Many verbs can occur in different sentence structures. For example, some ditransitive verbs — verbs that have both a direct and an indirect object, like "give" — are flexible as to the object ordering, as in sentences like "Jill gave the duck a cracker" and "Jill gave a cracker to the duck." The tendency for a verb to occur in one structure more frequently than another is called verb bias. Listeners use verb bias to anticipate how a sentence will progress. However, how a sentence progresses may go against verb bias and violate listener's predictions. Inaccurate predictions need to be revised in order to match the intended meaning of a sentence. Children under eight have immense difficulty with revising predictions. This difficulty is linked not only to the misinterpretations of sentences but also to the delayed acquisition of passive sentences and sentences with relative clauses. This dissertation project investigates whether preschoolers can use disfluencies in a speaker's utterance (e.g. thee uhh) to anticipate structures that go against verb bias, and whether using disfluencies in this way can facilitate language development. The relationship between children’s cognitive development and the use of disfluencies will also be explored. Gaining insight into these questions will advance our understanding of children's language development, language processing, and the impact of cognitive development on these processes. Furthermore, the results of this project will have implications for developing targeted language interventions in cases of developmental language delay.
The research project will use behavioral and computational methods to address these questions. Behavioral studies will measure children's expectations for ditransitive sentences with and without disfluencies by using eye-tracking procedures. Children's cognitive development will be measured through a well-validated cognitive flexibility task. Throughout the behavioral studies, the expectedness of the ditransitives will be manipulated within and across sentences. A computational cognitive architecture model incorporating disfluencies as a cue for low probability structures will then be used to map children's developing understanding of ditransitives and will be validated against data from the behavioral studies. This model will be used to conduct rigorous examination of subtle and fine-grained interactions between language specific and general cognitive mechanisms. In future work, the model can be applied more broadly to study the development of children's language processing abilities and to design targeted and tailored language interventions.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
