Distributed Representations of Lexical Sets and Prototypes in Causal Alternation Verbs

Lexical sets are sets of words filling a specific argument position of a verb in one of its senses. They can be extracted from corpora automatically. The purpose of this paper is demonstrating that the properties of lexical sets are mirrored by their distributed vector representations. This provides insights onto many linguistic phenomena, such as the causative-inchoative alternation in the verbal domain. A first experiment aims at investigating the internal structure of the sets, whose meanings are known to be radial and continuous categories cognitively. In particular, we establish an equivalence between the prototype of a set of words and the centroid of a set of vectors. A second experiment shows that the distance between the intransitive subject set and transitive object set is correlated with the spontaneity of the event expressed by the verb, defined in terms of morphological coding and frequency.