A Multidimensional Analysis of Human and ChatGPT-Generated English Translations of Arabic Film

Abstract

This study analyzed lexico-grammatical variations between two text types: human-written and machine-generated, using Biber's multidimensional analysis. It explores the effectiveness and limitations of AI-driven translation systems in maintaining the quality of film translations. It aims to add to the current discussion on the impact of AI in the field of translation. The research methodology involves selecting films from the Middle East and collecting their translations, both human-written and generated by ChatGPT. Biber's multidimensional analysis framework analyses the translations across dimensions such as involved versus informational discourse, narrative versus non-narrative concerns, explicit versus situation-dependent, overt expression of argumentation/ persuasion, and abstract versus non-abstract discourse. The findings of the analysis reveal similarities and differences between human and ChatGPT translations. Human translations are more involved, situation-dependent, argumentative, non-abstract, and less non-narrative than the translations generated by AI. However, further improvements and refinements in AI translation models could help bridge the gap between human and AI translations. The results gained from this comparative analysis offer insight into improving AI-driven translation systems, leading to more effective cross-cultural communication through film. This research will potentially contribute to the advancement of the field of translation studies by bridging the gap between human and AI translations. It provides valuable implications for the future development of AI technologies in film translation.