AI and the translation of phraseology: A comparison between NMT and GMT

This dissertation examines the role of Artificial Intelligence (AI) in translating phraseological phenomena, with particular emphasis on idioms, collocations and proverbs, which often carry nuanced, culturally embedded meanings. The primary objective is to compare Neural Machine Translation (NMT) and Generative Machine Translation (GMT) to understand how each approach handles the linguistic challenges posed by non-literal expressions. This study aims to identify how effectively each model preserves figurative meanings and cultural nuances, with a focus on determining specific patterns in translation accuracy and contextual adaptation. Through a series of practical examples, this research analyzes the differences in performance between NMT and GMT in translating phraseological expressions, exploring their respective strengths and weaknesses in managing culturally rich expressions.