Korean Learning and Technology: Korean-to-English machine-translated errors

27/05/2024

Several studies have investigated the educational benefits of machine translation, specifically in the context of second-language writing. This is due to the broad accessibility and quick advancements in the quality of online machine translation tools, such as Google Translate. Many individuals learning a second language have said that they utilise machine translation as an additional tool to aid in their language acquisition process.

Recent research has compared students' self-written second language texts to written products assisted by machine translation. The results of this comparison generally indicate that the use of machine translation is advantageous. Machine translation increased the text's accuracy, fluency, and vocabulary, facilitated self-directed learning, and increased learners' metalinguistic awareness.

The majority of these works, however, focused on the ultimate product of machine translation use and did not explore how learners analyze or process the machine-translation output.

Despite the significant improvements in the accuracy of Neural Machine Translation (Nmachine translation) in recent years, errors are still present, particularly in input composed of complex embedded sentences, cultural references and idioms, and pragmatic expressions and discourse subtleties. Additionally, the quality of the translation is contingent upon the linguistic distinctions between the source language (SL) and the target language (TL), and the output of machine translations frequently necessitates post-editing or revision to achieve a satisfactory level of quality. In reality, the utilization of machine translation output requires the capacity to analyze and identify production errors and edit or revise them accordingly. However, it has been noted that specific learners, particularly those with low second language proficiency, may heavily rely on machine translation and accept its output, including its errors, without critical analysis.

These concerns are expected among learners with little ability; it was also noticed that undergraduate students specialising in translation tend to prioritise small-scale faults without considering the context. Teachers reluctant to incorporate this technology into second language classes are concerned that learners' undue dependence on and acceptance of machine translation output may hinder language acquisition by depriving them of cognitive processes involved in critical thinking.

According to research on second-language sentence processing, second-language learners frequently struggle to integrate morphosyntactic information and often depend on lexical cues. However, whether this characteristic of second-language processing also applies to the processing of L1-to-second-language translated sentences has yet to be determined. In other words, would the learners experience more difficulty identifying and resolving errors resulting from crosslinguistic differences in syntax and morphology than semantics? Given the extensive implications of second language proficiency, it is reasonable to anticipate that it will significantly determine the learners' approach to machine translation errors.

Conducting research

The study sought to enhance the reading comprehension of Korean students by using the suggested reflective e-learning strategy. The control group, consisting of 50 participants, engaged in traditional e-learning for 13 weeks. On the other hand, the experimental Korean group, including 51 participants, participated in reflective reading-based e-learning. A quasi-experimental design was used in the inquiry. The experiment evaluated reading comprehension, reflective thinking, and active participation in the English language. This included measuring the time spent reading, the number of highlighted passages, the score on a quiz, and the number of notes taken. The results indicate that the reflective reading-based e-learning strategy can enhance Korean learners' reading comprehension and reflective thinking abilities and encourage active participation. These findings might be helpful for educators in developing techniques to improve the reading comprehension skills of English students and encourage active involvement in e-learning systems.