語音異常兒童的語音區辨及聲學調整對其聽知覺的影響

Speech Discrimination of Mandarin- Speaking Children with Speech Sound Disorders

語音異常（speech sound disorders， SSD）兒童常有語音區辨困難，本研究 之目的在探討SSD 兒童的語音對比區辨及調整聲學線索對語音區辨的影響。本 研究有102 位參與者，包括31 位六足歲的SSD 兒童和31 位年齡、性別配對無 SSD 的兒童為控制組；另有40 位正常成人為控制組，作為和兒童聽辨表現比較 的基礎。實驗一的刺激材料是52 對具最小音素對比的華語單音節音，含13 類語 音對比。實驗二的材料是將實驗一的刺激用聲學再合成調整聲學特徵性，比較聲 學調整對三組聽者區辨的影響。聲學調整方式是在摩擦噪音強度、共振峰轉折 帶、VOT(voice onset time)、噪音遮蔽和梅爾倒頻（MFCC）濾波等方面做強化或 減弱對比的調整。結果發現SSD 兒童的語音區辨正確率顯著低於兒童控制組， 兩個兒童組的區辨正確率皆顯著低於成人組；三組反應時間(RT) 亦達顯著差異， SSD 兒童組RT 顯著慢於兒童控制組和成人組，成人組RT 最短。在實驗二無論 是正確率或RT，三組聽者對於強化類刺激的反應皆顯著優於減弱類刺激。SSD 組對於強化類刺激，RT 顯著較快於實驗一的原始刺激，顯現增益效果，但在正 確率未見提升效果；對於減弱類刺激和實驗一原始刺激相較則有顯著降低效果。 語音區辨的正確率和RT 各與兒童構音測驗分數呈中度相關（r = .50，r = -.64， p < .001）。可知六歲兒童的語音區辨，無論在正確率和RT 皆不及成人，對於一 些較難的語音對比區辨習得應是在六歲之後；SSD 兒童語音區辨能力較一般兒童 為弱。由於聽知覺發展在學習「說」之前，推論語音區辨的困難可能是造成兒童 SSD 的原因之一。

Purpose: Preschool children with speech sound disorders (SSD)， previously called “articulatory/phonological disorders，” frequently show deficiency in speech perception， specifically in speech–sound discrimination. The purpose of this study was to investigate the speech discrimination ability of Mandarin-speaking children with SSD and to explore the effects of acoustically modified stimuli on speech discrimination. Methods: The participants consisted of 31 preschool children with SSD and two control groups. One of the control groups consisted of 31 normal preschool age-matched non-SSD children and the other control group consisted of 40 normal young adults. In Experiment I， the stimuli comprised 52 pairs of Mandarin monosyllables consisting of 13 phonetic minimal pair contrasts for speech discrimination testing. The discrimination task involved judging whether the two presented monosyllables were perceived as the same syllable sound. The children were tested individually. The speech stimuli used in Experiment II were acoustically modified from the stimuli in Experiment I. The acoustical modification involved altering the time or frequency domains of the target speech segments through resynthesis; specifically， the noise duration， VOT， and vowel transition duration were modified， and noise masking and MFCC filtering were applied. Findings: The results of Experiments I and II indicated that the SSD group exhibited significantly lower correct rates and longer reaction times (RTs) than did the child control group. Additionally， the adult group exhibited the highest correct rates and the shortest RTs. The articulation scores of the children were moderately correlated with the correct rates and RTs of the speech discrimination task (r = .50， p < .001; r = −.64， p < .001， respectively). Retroflex contrast， manner contrast at the alveolar place of articulation， and tone contrast were the three most difficult contrasts for the children with SSD. Aspiration contrast and fricative–affricate contrast were the two contrasts with the greatest differences between the two groups of children (SSD vs. normal). In Experiment II， the stimuli with acoustical enhancement significantly reduced the RTs in the SSD group， but exhibited no facilitating effect on the correct rates. For acoustically reduced stimuli， the two groups of children had lower correct rates than they did for the enhanced stimuli and original stimuli. The noise masking had a greater effect on the children with SSD. Conclusions/Implications: The results suggest that the ability of speech discrimination of the preschool children remains lower compared with that of adults， and that the development of speech perception is ongoing in preschool children. Difficulty in speech discrimination may be a crucial factor for SSD. Some acoustical modification can influence the discrimination of speech in children.