Human vs. ChatGPT’s translation strategies: A comparative study of religious terms in children’s literature from English into Arabic

Abstract

Largely because of the emergence of neural machine translation, which has produced notable gains in translation quality, the field of machine translation has advanced quickly in recent years. At the same time, artificial intelligence has shown amazing advancements; its uses now include translation assistance, which draws great attention in many different sectors. As culture is one of the main problems faces the translators, religious terms and expression need to be carefully translated as they are very critical and sensitive. This paper compares the performance of GPT-3 with that of human translators by examining the techniques used by GPT-3 in translating religious and culturally specific items. Religious terms are taken from three children’s stories addressing particular cultural concerns for this purpose. The translated religious terms and expressions extracted from the Arabic version for both professional human translator and GPT-3 have been analyzed and examined. The study categorizes the chosen items using established theoretical frameworks based on translation strategies: direct translation, retention, specification, generalization, substitution, and omission. The translation performance of GPT-3 is compared with that of the human translator using a descriptive analysis technique. The results indicate that GPT-3 appears to rely on a limited number of strategies, whereas the human translator employs a greater variety of techniques.