Detecting Word Usage Errors in Chinese Sentences for Learning Chinese as a Foreign Language

Detecting Word Usage Errors in Chinese Sentences for Learning Chinese as a Foreign Language Yow-Ting Shiue and Hsin-Hsi Chen Department of Computer Science and Information Engineering, National Taiwan University No. 1, Sec. 4, Roosevelt Rd., Taipei 10617, Taiwan E-mail: orina1123@gmail.com; hhchen@ntu.edu.tw Abstract Automated grammatical error detection, which helps users improve their writing, is an important application in NLP. Recently more and more people are learning Chinese, and an automated error detection system can be helpful for the learners. This paper proposes n-gram features, dependency count features, dependency bigram features, and single-character features to determine if a Chinese sentence contains word usage errors, in which a word is written as a wrong form or the word selection is inappropriate. With marking potential errors on the level of sentence segments, typically delimited by punctuation marks, the learner can try to correct the problems without the assistant of a language teacher. Experiments on the HSK corpus show that the classifier combining all sets of features achieves an accuracy of 0.8423. By utilizing certain combination of the sets of features, we can construct a system that favours precision or recall. The best precision we achieve is 0.9536, indicating that our system is reliable and seldom produces misleading results. Keywords: Grammatical error detection; HSK corpus; Second language learning. 1. Introduction Recently, more and more people select Chinese as their second language. Developing grammatical error detection and correction tools for Chinese language learners is indispensable. The flexibility of the Chinese language makes error detection more challenging than other languages. According to the analysis on the HSK dynamic composition corpus created by Beijing Language and Culture University, word usage error (WUE) with error tag CC, is the most frequent type of error at the lexical level.1 In the HSK corpus, the CC type errors are further divided into four major subtypes. The descriptions of the subtypes are shown as follows, each in terms of a pair (misused form, correct form).2 (1) Character disorder in a word, e.g., (先首, 首先) (first of all) and (眾所知周, 眾所周知) (as we all know). (2) Incorrect selection of a word, e.g., 雖然現在還沒有 (實踐, 實現), … (while it is not yet implemented, …). (3) Non-existent word, e.g., (農 作 品, 農 產 品) (agricultural product). (4) Word collocation error, e.g., 最好的辦法是兩個都 (走去, 保持)平衡 (The best way is to keep both balance). In Chinese, segmentation is a fundamental problem. When characters in a word are disordered, e.g., “首” and “先” are exchanged in the word “首先”, the resulting form may not be a word. Thus, they may be segmented into a sequence of characters by a dictionary-based segmentation system. In word collocation error, both the misused form and the correct form are real words, but the latter collocates with other words in the given sentence and the former does not. CC (1) and (3) are similar in that the misused forms are 1 http://202.112.195.192:8060/hsk/tongji2.asp 2 http://202.112.195.192:8060/hsk/help2.asp not in a dictionary. Likewise, CC (2) and (4) are similar. The misused forms are in a dictionary. …

Content truncated to 500 words