An Intelligent Large Language Model Framework for Context-Aware Translation of Natural-Phenomenon Connotations from English into Arabic

Abstract

The role of connotations as cultural and emotional associations that extend beyond their literal meaning, can present challenges in cross-cultural translation. This study investigates whether computer-assisted translation can accurately convey the English connotations of natural phenomena in Arabic. Large language models were used in the study to achieve the best possible outcome, '' Google Translate, DeepL and Translate Modern' to translate culturally bound expressions. The findings revealed that the Large Language Models (LLMs) consistently failed to translate English culture-bound expressions appropriately into Arabic .scoring 0% cultural equivalence   showing largely dependence on literal translation and paraphrasing, which weaken of idiomatic function of cultural expressions  resonance. This  confirms aligns previous studies findings.