Research on Readability Grade Formula Based on HSK Compositions
Yongquan Li · 2022 4th International Conference on Natural Language Processing (ICNLP) · 2022
The Chinese language proficiency test (Hanyu Shuiping Kaoshi, HSK) is an international Chinese proficiency standardized test. Composition research has always been a hotspot in HSK research. However, the readability of HSK compositions has not been systematically researched. In this paper, 5498 compositions were selected from the HSK examination papers over the years, and then rated by experts according to the differences of readability grades. A standard corpus of HSK compositions marked by experts was constructed. By means of multiple linear regression, the readability grade formula of HSK compositions was derived. The formula shows that the predictive ability of vocabulary features is stronger than that of character features, and the predictive abilities of Chinese character writing features, word length features, syntactic features and emotional features are poor.