- 學位論文
閩南語聲調感知的辨識能力和區辨能力
Identifiability and Discriminability of Tone Perception in Taiwanese Southern Min
摘要
閩南語聲調感知和類別覺知、聲調中和性、語言經驗、聲調任務有關係。過去文獻指出平音是無範疇性(non-categorical)，因為此在語音的連續體中呈現線性變化。此外，閩南語的平音和促音被認為是不完全中和(incomplete neutralization)，其中變調位置比起非變調位置有較高的辨識值和區辨值。再者，在聲調感知中，母語人士表現比非母語人士好。最後，辨識任務屬於高層次的語言處理，而區辨任務是低層次思考過程，所以前者比後者還要難。我們想探討使用閩南語雙字詞當刺激材料時，在年輕族群中，語言經驗、聲調位置、和聲調值是否影響聲調感知。 由邏輯式迴歸(the logistic regression)和四因子變異數(the four-way ANOVA)的分析發現，平音的辨識能力和區辨能力比促音好，此兩個音在變調和非變調位置也有顯著差異，再者，只有平音受到語言經驗影響，促音則無，最後區辨任務比起辨識任務有較高的正確率和較長的反應時間。 從實驗結果顯示：語言經驗、聲調位置和聲調值顯著地影響閩南語聲調感知。第一：平音的辨識能力和區辨能力都顯著高於促音，推測年輕族群可以歸類平音，而促音則無法。第二：音調值的辨識能力和區辨能力在變調位置和非變調位置有顯著差異，證明平音和促音是不完全中和。第三：語言經驗非助於母語人士在判斷促音時表現得最好。最後：區辨任務比起辨識任務有較高的正確率和較長的反應時間，所以區辨任務比辨識任務還要容易。
並列摘要
Regarding TSM tone perception, issues such as categorical perception, tone neutralization, language experiences, and different experimental tasks have been widely examined. Previous studies have shown that a linear function on the continuum of level tones is perceived non-categorically in the identification and discrimination tasks. Besides, both level tones and checked tones have been claimed to be incompletely neutralized with their sandhi tones, as higher perception scores were discovered in the sandhi position than the citation position (Kuo, 2013; Chen, 2016). Typically, native speakers (NT) performed better than non-native speakers (NNT) in tone perception (Lee, Vakoch, & Wurm, 1996). Moreover, the identification task involving higher-level linguistic processing by focusing on pitch value is more challenging than the discrimination task, including lower-level sensory processing by focusing on phonemic differences (Liberman et al., 1957). This thesis utilized TSM disyllabic words with complete endpoints to examine the effects of language experience, target (sandhi-citation) position, level-checked distinction on the tone perception among young listeners. Our logistic regression analyses and the four-way ANOVA revealed that higher identifiability and discriminability occurred in the level tones rather than the checked tones. Besides, significant differences in tone syllable position were found in both level tones and checked tones among the two tasks. Then, only level tones were significantly affected by language experience, while checked tones were not. Finally, higher accuracy and longer RT appeared in the discrimination task than in the identification task. Based on the results of these two experiments, language experience, target position, and target tone (pair) value significantly affected TSM tone perception. From the identifiability and discriminability, we speculate that level tones could be categorized while checked tones could not. Second, both level tones and checked tones are incompletely neutralized because of significant differences between the sandhi position and the citation position. Then, language experiences did not facilitate native groups’ performance in checked tones or identical checked-tone pairs. Finally, the discrimination task with longer RT and higher accuracy is less challenging than the identification task.