Developmental dyslexia is a reading disorder often characterized by reduced awareness of speech units. Whether the neural source of this phonological disorder in dyslexic readers results from the malfunctioning of the primary auditory system or damaged feedback communication between higher-order phonological regions (i.e., left inferior frontal regions) and the auditory cortex is still under dispute. Here we recorded magnetoencephalographic (MEG) signals from 20 dyslexic readers and 20 age-matched controls while they were listening to ∼10-s-long spoken sentences. Compared to controls, dyslexic readers had (1) an impaired neural entrainment to speech in the delta band (0.5-1 Hz); (2) a reduced delta synchronization in both the right auditory cortex and the left inferior frontal gyrus; and (3) an impaired feedforward functional coupling between neural oscillations in the right auditory cortex and the left inferior frontal regions. This shows that during speech listening, individuals with developmental dyslexia present reduced neural synchrony to low-frequency speech oscillations in primary auditory regions that hinders higher-order speech processing steps. The present findings, thus, strengthen proposals assuming that improper low-frequency acoustic entrainment affects speech sampling. This low speech-brain synchronization has the strong potential to cause severe consequences for both phonological and reading skills. Interestingly, the reduced speech-brain synchronization in dyslexic readers compared to normal readers (and its higher-order consequences across the speech processing network) appears preserved through the development from childhood to adulthood. Thus, the evaluation of speech-brain synchronization could possibly serve as a diagnostic tool for early detection of children at risk of dyslexia. Hum Brain Mapp 37:2767-2783, 2016. © 2016 Wiley Periodicals, Inc.

Cortical oscillations phase-align to the quasi-rhythmic structure of the speech envelope. This speech-brain entrainment has been reported in two frequency bands, that is both in the theta band (4-8 Hz) and in the delta band (<4 Hz). However, it is not clear if these two phenomena reflect passive synchronization of the auditory cortex to the acoustics of the speech input, or if they reflect higher processes involved in actively parsing speech information. Here, we report two magnetoencephalography experiments in which we contrasted cortical entrainment to natural speech compared to qualitative different control conditions (Experiment 1: amplitude-modulated white-noise; Experiment 2: spectrally rotated speech). We computed the coherence between the oscillatory brain activity and the envelope of the auditory stimuli. At the sensor-level, we observed increased coherence for the delta and the theta band for all conditions in bilateral brain regions. However, only in the delta band (but not theta), speech entrainment was stronger than either of the control auditory inputs. Source reconstruction in the delta band showed that speech, compared to the control conditions, elicited larger coherence in the right superior temporal and left inferior frontal regions. In the theta band, no differential effects were observed for the speech compared to the control conditions. These results suggest that whereas theta entrainment mainly reflects perceptual processing of the auditory signal, delta entrainment involves additional higher-order computations in the service of language processing.