| In China, many English learners are not good at English writing because there are huge differences between English and Chinese. They need more practice to improve their writing skill. Traditional English writing training has lots of shortcomings. As a result, writing training is ineffective.Automated English essay scoring system can effectively overcome these shortcomings, and benefit both teachers and students. There are several systems used in abroad. They all show good performance. However, these systems are not suitable for Chinese English learners because they are designed for native English learners.Because of the differences between Chinese English learners’essays and native learners’essays, the scoring strategy to deal with content and expression must be different, and the judge model should be rebuilt.We analyze some CET-4 and CET-6 articles and extract some text features such as words, phrases, and paragraphs. According to the correlation between these features and scores, we do multiple regression analysis based on the features recombination and extract part of features those are contributed to the quality of an essay. In the end, we build a reasonable model to distinguish high quality essay from others and solve the problem of low differentiation scores in automated essay scoring system.The experimental results show that some text features have high correlation with essay quality and can reflect it stably whatever the title is. In addition, the model proposed in this paper has good performance and can be considered as a reference in automated essay scoring system for Chinese English learners. |
PDF Full Text Request