Home // International Journal On Advances in Intelligent Systems, volume 11, numbers 3 and 4, 2018 // View article
Authors:
Koichi Kawamura
Harumi Kashiwagi
Min Kang
Keywords: Error Detection; Sentence Pattern; Global Error; Parser; Source Language; Criteria for Error Determination
Abstract:
Automatic error detection systems for English writing have been improving since they were first introduced and are being applied to foreign language learning. However, these systems mainly focus on local errors, such as grammatical aspects in the target language and ignore the meaning intended in the source language. As a result, it is quite difficult to detect global errors using existing error detection systems. In this paper, we propose a new automatic error detection system to solve this problem. In order to determine whether the structure of an English sentence is in error or not, criteria for error determination must first be defined. Our system is based on the idea that criteria for error determination are created by the correspondence relation between Japanese and English using sentence patterns. In order to evaluate our system, by way of illustration, seven sentence patterns based on two grammar categories and four POS (part of speech) categories were selected. Automatic error detection using these seven sentence patterns was carried out on 100 Japanese sentences with subjects and their corresponding English sentences. As a result, we concluded that, using the sentence patterns in the source language, automatic error detection is effective when based on our criteria for error determination.
Pages: 202 to 211
Copyright: Copyright (c) to authors, 2018. Used with permission.
Publication date: December 30, 2018
Published in: journal
ISSN: 1942-2679