Development and Evaluation of a System for Grading Japanese Reading Materials by Level of Difficulty

本研究旨在提高基于网络的阅读支持系统的整体质量和实用性，该系统最初是为世界范围内将日语作为外语学习的学习者而开发的。为了改进目前的系统，重点是根据词汇难度和句子复杂性确定难度水平的基准。对来自不同语言背景的JFL学习者进行了调查，以确定哪些词汇项和语法结构对不同熟练程度的学习者造成了理解困难。研究发现，通过同时考虑单词难度和句子复杂性，可以更全面(全面)地确定总体难度水平。此外，还使用了数据挖掘技术来确定在线辅导词典的使用趋势。这类数据对于调整词典中包含的信息以及深入了解真正的学习者如何使用词典是必要的。这项研究的结果将被纳入在线阅读支持系统，并向世界各地的教育工作者和学习者提供。

The present research is aimed at improving the overall quality and usefulness of a Web-based reading support system originally developed for learners world-wide who are studying Japanese as a foreign language(JFL). In order to improve the present system, attention was focused on determining the benchmark for levels of difficulty in terms of vocabulary difficulty and sentence complexity. JFL learners from a wide variety of language backgrounds were surveyed to determine which vocabulary items and grammatical structures pose comprehension difficulty for learners at different levels of proficiency. It was found that by taking both word difficulty and sentence complexity into account, a more well-rounded(comprehensive) determination of the overall level of difficulty could be made. Furthermore, techniques of data mining were used to determine trends in usage of the on-line Tutor dictionary. This kind of data is necessary for making adjustments in the information covered in the dictionary and for providing insights into how real learners use the dictionary. The results of this research will be incorporated into the on-line reading support system and made available to educators and learners throughout the world.