Developing a program for language teaching with parsed corpora
Grant-in-Aid for Scientific Research (C)
|Allocation Type||Multi-year Fund |
Basic Section 02060:Linguistics-related
|Research Institution||Hirosaki University |
バトラー アラステア 弘前大学, 人文社会科学部, 准教授 (90588873)
|Project Period (FY)
2019-04-01 – 2022-03-31
Granted (Fiscal Year 2019)
|Budget Amount *help
¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
Fiscal Year 2021: ¥390,000 (Direct Cost: ¥300,000、Indirect Cost: ¥90,000)
Fiscal Year 2020: ¥390,000 (Direct Cost: ¥300,000、Indirect Cost: ¥90,000)
Fiscal Year 2019: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
|Keywords||grammatical analysis / parsed corpora / language teaching / English / Japanese|
|Outline of Research at the Start
This research will apply a grammar "toolkit" developed from parsed corpora of English and Japanese to real textual materials for language teaching purposes. This will provide students with grammatical analysis skills for criticising and improving their own language use.
|Outline of Annual Research Achievements
The implementation plan is to develop a program for language teaching with parsed corpora.
The components are:
1) a grammar textbook focused on English language learning for Japanese students at university level,
2) a large grammatically analysed corpus of English, also linked to Japanese language analysis for purposes of comparison, and
3) the development of a "toolkit" for analysis creation, for students to start analysing their own written language.
The goal is to empower students to critically analyse their own language use and be drawn to explore wider insights from the grammatically analysed corpus.
|Current Status of Research Progress
Current Status of Research Progress
1: Research has progressed more than it was originally planned.
The first year of the project has seen development in all three components of the project.
An initial draft of the textbook has been prepared with chapters introducing 1) Words 2) Phrases 3) Clauses and 4) Subordination and coordination of clause content.
A lot of work has gone into the creation of an English corpus, and to the development of the scheme that underpins the annotation, which has been optimised to support the textbook description.
The "toolkit" has been developed as a post-processor that is able to convert the output of the Stanford CoreNLP parser, a state-of-the-art statistical parser, to correspond to annotations of the corpus/textbook description.
This conversion was described in a conference presentation (LENLS 16), and is currently available as a demo on the web.
|Strategy for Future Research Activity
Future plans are to further develop the three components. The textbook will be enlarged and trial lessons undertaken.
Annotation of the English corpus will aim to grow the overall size.
A key part of language learning is to gain knowledge of vocabulary. To support vocabulary learning, the coming year will start adding word sense disambiguation information.
So far the "toolkit" has involved adding post-processing to a statistical parser. This buys wide coverage and a way to manage ambiguity. However, this does lead to an unpredictable range of parsing errors, requiring students to be sensitive to miss-analysis. The future plan is to re-orient the toolkit around a logic based grammar approach with predictable properties. The new challenge is to increase coverage for unconstrained input.
Report (1 results)
Research Products (4 results)