Exploring human vs. AI-powered translation to metonymic expressions: A case study of the Holy Quran

Translating metonymic expressions presents a significant challenge for translators, as it requires maintaining both figurative meaning and accurate reference. The study compares AI vs. human translation using Larson's (1998) model as the analytical framework. The human translation is represented by Saheeh International, and AI-powered translation is represented by Monica application in rendering metonymic expressions in the Holy Quran. Using Larson's (1998) three-way model for metonymic translations, the study analyzes a corpus of 37 metonymic expressions of the Holy Quran as compiled from Islam Web. The findings indicate that both Saheeh International and AI translation predominantly rely on non-figurative language when rendering metonymic expressions, accounting for 44.1 and 59.9 %, respectively. While Saheeh International demonstrates a slightly greater tendency to preserve metonymy in selected instances or employ explicit equivalents, the AI translation exhibits a lower frequency of such strategies. The results also showed a statistically significant difference between the two translation methods, with Saheeh International achieving a mean score of 1.32 compared to 1.10 for AI-powered translation with a P-value of .036. These findings underscore the complexity of translating metonymic expressions in the Holy Quran, highlighting the intricate balance required between preserving the original figurative meaning and ensuring semantic clarity. The study suggests that while AI translation continues to advance, human translators exhibit a nuanced ability to navigate metonymic shifts, making their role essential in preserving the rhetorical depth of sacred texts.