Earticle

This study examines the potential and limitations of generative AI in subtitle translation from the perspective of linguistic economy. It compares the official Korean subtitles of the Russian film <Evil Boy> with ChatGPT-generated translations produced under general and subtitle-specific prompts. A total of 663 subtitle segments were analyzed in terms of subtitle length and linguistic economy strategies. The results show that subtitle-specific prompts produced significantly shorter translations than general prompts, indicating that prompt design can improve linguistic economy to some extent. However, AI-generated subtitles remained significantly longer than the official subtitles and tended to retain the source-language sentence structure. Context-dependent strategies such as omission, sentence integration, and reduction of repetitive expressions were also less frequently observed in AI-generated subtitles. These findings suggest that generative AI has potential as a tool for subtitle drafting, but human intervention remains necessary to ensure readability, naturalness, and linguistic economy. This study provides empirical evidence of both the potential and limitations of generative AI in audiovisual translation.

This study examines the limitations and applicability of machine translation (MT) in musical theatre lyric translation through comparison with a human translation. Drawing on domestic and international lyric translation principles, an analytical framework was developed and applied to 37 numbers from 19 English-to-Korean licensed musicals ranked among the KOPIS top 10 from 2015 to 2024, using Google Translate and Papago. Results show that human translation largely adheres to formal principles such as rhythm, accent, and rhyme, and accurately reflects sociocultural and intra textual context. MT, by contrast, fails to meet formal requirements and often cannot convey contextual meaning. However, MT's tendency to preserve the literal meaning and form of the source text yields results comparable to human translation in lyrics without implicit contextual meaning, suggesting potential applicability. Given the collaborative, multi-stage nature of musical translation, post-editing MT outputs through formal adjustments and content-level revisions reflecting sociocultural and intra textual context could enhance both translation efficiency and output quality.

This study examines instructor-mediated stepwise translation activities in an AI-based undergraduate Chinese-to-Korean translation course. The course treated AI outputs as materials for comparison, discussion, and translation decisions rather than as final translations. Each three-session unit was organized around a pair of difficulty-graded texts on the same topic and two activities: AI-assisted revision of learner translations and post-editing of AI-generated draft translations. The data consist of course materials, survey responses from 15 learners, and selected student assignments. The analysis shows that the graded text pair scaffolded source-text comprehension and AI-output evaluation. Learners perceived AI-assisted revision of their own translations as more helpful for learning, while they viewed post-edited AI draft translations as more complete final products. The comparison between lecture plans and student outputs shows how instructor mediation was reflected in learners' prompt use and translation decisions. This study shows that AI-based translation instruction gains educational value when AI use is embedded in instructor-guided activities.

This study examines the translation and transcription of the Russian proper nouns ‘Pushkin’ and ‘Vladivostok’ in Korean news articles. Although a standardized system for Russian loanword transcription was established by the National Institute of the Korean Language in 2005, the same proper nouns continue to be rendered in multiple ways in the media. Rather than treating such variation as simple deviation from the norm, this study considers the rendering of proper nouns as a form of translation strategy and investigates how they are actually used in news articles. The results show that standardized forms are widely used and tend to dominate in terms of frequency, while non-standard variants continue to be used in specific contexts. These findings suggest that the translation and transcription of Russian proper nouns cannot be evaluated solely in terms of adherence to prescriptive norms. By analyzing large-scale data from news articles, this study expands the scope of research on proper noun translation and provides a basis for examining the gap between normative standards and actual language use.

This study examines the potential of large language models (LLMs) as translation assessment tools in graduate-level translator education by comparing human scores with those from ChatGPT-4o and Gemini 3.5. Utilizing a shared rubric, the study analyzes English-Korean and Korean-English translations produced by ten graduate students. Statistical analysis revealed strong, significant correlations between human and LLM mean scores in both directions (r = .785 for English-Korean; r = .863 for Korean-English). However, analysis by rubric category revealed significant differences in the terminology category, whereas content, syntax, and vocabulary & register showed none. Qualitative analysis suggests that LLMs tended to penalize proper-name, title, and spelling errors more strictly, while human assessors relied more on holistic judgment considering the broader translation and assessment context. The results suggest that LLMs can support preliminary scoring, formative feedback, and learner self-review, provided that explicit rubrics and carefully designed prompts are used with human oversight.

