Jump to ContentJump to Main Navigation
The Architecture of DemocracyConstitutional Design, Conflict Management, and Democracy$

Andrew Reynolds

Print publication date: 2002

Print ISBN-13: 9780199246465

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0199246467.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

Constitutional Asymmetries: Communal Representation, Federalism, and Cultural Autonomy

Constitutional Asymmetries: Communal Representation, Federalism, and Cultural Autonomy

(p.141) 6 Constitutional Asymmetries: Communal Representation, Federalism, and Cultural Autonomy
The Architecture of Democracy

Yash Pal Ghai

Oxford University Press

Abstract and Keywords

Examines and compares three constitutional mechanisms for the political recognition of ethnic minorities and their potential to ameliorate or aggravate ethnic tensions—communal representation through the electoral process (with examples from Cyprus, India, Bosnia‐Herzegovina, and Fiji), territorial autonomy through asymmetric federalism, and cultural autonomy.

Keywords:   asymmetric federalism, communal representation, cultural autonomy, ethnic tensions, territorial autonomy

The New Constitutionalism?

It has been argued that the modern state, with its lineage of the market oriented and homogenising regime, built on the principle of individualism and equal citizenship, is inherently incapable of dealing with ethnic and social diversity that characterizes most countries. Constitutionalism associated with the modern state was concerned at first with limits on power and the rule of law, to which were later added democracy and human rights. It is argued that constitutionalism is not primarily concerned with the relations of groups to the state, or relations between groups.

Noting different communities or groups who are seeking constitutional recognition of their cultural or social specificity—immigrants, women, indigenous peoples, religious or linguistic minorities—James Tully concludes that what they seek is participation in existing institutions of the dominant society, but in ways that recognise and affirm, rather than exclude, assimilate, and denigrate, their culturally diverse ways of thinking, speaking, and acting. He says that what they share is a longing for self‐rule: to rule themselves in accordance with their customs and ways (Tully 1995: 4). The modern constitution is based on the assumption of a homogenous culture, but in practice it was designed to exclude or assimilate other cultures and thus deny diversity (Tully 1995: 58).

He argues that a constitutional order, which should seek to provide a framework for the resolution of issues that touch on the concerns of the state and its various communities, cannot be just if it thwarts diverse cultural aspirations for self‐government (Tully 1995: 6). Symmetries of power, institutions, and laws which define (p.142) the modern state are inconsistent with the diversity of forms of self‐government that Tully considers necessary for a just order in multi‐ethnic states. The necessity of a constitution which is based on mutual recognition of diversity is reinforced by the consideration that there is no escape from multi‐ethnic states as the alternative of over 1,500 ‘nation states’ is not feasible. Such a constitution should be ‘a form of accommodation’ of cultural diversity, of inter‐cultural dialogue in which the culturally diverse sovereign citizens of contemporary societies negotiate agreements on their forms of association over time (Tully 1995: 30).

A similar approach is taken by Bikhu Parekh, who argues that the theory of the modern liberal state presupposes a culturally homogenous society and becomes a source of disorder, injustice, and violence when applied to culturally heterogeneous societies. He identifies various institutional and structural features of the modern state that impose uniformity and ignore diversity. The organizing principle is state sovereignty, which justifies the centralization of power and displaces local and group sites of power. This sovereignty operates on a territorial basis, with hard boundaries. Rules for the exercise of this sovereignty are biased towards majoritarianism, stifling the voices of minorities. Much of his criticism is encapsulated in his view of sovereignty as ‘a rationalised system of authority, is unitary and impersonal in nature, is the source of all legal authority exercised within the state, is not legally bound by the traditions, customs and principles of morality, and is not subject to a higher internal or external authority’ (Parekh 1997: 183). People relate to the state through the concept of citizenship, based rigidly on equal rights and obligations of all persons, premised on loyalty to the state, and acknowledging no distinctions of culture or tradition. Citizens have rights but these are rights of individuals, based on an abstract and uniform view of the human person. The state operates through the medium of the law, but it is the law created by the state, rather than pre‐existing bodies of customs or local law. The state favours the uniformity of structures and seeks to achieve the homogenization of culture and ideology, propagating them as universal values. The domain of the state is the public space, with an ever‐shrinking area of private space, which alone allows some expression of cultural diversity.

The specificity of this system, despite its claims of universality, is demonstrated by both Tully and Parekh by contrast with pre‐modern polities. These polities cherished cultural diversity. It was no function of the state to impose moral or religious order, much less (p.143) to impose conformity. The public sphere was narrow and the private extensive, allowing ample space for diverse cultural and religious traditions. Nor did the centre aim towards a tight or detailed regulation of society, but was content with a large measure of decentralization, frequently based on cultural communities. It accepted pre‐existing bodies of customs and laws. There were multiple layers of authority and borders were porous, adding to the flexibility of the polity. Similar accounts of the diversity and flexibility of pre‐modern or pre‐colonial polities have been presented by other authors (for example, Kaviraj 1997; Tambiah 1992).

It is not my purpose to engage directly with this thesis—except to remark that it exaggerates the uniformity in the modern state and the flexibility and diversity in the pre‐modern. Pre‐modern China's experience, where the centralization of authority and the confucianization of the emperor's subjects were vigorously pursued, seems inconsistent with the picture sketched by Tully and Parekh. Several modern states have different categories of residents, there are differential spatial distributions of power, and religious and cultural affiliations are recognized for many public purposes. Many multi‐ethnic states recognise diversity through a variety of devices, including differential citizenship rights as in Israel (Peled 1992), Malaysia, and Fiji. Even ‘modern states’ like the US, Canada, Australia, and the Nordic countries had less than a uniform system of laws, citizenship, or institutions when they dealt with indigenous communities. If Lijphart (1977) is right about the prevalence of consociationalism in several parts of Europe, then also the monopoly of the centralized modern state is questionable. Several recent instruments and recommendations of the Organisation of Security and Cooperation in Europe and the Council of Europe seek to promote linguistic and religious diversity: decentralization, cultural councils, special voting rolls, language rights, and so forth. The general international law has come to recognize various categories of collectivities, such as minorities and indigenous people, with varying group rights. Even the regime of human rights, castigated for its obsession with the individual, has increasingly recognized group entitlements (Ghai 2000a). There is considerable flexibility in the design of states, such as Bosnia‐Herzegovina—perhaps in response to the kinds of criticisms levelled at the modern state by Tully and Parekh.

Nor is recognition of diversity always a virtue. The colonial state was par excellence a state of diversity and discrimination, deeply acknowledged, indeed entrenched, in constitutional and legal systems. The organization of the apartheid regime in South Africa (p.144) which ‘gloried’ in racial and cultural diversity used these distinctions to build its edifice of oppression. Jewish control over Israel is maintained through various legal institutions and distinctions which discriminate against Arabs or fragment the political community. More benignly, the essential principle for the organization of the political, social, and economic system of colonial Fiji was race: legislative representation and participation in the executive was allocated racially; indigenous Fijians had their own system of administration and the right to review legislative proposals before they reached the legislature, and there were several institutions to safeguard Fijian customs and laws. The division of labour was also structured along racial lines. Many features of the colonial system survived into the independence period, not always with positive effects on racial harmony. The separation of the political and economic organization of indigenous peoples in the US, Canada, Australia, and much of Latin America had the effect, as was the intention, of marginalizing them.

