Preserving Poetic Effect in Human–Machine Collaborative Translation of Mo Yan’s Red Sorghum Family

Abstract

The swift progress of artificial intelligence, exemplified by neural network models like GPT-4o, transcends literary boundaries, heralding novel human-machine collaborative models. The challenge of achieving translation efficiency without compromising poetic effect persists. Machine translation often overlooks the cultural imagery, rhetorical nuances, and emotional connotations that are essential to a literary text’s poetic appeal. Thus, this study assesses the efficacy of the collaborative translation mechanism in overcoming limitations, employing Mo Yan’s Red Sorghum Family as a case study. A corpus of one hundred high-poetic-density sentence segments was selected for a comparative analysis across three translation modalities: machine translation, human-machine collaboration, and Goldblatt’s authoritative translation. Quantitative methods and case studies are utilised in the study to demonstrate the inclination of machine-generated text to flatten imagery, weaken rhetorical effects, and disrupt emotional continuity. The paper presents a collaborative approach in which machines process semantics and flag cultural terms. Human translators then intervene with cultural compensation, rhetorical correction, and emotional adaptation. The model significantly reduces the gap with professional translations, providing practical guidelines for translating rural Chinese literature in the AI era.