Translational Shifts in AI-Assisted Translation: A Corpus-Based Comparative Study of Arabic–English Academic Writing

This study investigates translational shifts in AI-assisted Arabic–English academic writing through a comparative corpus-based approach. A parallel corpus consisting of Arabic source texts, human translations, and AI-assisted translations was compiled and analyzed. The research focuses on key discourse-level shifts, including explicitation, nominalization, and lexical density. Findings indicate that AI-assisted translations exhibit higher levels of explicitation and nominalization, as well as increased lexical density, compared to human translations. These results suggest that AI-assisted translation tends to standardize English academic register, potentially affecting rhetorical diversity and source-text representation. Implications for translator agency, pedagogical practices, and editorial standards are discussed. The study contributes to understanding the impact of AI on academic translation and provides empirical evidence for corpus-based analysis of machine-mediated translation.