However, it is not my contention that the political recognition of diversity is always fragmenting or oppressive. Special regimes for communities based on sensitivity to their vulnerabilities, or the recognition of the centrality of cultures to them, or of past injustices, have contributed to justice as well as improvement in inter‐ethnic relations. Whether the political recognition of diversity is fair or beneficial depends on the context, the preferences and aspirations of the various communities, and the forms that political recognition takes. Moreover, support for it depends on differing theories of ethnicity. What I do in this chapter, in order to make a preliminary assessment of potential of political recognition to ameliorate or aggravate ethnic tensions, is to examine three principal devices: communal representation, asymmetrical federalism, and cultural autonomy.

Communal Representation

It is only in recent years that consideration has been given to the adaptation of electoral systems to minority representation. Two widely different approaches have been advanced. The first focuses on ensuring representation of a minority by members of the minority, ideally proportional to their size of the population, either through electoral systems which will facilitate this or, if necessary, by a system of separate representation. The second approach is less concerned with direct minority representation than it is with its (p.145) political integration. There is not always a clear distinction between the two approaches, for some methods which may be used for direct minority representation are compatible with its integration, such as proportional representation. In practice the distinction between systems which do and do not provide for separate minority representation is greater.

It is generally accepted that most traditional electoral systems, which constitute the most numerous of the world's electoral systems, are not conducive to minority representation. Sometimes the plurality—‘first–past–the–post’—system can yield minority representation if a minority is concentrated in a constituency. If minorities are politically well integrated with the majority, their members may well be elected in such a system, as with the Jewish community in Britain. Majoritarian systems are even less favourable to minorities, for a candidate needs to secure at least 51 per cent of the votes to win. Proportional representation (PR) systems, which aim at relating the number of representatives to the votes cast for particular candidates or parties, are more favourable to minorities. Sometimes these traditional systems are adjusted to favour minorities: in the southern States of the US, constituencies may be gerrymandered to create a black majority. In many countries constituencies are to be drawn to reflect ‘community of interests’, which in some instances, as in colonial Mauritius, were taken to ensure adequate representation of two major communities, Hindus and the General Population (de Silva 1998: 77–8). In Nepal, which like the US has plurality voting, parties are required to nominate at least 5 per cent of candidates from among women. In Poland and Spain, which follow the PR system, a minority party does not need to cross any threshold in order to win seats. However, these adaptations have not resulted in significant representation of minorities, prompting consideration of systems for separate and guaranteed representation of minorities—typical forms of electoral asymmetry

The pre‐occupation with minority representation in recent years, particularly as part of complex constitutional schemes for the governance of multi‐ethnic territories, has led to provisions in many of them for separate representation for ethnic groups, particularly for minorities—called here ‘communal representation’—as in Bosnia‐Herzegovina, Hungary, Romania, Slovenia, Croatia, and Finland. Communal representation was the corner‐stone of British colonial system, but at independence most former colonies abolished this system of separate representation, the outstanding exceptions being Cyprus and Fiji. It is also to be found in China—where (p.146) minorities are deliberately over‐represented—New Zealand, and Samoa. The revival of communal representation calls for an assessment, which I do by examining a number of experiences.


From the very start of representative politics in Cyprus, Britain introduced communal representation. Such was the bitterness between the dominant Greek community and the minority Turkish community, fuelled no doubt by their ‘kin states’, that independence could be secured only through an intricate constitution built around far‐reaching consociational principles. De Smith (1964: 285) commented that the constitution was ‘weighed down by checks and balances, procedural and substantive safeguards, guarantees and prohibitions. Communalism has run riot in harness with constitutionalism’. The Greek—including Maronites—and the Turkish communities were treated as separate entities, and the entire system of representation, government, administration, and social services was based on proportionality, with Greeks counting for 70 per cent and Turks 30 per cent of the population. This entailed a slight overrepresentation of Turks. The House of Representatives consisted of 50 members, of which 35 were Greek and 15 Turks, elected on a communal basis. The president of the House had to be a Greek, the vice‐president a Turk. In addition to the full House, there were also communal chambers of Greeks and Turks members respectively, which had wide law‐making powers in educational, religious, and personal affairs, and other matters delegated to them by the House. The president of Cyprus had to be a Greek, elected by Greek voters, and the vice‐president a Turk who was elected by the Turks, each with his own special powers. Ministerial posts were also divided among the two communities; Greek ministers were appointed—and removed—by the president; the vice‐president performed similar functions in relation to Turkish ministers. As is well known, the system produced extreme rigidity; it was resented by the Greeks for giving disproportionate powers to Turks, and Turks resented the permanent dominant position of the Greeks. The Greeks wanted to change the constitution; the Turks boycotted arrangements agreed at independence, over which, it should be stated, the decisive influence has been of metropolitan powers. Cypriot politics were also complicated by political and military interventions by Greece and Turkey, which eventually spelled the end of the republic as described above.

(p.147) India

The Indian National Congress which led India to independence was opposed to ethnic electoral rolls and representation which the British had introduced in 1909. It would have been willing to contemplate them after independence if Pakistan had not been carved out of the subcontinent as the homeland for Muslims; separate rolls had been devised primarily to secure representation for Muslims. Austin (1972: 144) says that the ‘members of the Constituent Assembly had one predominant aim when framing the Legislative provisions of the Constitution: to create a basis for the social and political unity of the country’. He summarizes the situation at independence as follows: ‘ . . . not only did the provinces lack even a semblance of popular government . . . but the small electorate that existed was itself thoroughly fragmented . . . split into no less than thirteen communal and functional compartments for whose representatives seats were reserved in the various parliamentary bodies’. Similar distinctions were applied in the indirectly central legislature. Austin (1972: 144–5) concludes: ‘Quite evidently, the members of the Constituent Assembly could not pursue the goals of national unity and stability by perpetuating a system of government that accentuated existing cleavages in Indian society and tended to create new ones.’

The Constituent Assembly did agree to one form of special representation, for scheduled castes and tribes, as part of the package of affirmative policies for these communities. Seats are reserved for each of these communities in proportion to their share of the population in both the lower house at the national level and in the States (Arts 330 and 332 respectively of the Indian Constitution).1 The scheduled castes constitute about 15 per cent of the population, and the scheduled tribes 7 per cent, so that they enjoy significant guaranteed representation in the lower houses. The law also provides for smaller election deposits for candidates from scheduled castes and tribes.

All registered voters may vote in constituencies in which seats are reserved for scheduled castes or scheduled tribes, so that only members of these groups may be candidates; but there is no similar restrictions on voters. The Delimitation Commission, an independent body, determines in which constituencies seats will be reserved for them. Constituencies are first formed on the basis of the normal criteria of contiguity, communications, absence of natural barriers, (p.148) and cultural and ethnic homogeneity. The Commission then earmarks constituencies for scheduled castes and tribes based on the concentration of their population. Since tribes still live in particular areas, these constituencies contain a high proportion of their population, more than 50 per cent in more than half constituencies so reserved, so that the bulk of scheduled tribes—about 70 per cent—would live in such constituencies. As scheduled castes are more dispersed than the tribes, constituencies reserved for them have to be spread through the country, and to be located, as far as possible, in constituencies in which ‘the proportion of their population to the total is comparatively large’.2 Thus constituencies with reserved seats for scheduled castes contain proportionately fewer of them than is the case with scheduled tribes. The largest contain about 30 per cent.

