The involvement of phonological information during spoken language prediction: evidence based on Chinese tone sandhi

Can listeners predict phonological information? Do they use phonological cues for prediction? Previous studies on prediction have often conouded phonological and orthographic factors, making it unclear whether phonological prediction exists. Additionally, previous studies have primarily focused on abstract semantic or syntactic cues, leaving a gap in research on whether listeners use low-level features (e.g., phonology) for prediction. This study used a visual world paradigm (VWP) experiment and an event-related potentials (ERPs) experiment to look for evidence of phonological information involvement in prediction based on Chinese tone sandhi (tones are a purely phonological feature that is perfectly separated from orthography). The results showed that, in the VWP experiment, when listeners heard tone-bearing words in a sentence, they could predict the tone of the upcoming word based on the tone of the tone-sandhi-bearing word, indicating that listeners can not only predict upcoming tonal information, but also that the prediction process uses tonal information. In the ERPs experiment, tone-bearing words that violated the predicted tone induced a greater negative wave, confirming the robustness of tonal prediction. These results provide empirical evidence for the involvement of phonological information in language prediction processing.