Word choice is coupled with grammar: The case of kinship

Is it true that ‘grammars code best what speakers do most’, as Du Bois suggested already in 1985? And do members of different cultures privilege talk about kinship relations to differing degrees? Could this be linked to the differential centrality of kinship to the vastly diverse grammars of the world’s languages? We bring these three questions together in this study. Kinship relationships are central to the organisation of all human societies. Yet, we know little about what influences the frequency of reference to kin in spontaneous language use. In this study, we test factors that potentially associate with frequency of kin-term usage across 22 typologically diverse languages. Using a communicative task involving the description, arrangement and narration of pictures, we establish the proportion of human reference expressions formulated as kin terms in interactive language use and show that it does indeed differ across languages. We then examine candidate predictors for the usage frequency of kin-based lexical formulations (choice of word categories) by looking at (i) a structural variable (‘kintax’, the presence of grammatical means for encoding kinship in the language), (ii) a lexical variable (complexity of a subset of each language’s kinship term system), (iii) a discourse variable (type of interaction a language user engages in at the time of speaking or signing), and (iv) a topicality variable (type of stimulus a language user references at the time of speaking or signing). We find all these to be relevant, but we focus on structural kintax. This provides a way of addressing Du Bois’s hypothesis about the interaction between language structure and language use in human communication: the grammatical imprint of a language correlates with language users’ word-choice patterns. We show that the more grammar you have for this semantic domain, the more you talk about it.