There is considerable opposition from other communities to the designation of the constituencies in which they live as reserved constituencies, as it deprives their members of the right to the seat. Galanter (1984: 48) says that, on the whole, constituencies reserved for the scheduled castes ‘tend to be political backwaters—slightly less urban, with less newspaper circulation and a slightly greater percentage of agricultural labourers’. Scheduled tribe constituencies tend to be more isolated and less urban than general constituencies, given their habitat.

The effect of the reservations is to ensure the representation of these two communities who are otherwise politically and economically marginalized. This is particularly important for the scheduled castes since they are dispersed throughout the country, and nowhere in a majority. Given that over 20 per cent of seats are held by the members of these communities, all major parties have an interest in promoting candidates from them. The candidates have likewise an incentive to project their appeal beyond their own communities, particularly in the scheduled caste constituencies, where their own community would be in the minority. This has helped to integrate them into the constitutional and political system, but this result has perhaps been achieved at the expense of abandoning particular advocacy of the claims of their own community. The present system replaced an earlier one in which several constituencies had two seats, one reserved for the scheduled castes, for which they alone could vote, in addition to their vote for the other seat. In this way they controlled the election of one seat in such constituencies, a possibility which is remote under the current arrangements. (p.149) Nevertheless, there are parties which are based predominantly on their support, particularly at the State level (Mendelsohn and Vicziany 1994), where their members have achieved high office. Reservations have given the two communities considerable political influence. It has facilitated their entry into the government and their lobbying has been crucial for the maintenance and improvement of other affirmative action policies, which for the most part are authorized but not mandatory; some of these are discussed in the section on economic and social rights below.


The Federation of Bosnia‐Herzegovina is composed of two Entities: Bosnia and Srpska. Most powers are vested in the Entities, the federation being left largely with those powers which are necessary to constitute and exercise external aspects of state sovereignty. The constitution is built around the concept of ethnic communities as separate corporate bodies. Arrangements for representation and power‐sharing take the communities as building blocks, carrying forward the proposition stated in the preamble of the Constitution that Bosniacs, Croats, and Serbs are ‘constituent peoples’ of Bosnia and Herzegovina, ‘others’ and ‘citizens’ being mentioned only in passing. Zoran (Pajić 1999: 38) implies, critically, that this makes these three communities, rather than the people as a whole, the source and bearers of sovereignty.

The parliamentary assembly consists of five Croats and five Bosniacs from Bosnia and five Serbs from Srpska; they are elected by voters of their own communities (Art. 4). Nine of them constitute a quorum, so long as there are at least three from each community. The House of Representatives is constituted on the same principle and in similar proportions. The result of these arrangements is that politics are entirely communal, and almost perforce all political parties are ethnically based. Parties get together in parliament or government only after the elections. The system creates incentives for parties and their leaders to intensify appeals to narrow ethnic interests, linked to their kinfolk in other states, which does little for the unity of the country. In the 1996 elections, the most extreme ethnic party in each community won, leaving their leaders the impossible task of finding a common purpose. Recent local government elections show some erosion of support for nationalist as opposed to multi‐ethnic parties, but hopes that this trend would also be reflected in the national elections have been disappointed.

(p.150) Fiji

One of the most difficult questions that the leaders of the different ethnic communities had to resolve at independence in 1970 was the electoral system. Since independence Fiji has experienced three different electoral systems, and is about to design a fourth. The 1970 constitution was based on the separation of races. Its centrepiece was the electoral system dominated by communal seats and communal voting. Although there was provision for national seats, their structure was still based on a racial allocation of seats, while the logic of communal seats prevailed over the logic of national seats which was intended to provide a basis for non‐racial politics. There were 52 members, 27 elected on communal franchise: 12 by indigenous Fijians, 2 by Indo‐Fijians, and 3 by General Electors, principally Europeans and their part‐descendants and Chinese. Twenty‐five ‘national seats’ were allocated communally—10 each to indigenous Fijians and Indo‐Fijians and 5 to the General Electors—but as all the voters in the country voted for them the system came to known as ‘cross voting’. Every voter had three votes in the contests for national seats, each to be cast for members of different ethnic groups. This formula gave a slight over‐representation to indigenous Fijians and to the General Electors, who had traditionally allied themselves with the indigenous Fijians. The rationale of national seats was to politically integrate ethnic groups, promote inter‐ethnic parties, and prepare the transition to complete common roll. However, the logic of the system was dictated by the communal rather than the national seats.

Political parties were organized essentially on ethnic lines in order to compete for communal seats. There was one dominant party for each of the communities. While the need to contest national seats compelled each of the major parties to extend its appeal beyond the community it principally represented, for the most part this was not successful: each party was content to field a few candidates from other ethnic groups. National seats were decided principally by communal votes; thus, indigenous Fijian candidates sponsored by the dominantly Indo‐Fijian National Federation Party were successful in areas dominated by Indo‐Fijians as a result of Indo‐Fijian votes, and so on. This was possible because of the concentration of the two major communities in different constituencies. Consequently, national seats won by candidates who relied on the votes of the other sponsoring community had little support in their own community, while those who relied on votes from their own communities had little support in other (p.151) communities. In this way cross‐voting seats became an extension of communal seats.

The Alliance Party—the dominant party of indigenous Fijians—was a partial exception to this trend. It attracted a significant percentage of Indo‐Fijian votes, especially for the cross‐voting seats, in which it often achieved over 20 per cent of the vote. By contrast, the National Federation Party commonly gained less than 5 per cent of indigenous Fijian votes. However, the Alliance Party had to maintain its support among indigenous Fijians if it was to remain a serious political contender, especially as militant indigenous Fijian parties were bidding for the support of its principal electorate. The logic of the system compelled the Alliance Party to progressively champion exclusively indigenous Fijian interests. The disregard by it of Indo‐Fijian interests gradually led to the attrition of its Indo‐Fijian support, so that by 1997 it had lost most of it and the parties had settled back to relying on their old ethnic constituencies.

The 1990 constitution, adopted by the military government following the 1987 coups, abolished the national seats. It not only removed any vestiges of cross‐voting, completing the separation of ethnic groups—and making politics almost totally racial—but also aimed to ensure the permanent and undisputed rule of indigenous Fijians. It gave a disproportionately large representation to them in both houses of parliament, in the House of Representatives 37 out of 70 seats being reserved for them. In addition it provided that a prime minister had always to be an indigenous Fijian. It also dispensed with the rather awkward, residual agenda of the 1970 constitution that the ultimate aim was the development of a multi‐racial Fiji. The sidelining in this way of the Indo‐Fijians had the predictable effect of releasing factionalism within the Fijian community that had been largely contained under the more balanced allocation of communal seats in the 1970 constitution.

