Lexical Variation in English–Italian News Translation: A Comparative Study of Google Translate and ChatGPT-5.2

This study investigates stylistic features in English–Italian machine translation (MT) by comparing the outputs of Google Translate (GT) and ChatGPT-5.2 (GPT). The analysis adopts a quantitative corpus- based approach to studying news reports and opinion articles for which parallel translations produced by both systems were collected via API. In particular, the focus is on the lexical variety of the trans- lation equivalents of three high-frequency mental verbs: know, think, and believe. Concordance lines were analyzed and annotated according to the equivalents used, enabling a systematic comparison of lexical range across systems and the two news genres. The results show that the two systems seem to adopt similar translation solutions, but they differ in the quantitative distribution of translation equivalents. While GT exhibits greater sensitivity to morphological variation, GPT tends to rely on a more restricted set of equivalents and a wider range of phrasal and clausal reformulations. These findings contribute to studies on machine translationese (MTese) for the English–Italian language pair, with implications for the study of variation in MT and Artificial Intelligence (AI) literacy.