This research project explores the use of neural machine translation (NMT) systems and large language models (LLMs) in translating literary texts. Literary works are characterized by expressiveness, idiomatic expressions, emotionally charged language, and rich cultural references, making their translation particularly challenging.

Given that literary translation is a highly time-consuming and costly process, the research team investigates whether machine translation can produce translations of sufficient quality to allow post-editing by professional translators. The aim is to facilitate and enhance the global system of intercultural literary exchange, especially for disseminating works written in lesser-used or less widely spoken languages.

Main Research Objectives

This research project aims at:

• Evaluation of Translation Quality: Evaluating the performance of Translated's adaptive ModernMT system in translating from Greek into English the flash story collection Tango in Blue Nights (2024), using both automatic and human evaluation methods.
• Investigation of Stylometric Approaches: Examining the use of stylometric methods as an innovative tool for evaluating translation quality and comparing them to conventional methods.
• Development of a Specialized Post-Editing Protocol: Designing and testing a post-editing protocol specifically tailored to the needs of literary and other creative texts. 
• Production of a Ready-to-Publish Translation: Producing an English translation of the short story collection Tango in Blue Nights (2024) by Vassilis Manoussakis, ready for publication. 

Scientific Supervisor: Associate Professor Vilelmini Sosoni.

To learn more, please visit the project's page.