從產生和感知認識聲調

Understanding Tone from the Perspective of Production and Perception

認真考量語音產生和感知上的種種限制，將有助於我們對聲調的認識。具體的說，音高變化的最大速率和咽喉動作之間的協調，對詞彙聲調的產生加上了一些無法超越的限制；而儘管人類的感知系統可以很有效率的處理高速變化的聲學事項並解決因發音限制所導致的扭曲訊息，但在感知上能復原多少也仍有諸多限制。有鑑於這些限制，聲調產生的「目標近似模型」乃應運而生。這個模型將基頻的產生視為對目標音高的漸近過程，而這些目標音高就是由語言規則所推導的個別聲調。由於能對發音時的聲調變體與目標音高的聲調交替作出明確的區別，「目標近似模型」得以幫助我們認識許多與聲調有關的問題。關於聲調及聲調起伏分布的本質、不同連調變化現象的辨別、輕聲的目標音高和發音方式等等的問題，也都可以從「目標近似模型」中獲得啟發。

Our understanding of tone can be significantly improved if we take the constraints of speech production and perception seriously. In particular, the maximum speed of pitch change and the coordination of laryngeal and supralaryngeal movements impose certain impassable limits on the way lexical tones are produced. At the same time, although the human perceptual system is highly proficient in processing fast-changing acoustic events as well as resolving distortions due to articulatory constraints, there are limits as to how much undershoot can be perceptually reversed. The understanding of these constraints has led to the Target Approximation model of tone production. The model simulates the generation of F0 contours as a process of asymptotically approximating underlying pitch targets that are associated with individual tones via language-specific rules. The application of the Target Approximation model helps the understanding of various tonerelated issues by providing explicit criteria for distinguishing between tonal variations due to articulatory implementation and those due to alternation of the tonal targets. In light of the Target Approximation model, new insights are offered regarding the nature and distribution of contour tones, distinction between different kinds of sandhi-like tone phenomena, and the target and manner of implementation of the neutral tone.