Prediction error cost exists in the reading processing of Chinese native speakers and advanced Chinese L2 learners

This study applies the paradigm of self-paced reading to examine the Context Predictability Effect in the processing of Chinese and detect whether there is a prediction error cost. Context constraint strength (constraining and neutral) and word predictability (predictable and unpredictable) were strictly manipulated. The statistical results suggest that: (1) There is a Context Predictability Effect for Chinese native speakers in reading processing, which is consistent with most previous studies; (2) There is also a Context Predictability Effect for advanced Chinese L2 learners; (3) Both Chinese native speakers and Chinese L2 learners have a prediction error cost in reading processing, a finding different from those of much previous research. (4) Chinese L2 learners are significantly slower than Chinese native speakers when they conduct predictive reading processing. This paper is very enlightening in that it identifies the existence of a prediction error cost in Chinese L2 processing by means of behavioral experiments, providing evidence for the hypothesis of Lexical Prediction. In a strongly predictive setting, when encountering a plausible but unpredictable word, the brain must expend extra effort to suppress, revise, or reanalyze the material, and this may account for the prediction error cost.

Automated summary

This study examines the Context Predictability Effect in the processing of Chinese using self-paced reading and investigates if there is a prediction error cost. The study manipulates context constraint strength and word predictability. The results show that Chinese native speakers and advanced Chinese L2 learners both exhibit a Context Predictability Effect and a prediction error cost in reading processing, which is different from previous research findings. Additionally, Chinese L2 learners are slower than Chinese native speakers during predictive reading processing. This paper is significant as it provides behavioral experimental evidence for the existence of a prediction error cost in Chinese L2 processing, supporting the hypothesis of Lexical Prediction.