Phenotypical, Linguistic or Religious? On the Concept and Measurement of Ethnic Fragmentation

Malaysian Journal of Economic Studies, Jun-Dec 2003 by Yeoh, Kok Kheng

An equally important point to note is that there are other socio-economic reasons behind ethno-linguistic and ethno-religious divides. This is especially the case in Brazil and Spanish-speaking America where social definition is relatively fluid, reflected in the Brazilian proverb: "A rich black man is a white and a poor white man is a black" (Mason 1970:122). It is probably in this light that Hoetink had chosen the attribute 'socio-racial', which reflects the concept of 'social race' (vis-à-vis 'biological race') expounded by Wagley (1959). Similar concerns are covered by Gordon's concept of 'ethclass' as "the portion of social space created by the intersection of the ethnic group with the social class [which] is fast becoming the essential form of the sub-society in America" (Gordon 1978: 134), and by Bonacich's 'split labour market theory' as a 'class' approach to race and ethnicity (Bonacich 1972; 1979). These are summarised in Rex's comment that "the large communal quasi-groups which are called ethnic and racial are the collective entities which are brought together in systems of class, estate, status group domination, caste and individual status striving ... [and] what we call 'race and ethnic relations situations' is very often not the racial and ethnic factor as such but the injustice of elements in the class and status system" (Rex 1986: xiii). Although social classes may not be as precisely bounded as ethnic groups, both represent forms of demographic diversity which serve as a means of group identification, an arena for the confinement of group relations and a carrier of cultural patterns of behaviour (Gordon 1978).

3. An Index of Ethnic Fractionalisation

Data for computing the ethnic fractionalisation index (EFI) are drawn from various sources, including the individual studies of Katzner (1995), MRG (1990), Kurian (1990), Gunnemark and Kenrick (1985), Malherbe (1983), annuals such as the EWYB4, RSW5, WABF6, CIA's World Factbooks7, as well as many other references on individual countries/ regions. The first two categories are mainly concerned with the numerical dimension. The last category is particularly important since it concerns the socio-political and historical background which directly affects the definitions of ethnicity.

The source of data for the computation of the EFI (Table 2) is broader than that of previous studies on public policy and ethnicity, e.g. Mueller and Murrell ( 1986) and McCarty (op.cit.). Mueller and Murrell relied on Taylor and Hudson (1972)8 which computed three different sets of indices based on data from Roberts (1962), Muller (1964) and the Atlas Narodov Mira9 respectively, none of which are employed here since they are relatively dated. McCarty's source of data for his ethnic and religious 'variance' is the World Factbooks. However, a close scrutiny of this source reveals its major weaknesses, viz. the tendency to employ broad categories such as 'Caucasian', 'African', 'white', 'black', 'Nilotic', 'Mongoloid', 'Indo-Aryan', 'Dravidian', 'Hamitic' and the like, as well as the focus on 'official' languages and commercial linguae francae rather than 'home' languages. Computation based on such broad categories would result in the gross underestimation of heterogeneity. Therefore it is necessary to broaden the source of data to achieve more detailed breakdowns of racial, ethno-linguistic and ethno-religious categories.