語境線索對言語焦點表達與焦點感知的影響

The Influence of Contextual Cues on the Production and Perception of Speech Focus

韓采妮(Tsai-Ni Han)

指導教授 ： 黃國祐

中山醫學大學/醫學科技學院/語言治療與聽力學系/碩士(2020年)

https://doi.org/10.6834/csmu202000029

摘要

研究目的：在溝通障礙領域中探討副語言的研究較少，然而臨床上有許多溝通障礙的個案在副語言的表達和感知上都有缺陷。副語言中與語言最相關的面向便是言語韻律，其中又以言語焦點或言語重音最為重要。過去的研究中用以誘發言語重音的方式不盡相同，且有關台灣華語的研究十分稀少，因此本研究將進行不同誘發方式的比較，探討台灣華語焦點的表達與感知。
研究方法及資料：本研究透過操弄誘發情境與焦點位置偵測兩者對言語重音的影響。參與者為12位說話者與40位聽辨者，男女各半。實驗目標句是「佳欣今天開車」，句中有四種不同焦點位置，主詞、時間副詞、動詞與不特別指定位置的廣泛焦點。實驗分成表達作業與知覺作業兩部分。表達作業中包含三種不同情境，由疑問句誘發焦點的情境甲；由糾正錯誤誘發焦點的情境乙；以及直接唸讀語句並強調指定詞彙的情境丙。十二位說話者會在三個情境中說出包含不同焦點位置的語料，之後從十二位說話者中選出最具代表性的男性與女性語料進行知覺實驗。在知覺實驗中每個題目有一句疑問句與一句答句，疑問句包含誘發四種焦點位置的問句「誰今天開車」（主詞焦點）、「佳欣什麼時候開車」（時間副詞焦點）、「佳欣今天做什麼」（動詞焦點）、「發生了什麼事」（廣泛焦點）；而答句則是取自表達作業中的語料。聽辨者必須判斷答句中的言語重音位置與問句中的語意導引是否合適。知覺實驗使用雙因子變異數分析，自變項為誘發情境及焦點位置，依變項為反應正確率。聲學分析使用Praat分析語詞的平均基頻、時長與振幅。
研究結果：知覺實驗結果顯示聽辨者判斷焦點位置的平均正確率是65.4%，在誘發情境效應上情境乙與丙的正確率顯著高於情境甲，在焦點位置效應上句首與句尾焦點的正確率顯著高於句中與廣泛焦點。研究結果亦發現誘發情境與焦點位置具有交互作用：當焦點在句首及句中位置時情境丙的正確率高於甲、乙；而當焦點位於句尾時情境乙的正確率高於甲、丙。聲學分析結果發現語句重音有提高焦點詞之基頻、振幅與時長的效應。焦點位置對語句重音聲學特徵亦有影響，基頻與振幅在句首、句中焦點位置上的增加較為明顯，而時長在句尾焦點位置有較明顯的增加。此外，誘發情境對基頻的影響並無一致趨勢，而時長與振幅在情境甲、乙中獲得的增益多於情境丙。
結論與建議：誘發情境與焦點位置對言語焦點的辨識與聲學特徵都有影響，但是知覺表現與聲學分析結果並不全然一致。基頻、時長與振幅都是語者可以表達焦點的聲學特徵，但是聽者在辨識焦點時可能更為依賴基頻與時長線索。本研究是國內首次針對台灣華語的語句焦點議題進行知覺和聲學的研究，期待未來能有更多研究討論這個議題。

關鍵字

聲學分析 ； 言語知覺 ； 言語焦點 ； 言語情境

並列摘要

Objective: In the discipline of communication disorder, there are fewer studies investigating paralanguage abilities, comparing to language abilities; however, many patients with communication disorders have deficits in the production and perception of paralinguistic messages. Among all aspects of paralanguage, speech prosody is the one with the strongest relationship with verbal language, and the most important function of speech prosody is speech focus, also called accent. In previous studies, various types of contexts were used to elicit focus, and there were only a few studies investigating speech focus in Mandarin. Hence, this study aimed to carry out a comparison of different focus elicitation contexts and explore speech focus in Taiwan Mandarin.
Methods and Materials: The current study manipulated speech contexts and focus position to explore their impact on speech focus. There were 12 speakers and 40 listeners. The target sentence was “佳欣(name. Jia-shing) 今天(today) 開車(drives car),” and the sentence showed 3 narrow focus position: subject, time, action; and a broad focus with unspecified focus position. In the expression task, speakers have to produce the target sentence with different focus positions in three contexts. In Context A, the focus was elicited by a question; and in Context B, it elicited a corrective focus; as for Context C, the speaker was asked to read aloud sentences and emphasize the words underlined. Among the 12 speakers, the speech samples of the most representative female speaker and male speaker would be used as test stimuli in the perception task. In the perception task, every trial included a question and an answer. The questions would elicit four different focus positions, such as “who drives the car today (subject focus),” “When does Jia-shing drive the car,” “what does Jia-shing do today” and “what happened.” The answers to these questions were the speech samples produced by the 2 speakers. The listeners have to decide whether the focus accent position in the answer and the meaning of the question are matched or not. After the perception task, we performed a two-way ANOVA, with focus elicitation contexts and focus positions being independent variables and accuracy as a dependent variable. As for the production task, Praat was used to analyze the average fundamental frequency (F0), duration, and amplitude of words in the speech samples produced by the two most representative speakers.
Results: The average accuracy rate in the perception task was 65.4%. The context effect showed a higher accuracy rate in Context B and Context C than in Context A; and the focus position effect displayed a higher accuracy rate on sentence-initial focus and sentence-final focus, comparing to sentence-middle focus and broad focus. The result also revealed an interaction between context and focus position, the sentence-initial focus and sentence-middle focus were more easily identified in Context C; and the sentence-final focus was more easily identified in Context B. The results of the acoustic analysis showed that speech focus has an effect on increasing F0, duration, and amplitude of the target words. F0 and amplitude showed a greater increase in sentence-initial focus and sentence-middle focus, and duration displayed a greater increase in sentence-final focus. On the other hand, the contexts did not show a consistent effect on F0, but the duration and amplitude of target words both displayed more increase in Context A and Context B.
Conclusion and Suggestion: Both elicitation contexts and focus position had an impact on speech focus identification and production, but the results in the perception task were not completely consistent with the results of acoustic analysis. Speakers could manipulate F0, duration, and amplitudes to express focus, but listeners may rely more on F0 and duration to identify speech focus. The present study is the first in Taiwan to investigate speech focus perception and acoustic characteristics. We hope that there would be more research on this topic in the future.

並列關鍵字

acoustic analysis ； speech perception ； speech focus ； speech context

參考文獻

鄭靜宜（2013）。話在心·口難言：運動性言語障礙的理論與實務。台北：心理。

Boersma, P. (2001). Praat, a system for doing phonetics by computer. Glot International 5:9/10, 341-345.

Branco, A. U., Fogel, A. (2014). Metacommunication as a source of indeterminism in relationship development Dynamics and Indeterminism in Developmental and Social Processes (pp. 73-100): Psychology Press.

Braun, B. (2015). What causes the activation of contrastive alternatives, the size of focus domain or pitch accent type? Paper presented at the ICPhS 2015: 18th International Congress of Phonetic Sciences, Glasgow.

Braun, B., Tagliapietra, L. (2010). The role of contrastive intonation contours in the retrieval of contextual alternatives. Language and Cognitive Processes, 25(7-9), 1024-1043.