Electronic Theses and Dissertations



Document Type


Degree Name

Doctor of Philosophy


Communication Sciences & Disorders

Committee Chair

Gavin Bidelman

Committee Member

Eugene H Buder

Committee Member

Deborah Moncrieff

Committee Member

Stephanie Huette


Speech communication requires grouping continuous acoustic cues into meaningful phonetic units through categorical perception (CP). CP can be measured both behaviorally and neurally through electroencephalography (EEG). The perception of speech sounds is not static but can be modulated by stimulus context, such as by presenting token steps of a categorical continuum in sequential acoustical order (e.g., moving from low first formant (F1) frequency to high F1 frequency). Serial presentation order causes perceptual shifts, particularly around the categorical boundary, so an identical stimulus can be perceived differently by the same listener. Such perceptual warping of the acoustic-phonetic space can be described as a perceptual nonlinearity. This dissertation aimed to explore the neural mechanisms of these perceptual phenomena, especially when and where in the brain they occur, through the lens of speech categorization. In complementary studies using EEG and behavioral measures, this dissertation evaluated top-down influences on cortical (Study 1) and subcortical (Study 2) speech processing, manipulating the sequencing of speech stimuli to bias listeners’ categorical percepts. In both studies, vowel stimuli were presented in random, forward (low to high F1), and backward (high to low F1) orderings as listeners categorized tokens as /u/ or /ɑ/. Behaviorally, in both studies, listeners’ categorical boundaries shifted for individual listeners, indicating the effectiveness of the tasks in warping perception. Analysis of cortical event-related potentials (ERPs) showed that late (~300 ms) brain activity localized to superior temporal gyrus and middle/inferior frontal gyri predicted the magnitude of listeners’ perceptual warping effect (i.e., hysteresis/enhanced contrast). Analysis of subcortical frequency-following responses (FFRs) revealed changes in the fundamental frequency (F0) amplitude indicative of category-level coding, as well as a biasing of responses to ambiguous tokens towards the listeners’ phonetic label. Both studies indicate top-down processing shapes the neural encoding and categorization of speech throughout the auditory system, including in the brainstem. Taken together, the results of this dissertation reveal the context surrounding the stimulus continuously shapes neural responses to speech in cortex and brainstem in a predictive, behaviorally-relevant manner. Such top-down modulation at multiple levels of speech representation might aid the perceptual system by reducing ambiguity in speech signals.


Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to ProQuest.


Embargoed until 6/22/2023

SuppPubS3.tif (1375 kB)
Formats of the Stimuli

SuppPubS2.tif (38318 kB)