Lee, E. 1 & Snedeker, J. 2
1 Yonsei University
2 Harvard University
By 2 years of age, children rapidly use disfluencies to predict new referents (Kidd et al., 2011). We examined whether this could reflect their ability to draw inferences about the speaker?s referential intention online, or simply their sensitivity to distributional statistics (i.e., disfluencies are likely to precede novel referents). To tease apart these two possibilities, we built a new situation in which disfluency interpretation was associated with visual processing difficulty instead of with referential novelty. We monitored both adults? and children (mean age=5;8)?s eye-movements to the visual display consisting of four objects (e.g., yellow toaster, green toaster, yellow vest, green vest) as they listened either to fluent (Click on the yellow toaster.) or disfluent instructions (e.g., Click on thee uh yellow toaster.). Visual processing difficulty was manipulated by making the target object either clear or blurry. While adults fixated more at blurry objects during the disfluency and at clear objects in fluent instructions, children simply revealed an overall preference for clear objects, indicating that unlike adults, they failed to associate disfluencies with speaker difficulty. The data suggest that children?s disfluency interpretation is not as flexible as adults? and that it might be mediated by experience rather than by metacognitive processes.