If verb complementation patterns are learned through experience (e.g., Seidenberg & MacDonald, 1999), biases for flexible verbs should be modifiable even in adults. Two experiments investigated the impact of a single encounter with a verb and one of its complements on later processing of the verb. The verbs tested were optional transitives (e.g., FIGHT), verbs that take sentence or NP complements (REVEAL), and verbs that take sentence or PP complements (JOKE). In Experiment 1, speakers showed significant and equivalent compliment priming in production for all three verb classes over one unrelated intervening sentence. For example, given an incomplete sentence to complete ("After being dismissed, the mayor's aide revealed"), speakers were more likely to produce a sentence complement if they had recently completed another sentence fragment where the verb was used with a sentence complement ("Before dying, the victim revealed that") rather than a NP ("Before dying, the victim revealed the name of"). Because different types of complements are associated with expression of different types of arguments (i.e., entities, propositions), the observed priming effect may combine structural (syntactic) and argument (semantic) priming. The short-term priming effects set the stage for investigating longer-term consequences of a single exposure to a flexible verb with one of its complements in the second experiment.
If verb biases are modifiable, recent use of a verb with a NP complement should strengthen the verb's NP bias. Consequently, the verb's transitivity should be more quickly and accurately recognized in both production and comprehension. Participants studied pairs of sentences for immediate recall. Verbs' complements varied as in Experiment 1. The second half of the experiment was introduced to participants as an entirely new one. Participants viewed an isolated verb for up to four seconds, saying the word THEM aloud if (or when) they realized that THEM could sensibly follow the verb. If the verb was not transitive, participants remained silent. Despite a twenty minute interval between exposures, participants were significantly more accurate in recognizing verb transitivity when verbs were previously studied in sentences with NP complements. Participants were also faster to recognize the transitivity of primed optional transitives. The results support the previously untested assumption that verb representations are modifiable by a single use.
References
Seidenberg, M. S., & MacDonald, M. (1999). A probabilistic constraints approach to language acquisition and processing. Cognitive Science, 23, 569-588.