基于依存树库的英汉动词对比研究

发布时间：2018-08-26 19:58

【摘要】：树库是经过句法标注的语料库。树库是近年来语言研究的最新方法,使用依存语法建立的树库,直接反应了词与词之间的关系,与传统的短语结构语法相比,词与词之间的结点更少,可以方便计算机语言处理。但是国内利用树库的研究成果还很少。动词作为最重要的词类,其研究效果的好坏能在很大程度上决定语言研究的深入程度。本研究以依存语法和配价语法为基础,结合概率配价模式理论,通过建立依存树库的方法,研究了英语和汉语动词作为支配词和被支配词时依存关系的句法分布和英汉两种语言动词句法功能的异同点。还利用准确的数据,绘制了英汉动词概率配价模式图,为概率配价模式理论提供了实证依据,丰富并发展了该理论。本研究采用《新概念英语》一至四册的英语课文及其相应的汉语译文为语料,建立了一个新概念英语语料库。该语料库共包含英文4356句,合计69087个英语单词(含标点符号),与之对齐的汉语句子4356句,合计70576个中文词。语料全部经过了分词、词性标注和依存关系标注,依存关系的确定利用了斯坦福解析器的自动语法分析功能,并在完成后进行了人工校对以保证解析结果的准确性。最终录入Excel表格进行统计。研究结果显示,英语和汉语动词作为支配词和被支配词的依存关系分布既有共同点,又存在较大的差异。英汉动词从整体语法功能上来讲,能够充当的语法成分都较多。
[Abstract]:The tree database is a corpus of syntactic tagging. Tree base is the latest method of language research in recent years. The tree database based on dependency grammar directly reflects the relationship between words and words. Compared with traditional phrasal structure grammar, there are fewer nodes between words and words. It can be easily processed by computer language. However, there are few achievements in using tree banks in China. As the most important part of speech, the effect of verb research can largely determine the depth of language research. On the basis of dependency grammar and valency grammar and the theory of probabilistic valency pattern, the method of establishing dependency tree library is introduced. This paper studies the syntactic distribution of the temporal dependence of English and Chinese verbs as dominant words and the similarities and differences of syntactic functions between English and Chinese verbs. By using the accurate data, we draw the verb probabilistic valency pattern diagram, which provides an empirical basis for the theory of probabilistic valence pattern, and enriches and develops the theory. In this study, a new concept English corpus was established by using English texts and their corresponding Chinese translations from volumes 1 to 4 of New concept English. The corpus consists of 4356 English sentences, a total of 69087 English words (including punctuation marks), 4356 Chinese sentences aligned with them, and a total of 70576 Chinese words. The corpus is all processed by word segmentation, part of speech tagging and dependency tagging. The determination of dependency takes advantage of the automatic parsing function of Stanford parser, and is verified manually to ensure the accuracy of parsing results. The final input into the Excel form for statistics. The results show that the distribution of the dependency between English and Chinese verbs as the dominant word and the controlled word is not only common but also quite different. English and Chinese verbs, from the overall grammatical function, can serve as more grammatical elements.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：H314;H146

【相似文献】