Speech feature encoding in the human brain

The human superior temporal gyrus (STG) is a high-order auditory cortex area that is critical for speech perception. Over the past decade, cortical recordings directly from the human brain procedures have provided the most detailed information yet on the representations and computations that give rise to speech perception. Inspired by Ken Stevens’ theory of distinctive features, we have discovered several defining aspects of cortical processing of speech: (1) representation of acoustic-phonetic features, (2) encoding of prosodic cues that demarcate the timing and magnitude of syllables (peak derivative of the amplitude envelope) and sentence/phrasal onsets, and (3) speaker normalized representation of vowels in 2-dimensional formant space. Perceptual studies on phoneme restoration and selective attention have demonstrated that cortical processing in the STG is directly related to our conscious perception of speech sounds and not a veridical read-out of the stimulus. Our new work seeks to address phonological specialization in the STG based on language experience, and the representation of words.