المدونة
Volume 12, Numéro 1, Pages 517-540
2025-06-20
On The Assessment Of Chatgpt-4 Translation Accuracy Between Arabic And English With Special Reference To Polysemous And Culture-bound Items
Authors : Benyahia Wissam .
Abstract
The underscored advancement of Open AI’s large language model (ChatGPT) has earned immense attention in the field of translation. By assessing ChatGPT-4 translation performance, this study discusses the translation accuracy between the distant language pair of Arabic and English and probes the strengths and challenges raised by the distinctiveness of translating polysemous words and culture-bound references based on a human user experience evaluation. The study involves samples from various text typologies to maximize the validity of the findings. Subsequently, in the course of evaluation and quality assessment, the results indicate that ChatGPT-4 demonstrates high accuracy when translating polysemy in various contexts and resorts to literal translation when rendering a few cultural references such as idiomatic expressions and proverbs. The study suggests further fine-tuned training to enhance translation consistency and performance, particularly in regards to unrelated language pairs such as Arabic and English to assure language transfer accuracy and offer insights for more extensive research on other complex issues of language.
Keywords
Translation Accuracy ; ChatGPT-4 ; Polysemy ; Cultural References ; Translation Quality Assessment
Les articles similaires
Keji Felix Faniran
.
pages 185-198.
Benyahia Wissam
.
pages 139-155.
Benzehra Radia
.
pages 31-43.
Lahlou Hassina
.
Boucharif Nabila
.
pages 228-240.
