Linguistic and Cultural Challenges in Translating Idioms with Artificial Intelligence

Abstract

AI models, such as ChatGPT4, DeepSeek, and Reverso, have transformed the translation landscape. Although these models offer rapid and accessible solutions for bridging linguistic barriers, their efficacy in capturing the nuances and cultural intricacies of languages remains a subject of debate. This study focused on critical aspects, including translation accuracy, adequacy, cultural adaptation, and the handling of linguistic nuances, in the English-Arabic context through a comprehensive comparative analysis. A corpus (idioms) of the two languages' counterparts was analyzed to unveil the disparities that existed between AI and human translations of idiomatic expressions. Findings showed that while AI has made significant strides in quickly rendering large volumes of text, a nuanced understanding of linguistic and cultural aspects is still present, and that Reverso, ChatGPT4, and DeepSeek fall short in rendering idiomatic expressions from Arabic into English. With these findings, the study contributes to a deeper understanding of the capabilities and challenges of AI tools in handling complex corpora, such as idioms, of English-Arabic language pairs, offering insights into potential improvements in AI models.