The frequency dependence of EEG-speech envelope tracking does not reflect the rates of speech units or pitch

During speech listening, neural activity partly synchronises to fluctuations in the acoustic amplitude envelope of the stimulus. This neural tracking of the speech envelope has frequently been linked to the processing of acoustic features such as syllabic rhythmicity and pitch periodicity. However, it has not yet been investigated to which degree the frequency dependence of neural envelope tracking reflects the rate of acoustic and linguistic speech units. Here, using a large dataset of electroencephalographic (EEG) responses from participants who listened to naturalistic speech, we quantified neural tracking of the speech envelope across a wide range of modulation frequencies using coherence analysis. The coherence profile exhibited distinct peaks in the low-delta band (0.2-2 Hz), the theta-alpha band (4-15 Hz), and at higher frequencies near 45, 95, and 175 Hz. We show that this structure is independent of the rates of phonemes, syllabes and words as well as of pitch periodicity. Using temporal response functions (TRFs), we further show that neural envelope tracking in the gamma frequency band (30-250 Hz) is primarily driven by two clusters of neural generators with latencies of approximately 8 ms and 25 ms, likely located in the rostral brainstem and across the thalamocortical pathway, respectively. Together, these results highlight the interplay of different neural mechanisms and sources in shaping the neural envelope tracking, and lead to important methodological and interpretative considerations for studies that assess neural envelope tracking within narrow frequency bands.