L2 Prediction during complex sentence processing

Abstract

Recent studies have found that proficient second language (L2) listeners are able to predict upcoming linguistic information to the same extent as first language (L1) listeners during simple sentence processing, particularly when semantic cues are given and/or few cognitive resources are required for language processing. These findings may suggest that L2 listeners use the same mechanisms as L1 listeners for prediction. Yet, it has not been fully specified under which conditions L2 listeners can use predictive mechanisms. To address this issue, we investigated whether advanced L2 listeners make predictions while processing more complex constructions that are cognitively more taxing. Specifically, we investigated prediction in sentences containing a relative clause that can modify either of two noun phrases. In an eye-tracking study using a visual world paradigm, L2 learners listened to sentences containing a semantically biasing verb or a neutral one (e.g., “I know the friend of the dancer that will open/get the present”). We measured L2 listeners’ prediction by comparing the fixations to target objects (e.g., present among non-openable objects) between the two experimental conditions. Results showed that L2 listeners, similar to L1 listeners, made significantly more anticipatory looks to the targets in the semantically biasing condition than in the neutral condition, though their prediction started a bit (180 ms) later than L1 listeners’ prediction. These findings suggest that L2 speakers can use prediction mechanisms even during complex sentence processing and provide further evidence for the claim that there is no fundamental difference between L1 and L2 speakers, but that cognitive resources matter for prediction.