Abstract. The recent interest in utilizing a large collection of authentic electronic language resources, called a corpus, has increased in tertiary institutions to facilitate student language learning. The examination of a corpus has made it possible for students to analyze multiple language examples and the patterns of co-occurred words. This approach, referred to as 'data-driven learning', assists students in becoming more aware of language use and patterns of various forms. This paper outlines the research conducted for the purpose of examining how Japanese students perceive the efficacy of exploring different types of corpora: one corpus familiar to students, and the other, the Brown Corpus which was unfamiliar to students. Specifically, the focus was on examining the relationships between the students' proficiency levels and their perceptions of using different types of corpora.