Acquisition of Speech Perception by Second Language Learners : Learning Process of Japanese Special Phonemes

本研究的主要目的是从“言语知觉研究中的学习范式”这一新的视角来描述和分析日语作为第二语言学习中的言语知觉习得过程，并阐明学习的层次性和顺序性结构。(1)理论研究：讨论了前人关于二语言语知觉学习实验研究中的几个理论和方法问题，并提出了解决这些问题的新观点和新方法。(2)实验研究采用日语和美国英语的跨语言设计，从实验上阐明了二语言语知觉的特点。以日语为母语的人在音段水平上有各种各样的困惑(英语元音，/r/-/i/等等)；美国英语的人在音段水平上有困难，而是在超音段水平上(即日语Mora)。美国英语和汉语的母语者t…下了更多的雨，以识别日语的短/长元音和特殊的音素/q/(Moraic Gminata)。两组受试者在训练中都表现出稳定的进步，并在成绩测试中表现出广泛的泛化。使用自然标记的实验室训练对不熟悉的音素环境和说话者的泛化是有效的。音素语境中的各种样本促进了知觉学习，但足够数量的训练试验弥补了样本中不充分的变异性。阐明了分析动态和可塑性习得过程在二语言语知觉学习研究中的重要性。(3)展望为了支持个体学习者的自主学习，有必要开发一个日语言语知觉计算机辅助学习系统。在这样的学习系统中，训练刺激集应该在声学/语音空间和训练的时间/序列结构上进行优化。为了为以各种语言为母语的人建立一个有效的培训系统，对这些语言进行跨语言实验研究将是必不可少的。较少

The principal purpose of this project was to describe and analyze the acquisition processes of speech perception in learning Japanese as a second language (L2) from a new viewpoint, called here "learning paradigm in speech perception studies", and to clarify the hierarchical and sequential structures of learning.(1) Theoretical researchSeveral theoretical and methodological problems in the previous experimental studies on L2 speech perception learning were discussed, and new viewpoints and methods were proposed to overcome these problems.(2) Experimental researchCharacteristics of L2 speech perception were experimentally clarified by using a cross-language design between Japanese and American English. Native speakers of Japanese had various confusion at the segmental level (English vowels, /r/-/I/, and so on ) ; those of American English had difficulties not at the segmental level but at the suprasegmental level (i.e.Japanese mora).Native speakers of American English and Chinese were t … More rained to identify Japanese short/long vowels and the special phoneme/Q/ (moraic geminata). Both groups of subjects showed stable improvements in the training and extensive generalization in the achievement tests. The laboratory training using natural tokens was effective for generalization over unfamiliar phonemic contexts and talkers. Various exemplars in the phonemic contexts facilitated the perceptual learning, but sufficient amounts of training trials compensated for the insufficient variability in the exemplars. The importance of analyzing the dynamic and plastic acquisition processes in the studies of L2 speech perception learning was clarified.(3) ProspectsIt is necessary to develop a computer-assisted learning system for Japanese speech perception in order to support autonomous learning by individual learners. In such a learning system, training stimulus sets should be optimized both in acoustical/phonetical space and in temporal/sequential structure of training. To develop an effective training system for native speakers of various languages, cross-language experimental studies for those languages will be indispensable. Less