abstract

This paper takes Lu Xun’s A Madman’s Diary (1918), widely recognised as the first modern Chinese short story, as its research subject. Although numerous English translations of this work have been produced over the past century, few existing studies carry out systematic stylistic comparisons across different translated versions. Selecting three English translations-the 1960 version by Yang Xianyi and Gladys Yang, the 1990 version by William Lyell, and the 2009 version by Julia Lovell-this study builds a self-compiled parallel corpus consisting of 285 aligned Chinese-English sentence pairs. It conducts quantitative comparisons through multiple corpus indicators including type-token ratio (TTR), standardised type-token ratio (STTR), lexical density, mean sentence length, dialogue proportion, expansion ratio and Flesch Reading Ease from four dimensions: lexical richness, syntactic structure, dialogue representation and translational expansion, supplemented by qualitative close textual reading to quantitatively identify stylistic divergences among the translations. The results reveal markedly distinct translational styles across the three versions. Julia Lovell’s translation achieves the highest TTR (0.442) and lexical density (0.489), featuring superior lexical diversity and textual information density. William Lyell’s translation, with a total word count of 1,933, a dialogue proportion of 40.7% and 22 exclamation marks, is the lengthiest text and adopts a dramatic, performance-oriented translating strategy. The Yangs’ translation only has a dialogue proportion of 9.4% and the highest expansion ratio of 0.708, with 16 one-to-many sentence splits; it prioritises the explicitation of narrative content while downplaying dramatic expressions. The Flesch Reading Ease scores of all three translations range from 66.9 to 67.7, indicating comparable readability. This research verifies that corpus stylistic approaches can explicitly expose translators’ choices that are imperceptible through conventional impressionistic reading, and also provides an empirically grounded analytical framework for comparing multiple retranslations of canonical literary texts.