THE INTEGRATION OF LARGE LANGUAGE MODELS IN LITERARY TRANSLATION: ETHICAL, AESTHETIC, AND PROFESSIONAL IMPLICATIONS

Abstract

The integration of Large Language Models (LLMs) into literary translation represents one of the most significant technological disruptions in translation history, fundamentally challenging established theoretical frameworks that position literary translation as an inherently human creative practice. This study examines the ethical, aesthetic, and professional implications of LLM integration in literary translation through a systematic analysis that combines theoretical frameworks with empirical evaluations of translation quality, translator interviews, and close textual analysis. Findings reveal that contemporary LLMs, including GPT-4 and Claude Sonnet 4, achieve strong performance in baseline linguistic accuracy, scoring within 10–15% of professional human translators on semantic accuracy measures. However, significant limitations emerge in literary-specific dimensions, with LLM translations scoring 30–40% lower than human translations in preserving authorial voice, managing register variation, and handling figurative language, wordplay, and cultural specificity. Translators report productivity gains of 30–50% when using LLMs for initial draft generation, yet express concerns about professional identity, skill atrophy, and economic pressures driven by declining per-word rates. Ethical analysis identifies contested domains including authorship and attribution, cultural representation bias, environmental costs of AI deployment, and equitable compensation for translators whose creative labour contributed to LLM training data. Aesthetic evaluation demonstrates that while LLMs produce linguistically competent translations, they frequently lack the stylistic vitality, rhythmic sophistication, and affective resonance that characterise distinguished literary translation. The study concludes that optimal LLM integration requires frameworks explicitly designed for human-AI collaboration, positioning AI as a productivity tool while preserving essential human roles in cultural mediation, interpretive judgment, and aesthetic decision-making. Professional associations, educational institutions, and technology developers must collaboratively establish ethical standards, training curricula, and economic models that sustain literary translation as a vital humanistic practice in an increasingly automated landscape.