The 1997 constitution, overthrown in May 2000 but reinstated by a decision of the Court of Appeal in March 2001, largely abandoned that approach, but it continued with significant reliance on communal representation. It provided for 25 open seats in the House of Representatives, out of a total of 71, which were open to candidates of any ethnic group and for which all voters resident in the constituency could vote, and 46 communal seats—to be voted communally—divided between the ethnic communities. The voting for these, as for communal seats, was by the alternative vote (AV) system. While in communal seats this method of voting served principally the purpose of ensuring that the winning candidate enjoyed clear (p.152) majority support, its purpose in open seats was to provide incentives for political parties to cooperate across racial frontiers. Under the AV system, a voter has to declare his or her preference among all the candidates. Since a winning candidate has to have an absolute majority, the second and subsequent preferences of a voter can be crucial in determining the result, for if no candidate obtains an absolute majority after the count of first preferences, the second—or subsequent—preferences of those who voted for candidates who come bottom of the poll on first count have to be taken into account. This method thus opens up possibilities of arrangements between political parties for the trade‐off of the second and subsequent preferences of their supporters. The trade‐off can of course take place between parties of the same ethnic group, but, since it may be assumed that they would in fact be in competition in constituencies where the members of that ethnic group predominate, the trade between political parties of different racial groups makes better political sense. The logic of the system might well have led to multiracial parties, as was the expectation of the Reeves Commission which recommended it. Additionally, it was expected that candidates with moderate views would have an advantage over those espousing extreme views, as they would have a chance of capturing more second preferences (FCRC 1996: 9.150–2 and 10.31–109; see also Horowitz 1997b who advised the Commission on the electoral system). The result would be either a coalition government or, better still, government by a party with multiracial membership.

The results of the first general election seemed to have vindicated some of the assumptions of the Reeves Commission. Two broad coalitions of communal parties were formed and contested the elections. However, it was not only moderate parties with conciliatory policies that tended to trade preferences, as the Reeves Commission had envisaged. In fact, the more ethnically conciliatory coalition lost the election. In particular it seems that a basic assumption of the new system was not borne out by the results. The assumption is that, if a party is extremist, it may get a significant proportion of first preferences, but that that the more moderate parties would get the second and subsequent preferences. Doubts about this assumption in a multi‐ethnic state were expressed before the adoption of the system (Ghai 1997). As an acute observer of the results has commented:

Where racial polarization is particularly sharp, it is easy to envisage a situation where a majority of ethnic group's first preferences are picked by the militant flank party, which also attracts, at the second, third or (p.153) subsequent count, the preference votes from eliminated more moderate parties representing the same ethnic group. Here the AV system could serve, not as a vehicle for inter‐ethnic compromise, but as a means of cohering a politically fragmented ethnic group around an extremist position. (Fraenkel 2001:15)

Few first preferences in open seats were cast across ethnic divides, but some transfer did take place through second and subsequent preferences—although it is widely believed that this resulted from voters' ignorance of the voting procedures. A major party with predominant Indo‐Fijian support failed to secure a single seat, although its share of the communal vote was over 32 per cent. It does not seem therefore that the electoral system led to cross‐ethnic voting on any scale, nor to any proportionality, but the logic of open seats on the AV system did lead to multi‐ethnic coalitions (Prasad 2000). Unfortunately, it is not possible to make a reliable assessment of this interesting system, for it was tried only once, when knowledge of its operations and of electoral strategies to exploit the system was limited. But the experience may reflect limits of electoral designs and the ability of voters, particularly in rural areas, to handle elaborate voting systems.


Where both the majority and minorities are agreed that minorities should be represented separately, there may be no objection to communal electorates. There may indeed be cases of minorities which are so small, or where politics are so communalized, that the only realistic prospect of representation is through a communal roll. But the case studies examined here raise doubts whether separate representation in general is desirable. A particularly acute observer of constitutional politics, Stanley de Smith, concluded that communal representation ‘tends to magnify existing communal differences, in as much as communities are stirred to fuller self‐consciousness and electoral campaigns are dominated by appeals to communal prejudices; and new communities discover themselves as further claims to separate representation are lodged’ (de Smith 1964: 118). It is exceedingly hard to establish national parties, necessary for political integration, when voting is communal. Religious, particularly Hindu‐Muslim, conflict in British India is often attributed to the introduction of separate representation for Muslims in 1906.

Communal forms of representation often irritate and provoke majority groups, although this is not in itself a reason for not (p.154) adopting them. Members of minorities have fewer prospects of high office if they rely on their own separate parties and representation than if they were members of national parties—unless there are provisions for power sharing. Communal representation also tends to obscure social and economic interests that sections of different communities have in common. Moreover, leaders and parties of the majority party have little incentive to woo electors of minorities or to design policies to suit them. Within the community itself, the interests of the wealthy elite tend to dominate. Commenting on the consequences of communal representation in Lebanon, which applied it for 40 years from the early 1940s, Rabinovich (1985: 26) claims that Parliament reflected the ‘web of relationships among the traditional foci of power . . . in upholding the principles of confessionalism and playing down the notions of nation and class, the Lebanese political system acquired an archaic complexion, and from the mid‐1960s, it found itself challenged by the political attitudes prevailing in most of the outside world’. On the other hand, the Indian post‐independence experience shows that arrangements for the representation of scheduled castes has integrated them politically with other Indians while giving them special influence on national policies. Therefore, when it is desirable or necessary to give special representation to an ethnic group, preference should be given to integrating it within the general electoral system, such as mandatory nomination of a minimum number of minority candidates.

The largely negative record of communal representation strengthens the case for electoral systems designed to integrate different communities, creating incentives for political parties to broaden their appeal in order to attract votes from all communities, so that representatives have a broad base of support. The aim, therefore, is not so much to ensure direct minority representation but to ensure that those who are elected are likely to enjoy the support of minorities and thus be moderate in their policies. The rules for the election of the Nigerian president under the 1982 constitution were based on this approach, which has recently been used in Kenyan presidential elections. The AV system in Fiji's 1997 constitution was adopted for the same reason, whose success might have been greater if communal representation had been eliminated or reduced. Singapore provides an example of a system which both secures minority representation and integrates communities. A number of constituencies, called the Group Representative Constituencies (GRC), return either three or four members. A political party which wants to contest in these constituencies has to present a slate of three or four candidates, of which (p.155) at least one must be from a minority. Electors vote for the slate rather than individual candidates. This system is justified on the basis that it secures the election of some minority candidates, although its opponents criticized it as an attempt by the government to stifle opposition parties who would have difficulty in securing enough qualified candidates given the constraints under which opposition parties operate. It is certainly the case that so far all GRC seats have been won by the ruling party (Thio Li‐am 1997: 107–8). In order to encourage the political integration of Maori, in 1993 New Zealand altered its system of representation for them by giving them the option to vote on a separate electorate or common roll—the more Maoris opted for the common roll, separate representation would be correspondingly reduced. A substantial number of Maori opted for the common roll, and the community secured higher representation than in the past.

Territorial Autonomy

Territorial autonomy is a device to allow ethnic or other groups claiming a distinct identity to exercise direct control over affairs of special concern to them while allowing the larger entity to exercise those powers which cover common interests. Autonomy has become the most sought‐after and resisted device for conflict management. However, autonomy is controversial, and many conflicts themselves are about the demand of autonomy and resistance to them. At other times autonomy seems to offer a way out of conflict or the transformation of the conflict. The promise to consider or negotiate autonomy has been used successfully to bring truce among warring parties. Autonomy has sometimes secured a breathing space as an interim or even ambiguous expedient while longer‐term solutions are explored and negotiated. Autonomy has been used to separate as well as to bring people together. In recent years it has been seen as a panacea for cultural diversity, and as, under the influence of identity politics, the realization of the extreme heterogeneity of states dawns on us, autonomy seems to provide the path to maintaining unity of a kind while conceding claims of self‐government.

