國語中雙音節詞輕重音的聲學特徵

Some Stress Related Acoustic Features of Disyllabic Words in Mandarin Chinese

才又研究的目的在於自聲學特徵的觀點探討國語中雙音節詞的輕重音問題。研究範圍在重音形式方面包括重輕型（stress-weak，（SW）及一般型（normal-stress，（NS））二種，其中SW又包括了由詞素本身的輕聲以及由詞彙結合所引起的弱化及輕聲二種現象；以及兩種重音型都存在的詞組。在聲學語音特徵方面則包括了基頻軌跡（含調型與調值）、音節長度及強度三方面。發音的方式是朗讀，包括正常語調及加重語氣兩種。受試者為一以北平話為母語，但在臺灣地區居住了四十年的女性，測試結果顯示出受試者在不知道測驗性質時的表現，以一般型重音使用最多，就正常語調而言，在全部受試詞組中，含有詞素輕聲的部分，有75%是以一般型重音發出；而含有詞彙組合弱化輕聲的部分，有54%以一般型重音發出。而兩種重音型皆可的詞組中，有72%是以一般型重音發出。不過，當受試者在得知實驗性質以後，她的表現有很大的轉變：凡含有詞素輕音及詞彙組合弱化輕聲的詞組，全部（100%）以重輕型發出，而兩種重音型皆可的詞組中，則有96%是以重輕型發出。就加強語調而言，一般型仍為使用最多的重音型，即在所有受試的加重語調詞組中，有93%是以一般型發出。本研究並分析了所有受試狀況的基頻軌跡，包括調型、音節開始、中段及結尾的頻率。而各項受試詞組中的平均音節長度以及音強也加以分析。本研究發現，正常語調與加強語調最大的差異是在音節長度這部分。此外，本文並討論了目前在臺灣所通行的國語的重音型趨勢，似以一般型為最常用的形式。

The aim of the present study is to investigate acoustic features related to stress patterns of disyllabic structure in Mandarin Chinese (MC). The stress patterns under investigation include the stress-weak (SW) pattern and the normal-stress (NS) pattern in both non-emphatic and emphatic speech produced in citation forms. (The SW patterns include both morphemically determined weak stress as well as lexically determined weak stress.) Test tokens also include disyllabic words that could be produced with either stress pattern. The acoustic features being examined include fundamental frequency (FO) patterns, syllable duration and amplitude. The subject (S) was a female native speaker of Peking MC who had resided in Taiwan for nearly 40 years. Overall results of the S's performance regarding non-emphatic speech with no knowledge of the nature of the experiments yield a preference towards the NS pattern, i. e., 75% of the morphemically determined SW pattern was produced with the NS pattern, and 54% of the lexically determined SW pattern was produced with the NS pattern. 72% of the tokens that can be pronounced with either the SW or NS pattern was produced with the NS pattern. Nevertheless, S's performance changed towards the SW pattern after learning about the aim of the study. For both morphemically and lexically determined weak stress, 100% of the tokens were produced with the SW pattern, whereas 96% of the test tokens that can be pronounced with either stress pattern was produced with the SW pattern. However, test tokens in all conditions were again produced with the preference of the NS pattern in emphatic speech, namely, 93% of the emphatic data. Analyses of the F0 patterns include overall F0 shape, and the syllable beginning, middle and final frequency in all conditions. Mean syllable duration of the test tokens in each condition and the ratio of the two syllable in each condition were derived. The amplitude of syllables across all test tokens was also derived. It is found that the major difference in terms of acoustic features between non-emphatic and emphatic speech lies in the domain of syllable duration. A general tendency for MC spoken in Taiwan to adopt the NS pattern over the SW pattern is also discussed.