The research project examines machine-translated language, or machine translationese (MTese) (Daems et al. 2017) in news translation from English into Italian. While MTese has been widely studied in neural machine translation (NMT), the linguistic features produced by large language models (LLMs) have only recently garnered scholarly attention (Kong & Macken, 2025), and the English-Italian pair remains largely unexplored. The project compares MTese produced by NMT and LLMs across news genres and examines how raw machine-translated output differs from comparable original Italian news, investigating potential tendencies towards standardisation. As AI is increasingly adopted in newsrooms for the circulation of multilingual news information, this concern is particularly relevant if translations circulate as publishable products available to both professionals (journalists, editors, journalist-translators) and lay readers, and post-editing is conducted with limited awareness of translation issues. The project therefore provides empirical data on the features of MTese with implications for translator training, post-editing, computational modelling, and AI literacy.
This is a joint Digital Humanities PhD project with the universities of Genoa and Turin.