Artificial intelligence accuracy in translating resistance literature from Arabic into English: Google Translate and ChatGPT as a model

Abstract

Resistance literature serves as a repository for recounting the tragic experiences of injustice and persecution faced by individuals or communities, intertwining themes of resistance, freedom, and identity with profound emotional depth. This study scrutinizes the distortions inherent in translating such literature using AI, particularly Google Translate and ChatGPT, employing Berman’s twelve tendencies as a framework. Focused on the short story “Until We Return,” translated from Arabic to English, the research illuminates how machine-based translations, lacking nuanced understanding and context, may compromise the essence of resistance literature. The study unveils that both Google Translate and ChatGPT significantly alter the original text, demonstrating prevalent tendencies such as rationalization (36% in both), qualitative impoverishment (32% for Google Translate, 26% for ChatGPT), and destruction of rhythms (10% for Google Translate, 11% for ChatGPT). Additionally, both platforms exhibit quantitative impoverishment, destruction of linguistic patterns (7% each), and expansion (5% for Google Translate, 8% for ChatGPT). The lowest percentages are attributed to clarification (2% for both), with Google Translate lacking the destruction of underlying networks of signification and ChatGPT registering it at 3%. Ennoblement remains minimal (0% for Google Translate, 1% for ChatGPT). Conclusively, the study underscores the imperative for further exploration of alternative translation techniques for resistance literature. It emphasizes the paramount importance of preserving the integrity of the original text to ensure accurate representation and homage to the voices of the oppressed and marginalized. Prioritizing fidelity in translation is essential to safeguarding the essence of resistance literature amid technological advancements.