Preserving poetic meaning in translation: A comparative study of human and machine translation

Abstract

This study aims to evaluate the extent to which poetry can be translated while preserving its salient features, including metaphor, diction, structure, and tone, through a comparative analysis of Human Translation (HT) and Machine Translation (MT), focusing on two widely used MT tools: ChatGPT and Google Translate. Denise Levertov’s Listening to Distant Guns is selected as the sample text, with Samer Abu Hawash’s (2009) Arabic translation serving as the HT benchmark. The study adopts a qualitative approach, drawing on Newmark’s translation theory as its theoretical framework. The findings revealed that HT is more effective in poetry translation due to its ability to capture emotional nuance and subtle shades of meaning associated with figurative language. While HT showed frequent uses of nominalization to highlight the focus of readability, the MT output showed monotony of style, following a literal approach with consistency in tone and grammar. Google Translate proved more accurate in its rendition whereas ChatGPT produced more flowing and smoother output. However, both MT tools fell short of capturing the subtleties and metaphoric overtones inherent in poetic discourse. The study concludes that HT outperforms MT in poetry translation due to machines’ limited sensitivity to stylistic nuance, figurative language, and affective meaning. The study recommends using a hybrid approach that combines human expertise with machine technology. While human translators ensure cultural and stylistic adequacy, MT provides preliminary drafts that require substantial human refinement.