A Discourse Structure Analysis of Technical Japanese Texts and Its Implementation on the WWW

Abstract

This paper deals with a discourse structure analysis of technical Japanese texts for developing a Japanese writing Computer Assisted Language Learning (CALL) system whose goal is to assist students in learning to write technical Japanese texts. To analyze discourse structures of technical Japanese texts, cohesive expressions are used as cue words. The rules for analyzing texts are based on micro-level and macro-level information, namely cohesive expressions and headlines. A CALL system for helping foreigners to learn to write technical Japanese texts is under development using natural language processing (NLP) techniques. This paper describes a completed part of the work, which is a CALL system that can be used for automatically detecting headlines and cohesive expressions of technical Japanese texts on any World Wide Web (WWW) browser. This approach can be considered as a new means of language learning for the future. Furthermore, a system evaluation is conducted to evaluate the performance of the system. The results of this evaluation show that the system obtained a high degree of accuracy on extraction of cohesive expressions and headlines by using the revised rules set proposed in this study.