Sensitivity to honorific agreement: a window into predictive processing

Abstract Language processing involves active engagement with linguistic cues to predict future information while integrating available information in a bottom-up manner. Recent research in this area suggests that individual differences play a role in predictive processing. However, the exact ways in which individual factors contribute to these differences remain a topic that warrants further investigation. Additionally, most previous studies in this area have focused on English and European languages. The current study addresses these gaps, making use of the fact that the Korean honorific verbal suffix can be a predictive cue for an upcoming honorable referent, using both an offline sensitivity task and an online webcam-based eye-tracking task. The offline data analysis using hierarchical clustering reveals that Korean speakers exhibit individual differences in their sensitivity to the honorific verbal suffix. The eye-tracking data analysis using generalized additive mixed modeling shows that individuals with higher sensitivity tend to display more robust and faster predictive processing patterns. Theoretically, our results provide insight into the underlying mechanisms of predictive processing, highlighting the significance of individual factors. Meanwhile, our online data collection method offers a more inclusive approach to collecting psycholinguistic data.