Evaluating Students’ Proficiency and Challenges in Post-Editing Idiomatic Expressions in Literary Texts for Yandex Neural Machine Translation

Abstract

The use of machine translation (MT) has become a common practice among translators to produce fast and efficient translations. However, significant challenges persist when MT is applied to literary texts, which are often rich in sarcasm, metaphor, irony, idioms, and other forms of linguistic ambiguity. This study aims to evaluate students’ ability to perform post-editing on literary texts translated by Yandex Neural Machine Translation (NMT). The research subjects were students from the English Education Study Program at the University of Mataram who enrolled in the “Translation and Interpreting” course during the fifth semester of the 2021/2022 academic year. Data were obtained from a post-editing assignment based on the short story The Theft of the Mona Lisa, completed by 30 students. Additional data were gathered through direct observation during the learning process. The analysis revealed that the post-edited texts produced by students were generally acceptable in quality. Although minor errors related to less relevant information were present, the translations were largely comprehensible to readers. These findings support the argument that careful post-editing can significantly enhance the quality of literary translations generated by Yandex NMT. The greater the effort and time invested in the post-editing process, the higher the resulting translation quality.