2007 Fiscal Year Final Research Report Summary
RESEARCH INTO ERROR ANNOTATION FOR CORPORA OF JAPANESE-SPEAKING LEARNERS OF ENGLISH
| Project/Area Number |
17320088
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Foreign language education
|
|---|
| Research Institution | Tokyo University of Foreign Studies (2007)
Meikai University (2005-2006) |
|---|
Principal Investigator |
TONO Yukio Tokyo University of Foreign Studies, Graduate School of Area and Culture Studies, Associate Professor (10211393)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
ABE Mariko Takasaki City University of Commerce, Department of Economics, Associate Professor (90381425)
|
|---|
| Project Period (FY) |
2005 – 2007
|
| Keywords | Corpus linguistics / Learner corpus / SLA / proficiency level / Error Analysis / Interlanguage / Contrastive Analysis / Developmental criteria |
|---|
| Research Abstract |
The projects aimed to achieve the following three goals: (a) the critical survey of available error annotation schemes for learner corpora, (b) the design and implementation of our original error annotation scheme for the JEFLL Corpus, and c the compilation of a parallel version of the JEFLL Corpus, consisting of the original essays and their corrected version by a native speaker.
For the first two years, we did some major literature review and the survey of error annotation schemes and tagging systems used for learner language analysis. Then we revised our original version of error annotation scheme by integrating multi-layered & multi-modal annotation models. We also developed a stand-off annotation tool, using a open-source generic annotation tool, MMAX2.
At the same time, we have completed the proofread version of the JEFLL Corpus (a corpus of free compositions by secondary school Japanese learners of English), which has a set of the original data and its corrected version in parallel. With this resource, we will be able to accurately construct the error-tagged version of the JEFLL Corpus. Since we had to terminate the project in three years, instead of four, we could not make the sample error-tagged version of the JEFLL, which should be the goal for the future project.
|
Research Products
(50 results)
