Application of modern classification methods in the study of bilingualism

Abstract The main goal of the present paper is to demonstrate the usefulness of some modern cluster analytic techniques in linguistics by means of data from a longitudinal study of language shift of a Romanian community living in Hungary (Borbély 2016). Based on this sample we attempted to explore statistically valid and linguistically meaningful homogeneous types, which characterize bilingual adult persons of a Romanian community in Hungary. Cluster analyses identified seven main types of language shift. It was also shown that the way from the monolingual Romanian to the monolingual Hungarian state is not a linear process. This may give a chance to a successful intervention to slow down the process of assimilation and language shift. Our analyses demonstrated also the usefulness of the MORI indices (Vargha Bergman & Takács 2016) in determining a proper cluster number, and in the validation of a cluster structure, mainly with the application of a correlated random multidimensional normal data set.