日本語韻律学習のための音声アーカイブ構築 (Nihongo inritsu gakushū no tame no onsei ākaibu kōchiku)/Development of Sound Archives for Learning Japanese Prosody

Opportunities for Japanese learners to communicate in Japanese are ever-increasing, both inside Japan and abroad. As such, an important task for the teaching of Japanese is to research, establish, and disseminate effective methods for pronunciation teaching, as well as to continue developing pronunciation teaching materials and learning support tools. Indeed, the need for such resources is quite urgent. In particular, it has been pointed out that prosody (e.g., pitch accent and intonation) is important in order for Japanese learners to acquire natural Japanese pronunciation for fluid communication. Toward that end, the teaching of word accent alone is insufficient, and research at the sentence level (i.e., how each word is realized in a sentence) is also necessary. The project reported on in this presentation involves the large-scale collection of speech data relating to sentence-level prosody from several countries, representing Japanese learners with various languages of the world as their native languages. With special emphasis on prosodic features that are considered difficult to realize, data are being collected on a large scale, both cross-sectionally and longitudinally, in order to clarify both problems specific to certain native languages as well as problems shared in common. At the same time, an online archive of this learner speech, organized by native language and phonetic feature, is being constructed. In the near future, it will be possible to access and consult this resource from anywhere in the world. The resulting platform will be offered as a cornerstone for future research and development on Japanese pronunciation teaching. This presentation introduces the significance of these materials for Japanese pronunciation teaching, provides examples of specific ways to make use of them, gives an overview of the data collection conducted preliminarily as a first step toward the construction of the archive, and reports on the results of having analyzed these pilot data. Materials were prepared for each of the following six areas of difficulty in learners' Japanese prosody acquisition: (1) the intonation of interrogative sentences, (2) prominence/focus, (3) downtrends (i.e., downstep and declination), (4) syntactically ambiguous sentences, (5) emotional speech, and (6) vowel devoicing. At present, recordings have been conducted for speakers with Chinese, Russian, Korean, and Italian as native languages. In the pilot experiments, it was observed that all learners have difficulties with the production of the above prosodic features. For example, the pitch accent for a word may be pronounced appropriately for a word in isolation but become prosodically unnatural in a sentence, or distinctions in prominence or ambiguous sentences may be realized in ways that differ from Japanese native speakers. This presentation will show concrete examples (mainly using data from Italian speakers) and report on a comparison of them with speakers of other languages.