Children’s Literature in Machine Translation: How Neural Machine Translation and Large Language Models translate features found in Children’s Literature

Abstract

The present dissertation explores the potential of Machine Translation (MT) in comprehending and accurately translating specific features found in children's literature, with a focus on the book "Φρικαντέλα, η μάγισσα που μισούσε τα κάλαντα” [Fricadella, the witch that hated Christmas carols] by Eugene Trivizas. The aforementioned text that has been chosen for this study stems from the author's recognition both nationally and internationally, as well as the abundance of linguistic features present in the book. The project aimed to determine whether MT can effectively grasp and translate features such as neologisms, wordplay, repetition, sentence length and complexity of rhyming, animal sounds, and other features present in children’s literature. The project was carried out using three different types of software. Specifically, Google Translate, DeepL, and Chat GPT from OpenAI, were employed to translate extracts containing the above-mentioned features. The methodology involved translating the selected extracts using the aforementioned MT tools followed by an analysis to assess the quality of the translations. The extracts were subjected to post-editing processes to evaluate the potential for improvement and refinement in MT outputs. Despite the potential of MT, it has faced scepticism from translators, especially in the domain of Literary Translation. However, with proper use of post-editing, MT could become an invaluable tool for translators. By identifying which features MT can effectively understand and translate, this research aimed at facilitating faster and more efficient work for translators. In conclusion, this research seeks to shed light on the capabilities of MT in handling the unique features of children's literature and its potential to support translators in their work. By understanding the strengths and limitations of MT, translators can harness its power to streamline their processes and deliver high-quality translations more effectively in the future.