Barreiros, Ana Teresa Marques Ivo Gaspar2026-03-272026-03-272026-02-03http://hdl.handle.net/10400.5/117768Mestrado em Tradução.Este trabalho tem como objetivo analisar de que forma a inteligência artificial consegue desempenhar tarefas de um tradutor literário. Para o desenvolvimento da experiência levada a cabo neste projeto utilizámos seis excertos da obra O Monte dos Vendavais, de Emily Brontë, e a respetiva tradução humana editada em Portugal como tradução de referência. O primeiro capítulo descreve a história da inteligência artificial até aos dias de hoje, passando pela análise de conceitos como large language models, aprendizagem automática e de que forma a memória nestes sistemas contextualiza as respostas dadas. De seguida apresentamos as duas ferramentas utilizadas no nosso projeto: o Google Translate, uma ferramenta de tradução automática, e o Gemini, um large language model. O segundo capítulo aborda a obra que foi objeto de estudo, começando por apresentar uma breve biografia do percurso pessoal e literário de Emily Brontë e analisando a ligação entre a autora e a localização onde viveu grande parte da sua vida que se encontra visível na obra. Também contextualizamos a obra no espaço vitoriano. Partimos depois para a definição de tradução literária, o papel das ferramentas de TAC neste meio, nomeadamente a janela de oportunidades que oferecem. O terceiro capítulo é dedicado à metodologia usada para desenvolver a experiência, onde apresentamos a área de engenharia dos prompts, bem como aspetos da teoria da tradução em particular no que respeita à visibilidade do tradutor, do seu papel no mundo e que estratégias utilizaremos para analisar os excertos. O quarto capítulo é inteiramente dedicado à análise dos resultados produzidos pelo Gemini, utilizando o zero-shot prompting, o one-shot prompting e o few-shot prompting, bem como ao realizar a pós-edição da tradução gerada pelo Google Translate.This study aims to analyse how artificial intelligence can perform the tasks of a literary translator. For the development of the experiment carried out in this project, we selected six excerpts from Wuthering Heights by Emily Brontë, along with the corresponding human translation in Portugal as the reference translation. The first chapter describes the history of artificial intelligence up to the present day, including an analysis of concepts such as large language models, machine learning, and the role of memory in contextualising responses within these systems. We then introduce the two tools used in our project: Google Translate, a machine translation tool, and Gemini, a large language model. The second chapter focuses on the literary work under analysis, beginning with a brief overview of Emily Brontë’s personal and literary path and analysing the connection between the author and the place where she lived most of her life, which is reflected in the novel. We also place the work within the Victorian era. We then move on to the definition of literary translation, the role of CAT tools in this domain and, in particular, the window of opportunity they offer. The third chapter is dedicated to the methodology used, where we present the field of prompt engineering, as well as aspects of translation theory, particularly regarding the translator’s visibility, their role in the world, and the strategies we will use to analyse the excerpts. The fourth chapter is entirely devoted to analysing the results produced by Gemini, using zero-shot prompting, one-shot prompting and few-shot prompting, as well as post-editing the translation generated by Google Translate.application/pdfporA Inteligência Artificial e a tradução literária : testando os limites da máquinamaster thesis204194601