| Summary: | Prediction is often assumed to play a crucial role during language comprehension. While some theories propose that prediction robustly affects at all levels of linguistic representation, empirical evidence suggests that the circumstances under which linguistic predictions occur appear to be limited, particularly when comparing prediction of phonological information to semantic information. To more directly explore these limits, we compared effects of semantic and phonological prediction in a visual world eye-tracking study. Participants heard sentences where the target word was either predictable (e.g., ?That dog looks so happy, wagging its tail ??) or unpredictable (?If there is one, click on the picture of the tail.?), while viewing objects that corresponded to either the target word (tail), a semantic competitor word (paw), a phonological competitor word (table), or an unrelated word (daisy). Target and semantic competitor objects attracted more fixations than unrelated objects well before the target word onset in predictable sentences and not in unpredictable sentences, suggesting that participants predicted semantic information. However, there were no predictive eye movements for phonological competitor objects. The results suggest that phonological prediction is not as robust as semantic prediction.
|
|---|