Inclusive Translation in Political Discourse: Challenges and Opportunities in the Representation of Gender through AI

With an emphasis on gender representation in the German-Greek language pair, this paper uses Large Language Models (LLMs) to analyse the opportunities and difficulties of inclusive translation in political discourse. Significant cross-linguistic and cultural distinctions make it more difficult to transmit inclusive formulations that are becoming more prevalent in political discourse. Greek is still limited by its strongly gendered grammatical structure, while German has a variety of inclusive methods, including gendered doublets, the gender star, and gender-neutral neologisms. This study investigates how state-of-the-art LLMs handle these structural and cultural differences in translation, evaluating the ability of LLMs to accurately and sensitively represent inclusive discourse using a corpus of current german political texts. The results highlight the need to critically analyse AI-mediated translation in situations where the political and social aspects of language are particularly prominent.