Using multivariate statistical techniques to analyze development of speaking and writing proficiency level

2014 Fiscal Year Final Research Report

• Project/Area Number: 24520631
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Foreign language education
• Research Institution: Chuo University
• Principal Investigator: ABE Mariko 中央大学, 理工学部, 准教授 (90381425)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: Learner Corpus

Outline of Final Research Achievements

This project investigated the overall patterns of variation across seven proficiency levels of 1,263 Japanese EFL learners and native English speakers. The methodological approach combined a learner corpus, language processing techniques, and multivariate statistical analyses to identify patterns of language use. The largest spoken learner corpus in Japan, NICT JLE Corpus, was used for the analysis. The study found interesting rising, flat, and falling frequency patterns in how several linguistic features are used in different oral proficiency levels. Some linguistic features (e.g., phrasal coordination and nouns) were frequently used by the learners at a low level. The frequencies of some others (e.g., contraction, pronoun it, and emphatics) increased as oral proficiency increased. The study identified a set of linguistic features that differentiate among second language proficiency groups as well as between non-native and native speakers of English.

Free Research Field

コーパス言語学