Top-down influence in young children's linguistic ambiguity resolution.

Area of Science:

• Psycholinguistics
• Developmental Psychology
• Computational Linguistics

Background:

• Language ambiguity presents a significant challenge for language users.
• Adults resolve syntactic ambiguities using diverse cues, but children appear less sensitive to top-down information.
• It remains unclear if this insensitivity is syntax-specific or a general feature of children's ambiguity resolution.

Purpose of the Study:

• To investigate whether children's resolution of lexical ambiguities relies solely on lexical associations or also incorporates top-down global plausibility.
• To compare children's ambiguity resolution strategies with a computational model based purely on lexical associations.

Main Methods:

• A picture choice task was employed to assess 4-year-olds' ability to resolve polysemes and homophones.
• Children's choices were compared against predictions from a Bayesian algorithm that used only lexical associations.

Main Results:

• The Bayesian algorithm's predictive power for children's choices was limited, indicating reliance on more than just lexical associations.
• A second experiment confirmed that children can override strong lexical associations and integrate top-down plausibility information.

Conclusions:

• Four-year-old children integrate top-down plausibility when resolving lexical ambiguities, not relying solely on lexical associations.
• Findings suggest that children's linguistic ambiguity resolution is more sophisticated than previously thought and has implications for psycholinguistic development models.