A Python-Based Method for Translation Analysis

This paper explores a python-based data-driven method for translation quality assessment. The method depends on ratio of medium and long words, which is produced based on the Google's Trillion Word Corpus. The hypothesis of the research is that texts with high ratio of medium and long words are considered as academic and appropriate for translation of Chinese classics. The seven translations of Zhuangzi's Inner Chapters given by seven famous sinologists (including A.C. Graham, H.A. Giles, Martin Palmer, Robert Eno, Brook A. Ziporyn, Feng Youlan and Burton Watson) are used as texts to test this method. It is found that Giles' and Eno's translations are the top two best translations for they have the highest ratios of medium and long words proposed in this method. And then two other methods are provided to test the result for cross validation, i.e. the ratio of moderate and high-difficulty words and Type-Token Rate (TTR). And the two methods also produced similar results with the method of medium and long words. It can be concluded that the method of medium and long words proposed in this paper can be used for the assessment of Chinese classics as Zhuangzi's Inner Chapters.