Sentence-Level Readability Assessment Based on Dependency Grammar

The research on the sentence-level readability assessment is of great significance in the field of linguistics research and language teaching. Under the guidance of the language system view, this research constructed a linguistic feature system based on dependency grammar, and its validity is verified. In the sentence-level readability task from international Chinese language education textbooks, the machine learning models of new dependency features and traditional features were constructed respectively, and the validity of the models was compared and verified. The results show that the new dependency feature system performs well and was incremental to the traditional scoring research results.