Multi-variate coding for possession: methodology and preliminary results.

ABSTRACT

In this work we are presenting a database structure to encode the phenomenon of differential possession across languages, considering noun possession classes and possessive constructions as independent but linked. We show how this structure can be used to study different dimensions of possession semantics, noun valence, and possessive constructions. We present preliminary survey results from a global sample of 120 languages and show that there is a universal semantic core in both inalienable and non-possessible noun classes. Inalienables are centered on body parts and kinship. Non-possessibles are centered on animals, humans, and natural elements.