节点文献
基于标注语料库以[S][P][O]为样本的句系研究
Sentence system studies of [S][P][O] on tagged corpus
【作者】 孙道功;
【Author】 Sun Dao-gong (Chinese Language and Culture School, Nanjing Normal University, Jiangsu,210097;
【机构】 南京师范大学文学院;
【摘要】 句系是语言中句样的总和。句样是在句干基础上选择各类语气类型形成的。文章首先制定句法、语义标注规范,建立标注语料库。基于大规模语料库,以[S][P][O]为典型样本,定量统计考察句型句模的对应关系,建立句干系统。语料库中, 句型[S][P][O]共对应了30种句模类型,形成了30种句干。并根据使用频度差异,把句干分成三个层级。在句干的基础上,基于真实文本,考察句干对语气的选择情况,建立[S][P][O]对应的句样层级体系。各种句干对语气类型选择的自由度并不一致, 形成的句样类型使用频度差异很大。其中,陈述语气是选择性最高的语气类型。并根据[S][P][O的形成句系时表现,总结出现代汉语句系的特点。
【Abstract】 Sentence system is the total sentence samples in a language. Based on sentence stem, sentence samples take shapes of various tone types. The paper firstly makes out norms for syntactic and semantic marking, and sets up a tagged corpus. Based on this large scale tagged corpus, taking the sentence patterns of [S][P][O] as the typical samples, it quantitatively investigates the corresponding relation between syntactic and semantic sentence patterns of modern Chinese, and establishes a sentence-stem system. There are 30 types of semantic patterns and 30 sentence stems which correspond to[S][P][O] in corpus. According to frequency difference, sentence stems are divided into three levels. Based on sentence stems and real text, how the tone fits the sentence stem is also investigated. Hence the [S][P][O] corresponding level systems are set up. For various sentence stems, free degree which the tone fits the sentence stem is indifference, and frequency of sentence samples are also not the same. Statement tone is the easiest tone to be selected when setting up sentence samples. We briefly summarize the characteristics of sentence system of modern Chinese according to [S][P][O].
【Key words】 Sentence system; semantic pattern; sentence stem; sentence sample; tagged corpus;
- 【会议录名称】 第三届学生计算语言学研讨会论文集
- 【会议名称】第三届学生计算语言学研讨会
- 【会议时间】2006-08
- 【会议地点】中国辽宁沈阳
- 【分类号】TP391.1
- 【主办单位】中国中文信息学会