Neurocognitive Correlates of Spectrally-Degraded Speech Recognition in Children

Abstract

The research reported in this dissertation was carried out to investigate the contribution of several core neurocognitive factors in speech perception when degraded and underspecified phonological and lexical representations of speech are presented to normal-hearing (NH) listeners. The first study successfully replicated Experiment 2 from Eisenberg, Martinez, Holowecky, and Pogorelsky (2002), which demonstrated the influence of word frequency and neighborhood density on recognition of vocoded speech.

The second study assessed relations between vocoded sentence perception and measures of auditory attention and short-term memory in NH children. Analyses revealed significant relations between performance on both sets of neurocognitive measures and vocoded speech perception tasks. These findings support the hypothesis that vocoded speech perception reflects not only peripheral processes, but cognitive processes as well.

The third study used the California Verbal Learning Test - Children's Edition (CVLT-C) to examine relations between verbal learning in memory and speech perception. In the first part, NH children were assessed on the CVLT-C and vocoded speech. In the second part, we investigated relations between speech perception and the CVLT-C in a group of children with cochlear implants (CI) and NH age-matched controls. Findings from this study revealed that that CI children and NH children processed verbal material in fundamentally different ways in a multi-trial free recall learning task.

Taken together, these studies provided new insights into the foundational underlying neurocognitive processes that support perception of spectrally-degraded speech in children - an area of research previously unexplored--and established the need for extending future research in hearing impaired children with cochlear implants to include new measures of cognition such as attention, verbal learning, and memory.