Improvement of difficulty estimation of personalized teaching material generator by JACET

We have been developing a personalized teaching material generator for beginner level learners of English. Our system keeps personalized vocabularies of learners and recommends suitable teaching materials to them by using their vocabularies. Once the learners choose materials, the system provides each learner with the personalized glossary of words which is expected to be unknown to him/her in order to assist his/her learning suitably. Our basic idea is the concept of “word set.” We consider the learners´ vocabularies and the reading materials as word sets and operate their sets by the set operations. For estimating the difficulty of materials for each learner, it is important to use a proper set of words. Our previous system used the general service list (GSL) and the academic word list (AWL) as word sets. In this paper, we adopted the JACET word list, which is a word list created by the Japan Association of College English Teachers based on the English education system in Japan. We refined our previous system by using JACET as the word set. In addition, the estimation method of the difficulty of materials is improved. Then, the estimation of the difficulty level of reading materials and that of the English skill levels of the learners and materials became more accurate. Moreover, according to many requests from the learners to refine our design of the web site for better usability, our site was improved by adopting a more friendly user interface. In addition, we conducted a questionnaire survey to confirm the improvement of our system.