This article attempts to present the usefulness and pedagogical value of using learner corpora. Using a very rich source of the data provided from learner corpora, this article prioritizes the specific reference to particular grammatical errors made by Japanese learners of English, presenting how leaner corpora can be utilized to reveal such mistakes. In order to understand Second Language Acquisition (SLA), this paper suggests that data from learner corpora also need to be carefully compared with a wider range of references (e.g., teachers' experiences, data from general reference corpus, and normative grammar presented in, for example, dictionaries) to improve EFL pedagogic grammar teaching for Japanese learners of English.