透過您的圖書館登入
IP:34.201.173.244
  • 期刊

母音共振峰頻率與舌構音超音波影像之相關性分析

Correlation between Formant Frequencies of Vowel and Ultrasonographic Imaging of Tongue during Articulation

摘要


背景:舌是人類說話,尤其是構音(articulation)過程中一個很重要的器官。由於普遍認為F1與F2分別由聲道(vocal tract)後側共振管及前側共振腔所產生,因此傳統上認為母音的第一共振峰(first formant, F1)與舌位的高低(height)相關,而舌位的前後移動則影響第二共振峰(second formant, F2)的共振腔大小。關於母音共振的產生,文獻上有以封閉管(closed-closed tube),盲管(open-closed tube),窄口管(closed narrow opening tube)等模型來解釋。然目前仍少有直接測量的數據以證明這些模型與聲學特徵上的直接相關性。本研究提出對軟組織顯像能力佳的舌部超音波影像(Iingual ultrasonogray)來實際測量母音構音時舌的實際位置,並輔以共振峰分析,採討兩者間的相關性,並與前人的測量結果做比較。方法:健康受試者,男性7名,女性7名,年齡22歲到32歲(平均27歲,中位數25歲)。首先請受試者持續穩定發聲母者/a/、/i/、/u/等最少10秒。發音同時以3.8百萬赫茲(MHz)的超音波在下頦中線截取舌之縱切(mid-saggital)影像,並且同步錄音。影像分析各母者之舌最高點延著舌面到舌骨之表面長度(L1)以及舌最高點到下頦的門牙端之直線距離(L2)。此外以聲學訊號分析,取得各母奇之F1及F2的頻率。統計L1、L2以及F1、F間的差異,以及L1與F1以及L2與F2間的相關性。結果:不同母者之F1、F2間有統計差異(p < 0.05, one-way ANOVA, post-hoc Tukey procedure)L1、L2間也有統計差異(p < 0.05)F1與L1間有統計上的顯著相關(r=-0.86, p <0.001, Pearson's correlation analysis),L2與F2問亦有統計上之顯著相關。(r=-0.80, p < 0.001, Pearson's correlation analysis)。結論:超音波舌影像可以用來即時顯示舌構音時的位置,進而觀察不同共振腔的改變對母音共振峰的影響。本研究結果發現舌頂點到舌骨上緣間的曲線長度和F1頻率有良好相關性,代表藉由超音波測量舌根長度可作為推估F1頻率的客觀方式。舌頂點到下門牙頂端的直線距離與F2頻率有良好相關性,代表測量口腔前端長度可作為推估F2頻率的有效方法。

關鍵字

母音 共振峰 超音波 構音

並列摘要


Background: The tongue plays an important role in human speech articulation. Current understanding of the articulatory-acoustic relationships in vowels are generally based on the simulation of the vocal tract as a uniform straight tube, and, conventionally, the tongue height and tongue retraction are considered to be closely related with the first and the second vowel formants, especially in the corner vowels which form oral constriction small enough to result in acoustic decoupling. However, these principles are not easy to be clinically applied. Therefore, this study aims to develop an objective method to measure the articulation movements of the tongue as well as providing a more specific explanations about vowel articulatory-acoustic relationships by ultrasound imaging and processing. Methods: Fourteen healthy participants (7 M and 7 F) were enrolled. The age was ranged from 22 to 32 years with the mean age of 27 years. The sustained vowels /a/, / i/, and /u/ for at least 10 seconds were obtained and digitally recorded. Meanwhile, a 3.8 MHz probe of ultrasound was adopted to acquire the ultrosonography of the tongue at the mid-saggital plane submentally. These images were then analyzed using self-designed software to extract for presenting the L1 the contour distance from the peak of the tongue to the hyoid bone. The L2 was also marked as the linear distance between the tongue peak and the top of the lower incisor. The first formant (F1) and the second formant (F2) were acquired by acoustic analysis using the shareware PRAATR version 5.3.71. Results: There were significant differences of mean F1 frequency and mean F2 frequency among the three vowels (p < 0.05, one-way ANOVA, post-hoc Tukey procedure). The L1 and L2 of the three vowels were also significantly different among the vowels (p < 0.05, one-way ANOVA, post-hoc Tukey procedure). Strong and significant correlations were revealed between the F1 and the L1 (r=-0.86, p < 0.001, Pearson's correlation analysis) and between the F2 and the L2 (r=-0.80, p < 0.001, Pearson's correlation analysis). Conclusions: The lingual ultrasonography provides a new objective means to explore the resonance mechanism of vowel articulation. The better correlation of the F1 with the length of posterior tongue surface than with tongue height may indicate that the F1 resonance corresponds more to the length of posterior tongue surface compared with the height of tongue. In addition, the slightly better correlation of F2 frequency with the length of anterior oral cavity than with the degree of tongue retraction may suggest that the F2 frequency depends on the size of the anterior oral cavity, which can be modified by the changes in anterior-posterior tongue movement.

並列關鍵字

vowel formant tongue ultrasonography articulation

延伸閱讀