Let’s See a Boy and a Balloon: Argument Labels and Syntactic Frame in Verb Learning

Abstract

It is by now well established that toddlers use the linguistic context in which a new word—and particularly a new verb—appears to discover aspects of its meaning. But what aspects of the linguistic context are most useful? To begin to investigate this, we ask how 2-year-olds use two sources of linguistic information that are known to be useful to older children and adults in verb guessing tasks: syntactic frame and the semantic content available in the noun phrases labeling the verb’s arguments. We manipulate the linguistic contexts in which we present novel verbs to see how they use these two sources of information, both separately and in combination, to acquire the verb’s meaning. Our results reveal that, like older children and adults, toddlers make use of both syntactic frame and semantically contentful argument labels to acquire verb meaning. But toddlers also require these two sources of information to be packaged in a particular way, into a single sentence that identifies “who did what to whom.”

Notes

¹ Note that Sparse Syntax means that the syntactic frame in which the verb is embedded is relatively uninformative. This frame does still convey some information, e.g., that the novel word is a verb, can describe an ongoing event, etc.

² To provide assurance that chance performance is an appropriate baseline for this task, we tested a different group of nine toddlers in a No Word condition, in which the visual stimuli were identical to Experiments 1 and 2, but no novel words were introduced. At test, toddlers were asked, “Which one do you like?” Seven toddlers performed at chance (0.50), one chose the Familiar Action scene on two of the six trials, and the other chose it on four of the six trials. These results support our use of 0.50 as chance performance.

³ Thanks to an anonymous reviewer for raising this last possibility.

Additional information

Funding

This research was supported by National Institutes of Health (NIH) Grants HD067485 to both authors and HD30410 to the second author.