ABSTRACT
Many comprehension studies of grammatical development have focused on the ultimate interpretation that children assign to sentences and phrases, yielding somewhat static snapshots of children's emerging grammatical knowledge. Studies of the dynamic processes underlying children's language comprehension have to date been rare, owing in part to the lack of online sentence processing techniques suitable for use with children. In this chapter, we describe recent work from our research group, which examines the moment-by-moment interpretation decisions of children (age 4 to 6 years) while they listen to spoken sentences. These real-time measures were obtained by recording the Children's eye movements as they visually interrogated and manipulated objects in response to spoken instructions. The first of these studies established some striking developmental differences in processing ability, with the youngest children showing an inability to use relevant properties of the referential scene to resolve temporary grammatical ambiguities (Trueswell, Sekerina, Hill, & Logrip, 1999). This finding could be interpreted as support for an early encapsulated syntactic processor that has difficulty using non-syntactic information to revise parsing commitmerits. However; we will review evidence from a series of follow-up experiments which suggest that this pattern arises from a developing interactive parsing system. Under this account, adult and child sentence comprehension is a “perceptual guessing game” in which multiple statistical cues are used to recover detailed linguistic structure. These cues, which include lexical-distribution evidence, verb semantic biases, and referential scene information, come “online” (become automated) at different points in the course of development. The developmental timing of these effects is related to their differential reliability and ease of detection in the input.
