Electrophysiological Assessment of Semantic Processing of Cochlear Implant Users Using an Audiobook

Cochlear implants (CI) afford listeners with limited spectral resolution, thereby constraining semantic processing. Prior research has predominantly used isolated words or sentences without context, a method that lacks ecological validity and may place greater demands on working memory (WM) due to the absence of contextual linguistic information. This study examined the semantic processing of experienced CI users (N = 21) and age-matched normal-hearing controls (NH, N = 21) using an audiobook. We measured neuroelectric brain activity while participants listened to a story in which we inserted an incongruent word every 45 s, herein, the E45 condition. WM and behavioral semantic judgment were quantified with a listening span test. Temporal response functions, linear models that characterize how continuous stimulus features are temporally mapped onto neural responses, were used to quantify speech neural tracking of audio features (envelope, word onset, word surprisal) during ongoing speech. Results showed reliable neural tracking of acoustic features and word surprisal. Compared to the Control condition, E45 elicited larger TRF-N400 responses in both groups. Although TRF-N400 amplitudes were comparable between CI and NH listeners, CI users exhibited delayed responses. In NH listeners, larger N400 effects were associated with greater semantic judgment accuracy, linking neural responses to behavioral performance. These results show the validity of an ecologically valid paradigm that does not require an explicit behavioral response to quantify semantic processing in CI users.