ABSTRACT
Free-association tasks have been used to investigate knowledge of verb-argument constructions (VACs). They demonstrated that English speakers have independent implicit knowledge of (i) verb frequency in the VAC, (ii) VAC-verb contingency, and (iii) verb prototypicality in terms of centrality within the VAC semantic network. They concluded that VAC processing involves rich associations, tuned by verb type and token frequencies and their contingencies of usage, which interface syntax, lexis, and semantics. However, the tasks they used, where respondents had a minute to think of the verbs that fitted in VAC frames like ‘he __ across the . . . ’, ‘it __ of the. . . ’, etc., were quite conscious and explicit. The current experiment therefore investigates the effects of these factors in on-line processing. The experiment had participants read aloud verb-VAC arguments like “leaned over”, “thinks over”, “knock over” as quickly as possible. These exemplars varied in their corpus-derived verb-VAC frequency, verb-VAC contingency, and verb prototypicality within the VAC semantic network. Responses were audio recorded and Audacity was used to measure the latency of naming the second word (the preposition) after simultaneous visual presentation of the verb-VAC arguments. A mixed linear model including random intercepts for participant and VAC, and random slopes for frequency, contingency, and prototypicality, explained 64% of the variance. It demonstrated that naming latency of the VAC preposition was affected by verb length, verb frequency in the corpus, verb-VAC frequency and semantic prototypicality. These results demonstrate the effects of usage on VAC knowledge, and particularly that VAC processing is sensitive to the statistical co-occurrence of verbs, VACs, and their meaning.
