Semantic similarity to high-frequency verbs affects syntactic frame selection

This paper investigates the effect of the high frequency of occurrence of a verb in a syntactic frame on speakers' selection of that syntactic frame for other verbs. We hypothesize that the frequent co-occurrence of a syntactic frame and a particular verb (what we call an anchor verb) leads to a strong association between the verb and the frame analogous to the relationship between a category and its best exemplar. Our Verb Anchor Hypothesis claims that verbs that are more semantically similar to the anchor are more likely to occur in that syntactic frame than verbs that are less semantically similar to the anchor. We tested the Verb Anchor Hypothesis on the dative alternation which involves the meaning-preserving ditransitive and prepositional frames. A corpus study determined that give was the anchor verb for the ditransitive frame. We then examined whether high semantic similarity to give increases the likelihood of an alternating verb (e.g. to hand) occurring in the ditransitive frame (Mary handed the boy a book) rather than in the prepositional frame (Mary handed a book to the boy). The results of several logistic regression analyses show that semantic similarity to give makes a unique contribution to predicting the choice of the ditransitive frame aside from other factors known to affect syntactic frame selection. Additional analyses suggest that the Verb Anchor Hypothesis might also hold for more narrowly-defined subclasses of alternating verbs.