GENDER BIAS IN HUMAN AND MACHINE TRANSLATION OF LITERARY TEXTS: A COMPARATIVE STUDY OF ENGLISH AND UZBEK.

Abstract

This article investigates the manifestation of gender bias in human and machine translation of literary texts within the English-Uzbek language pair, situating the analysis at the intersection of translation studies, gender linguistics, and artificial intelligence. While existing scholarship has extensively explored gender representation in language and, more recently, bias in machine translation, comparative studies that integrate both human and algorithmic translation practices in the context of underrepresented languages remain limited. In particular, the English-Uzbek pair offers a productive site for analysis due to significant differences in grammatical structure, cultural norms, and the linguistic encoding of gender.

The study addresses this gap by adopting a qualitative comparative methodology that examines selected literary texts alongside their human and machine translations. The analysis focuses on the transformation of gendered meanings, the reproduction of stereotypes, and the interaction between linguistic structures and cultural assumptions. Special attention is given to the ways in which machine translation systems, trained on large-scale datasets, may systematize and amplify implicit gender bias, in contrast to human translators, whose choices are mediated by interpretive and contextual awareness.

The findings demonstrate that gender bias operates differently across translation modes: while human translation tends to reflect context-sensitive variation, machine translation exhibits a higher degree of regularity in reproducing stereotypical patterns. These results contribute to ongoing debates on the ethical and epistemological implications of AI in language practices and underscore the necessity of integrating gender-sensitive approaches into both translation theory and computational system design.