A Corpus-Based Study of Metaphorical Uses of Language in English and Italian Popular Science Articles on Artificial Intelligence

The development and widespread use of Artificial Intelligence following the launch of ChatGPT in the fall of 2022 have fostered extensive media coverage of AI. Popular science magazines have played a significant role in communicating this topic to the general public, often relying on figurative language. In this context, metaphors constitute an effective strategy for making a complex and emerging technology such as AI more accessible to lay audiences by mapping familiar domains onto an unfamiliar one. Moreover, from a cognitive perspective, metaphors are more than stylistic or explanatory tools: they contribute to shaping the representation and interpretation of AI, as they may convey evaluative, ideological, personal and context-dependent perspectives. Despite the growing body of research on AI metaphors, little attention has been given to their translation in English-Italian popular science discourse. The present study addresses this gap by investigating the metaphorical representation of AI through a specialised English-Italian parallel corpus, consisting of English popular science articles published in Scientific American, Nature and Quanta Magazine and their Italian translations published in Le Scienze. Building on Conceptual Metaphor Theory and Systemic Functional Grammar, the research identifies and analyses the linguistic manifestations of underlying conceptual metaphors and examines how their lexico-grammatical framing is preserved or reshaped in translation. The findings indicate that anthropomorphism is a recurring feature of AI discourse, with personification representing its main realisation in both titles and article bodies. In the titles, personification is slightly less frequent in the Italian translations. Conceptual metaphors are generally preserved, with occasional shifts, removals and additions. The analysis of the article bodies reveals a gradual shift from operational to more intentional and socially oriented forms of AI agency.