This study examines the sentiment translatability of Kenneth Grahame’s The Wind in the Willows through comparative analysis of the English original and four Chinese translations—two by human translators and two by neural machine translation systems. By integrating the NRC emotion lexicon with the DUTIR Chinese sentiment dictionary, we construct four core metrics (polarity, intensity, density and complexity) to develop the Sentiment Fidelity Index (SFI), which quantifies how faithfully translations preserve the source text’s affective structure. Results reveal substantial differences in sentiment fidelity across translation approaches. Human translations demonstrate distinct strategies: Yang’s version achieves the highest SFI, maintaining 94% of the original’s emotional complexity and closely mirroring chapter-level emotional arcs, while Ren’s translation adopts an emotion-amplifying strategy, increasing overall sentiment intensity by 15% but reducing complexity. Both machine-translated versions exhibit systematic weakening effects, with 24% average loss in emotional complexity and tendency to neutralize strongly affective expressions. The analysis reveals a hierarchy of sentiment translatability, with joy showing the highest retention and disgust the lowest. While machine translation has advanced in syntactic accuracy, it continues to lag in conveying emotional nuance. This study broadens methodological scope in children’s literature translation research while providing theoretical grounding and empirical evidence for improving machine translation of sentiment-sensitive texts.
Liang et al. (Fri,) studied this question.