ABSTRACT
This chapter addresses the question: “What is translation when performed by a machine?” While Translation Studies (TS) has decades of expertise in conceptualising translation, defining the concept within machine translation (MT) research and development remains challenging. This contribution traces the evolution of translation concepts in TS, examining the discipline’s theoretical “turns” and comparing them to the translation notions embedded in different MT models, from rule-based to neural. As these models have evolved, MT has become a ubiquitous part of daily life, integrated into web browsers, smartphone apps, and social media, subtly shaping public perception to view translation as a simple, automated process not requiring human insight. By offering translation as an easily accessible, often free service, MT risks devaluing the cultural and economic value of human translation. TS can counter this trend by promoting inclusive, precise, and adaptive conceptualisations that emphasise the societal importance of human translation in transcultural communication and global knowledge exchange. Such an approach not only educates the public on the cultural and societal value of translation in general but also raises awareness of the respective strengths and limitations of various forms of translation, be it human, machine, or hybrid.
