| Adult language combines a complex mix of regular 'rule life' processes and more conservative, lexically based patterns. For example, verb argument structure constructions may generalize to new verbs (Arthur fugged the box to Ford → Arthur fugged Ford the ball) yet resist generalization with certain lexical items (Arthur carried the box to Ford → *Arthur carried Ford the box) Certain researchers have suggested that whether a particular verb can occur with a particular argument structure depends upon its semantic representation. However, frequency-based entrenchment effects in young children, and statistical effects in sentence processing, suggest that learners track verb-structure co-occurrences. This concurs with recent approaches to language acquisition which emphasize the role of statistical learning processes. The current work uses the Artificial Language Learning Paradigm to explore whether the relationship between verbs and argument structures can be acquired from input statistics.; Adult subjects were exposed to languages in which input distribution provided the only cue to verb sub-category. Four languages were explored, with differences among them in the degree to which their verbs exhibited lexically based versus language-wide patterns. After several days of exposure, learners took Production, Grammaticality Judgment and On-line Sentence Processing tests.; Across the different experiments, subjects proved able both to acquire verb sub-categorization, and to generalize. Behavior in the different tests was very consistent, and reflected both the statistical preferences of particular verbs and across-verb statistics. However, the tendency to generalize was affected by a third source of information: the distribution of verb types in the language. Learners exposed to languages in which the majority of verbs occurred in multiple structures were more likely to generalize than those exposed to a language in which most verbs were lexically constrained.; These results demonstrate that verb argument structure acquisition is driven by various distributional properties of the input, and that these purely formal phenomena may occur in the absence of semantic cues. In addition, exposing learners to artificial languages has been shown to provide a fruitful methodology for exploring the relationship between statistical language learning and statistical effects in language use and comprehension. |
