Tydskrif vir Geesteswetenskappe
versión On-line ISSN 2224-7912
versión impresa ISSN 0041-4751
SENEKAL, Burgert. Monolingualism, integration and assortative mixing in social networks: A literature review. Tydskr. geesteswet. [online]. 2015, vol.55, n.3, pp.356-372. ISSN 2224-7912. http://dx.doi.org/10.17159/2224-7912/2015/V55N3A3.
Afrikaans education institutions are increasingly coming under fire, because the use of Afrikaans is believed to lead to a lack of integration between different population groups in South Africa. Afrikaans is seen as a white language, while it is argued that black students attend classes in English, which means that the separation of students in terms of language also allegedly leads to a separation in terms of race. The proposal is then that by removing Afrikaans as a medium of instruction, South African education institutions will become integrated. This article critically engages with this view from the perspective of complex network theory by investigating community formation in networks. The concepts of modularity and transitivity or clustering, as developed within the field of complex network theory, are discussed with reference to examples of different types of networks. Modularity provides a way of quantifying whether distinct groups of nodes emerge in a network, while transitivity refers to the formation of triangles in a network. Modularity and transitivity are calculated for a number of real networks, and compared with Erdös and Rényi (1960) type network models where link formation occurs at random. It is shown that community formation is not peculiar to South African society or even social networks in general, but is a characteristic of complex networks, which include social networks as well as technological, biological and information networks. Community formation arises partly as a result of assortativity or homophily, which means that similar nodes tend to form links with each other. Homophily in social networks can be based on language, but also manifests on the grounds of race, class, income group, geographical location and so on. An overview is given of a number of studies that investigated community formation in social networks, including in terms of language, such as Blondel et al.'s (2008) study of language communities in Belgium, or Leskovec and Horvitz's (2007) study of the Microsoft Instant Messaging (IM) network. Although it is shown that quantitative evidence exists that language does lead to the formation of distinct communities, some studies of community formation in the monolingual environment of the United States are also discussed, including Moody's (2001) study of race relations amongst high school students, where it was shown that homogenous communities also emerge without language being a variable. In addition, some quantitative evidence is also provided for community formation based on age and geographical location. The conclusion the article arrives at is that integration will not be brought about by removing Afrikaans from education institutions, since the monolingual environment of the United States is still characterized by separate communities that are formed on the basis of race. The wide-spread occurrence of communities in networks indicates that it is a salient feature of networks, and that when removing one variable, another will still result in the formation of distinct communities. A brief discussion is also given of problems associated with the formation of social ties based on heterophily, i.e. where social ties are formed based on difference rather than similarity.
Palabras clave : complex networks; social networks; integration; modularity; clustering; transitivity; language communities; language policy; monolingualism.