Research Institute for Translation Studies (RITS)

Stylistic Shifts in the English Dubbing of The Last Fiction: Human and ChatGPT-Generated Translations

Document Type : Translation Studies

Authors

1 MA in Translation Studies, Department of English Language and Literature, Faculty of Foreign Languages, University of Isfahan, Isfahan, Iran

2 Associate Professor, Department of English Language and Literature, Faculty of Foreign Languages, University of Isfahan, Isfahan, Iran

Abstract
Audiovisual translation, like literary translation, requires not only transferring meaning but also preserving style. This study examines stylistic shifts in the English dubbing of the Persian animated film The Last Fiction by comparing the official human-dubbed version with two ChatGPT-generated translations using simple and detailed prompts. It identifies stylistic changes, evaluates Baker’s (1996) model, determines the most and least frequent translation strategies, and examines how prompt design affects AI-generated translations. The findings reveal that in the human-dubbed version, leveling out and normalization were the predominant strategies, indicating a strong tendency toward target-oriented language. In contrast, both ChatGPT outputs relied heavily on normalization. The results also show that the detailed prompt increases explicitation and simplification. Furthermore, the study shows that Baker’s (1996) model alone cannot account for all stylistic changes that occurred in translation. Thus, four additional strategies were offered—retention, exoticization, prosification, and deletion. The findings contribute to audiovisual translation research by extending Baker’s (1996) framework and indicating how prompt design can affect the stylistic quality of AI-assisted dubbing.

Keywords