Autonomy has several features which distinguish it from the model of the modern state described by Tully and Parekh. It is multi‐layered, state sovereignty is divided, there are many sites of power and authority—qualifying majoritarianism—there is explicit or implicit recognition of cultural or ethnic diversity, and there are various asymmetrical features, even in classical federations. However, not all federations are hospitable to this degree of diversity; there (p.156) has been considerable resistance to multi‐ethnic federations and asymmetry. Federations in liberal societies, it has been argued, are meant to reflect principles of equality—symmetry—and common values (Glazer 1977). Moreover, the diverse arrangements for territorial autonomy have begun to distinguish one state from another, so that the model criticized by Tully and Parekh is becoming less ‘universal’. Federal systems where one or more regions are vested with special powers not granted to other provinces are known as ‘asymmetrical’ (Stevens 1977; Watts 1994; Agranoff 1994; Boase 1994; Brown‐John 1994).

The best known form of autonomy is federalism, where all regions enjoy equal powers and have an identical relationship to the central government. Traditionally, federalism has not been used as a way to solve problems of ethnic diversity, except for Switzerland and Canada. Classical federalism, where all regions have equal powers, may not be sufficiently sensitive to the peculiar cultural and other needs of a particular community, and it is not surprising that Swiss and Canadian federations have various asymmetrical features, as have more recent multi‐ethnic federations, such as India, Spain, Russia, and Malaysia. The federal model may be unnecessary if the need is to accommodate only one or two minority groups. In these situations, special powers may be devolved only to a part of the country where the minority constitutes a majority; these powers are exercised by regional institutions. Normally very significant powers are devolved and the region, unlike in a federation, plays relatively little role in national government and institutions. This kind of autonomy is some times referred to as regional autonomy (Heintze 1998: 10–11) or federacy (Stevens 1977; Elazar 1987: 7). Examples of autonomous regions include Åland Islands (Finland), South Tyrol (Italy), Kosovo (the former Yugoslavia), the Cordilleras and Mindanao (the Philippines), Puerto Rico (US), Zanzibar (Tanzania), Hong Kong and Macao (China), Greenland and Faroes (Denmark), New Caledonia (France), and Scotland (UK). By its nature, this kind of regional autonomy is asymmetrical

A special instance of spatial organization of government is ‘reserves’, which were first used by European settlers in the Americas to isolate and dominate indigenous peoples, and were subsequently adopted in Australia, Africa, and parts of Asia. The apartheid policy of Bantustans was a modern version. However, in recent years the aspirations and historical claims of indigenous peoples have been recognized through the transformation of reserves into self‐governing areas, particularly in Canada and the Philippines, although the extent to which they can opt out of (p.157) national laws, which may be necessary for the preservation of their political and cultural practices, is variable.

A new but uneven element in the spatial organization of government is the emergence of international regional organizations in which national sovereignty has been traded for a share in participation and decision‐making in these organizations. Common policies over larger and larger matters are determined by the regional organization, so that a measure of control of the affairs of a national region has been transferred from national to supranational authority. The consequences are that the diminution of the salience of national sovereignty opens up possibilities of new arrangements between the state and its regions, the state feeling less threatened by regions in a multi‐layered structure of policy‐making and administration and the region being more willing to accept the national sovereignty which may be the key to its participation in the wider arrangements. This trend is most developed in the European Union, with its developing concept of the Europe of Regions (Bullain 1998), which is helping to moderate tensions between states and border regions previously intent on secession, as in Spain and Belgium and which has facilitated the interesting spatial arrangements for policy, administration, and consultation in the two parts of Ireland, each under separate sovereignty, which underlie the new peace settlement. Attempts to provide for unified Nordic arrangements for the Saami people, including a substantial element of autonomy, regardless of the sovereignty they live under, are another instance of similar kind (Hannum 1990: 256–62).

The Distinctiveness of Ethnically Based Autonomies

Ethnically based federations or regional autonomies have different structures and orientations from federations like Australia or the United States. Naturally, ethnic federations emphasize diversity and multiplicity of values. Such federations are more likely to be the result of devolution or disaggregation, as in Canada, India, Spain, Papua New Guinea, and Ethiopia. Because they start with a centralized structure and because there is unease about the political implications of devolution, national powers tend to be dominant, including, not infrequently, the power to suspend regional governments. It is worth noting that Indian federation emerged at a time when state building through central management and homogenization was the dominant paradigm; concessions to ethnicity were reluctant and grudging. Because, ideally, a region is supposed to provide ethnic or cultural homogeneity, the size of regions is likely (p.158) to be uneven and agreement on boundaries hard. There is also likely to be more emphasis on self‐rule than shared rule, particularly with regional autonomy. But, at the same time, there is likely to be greater regional representation at the federal level. Watts (2000: 45) says that ‘in both Canada and Switzerland the “representational syndrome” whereby different linguistic, religious and geographical groups are carefully represented within each of the federal policy making bodies, including the federal executive, is very marked’. Klug's (2000) account of South Africa establishes the same point, indeed, even more so since the principal form of ‘autonomy’ is not the exercise by regions of power in the regions but participation at the centre. In Ethiopia, a crucial role is ascribed to the House of Federation, composed of regional representatives, where relations between regions and the federation and between the regions themselves are mediated and resolved.

The division of powers is likely to be more focused on cultural matters, like education, religion and arts; and the normal tensions of federalism, like fiscal redistribution or regional influence, take on an ethnic dimension and become aggravated. Distinctions between the private and public spheres may be less sharp than in other types of federations. Inter‐regional mobility is likely to be contentious. The party structure may be different as there may be no great connection between national parties and regional parties. National parties often lack viable affiliates in particular localities; relations between centre and region depend significantly on this ‘asymmetry’. As G. Smith (1995: 9) notes, ‘such enduring asymmetry can be symptomatic of inter‐communal tensions and lead to pressures not only for greater autonomy but also for secession from federation’.

But, most of all, the major factor which distinguishes ethnic autonomy from classical federations is its asymmetrical features. Just as in liberal theory all individuals must be equally treated, so must regions in a federation. This approach is not very constructive when autonomy is used to acknowledge and manage ethnic differences. Asymmetry acknowledges the unevenness of diversities and opens up additional possibilities of awarding recognition to specific groups with special needs or capacities, such as indigenous peoples whose traditional culture is central to their way of life, or a minority linguistic group. Examples of asymmetry abound; China has at least four types of autonomy—economic zones, metropolitan cities, ethnic minorities, and special administrative regions—responding to different imperatives (Ghai 2000d). So has India, with its ‘standard provinces’, special arrangements for Kashmir, and provinces in (p.159) the north‐east, tribal areas, and union territories, each enjoying a distinct relationship with the centre.

