to geographic origin (Chinese), and one is a residual category containing people with their origins in France, Africa and/or Madagascar (general population). Nearly all of the latter are Catholics, but this cannot be a distinguishing criterion since most of the Chinese are also Catholics. A few of those who are classified as Hindus are also Catholics.

Many anthropologists have grappled with the problem of criteria for what is and what is not ethnicity. Abner Cohen (1974b) has taken an extreme position in arguing that London stockbrokers may be said to constitute an ethnic group; they are largely endogamous (at least to the extent of marrying within their class) and have a shared identity. Many other anthropologists would wish to delimit ethnic status to groups with a more obvious permanence in time and a clearer cultural identity based on fictive kinship, and would perhaps emphasise that ethnic identity sticks to the individual, that one cannot entirely rid oneself of it (Barth, 1969a). The general problem remains, nevertheless: where should we draw the boundary between ethnic groups and other groups, such as social classes?

Manning Nash (1988) has proposed, as the lowest common denominators for all ethnic groups, the metaphors of ‘bed, blood and cult’. By this he means that all ethnic groups consider themselves as biologically self-perpetuating and endogamous, that they have an ideology of shared ancestry, and that they have a shared religion. This kind of definition, whereby one denotes a number of presumedly objective criteria for ethnicity, has been challenged on many occasions (see chapter 3). Nonetheless, ethnic groups or categories generally have notions of common ancestry justifying their unity. But even this delineation can be contested within the group and from the outside, for how many generations does one have to go back in order to talk of shared ancestry?

Some ethnic groups use notions of ‘race’ or ‘blood’ in their ideology. Other groups rather emphasise criteria of cultural competence. Some groups do not allow outsiders to assimilate, whereas others do. However, they all have notions of shared culture in common; in this ethnic groups are distinct from classes.

The main problems which have been posed here deal with the relationship between ethnicity and culture, and the question of where an ethnic group ends and another begins. We shall return to both of these questions regularly in later chapters.

This chapter has argued that ethnicity is a product of contact and not of isolation, and it has also shown why the idea of an isolated ethnic group is meaningless. By implication, ethnicity entails both commonalities and differences between categories of people – both complementarisation and dichotomisation. The next chapter will show how social anthropologists conceptualise processes of ethnicincorporation and the maintenance of ethnic distinctions, or ethnic boundaries, through time.


Further reading

Epstein, A.L. (1992) Scenes from African Urban Life. Collected Copperbelt Essays. Edinburgh: Edinburgh University Press. Some of these essays were written in the 1950s, but they remain, with some of Mitchell’s work, among the finest studies of ethnicity at the level of interpersonal relationships.

Hannerz, Ulf (1980) Exploring the City: Inquiries Toward and Urban Anthropology. New York: Columbia University Press. Excellent introduction to the main schools of urban anthropology, many of them concerned with ethnicity, up to the late 1970s.

  1.  I owe this example to John Davidson.

2. This categorisation is based on the Bogardus social distance scale, used in research on ethnicity in American cities by the Chicago School. The original categories were: (1) Would marry. (2) Would have as a regular friend. (3) Would work beside in an office. (4) Would have several families in my neighbourhood. (5) Would have merely as speaking acquaintances. (6) Would have live outside my neighbourhood. (7) Would have live outside my country.

3. To some extent, this has changed since Mauritius was industrialised and increasingly culturally globalised from the mid-1980s onwards. Some implications of these changes for ethnicity will be suggested in chapter 9.