聽覺障礙兒童國語塞擦音聲學特質分析研究

A Study on Acoustic Characteristics of Mandarin Affricates of Hearing-Impaired Children

本研究主要在於探討聽力正常與回歸在普通班級中聽覺障礙兒童國語塞擦音的聲學特質，及聽覺障礙兒童塞擦音構音錯誤情形。茲以台南市八十六度學年回歸在普通班級之聽障兒童十六名，及聽常兒童三十五名為研究對象，以Dr.Speech＆RealSpeech語音分析軟體為研究工具，探討聽力正常與聽覺障礙兒童國語塞擦音聲學特質，及聽覺障礙兒童塞擦音構音錯誤情形。期以研究結果提供啟聰教育教師教學上，及聽障兒童語音研究上的參考。本研究主要發現如下：一、聽力正常兒童塞擦音聲學特質：（一）沖直條出現百分比：67.5％。以捲舌音出現沖直條百分比最高。（二）嗓音起始時間：20～40ms，平均115ms。可區分送氣音與不送氣音的特質（t=11.03，p<.001）。（三）擦噪音上升時間：0～180ms，平均55ms。舌尖音64ms，捲舌音50ms，舌面音52ms。可區分送氣音與不送氣的特質（ｔ＝6.59，P<.001）。（四）擦噪音段能量最大頻率：2381～4681Ｈｚ，平均3967Hz。擦噪音段能量最小頻率：1601～2762Hz，平均2042Hz。捲舌音擦噪音段能量最大與最小頻率分佈最為狹窄。（五）音調：277～292Hz。顯示音調並不因不同構音方式而有明顯的變化。二、聽覺障礙兒童塞擦音發音不清晰的聲學特質與構音錯誤情形：（一）塞擦音沖直條出現率較低。顯示此發聲腔道的動作控制精密度較差。（二）塞擦音嗓音起始時間較長，分佈範圍較為分散，可見在發塞擦音時，噪音時間拉長，而且有的聽覺障礙學生嗓音拉得較長，有的則太短；在時間控制上較差。（三）擦噪音段上升時間較長，變異性較大。顯示在發塞擦音時，擦噪音段時間拉長，而且有的聽覺障礙學生嗓音拉的較長，有的較短。（四）塞擦音擦噪音段能量最大與最小頻率差距較小，則做為不同構音部位參照指標降低。

The study aims to explore the acoustic characteristics of Mandarin affricates pronounced by normal children and mainstreamed hearing-impaired children. The subjects of the study included 35 normal children and 16 hearing-impaired children who returned to the general classed in Taiwan City in the academic year of 1997-1998. The speech analysis software by Dr. Speech & Real Speech was used as the tool to investigate the acoustic characteristics of Mandarin affricates pronounced by normal children and hearing-impaired children as well as speech errors of affricates made by hearing-impaired children. And it is our hope that the result of this study will be used for reference in the education of hearing-impaired children and in the research on the speech of hearing-impaired children. The mail findings of this study are shown as follows: 1.Acoustics characteristics of affricates pronounced by normal children: (1) Occurrence of release burst (%): 67.5%. The occurrence of release burst in retroflex is the highest in percentage. (2) Voice onset time: 20-400ms, 115ms on average. Divided into aspirated sounds and unaspirated sounds (t=11.03, p< .001. ). (3 ) Rise time of fricative noise: 0-180ms. 55ms on average. 64ms for apical-dental, 50ms for retroflex, and 52ms for alveolo-palatal. Divided into aspirated sounds and unaspirated sounds (t=6.59, p<.001). (4) Maximum frequency of energy in the fricative noise:2381-4681 Hz, 3967 Hz on average. Minimum frequency of energy in the fricative noise. 1601-2762 Hz, 2040 Hz on average. Retroflex have the narrowest distribution range in the maximum and minimum frequency of energy in the fricative noise. (5) Pitch: 277-292 Hz. It indicates that a significant change in the pitch will not occur due to different speech methods. 2.Acoustic characteristics of unclear speech and speech errors of affricates pronounced by haring-impaired children(1) The occurrence of release burst in affricates is lower in percentage, suggesting a worse accuracy in controlling the movement of the vocal tract. (2) There are a longer voice onset time and a wider distribution range of affricates. It is obvious that the voice onset time in pronouncing affricates is lengthened. Besides, some hearing-impaired students have longer voice, whereas some have too short voice, indicating a poor control of time. (3) The rise time for the fricative noise is longer, and the variation is bigger. This shows that the time for the fricative noise is lengthened while pronouncing affricated. And some hearing-impaired students have longer voice, whereas some have shorter voice. (4) Affricates have a smaller difference between the maximum and munimum frequency of energy in the fricative noise, and the reference index for different speech parts is lowered.