Eye movements reveal differences in children's referential processing during narrative comprehension

Abstract

Children differ in their ability to build referentially coherent discourse representations. Using a visual world paradigm, we investigated how these differences might emerge during the online processing of spoken discourse. We recorded eye movements of 69 children (6-11 years of age) as they listened to a 7-min story and concurrently viewed a display containing line drawings of the protagonists. Throughout the story, the protagonists were referenced by either a name (e.g., rabbit) or an anaphoric pronoun (e.g., he). Results showed that the probability of on-target fixations increased after children heard a proper name, but not after they heard an anaphoric pronoun. However, differences in the probability of on-target fixation at word onset indicate that the referents of anaphoric pronouns were anticipated by good comprehenders, but less so by poor comprehenders. These findings suggest that comprehension outcomes are related to the online processing of discourse-level cues that regulate the accessibility of entities.

Keywords: Coherence; Discourse comprehension; Language development; Latent class analysis; Multilevel logistic regression; Referential processing; Visual world.