English-Chinese Language Transfer Correction Incorporating Lexico-Syntactic Templates and Relative Position Modelling

• Main Authors: Matthew Harris, 韓天賜
• Other Authors: Chung-Hsien Wu
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/76405774977318847550

碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 96 === Language transfer is the phenomenon by which utterances produced by second-language learners are subject to the influence of their first language. This can result in errors being introduced into a sentence, such as incorrect word order, wrong lexical choice, the inclusion of redundant words, or the omission of necessary words. Much recent work has focussed on the correction of errors in English as a Second Language. However, despite the growth of Mandarin Chinese as a popular foreign language, there has been relatively little research on errors made by learners of Chinese. This study presents a discussion of the problem of English-Chinese language transfer and proposes techniques for the automatic correction of such errors. Two new language modelling techniques, a Relative Position Language Model and a Lexico-Syntactic Model are introduced to help overcome some of the limitations of traditional models. Evaluations on the proposed correction procedure show that using the premise of language transfer can be an effective way to correct error sentences produced by Chinese learners. It is also demonstrated that the two proposed models can outperform existing approaches in their ability to find optimal candidates produced by the correction procedure.