Asymmetry arises in various ways. Regional autonomy is by definition asymmetrical. Sometimes it is the result of a constitutional provision enabling regions to negotiate separately with the centre for autonomy and establishing a menu of which powers may be devolved—Spain, Papua New Guinea, Russia. Regions may make different use of concurrent powers. Regions may be endowed with the power to determine their own structures for the exercise of autonomy, leading to differences in constitutional arrangements. National laws may apply differentially for other reasons, the outstanding example being the ‘notwithstanding’ clause in Canada which enables a province to opt out of most provisions of the Charter of Rights under prescribed conditions, and another provision which limits the application of the charter in aboriginal areas by virtue of the supremacy of treaties between indigenous groups and the Crown. Asymmetry can also be used as a general technique for opting out of a scheme, or for a phased entry to full membership, as has happened frequently with the European Union. Other forms of asymmetry include special representation for a region at the centre—Quebec's entitlements to seats in the Senate and the Supreme Court—or special voting power given to the region at the centre, such as double voting or vetoes. Residents of a region may have special rights, at least in the region, that are not available to other citizens, as in the concept of permanent resident of the Hong Kong Special Administrative Region. The questions of the feasibility of negotiating and sustaining asymmetry are therefore fundamental to the design and operation of ethnic autonomy. The validity of the Russian federation depends on the acceptance and successful operation of autonomy: 88 units have differing relationships with Moscow.

These developments regarding federalism and autonomy have helped political leaders to devise arrangements for forms of self‐government to suit varying circumstances and contingencies. In addition there are variations in detailed arrangements within each category, such as in the division of powers between different layers of government, structures of government, the relationship between these structures at different levels, and the distribution of financial and other resources. While this flexibility is important in the negotiation process and facilitates compromises, there is the danger that it may lead to complex arrangements and systems, producing a lack of cohesion and problems of governability. When negotiations enter a difficult phase, there is the temptation to device some fancy (p.160) scheme which may produce a temporary consensus which is hard to operationalize: thus there is a conflict between immediate and long‐term interests. Federal or autonomy arrangements are inherently hard to operate, requiring both great administrative capacity and political skills, and the embroidery on classical systems that tough negotiations may lead to would undermine long‐term prospects of settlement by their sheer weight or complexity. A good example of this experience is the regional arrangements in Kenya's independence constitution (Ghai and McAuslan 1970), Papua New Guinea's system of provincial government established in 1976 (Ghai and Regan 1993), and even Spain's autonomous communities (Conversi 2000). Lack of resources is likely to negate large parts of Ethiopia's complex and complicated constitution of 1994, as Paul (2000) argues.

Perhaps it is not surprising that asymmetry has also become controversial; concerns about it have, for example, prevented a satisfactory resolution of Canada's constitutional problems. It is in Canada that the issue has been most extensively debated, in political as well as academic circles. Canadian scholars have argued that differences over asymmetry may be the undoing of ethnic or multinational federations; Milne (1994: 159) notes an ‘overwhelming’ hostility towards proposals for asymmetry in Canada (see also Kymlicka l998a, b). There is resentment in India at the privileged position of Kashmir (Kashyap 1990), although it has not emerged as a major political issue, perhaps because of Indo‐Pakistani conflict over Kashmir. It is said that President Habibie offered independence to East Timor because he was afraid that the UN proposals of autonomy would set a precedent for other provinces of Indonesia, and that it would be politically difficult to restrict the high degree of substantive and institutional autonomy to East Timor.

One objection to asymmetry is that it is administratively and politically difficult to manage. The centre has to deal with regions with varying degrees of devolution and different institutional structures. This can pose problems in states as well developed as Spain; it can be a nightmare in states with less efficient bureaucracies or with politicians not given to compromise, as in Papua New Guinea and Ethiopia. A consultancy firm which advised Papua New Guinea on the implementation of decentralization, expected to be asymmetrical under negotiated constitutional provisions, recommended the equal devolution of powers to all provinces, regardless of their capacity or willingness to assume these powers. If this proposal avoided one bureaucratic nightmare, it created another: (p.161) poorly equipped provinces struggling to carry out new responsibilities which they neither understood nor wanted. The result was continuing domination by central bureaucrats and a not inconsiderable degree of inefficiency (Ghai and Regan 1992).

But the political problems with asymmetry are even more decisive. I have already referred to the difficulty of conceding autonomy on a purely ethnic basis. The difficulty is greater if only one or two groups are to enjoy autonomy. If the national government is inclined to support autonomy, it may have to generalize the conditions for granting it. In Papua New Guinea in 1976, negotiations for autonomy were conducted between the national government and the representatives of Bougainville. The assumption was that the arrangements under negotiation were for Bougainville only; and, in fact, Bougainville leaders insisted that only their province was to be entitled to them, to recognize their distinctiveness. However, the government realized that parliamentary support for these arrangements could not be guaranteed unless all provinces were given similar options. Similar developments took place in Spain, where all provinces or groupings of provinces were given roughly the same options as the ‘historic territories’. Increasingly, Spain takes on the appearance of a federation, and a symmetrical federation at that. The devolution to provincial councils in Sri Lanka followed a similar trajectory, diluting the special claims of Tamils to autonomy. In Britain, following autonomies for Scotland and Wales, there is agitation for English regions. The tendency towards symmetry is, however, not universal. Sometimes there may be recognition of the historical claims of a community or the clear distinctiveness, and vulnerability, of a culture—as of Greenland, Faroes, Åland, Corsica, and the Cordilleras. A community may desire a greater measure of national integration than asymmetry would permit: the Swedish‐speaking community on the Finnish mainland rejected the offer of an Åland‐type of autonomy.

Asymmetry is particularly controversial when the region benefiting from it wants equal or even superior representation in central institutions. Logically, the region should not participate in decisions at the national level on areas which are within its autonomy, for then it would be making decisions for other regions, especially when the votes of its representatives hold the balance. When there is substantial asymmetrical autonomy, the moral or political right of the representatives of that region to count towards a parliamentary majority, thus determining the formation of the central government group, can be questioned. Claims might be made by the rest of the country that representatives from that region should be excluded (p.162) from holding ministries whose portfolios cover areas within asymmetrical autonomy or, indeed, that the number of ministries given to them should be severely restricted. If there is equal representation for the autonomous region, other provinces will resent it; if the representation is less favourable, the region will tend to look inwards, political parties will tend to become regional, and the region's integration with the state will weaken.

The conversion of asymmetry into symmetry would not necessarily be against the interests of the original claimants of autonomy. They would cease to be the object of envy and resentment. A greater number of beneficiaries would produce a more balanced state. It would also increase the capacity of regions to negotiate with the centre and extract higher benefits. But, for many groups, the exact amount of devolved power is less important than that they alone should enjoy some special powers as a way to mark their status. If the powers they have are generalized, they increase their own demands for more, not only leading to a higher level of general devolution than is desirable or desired but also pushing the special groups towards confederal solutions. They regard asymmetry as a proper recognition of their ‘distinct society’ status. This conflict, rather than bureaucratic problems of managing diversity, is the real problem besetting asymmetry. Kymlikca (1998a, b) points to a number of provisions about asymmetry that make it unsustainable. If a group insists on asymmetry and others do not concede it, the stalemate may result in attempts at secession. On the other hand, the concession of asymmetry merely encourages the demand for further powers and emboldens the group, having already won and operating with a large measure of autonomy, to go to the logical next step: separate statehood.

The future feasibility and viability of multi‐ethnic autonomy thus depends greatly on how asymmetry is negotiated. While the utility of asymmetry may be acknowledged, political and bureaucratic difficulties may limit its application. On the other hand, it must be noted that, outside Canada, the difficulties are more theoretical than practical. The Indian experience and that of other federations show that groups claiming or enjoying autonomy do not see themselves as nations alienated from others. They are strongly bonded to the wider nation and their representatives through regional or national parties, and play a full part in national institutions. But equally one must not underestimate the ability of politicians to erect these theoretical difficulties into real barriers to asymmetrical autonomy.

