Syntactic bootstrapping in the acquisition of attitude verbs

Kaitlyn Harrigan

Attitude verbs (e.g., think, want, hope) report mental states. Learning the meanings of attitude verbs may be difficult for language learners for several reasons; including the abstractness of the concepts that they refer to, and the linguistic properties. In this dissertation, we investigate the learning process for these words, by looking at an asymmetry that has been observed in the acquisition trajectory: want, which refers to desires, has been claimed to be acquired before think, which refers to beliefs. We explore this asymmetry in attitude verb acquisition in two ways: by comparing interpretation of think and want, controlling for several methodological differences in the way they have previously been tested; and by investigating children’s sensitivity to syntactic distribution in interpreting and learning attitude verbs.

We start with an observation that previous tasks comparing interpretation of think and want often tested these verbs under different experimental conditions. Tests of think required processing additional demands; including a conflict with reality, and a conflict with the child’s own mental state. In experiments 1-3, we test interpretation of want adding these additional task demands; and find that children are still adult-like in interpreting want sooner than they have reliably shown to be adult-like in interpreting think. In Experiment 4, we directly compare think and want in the same experimental context. We still find adult-like behavior with want and not think. These studies demonstrate that the observed asymmetry between think and want reflects a real acquisition asymmetry, and is not due to experimental artifacts.

After establishing in experiments 1-4 that the asymmetry between think and want reflects real acquisition facts, we explore children’s learning mechanism for attitude verbs in experiments 5 and 6. We test children’s sensitivity to syntactic distribution in hypothesizing an unknown attitude verb’s syntax. In experiment 5, we find that children use syntactic complement to interpret sentences with a potentially unknown attitude verb. In experiment 6, we show that they integrate syntactic information into their semantic representation for this new verb; and continue to hypothesize a meaning based on syntactic frame in future experiences with the same verb.