Measuring Similarity for Observation of Learners’ Syntactic Awareness in Web-Based Writing Environments

Min Kang, Koichi Kawamura, Shuai Shao, Harumi Kashiwagi, Kazuhiro Ohtsuki

Writing in foreign language is a struggle task for learners and revising learners’ writings is time consuming for teachers as well. For this reason, writing support systems have been proposed a lot and one of main functions is to automatically detecting and revising errors in learners’ writings. However, the detection technologies are in progress and the effectiveness of error revision feedback is in argument. Meanwhile, numerous efforts in teaching writing have been made to enhance learners’ writing proficiency and reduce errors. Reading is emphasized to be one of important strategies. However, for the problems what linguistic knowledge learners pay attention to and how they use the knowledge into their writings in web-based learning, there is few studies reported. In this paper, we performed a reading-to-write experiment in a web-based writing environment and analyzed reading materials and learners’ writings to explore how to observe learners’ awareness on syntactic structures in the materials. Sentence patterns that proposed in our previous works are introduced to categorize sentences and the syntactic similarities be-tween reading materials and writings are calculated. The experimental results revealed that students showed higher comprehension on content but poor attention on syntactic structures in reading activities if the structures are not significant salient. It is assumed that the similarity measure is effective to ob-serve students’ awareness on syntactic structures in materials and further works are needed to automatically observe the awareness.