Analyse comparative de la traduction de textes religieux de l'arabe vers le français et l'espagnol : traduction humaine versus intelligence artificielle (Chatgpt) dans l’ouvrage de Al Qarni «لا تحزن»

Translation, understood in its broadest sense, constitutes an essential tool for overcoming linguistic and communicative barriers in diverse social, cultural, and academic contexts. In recent decades, the development and widespread adoption of machine translation systems have significantly facilitated access to multilingual content, particularly in digital environments. However, despite technological advances, these systems still present notable limitations in specialized fields, such as religious texts, whose translation requires not only advanced linguistic skills but also a deep understanding of theology, culture, interpretive traditions, and the historical context in which such texts are embedded. This article aims to analyze and compare translations of Islamic religious texts from Arabic into French and Spanish, produced both by human translators and by the artificial intelligence system ChatGPT. The study adopts a comparative and qualitative approach, focusing on the evaluation of fundamental linguistic aspects, including fidelity to the source text, terminological precision, discursive fluency, grammatical accuracy, lexical adequacy, and semantic coherence. Furthermore, the particularities of Islamic religious discourse (characterized by strong symbolic, doctrinal, and cultural content) are taken into account, as they influence translation decisions. For the assessment of translations, specific linguistic criteria were applied, adapted to the type of text analyzed, to evaluate both the formal quality of the translations and their interpretative adequacy. The results indicate that, while ChatGPT demonstrates satisfactory performance in general text translation, its effectiveness significantly decreases when translating Islamic religious texts, particularly in the Arabic–Spanish language pair. The main issues identified relate to the inappropriate use of specialized lexicon and semantic interpretation errors, with a higher frequency and visibility of such errors in the Spanish translations compared to the French versions. Finally, the article offers a critical reflection on the advantages and limitations of using ChatGPT for translating religious texts, emphasizing the need for expert human intervention in contexts that require high cultural, hermeneutical, and doctrinal sensitivity.