Electroencephalographic evidence of vowels computation and representation in human auditory cortex

By exploiting the N1 component of the auditory event related potentials (AEPs) we measured and localized the processing involving the spectrotemporal and the abstract featural representation of the Salento Italian five vowels system. Findings showed two distinct N1 sub-components: The N1a peaking at 125-135 ms, localized in the primary auditory cortex (BA41) bilaterally, and the N1b peaking at 145-155 ms and localized in the superior temporal gyrus (BA22) with a strong leftward lateralization. Crucially, while high vowels elicited higher amplitudes than non-high vowels both in the N1a and N1b, back vowels generated later responses than non-back vowels in the N1b only. Overall, these findings suggest a hierarchical processing where from the N1a to the N1b the acoustic analysis shift progressively toward the computation and representation of phonological features. Introduction Speech comprehension requires accurate perceptual capacities, which consist in the processing of rapid sequential information embedded in the acoustic signal and in its decoding onto abstract units of representation. It is assumed that the mapping principles exploited by the human brain to construct a sound percept are determined by bottom-up acoustic properties that are affected by top-down features based on abstract featural information relating to articulator positions (Stevens 2002). Such features, called distinctive features, would represent the primitives for phonological computation and representation (Halle 2002). Therefore, one of the central aspects for understanding the speech processing mechanisms is to discover how these phonetic and phonological operations are implemented at a neuronal level to shape the mental representations of the speech sounds.