Alpha-Band Activity During Sustained Speech Production and Playback: EEG Evidence for Continuous Monitoring

Abstract

Speech production is unquestionably the most vital element in communication, and it depends heavily on a dynamic interaction between motor commands and sensory feedback. To produce speech, auditory feedback, in particular, plays an important role in detecting and refining mismatches between expected and actual vocal output through rapid corrections (Behroozmand & Larson, 2011). A widely utilized approach for examining this feedback control system is the pitch-shift paradigm, wherein perturbations (unexpected changes, such as shifts in voice pitch during vocalization) in the auditory feedback signal elicit compensatory vocal and neural reactions (Behroozmand et al., 2011). This feedback control mechanism has been traditionally studied using Event-Related Potentials (ERPs), particularly N1 and P2 components (Y. Liu et al., 2018). However, these paradigms primarily capture transient, event-locked responses, leaving the neural dynamics underlying sustained, unperturbed speech production less well understood. In particular, frequency-specific neural activity during continuous speech has received comparatively limited attention. In EEG data from previous experiments (Suchý et al., 2023), a recurring pattern of oscillatory activity was observed both before and after ERP responses, but only during vocalization, not during passive listening. The origin of this pattern remains unclear, and it is uncertain whether it reflects a functionally meaningful neural process or arises from methodological factors such as phase-locked averaging or the mixture of periodic and aperiodic components in the power spectrum. The dataset includes 64-channel EEG recordings of 20 participants producing sustained vocalizations of the vowel /a/ while receiving pitch-shifted auditory feedback through headphones. As the auditory stimuli are physically identical across conditions, the difference between self-produced and externally presented stimuli can be attributed to any neurological variations. This design enables a direct comparison between self-produced vocalization and passive listening under controlled conditions. This study, therefore, focuses on alpha-band activity (7 – 12 Hz) during sustained speech production and playback. Spectral analysis using Fourier-based methods was employed to quantify spectral power within a post-onset time window. The results indicate that alpha-band activity is present in both conditions but does not differ significantly between vocalization and playback. These findings suggest that alpha-band activity during sustained speech may not reflect a process uniquely associated with self-generated vocal monitoring. Instead, the previously observed rhythmic patterns may arise from more general neural dynamics or from the combined contribution of oscillatory and aperiodic components. Future work using approaches that separate these components may help clarify the functional role of frequency-specific activity in speech monitoring.