Beyond associations: Sensitivity to structure in pre-schoolers' linguistic predictions

Area of Science:

• Developmental Psychology
• Psycholinguistics
• Cognitive Science

Background:

• Children learn language by making predictions based on semantic associations.
• Previous research indicates 3-year-olds can generate predictions using semantic cues.
• It remains unclear if children integrate semantic associations with linguistic structure for predictions.

Purpose of the Study:

• To investigate whether preschool-aged children generate predictions by combining semantic associations with knowledge of linguistic structure.
• To determine if children's predictions are sensitive to grammatical constraints.

Main Methods:

• Recorded eye movements of preschool children listening to sentences.
• Presented sentences designed to elicit predictions based on verb meaning and grammatical role.
• Compared predictive looking behavior in structure-present versus structure-absent conditions.

Main Results:

• Children predictively looked at plausible patients (e.g., 'horse' after 'ride') but not implausible ones.
• This predictive looking was absent when linguistic structure was absent, even with strong semantic associations.
• Findings indicate children integrate semantic and structural information for prediction.

Conclusions:

• Young children's language predictions are sensitive to linguistic structure.
• Supports predictive-learning models of language acquisition.
• Demonstrates early integration of semantic and structural knowledge in language processing.