A Study on Syntactic Complexity and Topic Difference Influence in Second Language Writing Based on Dependency Treebank

Using Stanford parser as a tool, taking 1840 articles on two topics by 920 English learners of four levels after random selection as the research corpus, this study attempts to explore the development trend of syntactic complexity of learners of different levels and what influence divergent topic can have on it. Results show that multi-level learners generally prefer adverbial clauses (advcl) and complement clauses (ccomp) to relative clauses (acl: recl); adjective modifiers (amod) are the best in distinguishing the level of English learners, which can be used as an important basis for judging the level of L2; in addition, different topics of the same task type have no significant effect on L2 learners’ syntactic complexity. It can be concluded that most syntactic complexity measures at the clause and phrase level show a non-linear development trend. Based upon the findings, implications are provided for English textbook development and English pedagogical practices.