We witnessed remarkably rapid development of the linguistic study of intonation during the last 25 years or so. Two main factors that seemed to enabled this development were analyzed in the first half of the paper, namely, progresses in the generative grammatical treatment of intonation, and, development of digital speech processing technology. The last half of the paper was devoted to the discussion of current issues that requires further investigation; these include classification of paralinguistic information, definition of accentual phrase in Japanese, classification of phrase-final intonation contours, intonation of dialects, phonetic features other than the F0 that contribute to the perception of intonation, and, issues related to the techniques of digital speech analysis.

抄録全体を表示

日常対話音声の正しい理解や,対話における話者交替の制御,更には自然な対話音声の合成を行なうためには,韻律特徴の適切な操作は不可欠である。日本語では,句末の音調は,疑問・強い主張・意外性・促しなどの文のモダリティを表示する機能,フォーカスを表示する機能,大きな意味の区切りの表示,または発言がまだ終わっていないことを示す談話機能など,さまざまな役割を持つ。また,句末に終助詞が付くか否かによって,句末音調がもたらす機能が変わる。本稿では,対話における話者交替の談話機能に焦点を当て,句末の形態素や品詞などの言語情報を考慮し,句末音調の役割について論ずる。自然会話音声データを用い,300ミリ秒以上のポーズで区切った発話区間の句末において,形態素と談話機能のラベルを付与し,音調を表す音響的特徴を抽出した。分析の結果,多くの形態素では,句末音調と話者交替の談話機能に関連が見られたが,主に終助詞を含んだ一部の形態素では,関連が見られないことが示された。また,複数の品詞が同じ形態素で表現される場合があり,言語情報を用いる際の注意点について述べる。

抄録全体を表示

This report is a synthesis of the multiple studies that introduces systematic relations between prosodic feature of speeches and impressions of speakers' personality. These studies utilize speech stimuli synthesized by manipulating the speech-rate and fundamental frequencies. The results suggest that it is possible to approximate the impressions of personality characteristics from prosodic features in the synthesized speech. The effects of familiarity and naturalness in intonations on the personality impressions are also discussed in relation to the dialectal regions. Responding to the recent progress in the technology to manipulate speech sound, the report concludes with the future directions and possibilities to investigate the expressive speech sound.

抄録全体を表示

The purpose of this paper is to introduce characteristics of loudness control in realizing prominent syllables by Japanese learners of English. More specifically, the phonetic features peculiar to RMS (Root Mean Square) envelopes gained in the utterances by Japanese learners will be treated and pedagogical suggestions to correct these features will be made. After reviewing the previous studies on intonation, intonational characteristics by Japanese will be summarized in the order of tonality, tonicity, and tone. Then, the production experiment designed to compare phonetic realizations of prominent syllables by both native speakers of English and Japanese will be introduced. Finally, the problems with pronunciation training programs and software will be pointed out and suggestions for better training of rhythm will be put forward.

抄録全体を表示

Through searches on the Internet using Google, 396 adjective expressions that immediately precede the noun kutyoo (tone of voice) were identified as candidates for the major types of Japanese speech styles. Thirty-seven percent of them characterize the speaker's emotional, physical or social state, 31% describe the social or psychological relationship between the speaker and the hearer, and 17% concern the styles used in particular situations or of particular speakers. Comparisons with the results of other searches using Google performed 11 months later and with those using Yahoo! JAPAN Search revealed that there are 16 kutyoos that are constantly used with high frequency on the Internet. They are, in frequency order, tuyoi (strong), meeree (commanding), yasasii (gentle), teeneena (polite), kibisii (severe), odayakana (mild), ikari (angry), karui (light, not heavy), sizukana (quiet), tantantosita (plain), kitui (stern), otituita (calm), hagesii (violent), setumee (explaining), reesee (cool), and dantee (assertive) tones of voice. A multidimensional scaling analysis of perceived similarities among these kutyoos suggests that they can be grouped into 12 categories of speech style.

抄録全体を表示

本研究は,話し言葉において主要な統語境界を示す音声的手掛かりが,英語母語話者と日本人英語学習者でどのように異なるかを検討したものである。実験ではrainingとhomeが,文末または節末にそれぞれ3回と4回生起する談話を,8人のアメリカ英語話者(NS)と30人の日本人大学生(JS)が朗読した。rainingとhomeのピッチと長さ,その前後のポーズ挿入を音響分析した結果,JSはNSより頻繁に節末にポーズを挿入する傾向が観察された。またNSは文末より節末で大きくピッチを変動させ,談話中の位置と前後関係によりピッチパタンとピッチの高さを多様に変化させた。これに対してJSでは,文末・非文末(節末)の区別はなく,一貫してピッチの変動幅の比較的小さい下降調が観察された。またfinal lengtheningは,rainingとhomeの語末子音の/ŋ/と/m/において,JSの方がNSより有意に小さかった。JSのこれらの音声的特徴から,日本語のイントネーション・時間構造の干渉が考察された。

抄録全体を表示

This paper analyzes the accentuation of so-called alphabetic acronyms in Kinki Japanese with main focus on the effects that the syllable structure and familiarity of a word exert on its accent pattern. The data obtained from an original fieldwork show that the initial tone of alphabetic acronyms in Kinki Japanese is determined by their phonological length, i.e., number of moras involved, and their syllable structure. These phonological factors are quite similar to those that are responsible for the unaccented pattern observed in loanwords and alphabetic acronyms in Tokyo Japanese. This study has also revealed a certain correlation between the speaker's familiarity with a word and the ratio at which the word violates shiki-hozon, a tone preservation rule exhibited by compound words.

抄録全体を表示