(p.163) Cultural Autonomy

A major limitation of territorial devolution of power, its restriction to circumstances where there is a regional concentration of an ethnic group, can be overcome by ‘corporate or cultural autonomy’ whereby an ethnic group, dispersed geographically, is given forms of collective rights. There are different forms and uses of corporate autonomy. Rights or entitlements protected under such autonomy can be personal, cultural, or political. They can be entrenched or subject to the overriding authority of the government. They normally consist of positive and substantive rights and entitlements, but they can be negative, such as a veto. They form the basis of the communal organization of politics and policies and of the collective protection of their rights. The Cyprus constitution of 1970 was an example of expansive corporate autonomy, while the current constitution of Bosnia‐Herzegovina combines more traditional federalism with corporate shares in power and communal vetoes.

Cultural autonomy was a significant feature of old and modern empires. Modern examples include provisions in the constitutions or laws of Estonia, Hungary, Slovenia, and the Russian federation, which countries provide for the establishment of councils for national minorities that assume responsibility for the education and cultural affairs of the minorities (Eide 1998: 256–9). In principle, a council can be set up if a majority of the community desire it, as expressed in votes. Once established, its decisions bind members of the community throughout the state, except that a member can opt in or out of membership—the important principle of self‐identification is maintained. Within the areas in relation to which powers are vested in it, the council's regulations prevail over those of the state. The council has the power to levy a tax on its members and also receives subsidies from the state. It has authority over the language, education, and culture of the minority. The principal objective of the system is the maintenance or strengthening of the identity of the minority, based on language and culture. The objective is to take culture out of ‘politics’, and leave other matters to the national political process, in which minorities may or may not have a special status through representation. It is too early to evaluate their experience as the few councils established so far, often under external pressure, have existed for only a short period. However, it would seem that the distinction between culture and politics may be too simplistic, especially today when the survival of culture (p.164) is closely connected to the availability of resources and to national policy in several areas.

More central reliance on group autonomy through cultural councils is found in the developing constitutional dispensation of Belgium. In 1970 separate councils were established for Dutch, French, and German language speakers with competence over aspects of cultural and educational matters; their competence was considerably extended in the 1980s (Peeters 1994; Murphy 1995). In some new constitutions group autonomy is related to, or is part of a package of, federal or other devices for protection of ethnic communities, frequently in consociational arrangements, such as in Belgium, Bosnia‐Herzogovina, and Fiji.

Cultural autonomy can take the form of the application to the members of a community of its personal or religious laws, covering marriage and family, and occasionally land, particularly for tribal communities (see Ghai 1998: 52–9); for an historical account of its use in Europe, see Eide 1998). The application of personal laws, and thus the preservation of customary law or practices, is considered important for maintaining the identity of the community. When India tried, during the drafting of its constitution, to mandate a common civil code for all of the country, some Muslim leaders objected. The supporters of a common code argued that common laws were essential for national unity. The opponents argued that it amounted to the oppression of minorities and the loss of their communal identity. The result was that the constitution merely set a common code as an objective of state policy, and it is now a well‐established convention that the shariah will continue to apply to Muslims so long as they desire it.

The scope of the application of personal laws, quite extensive during the colonial period in Africa and Asia, is now diminishing under the pressure of modernization, although it is being reinforced in some countries committed to a more fundamentalist view of their religion. However, one place where regimes of personal laws still apply with full vigour is Israel, where each of the major religions has its own laws on personal matters (Edelman 1994, on which the following account is principally based). Israel has civil courts, military courts, and courts of 14 recognised religious communities. The principal and exclusive jurisdiction of religious courts is over matters relating to marriage and divorce, there being no civil marriage or divorce in Israel. These courts also resolve other personal and private‐law issues. Since legislative authority over these matters is rarely exercised, courts have a profound effect on shaping the country's political culture, involving rights of women, contacts between (p.165) members of different communities, and more generally the lives of Israelis. For the Jews, most matters of personal law fall exclusively within the rabbinical courts, while Muslims are subject to the jurisdiction of shariah courts applying the shariah. Although linked to and supported by the state, these courts are administered independently of the state. For the Muslims the presence of shariah courts has reinforced their sense of community and the values they want to live by, and helped in the social reproduction of the community—an important factor for a minority, many of whom live under foreign occupation. These conclusions corroborate an argument for cultural autonomy, namely, that it ‘supports political stability by providing non‐dominant (and unassimilable) groups with mechanisms that enable them to minimise the effects of their inferior position in the larger society’ (Jacobsohn 1993: 30).

But the separate regime of Muslim law has isolated Arabs from the mainstream of Israeli politics. For the Jews, the rabbinical courts have been deeply divisive, symbolizing the fundamental schism between orthodox and secular Jews. In both instances the courts give the clergy, committed to the preservation of orthodoxy, a specially privileged position. The law is slow to change in these circumstances, and can lag well behind social attitudes and social realities. In contrast to civil courts, which have sought to promote a democratic political culture based upon the rule of law, religious courts and personal regimes of laws have sharpened distinctions among Israel's communities, and retarded social relations among them and the development of a unifying political culture. Edelman (1994: 119) concludes that religious courts have emphasized group identity and solidarity at the expense of a unifying political culture: ‘Yet without a shared political culture and the concomitant sense of a shared national identity, the prospects for a sustained, peaceful national existence are not bright’. This view is not endorsed by Jacobsohn (1993: 37), who says that studies of Jewish public opinion in Israel reveal that shared ethnicity and a shared set of religious symbols are much more important than a shared set of values in providing unity for Israeli society. ‘Thus, the subordination of cultural aspects to individual liberties on the basis of the assertion that the latter are “principles” has less justification in a polity where cultural imperatives may legitimately demand principled consideration.’

One of the major problems with cultural/religious/legal autonomy of this kind is that it puts certain sections of the relevant community at a disadvantage. Edelman (1994) shows how both Jewish and Muslim women come off worse in their respective autonomous courts. In India, Muslim women are unable to benefit from the more (p.166) liberal legal regime that has applied to other Indian women after the reforms of the 1960s. One aspect of their disadvantage was illustrated in 1985 by the famous Shah Banu case,3 where the Supreme Court held that the maintenance that a Muslim divorced woman could claim from her former husband was to be determined under the general national law, which provided a higher amount than she would get under the shariah. This decision provoked a violent reaction from a section of the Muslim community, which considered that its identity was thrown in jeopardy. The government gave way to pressure from the Muslim clergy and other sections of the Muslim community and legislatively overruled the decision. The rise of Hindu nationalism is often ascribed to this ‘capitulation’ by the government to Muslim minority demands. In Canada the application of the customary law of Indian bands has also disadvantaged women; the UN Human Rights Committee has held invalid the law which deprived an Indian woman of her land and other community rights if she married an outsider, men who marry outside the community not incurring a similar liability.4 In South Africa, demands by traditional leaders for the continuation of customary laws were resisted by African women because of the discriminations against them, such as in relation to custody and inheritance. The South African solution was to provide for the application of customary law but subject to the Bill of Rights. The Canadian government is negotiating a similar solution for the band laws.

