SPECTRAL CONTRASTS PRODUCED BY CHILDREN WITH COCHLEAR IMPLANTS: INVESTIGATING THE IMPACT OF SIGNAL DEGRADATION ON SPEECH ACQUISITION

Abstract

The primary objective of this dissertation was to assess four consonants, /t/, /k/, /s/, and /ʃ/, produced by young children with cochlear implants (CIs). These consonants were chosen because they comprise two place-of-articulation contrasts, which are cued auditorily by spectral information in English, and they cover both early-acquired (/t/, /k/) and late-acquired (/s/, /ʃ/) manners of articulation. Thus, the auditory-perceptual limitations imposed by CIs is likely to impact acquisition of these sounds: because spectral information is particularly distorted, children have limited access to the cues that differentiate these sounds.Twenty-eight children with CIs and a group of peers with normal hearing (NH) who were matched in terms of age, sex, and maternal education levels participated in this project. The experiment required children to repeat familiar words with initial /t/, /k/, /s/, or /ʃ/ following an auditory model and picture prompt. To create in-depth speech profiles and examine variability both within and across children, target consonants were elicited many times in front-vowel and back-vowel contexts. Patterns of accuracy and errors were analyzed based on transcriptions. Acoustic robustness of contrast was analyzed based on correct productions. Centroid frequencies were calculated from the release-burst spectra for /t/ and /k/ and the fricative noise spectra for /s/ and /ʃ/. Results showed that children with CIs demonstrated patterns not observed in children with NH. Findings provide evidence that for children with CIs, speech acquisition is not simply delayed due to a period of auditory deprivation prior to implantation. Idiosyncratic patterns in speech production are explained in-part by the limitations of CI’s speech-processing algorithms. The first chapter of this dissertation provides a general introduction. The second chapter includes a validation study for a measure to differentiate /t/ and /k/ in adults’ productions. The third chapter analyzes accuracy, errors, and spectral features of /t/ and /k/ across groups of children with and without CIs. The fourth chapter analyzes /s/ and /ʃ/ across groups of children, as well as the spectral robustness of both the /t/-/k/ and the /s/-/ʃ/ contrasts across adults and children. The final chapter discusses future directions for research and clinical applications for speech-language pathologists.