A significant number of undergraduate T&I graduates do not work directly as interpreters or translators. Instead, they are employed in other fields where they can apply what they’ve learned in T&I. This suggests that they gained transferable competences through their major, and that these competences help them in their current jobs. To explore this, the researcher surveyed 17 faculty members and 131 graduates. The survey had two aims: to identify (1) which transferable competences graduates gained through their major, and (2) which other disciplines create the greatest synergy when combined with T&I. Graduates most often named productive English skills—such as presentations and writing in English—together with communication skills. Both faculty and graduates saw the strongest synergy when T&I was combined with business administration and then language technology. The findings also point to different needs for different career paths. This study is one of the few to survey undergraduate T&I graduates, with a focus on those now working in fields where they apply their transferable competences acquired from majoring in T&I. It contributes to our understanding of transferable competences in T&I and of interdisciplinary education.

This study compares the translation strategies used for culture-specific items in Korean fansubs and official subtitles of the Japanese anime The White Mage Who Was Banished From the Hero's Party is Picked Up by an S-Rank Adventurer: This White Mage is Too Out of the Ordinary! The analysis covers Episodes 1 to 10, focusing on honorifics, fantasy genre terms, magic-related expressions, wordplay, annotations, Chinese-character glosses, and metaphors. From aligned subtitle data, 919 candidate culture-specific items were extracted, and 310 cases were coded by item type and translation strategy. The findings suggest that fansubs relatively often use literal translation, borrowing, annotations, foreignizing strategies, and genre-specific expression preservation. Official subtitles more often employ semantic translation, generalization, domestication/localization, explicitation, and restructuring to improve readability. However, both subtitle types often share preservation-oriented strategies for familiar genre terms. The study interprets these differences through professional norms and community-generated norms, not as a hierarchy of translation quality.

This study examines a training course for early-career, part-time Korean-English interpreter trainers who may face challenges as educators due to limited knowledge of interpreting studies and pedagogy despite their professional interpreting experience. To address this issue, an eight-hour training course consisting of three sessions was developed and conducted for five interpreter trainers, aiming to enhance their teaching competencies. Efforts were also made to foster a professional learning community among participants. For data collection, two rounds of written interviews were conducted, one immediately after the course and the other after one semester, to explore what aspects of the training the participants found helpful and how they attempted to incorporate the content of the training course into their teaching practices. Results show that participants actively incorporated various theories and pedagogical approaches in their classes. Notably, the participants reported significant improvements in feedback practices, along with increased confidence and self-efficacy as interpreter trainers. Overall, the training course was found to have positive impact on the participants’ competencies as interpreter trainers.

This paper examines the Korea Coast Guard (KCG) interpreter training programs implemented between 2023 and 2025, which combined Korean-language theoretical instruction with language-specific interpreting practice. The post-training surveys showed that the interpreter trainees were highly satisfied with the training overall. However, the findings also identified several areas requiring further attention. Most trainees came from non-Korean-speaking backgrounds, underscoring the need for greater support in Korean language proficiency. Given the complexity of technical terminology and seafaring jargon, interpreters also require ongoing professional development to strengthen their domain-specific knowledge and interpreting skills. Furthermore, trainees consistently rated their interpreting competence more positively than did the trainers, suggesting the need for more systematic feedback and additional opportunities for skills development. Taken together, these findings indicate that interpreter training should extend beyond a one-time program. The study therefore argues that KCG should adopt a long-term approach to interpreters’ professional development and concludes by offering recommendations for future training programs.

This study examines how Korean-English freelance interpreters integrate ChatGPT into their preparation process and how such use reshapes traditional preparation practices. Based on semi-structured, in-depth interviews with six freelance interpreters, the study analyzes the practical uses, perceived benefits, and risk-management strategies associated with ChatGPT-based preparation. The findings show that ChatGPT is mainly used for glossary generation and anticipated script creation. The study also finds that ChatGPT contributes to greater preparation efficiency by reducing time spent on interpreting preparation. It also helps to ease psychological burden by providing interpreters with a sense of preparedness and increased predictability. Meanwhile, most interpreters were aware of potential problems driven by ChatGPT, and they performed cross-checking selectively depending on time constraints, information sensitivity, and the importance of the term or content. The study concludes that ChatGPT does not replace traditional preparation but functions as a cognitive support tool that restructures parts of the process. It further suggests that AI literacy, critical verification, and responsible use of AI tools should be systematically addressed in interpreter training.