Finally, I examine the system of cultural and administrative autonomy that the British established for indigenous Fijians in the late nineteenth century and which was subsequently entrenched in the independence and later constitutions. British policy was directed at the preservation of indigenous Fijians' social and economic structure, in view of the destruction of indigenous communities in the South Pacific. It was to be done principally through the preservation of communal responsibility, ownership, and hierarchy. At the same time Britain aimed to exploit the resources of the country, leading to the development of market and modern economic relations. It met the need for labour by recruiting indentured labour from India, and the need for land by very limited alienation of land, leaving most of it under customary ownership. It was therefore necessary both to preserve the Fijian traditional system and to insulate it from external influences. For this purpose a separate system of (p.167) native Fijian administration was established. It consisted of a Council of Chiefs, which was largely advisory to the governor, and a Fijian Affairs Board, consisting of senior bureaucrats and, later, indigenous members of the legislature, which had the power to review legislative bills, particularly for their impact on Fijians, before their introduction in the legislature, and to make regulations for the social and economic life of Fijians. The Council of Chiefs also used to nominate Fijian members of the legislature while representatives of other groups were elected. There were provincial and district councils with jurisdiction over only indigenous Fijians, in which a key role was created for chiefs. Fijian courts were set up to administer customary law and regulations of the Board and the Council. Later legislation provided for a special administrative body to deal with customary land, particularly leases to outsiders.

The Fijian administration could also be seen as a system of indirect rule. This is manifested in the key role played by colonial officials in it and the ways in which chiefs were bolstered and land system administered, sometimes contrary to customary ways. Furthermore, the system of Fijian administration was regulated and supervised by the Ministry of Fijian Affairs—as is still the case. There is not much doubt that these policies had the effect of protecting indigenous Fijians from exploitation and the destruction of their cultures and institutions, even though aspects of these were reconstructed by the colonial authorities.

But after independence the Fijian administration was used less for indirect rule than for maintaining the authority of the chiefs and the cohesion of the community. The Council of Chiefs acquired greater political authority, although its legal powers remained unaltered, and the provincial administration became a device to develop Fijian consensus, particularly on matters regarding relations with the Indo‐Fijians. It became increasingly a system which, from a communal base, affected and shaped the policies of the state, representing a reversal of the original purpose: it came to dominate the state rather than being an adjunct to it. But because it is a communal system, it is discriminatory and exclusionary. The Council of Chiefs has been mobilized by Fijian politicians wanting indigenous political hegemony; it attempted to legitimize all the coups d'état through its support. At the same time Fijian administration has prevented the full integration of indigenous people in the modern economy—it is the absence of a fair role for them in the economy which has been a cause of resentment and sometimes the justification for the coups. Moreover, the separate system has retarded fair and amicable inter‐ethnic relations and a more integrated political (p.168) system. Its original rationale no longer applies, particularly as market relations increasingly dominate life in Fiji and large numbers of indigenous Fijians have moved to urban areas. The ability of Fijian administration to bring cohesion to the indigenous Fijians has been seriously eroded, as many of the values it sought to promote have little resonance with the ordinary Fijians, who see it principally as a means to bolster chiefly privileges.


The cases studied here show that no simple judgement on the utility or justice of the political recognition of ethnic diversity is possible. Separate legislative representation has sometimes been worthwhile, as the Indian example shows; but mostly it has been harmful. Asymmetrical federalism has great capacity to respond to the varying circumstances and needs of ethnic groups. But it is hard to negotiate and sometimes hard to operate. Cultural autonomy can give a beleaguered community a sense of identity and moral cohesion, and assist in preserving its traditions. But as with other asymmetrical devices, it can cause injustice to both the members of the autonomous community and those outside it. All three can produce resentment and conflict.

Each of these devices has supporters and opponents. Even if it were agreed that one of them was the preferred approach, it may be hard to generalize about the usefulness of particular modalities. The choice between these options may depend, in many situations, less on their inherent merits than on circumstances and constraints. The objective circumstances as well as the aspirations of minorities vary from place to place and from time to time. For example, the size of the minority is a material factor: a substantial and economically well‐off minority might not require special rules for legislative representation, but a small minority might. Moreover, in the former case special rules might be resented or mistrusted by the majority, but not necessarily in the latter case.

The choice of approach and modalities would depend on the ultimate goals that the state and minorities have set themselves. The problem arises when there is no consensus either between the majority and the minority or within either group. A section of a minority may want to preserve its social structure and culture at all costs; another may wish to escape the constraints or even the oppression of the community and seek its identity in a cosmopolitan culture. The choice would also depend on the balance between (p.169) individual and communal rights. Nor are particular solutions valid for all times; they may need to be reviewed as the socio‐economic and demographic situation changes. It is worthwhile to caution against reifying temporary or fluid identities, which are so much a mark of contemporary times. There is a danger of enforcing spurious claims of primordialism and promoting competition for resources along ethnic lines, thereby aggravating ethnic tensions. Separate representation and institutions tend to lead to ethnic manipulation or extremism. Many proposals for diversity which have being made in recent years are untried; and, even when tried, it is too early to assess their success. Many of them are concerned excessively with conflict management, and perhaps not sufficiently focused on long‐term objectives.

Nevertheless, these studies highlight some aspects of the constitutional recognition of diversity that pertain to policy on this matter. Several examples of legal recognition of cultural diversity were imposed rather than sought by minority groups—for example, apartheid structures, or divide‐and‐rule mechanisms used by colonial authorities. Historically, diversity arrangements have been connected with discrimination and domination. Often, if culture or religion is constitutionally recognised, it is the culture of the majority, resulting in the domination of the culture of others—Sri Lanka, Malaysia. Separate cultural systems are also a way of privileging some members of a community, such as traditional elites—usually male—or the wealthy, over others. For similar reasons, emphasis and efforts that go into developing separate systems for separate cultures mean that urgent social problems, whether of a community or of all the people, may be neglected.

Cultural rights may also put at risk human rights. The literature on group rights has highlighted how the rights of both certain members of the cultural community and outsiders may be infringed in this way. While it is true that human rights may themselves be culture‐specific, the fact is that modern human rights are for the large part negotiated norms to which different traditions have contributed, and they do represent the aspirations of many people around the globe. Rules for cultural autonomy should be sensitive to the needs of individuals for more cosmopolitan identities, and, on the principle of self‐identification, provide a reasonable basis for ‘exit’. Nor should rules for cultural autonomy overlook the contingency and dynamic of cultures. Cultural autonomy tends towards conservatism and conservation at a time of rapid social and technological change when many of its assumptions are being undermined.

(p.170) Constitutional recognition of cultures tends to sharpen differences between communities. Many schemes for recognition have been controversial and divisive. Progress towards a fair and integrative constitution in Fiji in the 1990s was obstructed by the entrenchment of ‘diversity’ from the past; indeed, the entrenchment of Fijian administration ran counter to the integrative aspirations of other parts of the constitution and of many people. Cultures should be recognized in a way that builds bridges and increases mutual understanding and appreciation of cultures; we need more inter‐cultural than multicultural enterprises. Instead of a multiplicity of laws, we should work together for a genuine integration of laws, drawing together what is valuable in each culture. We ignore the implications of such recognition for inter‐ethnic relations at our peril.


(1) The following account draws heavily on Galanter (1984).

(2) Delimitation Act 1972, sec. 9(1)(c).

(3) (1985) 2 Sup. Ct. Cases 556.

(4) Sandra Lovelace v. Canada (Report of the Human Rights Committee. GAOR. Thirty‐sixth session, Supplement No. 40 (A/36/50), 166–75.