Computational Psycholinguistics Tokyo

プロジェクト名，担当者名: 対照言語学の観点から見た日本語の音声と文法 (文法研究班「名詞修飾表現」)
窪田悠介 (国立国語研究所理論・対照研究領域准教授); 大規模コーパスを利用した言語処理の計算心理言語学的研究
大関洋平 (早稲田大学理工学術院講師); コーパスアノテーションの拡張・統合・自動化に関する基礎研究 (係り受け班)
浅原正幸 (国立国語研究所コーパス開発センター教授)
開催期日: 2020年1月31日 (金) 13:00～15:00
開催場所: 早稲田大学西早稲田キャンパス 51号館 5階 12号室 (東京都新宿区大久保3丁目4-1)
交通アクセス
関連サイト: Computational Psycholinguistics Tokyo

プログラム

"Message-Oriented Phonology in Japanese: Word Duration and Pitch Peak" 橋本大樹 (東京大学)

It has been widely demonstrated that a word is pronounced with lower phonetic redundancy when it has higher contextual predictability. For example, when a word is predictable given a preceding word and when a word has higher contextual predictability given a following word, it is pronounced with shorter duration. Likewise, words with higher contextual predictability are produced with centralized formant values. This probability-oriented reduction is known as “probabilistic reduction.”

This phenomenon can neatly be captured by Message-Oriented Phonology (MOP). MOP hypothesizes that a speaker balances the efficiency and accuracy of message transmission. When a word is contextually predictable, it can be conveyed successfully to an addressee, the result of which is that the speaker improves the efficiency of the message transmission. On the other hand, when a word is less predictable, the message transmission is more likely to fail, and thus a speaker needs to invest more resource cost in a speech signal, with the result that the phonetic redundancy is increased.

The aim of this study is to explore whether probabilistic reduction can be extended to pitch values. Most previous literature discusses probabilistic reduction in relation to word duration, so therefore, to the best of my knowledge, this study is the first study to investigate the relationship between pitch values and contextual predictability of a word. It will be demonstrated that a word is pronounced with a higher pitch value, when it is contextually less predictable. This result is amenable to MOP.