On the creation of a learner corpus for the purpose of error analysis

概要

論文の詳細を見る
Learners with similar backgrounds have a tendency to make the same types of errors in L2 production. Such errors can be viewed as having the potential to inform pedagogical methodologies, in that they shed light onto which features of the L2 are the most problematic for particular learners. Analyzing such errors also provides insight as to why these learners tend to make these errors, thus furthering our understanding of how second languages are acquired.This study aimed to create a learner corpus for the purpose of error analysis to discover which errors occurred most frequently, and to examine why such errors occurred. Various CALL (computer aided language learning) methodologies were utilized to create an approximately 85,000 word learner corpus. Errors were corrected and classified, and error analysis was conducted on the most frequent errors found. This analysis revealed that interference from the learners' L1 was the source for the majority of errors, while cultural and metalinguistic knowledge also proved to be at fault for some particular errors.The results of this study should prove to be valuable for English language teachers and researchers in Japan, in that the most frequent English errors that Japanese learners produce were quantified and also discussed. Thus, teachers and researchers can be cognizant of which errors prove to be the most troublesome, and can better understand why they occur to help Japanese learners to avoid them.