語音知覺訓練對發音的影響:以華語聲調為例

Abstract

本文探討辨音能力對語音產製的成效，意即知覺訓練有助於發音學習。本文以華語聲調為材料，為英語母語者設計兩個實驗；主要實驗15天中，實驗組進行知覺訓練，控制組為仿說訓練。知覺訓練過程中，受試者每次聽完四位華語母語者錄製的帶調單音節即辨識調值，系統給予簡單對錯的回應型立即反饋；仿說訓練過程則要求受試者每次聽完相同排序的單音節立即仿說，但不提供反饋。兩組接受看字讀音的發音測驗，分別為訓練前、訓練終日及其兩週後。結果顯示，實驗組在訓練過程中，聲調辨識能力逐漸進步，且發音測驗在整體結果、訓練音（舊音）和非訓練音（新音）都顯著進步，控制組則沒有任何效果。後續實驗仿製主要實驗的仿說訓練，但訓練過程提供回應型立即反饋。結果顯示，發音準確度在訓練和測驗都有顯著進步。兩個實驗結果發現，反饋在語音學習的重要性，提供知覺訓練對於發音動作學習的明確證據，效果也符合理論預期的知覺與發音動作相聯關係。本文對研究發現進行發音學習與教學的討論，簡言之，本文建議知覺訓練可融入課外發音教學，人工智慧技術也許有助於糾正型立即反饋，進而提升發音教學成效。整體而言，本文鼓勵以知覺訓練的教學方式增強學習者的發音及改善二語習得能力。

This study explored how actively listening to speech sounds might enhance subsequent speech production, suggesting that perceptual training could improve pronunciation. Using Mandarin Chinese tones, two experiments were conducted with native English speakers. The main experiment involved one group receiving daily perceptual training for 15 days and another receiving a simple production training to serve as a control group. Perceptual training required the experimental participants to listen to tokens of Mandarin monosyllables, spoken by four native speakers, and identify their tones. Immediate accuracy feedback was given for each trial. Production training required the control participants to listen to the same set of syllables and repeat back each token out loud. No accuracy feedback was provided. Both groups were also given read-and-record tone production tests, once before the training and twice, immediately and two weeks after the training. For the experimental group, results showed gradual improvements in tone identification during training, as well as significant improvements in tone production for both the trained and non-trained syllables. In contrast, the control group showed no improvements in tone production on their tests. A follow-up experiment repeated the control group of the main experiment but provided accuracy feedback during production training. Significant improvements in tone production were observed during training and on the tests. These findings offer unequivocal evidence for the positive effect of perceptual training on speech production, while also highlighting the importance of feedback to learning. The effect is consistent with the prediction of theories that posit a connection between perception and production. Implications of these findings for pronunciation learning and teaching are discussed. In brief, incorporating perceptual training into pronunciation teaching outside the classroom is recommended, and artificial intelligence technology may eventually facilitate immediate corrective feedback, making for more effective pronunciation teaching. Overall, this research encourages a perceptual training approach to enhance the speech production of language learners and improve second language acquisition in general.