Parallel Dependency Treebank Based Contrastive Study Of English And Chinese Verbs

Treebanks are corpus with syntactic annotation which is an updated approach of language study. Based on dependency grammar, treebank can reflect the interrelationship of individual word. Comparing to the pervading phrase-structure approach, nodes between words are much less. Therefore, treebanks based on dependency grammar is more compatible to computational natural language process. But the present Chinese language study field faces a scarcity of researches incorporating treebanks.The study of verbs is always the most important part in studies concerning part-of-speech. Few studies of English and Chinese verbs and contrastive studies about them are accomplished by applying dependency grammar. This research takes dependency grammar and valency grammar as theoretical bases, integrating PVP theory, investigates distribution of dependency relations of English and Chinese verbs as governor and dependent and clarifies similarities and differences between English and Chinese verbs.Using accurate statistics, PVP graphs of English and Chinese verbs playing as governor and dependent are drew, which illustrate strong binding force of verbs. Furthermore, accurate data are generated and applied to the drawing of PVP graphs which is an improvement of PVP theory.Based on dependency grammar and valency grammar, incorporated with PVP theory, this research aims to make a comparison between English and Chinese verbs by using a parallel dependency treebank.A parallel treebank with the text of New Concept English as raw material is established. Taking the English text and its correspondent Chinese translation of New Concept English Volume1to4as raw materials, the treebank contains4356English sentences, counting to69087English words (punctuations included) and4356aligned Chinese sentences, amounting to70576Chinese words (punctuations included). All English and Chinese words are segmented, tagged with part-of-speech and labeled with dependency relations. The establishment of dependency relations is complemented firstly by Stanford Parser, which can tackle with both English and Chinese sentences automatically. Then these results are revised manually. The proved qualified results are typed into Excel to extract information about the syntactic distribution of English and Chinese verbs and the differences between them.It is indicated by the ultimate findings that for both English and Chinese the grammatical function of governor and dependent are dissimilar. Some grammatical functions are commonly shared by these two languages. With regard to verbs as a whole, for English and Chinese, the functions verbs can serve are relatively the same.