Analysis for Error Patterns of Consonant-Vowel Combinations in Hearing-Impaired Elderly

Abstract

Purpose: Many previous studies related to speech recognition of hearing-impaired elderly have found that the elderly have difficulty in monosyllable recognition under noisy situations, but there is still lack of their specific information to perceive the monosyllables in terms of error pattern. The purpose of the present study was to measure and analyze the prominent errors of consonant (C)- vowel (V) combinations as a function of noise levels in the hearing-impaired elderly. Methods: A total of twenty older adults with sensorineural hearing loss participated. While randomly presenting 122 syllables which consisted of the Korean CV and VC combinations under quiet and two signalto-noise ratios (SNRs, +6 and +3 dB) conditions, the participants were asked to respond the syllables heard in a way of either discrimination or identification. Error rate (or percent incorrect) and error patterns of the syllables were analyzed in each condition. Results: Substitution error of the initial consonant showed the highest rate regardless of the background conditions. As the medial phoneme, the vowel errors were decreased as the level of noise increased. In the final consonant, additional error rate of /k/ was abruptly increased as SNR decreased. Furthermore, it was an interesting pattern that the initial consonants such as /s/, /l/, and /t*/ had a high substitution error rate when combined with /i/, /o/, and /ɛ/ vowels, respectively. Conclusion: Based on the current results, the hearing-impaired older adults seem to have their unique error patterns although there is individual difference. We expect that these results will be fundamental data to understand characteristics of speech perception in the older adults at the monosyllables level.