以 CLEC 語料庫為基準設計實作與改良一個英語文法檢查系統

Abstract

在現在的社會中，小學生便開始學習英語了，而且英語的檢定考詴越來越普及、歐盟委員會亦提出將英語、法語、德語作為歐盟“共同專利體係”的官方語言，學習英語已蔚為風行。因應這些發展，學生從小就開始學習英語，以及從小就開始□加英語相關的檢定考詴。可是，在學習英文時，以英文作文為例，當學習者在完成一篇作文的時候，只能以人工的方式去批改。人工方式批改作文，不但費時，而且 因為作文的批改沒有標準答案，有可能出現評分標準不一的情況。即使目前有少數系統可以批改作文，正常率還是有待提升。

因此，本研究詴著去實作一個低出錯率、高檢查率的文法檢查系統。希冀先以基本文法規則為設計藍本，設計出不同的文法檢查模組後，再以CLEC 語料庫的錯誤類型分類去增加模組的反向的規則並加以驗證。期望以一致的角度，先將作文當中的拼字，文法錯誤的情形清楚的標示出來，以提供閱卷者在閱卷時評分的參考。

另外，對我們華人來說，由於英語不是我們的母語，我們在文法和習慣用法上也會發生一些與以英語為母語的學習者不同的錯誤。這些錯誤不一定能被一般以英語為母語的英語檢查系統檢查出來。因此我們希望建立一個專門以華人為背景的英語文法錯誤檢查系統。

Nowadays, elementary students start to learn English, taking English test becomes more and more popular, and The European Union proposed to make English, French and German as the official language of the “Common Patent Policy” of the EU patent. It is a trend for people to learn English due to the above reasons. Listening, Speaking, Reading ad Writing are the four basic skills in learning a language. But it is not easy for people to learn how to write. After wring a passage or a composition, the only way to check its correctness is to check it manually by a teacher. Checking a composition is time-consuming. And since there is no standard answer to the composition, the grading standard way vary. Even though some system can find the grammar errors in passages, their correctness still need improvement.

Therefore, we try to implement a grammar checker with low error rate and high error checking rate. We first take the basic grammar rules as the base to design several modules with different functions. Then we continue adding other rules to those modules according to the error classification of CLEC. Finally we examine our modules with passages in CLEC. By using the corresponding approaches, spelling errors and grammar errors can be marked and served as references for the graders.

Also, for Chinese people, since English is not out mother tongue, we will make some errors that native speakers would not. And these errors may not be found out by those grammar checkers which were implemented by native speaker. Therefore, we hope to implement an English grammar checker which was specially designed for Chinese.