汉语词汇测试自动命题研究

发布时间：2018-03-07 04:24

本文选题：二语教学　切入点：词汇测试　出处：《中文信息学报》2017年01期 　论文类型：期刊论文

【摘要】：为了提升汉语词汇测试的命题效率,该文从汉语语言特性和二语教学需求出发,对词语听力、多空词语选择、词语排序和单空词语选择四种词汇测试题型进行自动命题尝试,以满足不同语言信息、不同难度的词汇知识考查。在词语特征的提取上,构建了一个覆盖词音、词形、词义、语法、搭配、偏误各层次信息的词汇知识库,在句子特征的提取上,实现了语法项目自动识别、句子难度分析等算法,为自动命题中的题干句、目标词和干扰项选择提供依据。通过词句选择和语块合成等步骤,生成四种题型共计7 263道词汇测试题。人工测试数据显示,词汇测试自动命题的初步尝试取得了较好的效果,约58%的试题被评价为完全合理,经人工简单调整,试题接受率达到75.7%。
[Abstract]:In order to improve the propositional efficiency of Chinese vocabulary testing, this paper attempts to make automatic propositions on four types of lexical test questions, namely, word listening, multi-empty word selection, word sorting and single empty word selection, based on the Chinese language characteristics and the needs of second language teaching. In order to meet the needs of different language information and different difficulty of lexical knowledge, a lexical knowledge base covering all levels of information of word sounds, forms, meanings, grammar, collocations and errors is constructed in the extraction of lexical features. In the aspect of sentence feature extraction, automatic recognition of grammatical items, sentence difficulty analysis and other algorithms are realized, which provide the basis for the selection of topic dry sentences, target words and distractions in automatic propositions, and through the steps of sentence selection and chunks synthesis, etc. A total of 7 263 vocabulary test questions were produced. The manual test data showed that the preliminary attempt of automatic test of vocabulary test had achieved good results, and about 58% of the questions were evaluated as completely reasonable and adjusted by manual simple adjustment. The acceptance rate of test questions reached 75.7%.
【作者单位】：北京师范大学中文信息处理研究所;
【基金】：国家语委“十二五”科研规划项目(YB125-124)
【分类号】：H146.2;TP391.1

【相似文献】