Exploring user perspectives: Student evaluation of the evolution in neural machine translation from English to Korean and its implications

Since the advent of Google NMT in 2016, human translators have been overwhelmed by the concern about being replaced by machine translation. Although professional translators argue that the machine translation output is not refined enough to surpass human translators, their claims are sometimes emotional and based on incorrect perceptions, without verification and substantiation from user evaluation and specific quality evaluation data. Therefore, this study examines the evolution of NMT output from English to Korean diachronically and provides specific user evaluation data that can verify and substantiate the claims. Despite the steady improvement in NMT performance observed in recent years, it has been recognized that there is still a significant gap that must be bridged for NMT to achieve parity with human translation. Nevertheless, as the collaboration between NMT and human translators is expected to increase, it is essential to reinforce relevant systems, such as technological and legal support systems.