中文諷刺語氣表達及理解之間之聲音特色與年齡及句型對諷刺語音之影響
No Thumbnail Available
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
由於過去文獻顯示語音訊息為幫助分辨諷刺語氣的重要依據,此研究欲探討臺灣中文母語者其諷刺語氣之語音特色以及聽者從何正確判斷說話者之語氣。除了說話者的態度,句子的型態以及說話者的年齡對諷刺語氣之影響也被納入討論。研究中,首先進行的錄音實驗錄製了中文母語受試者在不同句子型態裡所表達的 三種態度(中性、真誠、諷刺)。接著,另一批受試者對錄下的句子進行語氣的判斷。 實驗結果顯示,相較於中性態度,諷刺語氣呈現較高的音調(mean F0)、較寬的音調全 距(pitch range)、較低的頻率擾動度(jitter)和音量擾動度(shimmer)、以及較慢的語速 (speech rate)。而與真誠態度比較之下,諷刺語氣則呈現較低的音調(mean F0)、較小的 音調全距(pitch range)、較低的頻率擾動度(jitter)和音量擾動度(shimmer)、以及較快的 語速(speech rate)。年齡則對諷刺語音的影響顯示於音調及音調全距,且對於音調的影 響僅出現於短句(keyphrases)。而在諷刺的語氣中,短句與其他句子型態相比,呈現較 慢的語速。另外,本研究也發現頻率擾動度、音量擾動度以及語速容易造成聽者混淆說話者所表達的語氣。說話者的諷刺語氣若含有較高的頻率和音量擾動度以及較快的語速,則容易被誤判為真誠的語氣。而說話者的中性語氣如有較低的音量擾動度也會讓聽者判斷為諷刺語氣。而在不同年齡的句子中,音調及音調全距則影響聽者的判斷。較高的音調及較寬的音調全距容易導致年輕說話的諷刺語氣被誤認為真誠。相反地較低的音調及較小的音調全距則會讓年長者的諷刺語氣被誤判為真誠。
Previous research has acknowledged prosodic information as one major component contributes to sarcasm detection. However, the voice quality of sarcastic speech shows no consistency cross-linguistically. This study focuses on the voice quality of sarcasm in Taiwanese Mandarin. Specifically, we investigate whether phrase types and age differences have effects on Taiwanese Mandarin speakers’ delivery and perception towards sarcastic utterances. Six voice quality parameters are examined, including mean F0, F0 range, jitter, shimmer, H1-H2, and speech rate.A sarcasm elicitation task, which uses a fully crossed 3 (attitudes) x 3 (phrase types) design, was adopted to record participants’ utterances of neutrality, sincerity and sarcasm. Then, a perceptual validation process helped identify the successfully recognized and misinterpreted attitudes produced by the speakers.Our results showed that Taiwanese Mandarin sarcasm featured higher mean F0, wider F0 range, lower jitter, lower shimmer, and slower speech rate compared with neutrality, but lower mean F0, narrower F0 range, lower jitter, lower shimmer, and slower speech rate than sincerity. Age difference can be seen in speakers’ sarcasm production strategies regarding F0 range and mean F0, while the difference in mean F0 was only observed in keyphrases. Phrase type effect can be seen in speakers’ sarcasm where keyphrases were produced more slowly than the other two phrase types.Vocalization of jitter, shimmer, and speech rate were found to be major causes for misinterpretation. Sarcastic expression with higher jitter, higher shimmer, and faster speech rate would be considered as sincerity. Sincere utterances with slower speech rate would be recognized as neutrality. Neutral expression with lower shimmer would be misjudged as sarcasm and would be misinterpreted as sincerity if it featured faster speech rate. Moreover, mean F0 and F0 range showed significant effects on misinterpreted expression for different age groups. The sarcastic utterances misinterpreted as sincerity produced by young speakers demonstrated higher mean F0 and wider F0 range. Lower mean F0 and narrower F0 range would cause elderly speakers’ sarcastic expression to be misjudged as sincerity.
Previous research has acknowledged prosodic information as one major component contributes to sarcasm detection. However, the voice quality of sarcastic speech shows no consistency cross-linguistically. This study focuses on the voice quality of sarcasm in Taiwanese Mandarin. Specifically, we investigate whether phrase types and age differences have effects on Taiwanese Mandarin speakers’ delivery and perception towards sarcastic utterances. Six voice quality parameters are examined, including mean F0, F0 range, jitter, shimmer, H1-H2, and speech rate.A sarcasm elicitation task, which uses a fully crossed 3 (attitudes) x 3 (phrase types) design, was adopted to record participants’ utterances of neutrality, sincerity and sarcasm. Then, a perceptual validation process helped identify the successfully recognized and misinterpreted attitudes produced by the speakers.Our results showed that Taiwanese Mandarin sarcasm featured higher mean F0, wider F0 range, lower jitter, lower shimmer, and slower speech rate compared with neutrality, but lower mean F0, narrower F0 range, lower jitter, lower shimmer, and slower speech rate than sincerity. Age difference can be seen in speakers’ sarcasm production strategies regarding F0 range and mean F0, while the difference in mean F0 was only observed in keyphrases. Phrase type effect can be seen in speakers’ sarcasm where keyphrases were produced more slowly than the other two phrase types.Vocalization of jitter, shimmer, and speech rate were found to be major causes for misinterpretation. Sarcastic expression with higher jitter, higher shimmer, and faster speech rate would be considered as sincerity. Sincere utterances with slower speech rate would be recognized as neutrality. Neutral expression with lower shimmer would be misjudged as sarcasm and would be misinterpreted as sincerity if it featured faster speech rate. Moreover, mean F0 and F0 range showed significant effects on misinterpreted expression for different age groups. The sarcastic utterances misinterpreted as sincerity produced by young speakers demonstrated higher mean F0 and wider F0 range. Lower mean F0 and narrower F0 range would cause elderly speakers’ sarcastic expression to be misjudged as sincerity.
Description
Keywords
諷刺語氣, 聲音特質, 句子型態, 年齡, sarcasm, acoustic features, prosodic features, age, phrase type