Machine Componential Analysis of Kinship Vocabularies: An Example

Abstract

The paper introduces a computer program to handle the task of componential analysis of kinship terms. Given the kin terms of a language with their attendant kin types, the program discovers all componential models, including the "simplest" ones (using the minimum number of dimensions and components in kin term definitions). As an illustrative example of the application of the program we use the kinship vocabulary of Catalan, a previously unanalyzed language. A completely unconstrained analysis of Catalan leads to an intolerably large number of alternative models but our simplicity restriction pertaining to choosing the minimum number of dimensions leads to a unique model. The generated componential model of Catalan uses four dimensions of contrast, viz. "sex", "generation", "distance" and "affinity", and coincides with that of Spanish. However, it is different from those of other Indo-European languages that also use for demarcation this very set of dimensions.