Files in this item



application/pdfLU-DISSERTATION-2018.pdf (10MB)
(no description provided)PDF


Title:Speech perception in children with reading disabilities
Author(s):Lu, Jie
Director of Research:Allen, Jont
Doctoral Committee Chair(s):Allen, Jont
Doctoral Committee Member(s):Uddin, Rizwan; Johnson, Cynthia; Wickesberg, Robert; Stine-Morrow, Elizabeth; Zhang, Yang
Department / Program:Nuclear, Plasma, & Rad Engr
Discipline:Nuclear, Plasma, Radiolgc Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Reading disabilities in children
Phoneme perception
Phonemic awareness
Source of reading disabilities
Abstract:Reading disability (RD) is typically viewed as a major obstacle in the development of literacy. This thesis proposes that the source of RDs is related to inadequate phonetic non-categorical processing skills rooted in early development. This view is supported by two experiments on children with reading disabilities in two tasks: a Syllable Confusion Oddball task (SCO) and a Nonsense Syllable Confusion Matrix task (NSCM). The SCO task tested children’s ability to select a different syllable (either a consonant vowel–CV or vowel consonant–VC syllable) from a string of three such syllables spoken by three different talkers. The NSCM task tested children’s ability to reproduce the exact syllable after hearing it. All of the natural CV and VC speech syllables used in both tasks were taken from a commercial database of 18 talkers. Experimental results were: First, regardless of pure tone hearing ability and higher language processing ability (at the level of words and sentences), children with RD (RDs) encountered significant difficulties in phonetic perception compared to a normal reading control group (RCs). Second, RDs had a speech perception problem with nonsense syllable identification, despite normal hearing for pure tones. The stark contrast in performance of the two tasks showed that the SCO task was more difficult than the NSCM task for the RDs. Based on this contrast, the hypothesis is that RDs are not able to retain the three syllables in their phonetic short-term memory, as required to produce an accurate outcome. Third, the clustering analysis for probability of error and entropy showed that the RDs had much greater diversity in responses than the RCs. Some RDs gave all possible responses of phones (maximum entropy), while a few RDs gave only one alternative guess for certain phones. It is reasonable to conclude that, while weakness in phonetic perception could separate the majority of the RDs from the RCs, there exist other factors which contribute to the difficulties in reading for some RDs. Finally, major similarities in confusion patterns for the two groups were plotted using directed graphs and stacked bar plots. For consonants, centers of confusion were mainly affricates and fricatives. The RDs also had additional confusion patterns related to stops and liquids. For vowels, perception of diphthongs and tense vowels was mostly intact for both groups. Confusions mostly occurred among certain “front” and “back” lax vowels. With the purpose of developing an automatic diagnostic tool with confusion pattern, methods were explored to rank phones by automatically block-diagonalizing confusion matrices.
Issue Date:2018-03-13
Rights Information:Copyright Jie Lu, 2018
Date Available in IDEALS:2018-09-04
Date Deposited:2018-05

This item appears in the following Collection(s)

Item Statistics