Computer-Assisted Second Language Learning by Going Back and Fourth in a Collection of Layered Learning Materials for Formative Learning

为了帮助日语作为第二语言的表达自主习得，我们设计并实现了一个系统：我们为学习者提供一组文本，每组文本中主要使用属于一个形成层的表达，并记录阅读文本的声音。作者或学习者编辑并在文本位置放置注释，在注释中描述了这些表达式的格式和语义类别。系统在各层中来回移动，检索学习者指定的表达式使用的文本位置，并显示这些文本部分，标记用法并突出显示与语音再现同步的文本部分，以及可选的语音的道德明智音高序列。通过这种方式，系统允许学习者比较相同表达的不同用例，或者在语义或格式上相关的用例，伴随着语音，并通过显示韵律来促进学习。因此，它有助于在情景验证的语境中区分表达的使用。该系统还在文本中提供句子填空任务，并诊断学习者的答案，引导他们使用与上述相同的机制来发现自己的错误。在实现过程中，我们提供了一组原型表达式笔记，每个笔记中描述了表达式的语法特征及其语义类别的格式，并使用XML技术开发了编写表达式笔记的辅助工具。我们还开发了表达式关系图，表示每个节点对应于一个原型表达式，该图为学习者提供了一个环境，使他们能够在格式或语义相关的节点之间导航，并有助于比较表达式的用例。在作文诊断中，我们改进了词汇化树邻接语法的用法。最后，我们引入了语音识别引擎，以同步高亮文本与语音再现，并显示道德明智的音高信息。少

For assisting autonomous acquisition of expressions of Japanese as second language, we designed and implemented a system as follows : We provide the learners with a group of texts in each of which those expressions belonging to one of the formative layers are mainly used, along with records of the voice reading the texts. The authors or learners edit and place notes to the text positions where such expressions are used as described in the notes of their formats and semantic categories. The system, going back and forth in the layers, retrieves the text positions where the expression specified by the learners are used and displays those text parts with marking the usage and highlighting the text parts in synchronism with voice reproduction and, optionally, mora-wise pitch sequence of the voice. In this way the system allows the learners to compare different use cases of the same expression or those related either in semantics or formats being accompanied by voice sound and promoting lear … More ners of speaking by way of displaying prosody. And thus it helps distinguishable usage of expressions in contexts with episodic validation. The system also provides for sentence-wise blank filling composition tasks in the texts and, diagnosing the learners' answers, guides them using the same mechanism as described above to discover their own mistakes. In implementation, providing a collection of prototype expression notebooks in each of which is described the format of an expression with syntactic features and its semantic category, we developed, using XML technologies, assistance of authoring expression notes. We also developed the expression relational map representing nodes each of which corresponds to a prototype expression and the map provides the learners with an environment in which for them to navigate between the nodes related in formats or semantics, being assisted of comparison between the use cases of the expressions. In diagnosis of composition we refined usage of the Lexicalized Tree Adjoining Grammar. Finally we introduced voice recognition engine in order to synchronize highlighting texts with voice reproduction and also to display the mora-wise pitch information. Less