Ph.D Thesis

Ph.D StudentHorev Nitza
SubjectProcessing of Temporal and Spectral Cues in the Human Brain
during Perception of Speech and Corresponding
Non-Speech Signals
DepartmentDepartment of Medicine
Supervisor PROFESSOR EMERITUS Hillel Pratt
Full Thesis text - in Hebrew Full thesis text - Hebrew Version


The purpose of the present study was to define general and speech-specific brain mechanisms involved in stop consonants perception, by examining how the brain processes linguistically relevant spectral and temporal information compared with the processing of corresponding non-speech signals. This was done by recording Auditory Evoked Potentials (AEPs) during active discrimination of spectral (/ubu/ - /udu/) and temporal (/ubu/ - /upu/) stimulus pairs and their analogous non-speech stimuli. In order to differentiate between acoustic effects and speech-specific effects, AEPs were also recorded during active discrimination of Sine-Wave Speech (SWS) stimulus pairs. SWS stimuli are auditory signals that can be perceived either as speech or non-speech, depending on listeners’ expectations about the nature of the stimuli, hence providing a tool to study neural speech specificity using identical acoustic stimuli.

Nineteen, healthy, right-handed, young native Hebrew speakers participated in this study. AEPs were recorded from 61 scalp electrodes in two sessions. In the first, “non-speech”, session, subjects discriminated spectral and temporal differences in non-speech stimuli and in SWS stimuli, unaware of their speech-like nature. In the second, “speech”, session, the same subjects discriminated between spectral and temporal speech pairs and between their SWS counterparts after learning to perceive the SWS as speech. The non-speech session was always the first, since after learning to perceive SWS as speech it cannot be perceived as non-speech again.

Perceiving the stimuli as speech, resulted in higher percentage of correct responses and in shorter reaction times. Electrophysiological responses indicated that speech and non-speech perception evoked significantly different voltage differences around the main scalp recorded AEPs peaks. Acoustic differences between speech and analogous non-speech stimuli were reflected in differences in N1 amplitude, recorded 100 ms after stimulus onset, whereas later processing stages were ascribed to the transition to “speech-mode” of processing.  sLORETA source estimation revealed stronger activation of auditory cortices to “speech” stimuli at early (around P2b) stages of processing, and at the time of a sustained negativity, stronger activation to “non-speech” stimuli, reflecting later processing stages. Differences were prominent mainly within the right superior temporal cortex and the right prieto-temporal junction.  

Our data provide additional evidence for a “speech mode” of processing, which engages different brain mechanisms, distinct from those involved in general auditory (non-speech) processing mode. We suggest that brain specialization for speech processing is not hemisphere- or region-specific, but results from specific and unique activation patterns of distributed neural networks in both hemispheres.