ABSTRACT
This paper aims to investigate the challenges that translators face while post-editing Machine Translation (MT) of idioms and expressions in literary works using Neural Machine Translation (NMT). Translating literary texts can be challenging for translators due to the cultural differences between the source text and the target text. However, the task becomes even more complex when translation technology, such as computer-assisted translation (CAT) tools, is utilised, which requires the translator to review and refine NMT-generated translations through post-editing (PE). The hypothesis of this study is that NMT technology lacks translation memory (TM) and a glossary of idioms and expressions, making it necessary for human translators to be involved to ensure precise meaning conveyance to the target reader. The data for this descriptive study were collected from the Iraqi novel "Murder of the Bookseller." The post-edited target text was evaluated using the Larson model, a comprehensive evaluation method that considers several aspects of translation quality. The findings of this study reveal that idioms and expressions present significant challenges during the post-editing process. Furthermore, the lack of appropriate TM and glossary in NMT technology makes it challenging for translators to ensure the accuracy and fluency of the translated text. Therefore, it is essential for human intervention to improve the quality of MT output, especially in the context of literary translations. Overall, this study sheds light on the importance of human intervention in post-editing MT for literary translations, highlighting the need for effective collaboration between human translators and NMT technology to produce high-quality translations.
Keywords: Computer-Assisted Translation, Literary Text, Machine Translation, Neural Machine Translation, Post Editing