An algorithm for generating a dictionary of Japanese scientific terms

Abstract

This article describes the implementation of algorithms for generating a dictionary of Japanese scientific terms originating from the English language. Such words are typically transliterated into katakana, one of the four distinct orthographies that commonly occur in Japanese texts. The effort required to learn katakana yields significant returns to readers of technical Japanese due to the high incidence of terms derived from English. The algorithms described here are able to automatically generate a katakana to English dictionary from raw Japanese text and its English translation, which in many cases is available in electronic form, with a reasonable degree of accuracy. The algorithm thus allows an instructor to generate specialized Japanese vocabularies from selected articles, so that he/she can individualize lessons for a particular student's technical interest and competence at reading katakana. The algorithm has been shown to be very effective in technical Japanese instruction and is currently used in a course on Japanese information processing for electrical engineering students at Purdue University.