Differential effects of overt, covert and replayed speech on vowel-evoked responses of the human auditory cortex

Recent magnetoencephalographic studies showed that speaking dampens voice-evoked activity in the human auditory cortex. To further characterize this audio-vocal interaction, neuromagnetic responses to short tape-recorded probe vowels were measured while subjects were vocalizing long (8 s) 'background' vowels either aloud or silently, or while both probe and background vowels were replayed from tape. Auditory cortex responses peaking at 100 ms (M100) were delayed and dampened bilaterally relative to a background-free control during both overtly spoken and replayed long vowels, identifying auditory interference as the main cause for these modifications. During covert speech M100 peaked later for matching than non-matching probe/background vowels in the speech-dominant left hemisphere. Thus, voiceless 'inner' speaking is sufficient to modify utterance-specific processing in the human auditory cortex.