Proceedings of the Annual Meeting of the Cognitive Science Society

Researchers studying semantic processing during continuous speech have proposed temporal response function (TRF)–based measures of semantic tracking as a complement to the N400 event-related potential (ERP). However, the interpretation of these measures and their relation to the well-established N400 remain unclear. Here, we asked whether semantic tracking during naturalistic podcast listening relates to cross-modal semantic priming, as reflected in the N400 elicited by intermittently presented visual word probes. EEG was recorded while participants listened to podcast excerpts containing time-locked visual probes that varied in probe–context relatedness. In parallel, TRF decoding models were used to estimate individual semantic tracking performance from the speech stream. We found that less related probes elicited larger N400 amplitudes, and N400 sensitivity to probe–context semantic distance scaled with semantic tracking performance. These results suggest that TRF-based semantic tracking and probe-evoked N400 priming reflect a shared discourse-level semantic representation during naturalistic listening.