Comparative Study of Chinese Translations of “Gitanjali” based on Natural Language Processing

Abstract

This paper employs the HanLP natural language processing model to conduct a quantitative text analysis of four Chinese translations of “Gitanjali”. The analysis examines text features such as character and sentence count, lexical properties, lexical density, and high-frequency words. The results indicate that, except for Wu Yan’s translation, the text features of the other translations are relatively similar in terms of data distribution. Wu Yan’s translation has higher character and sentence count, and usage of lexical properties than the other translations, but it has the lowest lexical density, indicating a lower information density. The study also reveals that the choice of lexical properties and high-frequency words in each translation is influenced by the translator’s personal language preferences. Although the types of high-frequency words are relatively fixed, their frequencies vary significantly. Based on these findings, the study further uncovers the intrinsic logic and structural relationships among different text features, offering significant insights for the study of specific literary and translation texts, as well as broader translation and linguistic research. This paper demonstrates the effectiveness of HanLP in literary and translation research, highlighting the potential of natural language processing models in large-scale text analysis.