Earticle

This study examines whether, when using the generative AI ChatGPT for literary translation, training it in specific habitus and strategies to construct a translator persona enables the model to move beyond its tendency toward literal translation and instead produce creative renderings aligned with that persona. To explore this, the paper focuses on Deborah Smith’s controversial translation of The Vegetarian, which sparked debate for employing a distinct habitus and feminist translation strategies that produced a work differing noticeably fromthe original. Before the habitus and strategy training, ChatGPT adhered closely to the source text in its translations; after the training, however, it generated outputs that diverged fromthe original according to the specific strategies, and these results were found not to differ significantly fromSmith’s translation. In addition, by investigating and applying prompt-design techniques, this study explores ways to employ AI as an IA tool for literary translation.

This study analyzes the evolution of AI translation technology from the perspective of the philosophy of language. AI translation has advanced to a stage where it infers context and meaning through NMT. This evolution has triggered a philosophical debate and fear regarding whether AI can encroach on the domain of pragmatics, including human intention and emotion. This debate is broadly divided into two main viewpoints: anti-AI and pro-AI. Meanwhile, Quine and Wittgenstein view meaning not as a fixed entity, but as something determined by usage, context, and communal consensus. This perspective is analogous to AI, however, and AI lacks deep consensus and social practice of a human community, thus showing limitations in translating pragmatic discourse that goes beyond the straightforward narrative level. Beyond these technical limitations, a critical challenge is that AI translation faces major hurdles: algorithmic bias, which amplifies social prejudices inherent in the training data, and linguistic homogenization, which threatens the diversity of minority languages. As a methodical solution, this article proposes defining AI translation as a distinct 'sociolect' (or 'social language') with its own unique rules and characteristics.

This study examines how close-up shots affect the use of appraisal resources in film dubbing, combining Systemic Functional Linguistics (SFL) and Visual Grammar within a multimodal translation framework. Focusing on Bong Joon-ho’sMemories of Murder (2003) and The Host (2006), the study investigates how visual proximity influences evaluative language in dubbed dialogue. Using the ELANmultimodal corpus tool, the films were segmented by shot and annotated with shot size information. Quantitative analysis shows that appraisal expressions increase markedly in close-up dubbing compared with mediumor long shots. Qualitative findings further reveal that evaluative language is strategically added to align with intensified emotional cues and facial expressivity. The results show close-up dubbing as a distinct translation practice where linguistic choices interact dynamically with visual elements. By integrating SFL-based appraisal theory and Visual Grammar, this research suggests the importance of multimodal coherence in audiovisual translation and provides an empirical foundation for future multimodal corpus-based translation studies.

This study examines the emergence and expansion of Hangul labeling on the covers of Arabic translations of Korean novels (2006–2025) within the frameworks of world-literature theory and paratextual analysis. Arabic translations of Korean literature, once rare, gained wider recognition following Please Look After Mom (2009), The Vegetarian (2016), and Han Kang’s 2024 Nobel Prize, which elevated the global prestige of Korean literature. These milestones transformed publishing practices in the Arab world, leading to the strategic use of Hangul as a marker of cultural identity and literary capital. Drawing on Casanova’s andMoretti’s world-literarymodels, Genette’s theory of the paratext, and multimodality studies, this research analyzed 46 Arabic editions. Hangul labeling, which had not appeared before 2018, was first introduced in TheWhite Book (2019) and rose to 42.4%of all titles by 2024– 2025. A survey of publishers showed that Hangul functions as a visual branding device designed to attract culturally engaged readers and strengthen global visibility. The study concludes that Hangul labeling represents a deliberate paratextual strategy that visually reflects the changing global position of Korean literature within the Arabic literary field.

This study examines the quantitative impact of fuzzymatch rates in translation memory (TM), a key function of computer-aided translation (CAT) tools, on translation productivity. In Experiment 1, the study compares CAT and non-CAT environments to assess how segment-based interfaces affect translation time and working patterns. Experiment 2 analyzes the influence of varying TMmatch rates on translation speed. The results show that while CAT environments had a limited effect on reducing overall translation time, they did affect patterns such as sentence merging and omission rates. High fuzzy match rates (above 85%) significantly reduced translation time, whereas lower match rates (60-74%) tended to increase cognitive load and editing effort. These findings offer empirical support for optimizing TM usage in CAT workflows, rate-setting practices in the translation industry, and translator training strategies. The study also identifies the need for follow-up research comparing TM-based translation with post-editing of machine translation output, and examining the effects of text genre and complexity on translation efficiency.

This study aims to investigate the impact of AI translation on undergraduate students' Russian writing and to explore ways to teach Russian writing using AI tools. This study examined the differences between texts written by learners themselves and texts generated using AI translation. Furthermore, it examined the differences in writing outcomes using AI translation depending on text genre. The analysis results are as follows: First, AI translation in Russian writing improves syntactic complexity regardless of text genre. Second, AI translation in Russian writing improves fluency regardless of text genre. Third, learners' approaches to AI translation differ depending on text genre. Fourth, AI translation does not improve grammatical accuracy in Russian writing. Fifth, AI translation does not improve vocabulary in Russian writing. Sixth, AI translation in Russian writing ismore effective in tasks involving more difficult text genres. Seventh, AI translation in Russian writing does not contribute to the completeness of the text.

This study conducts a bibliometric analysis ofmachine translation (MT) papers published in China from2017 to the first half of 2025, offering a comprehensive review of research trends and topic-specific characteristics. The results reveal a multilayered developmental trajectory: research that began with a technology-centered focus has progressively expanded to comparative error analysis, human collaboration, pedagogy, and, ultimately, translation ethics. The findings further indicate that ChineseMT research has moved beyond a purely technical-analytic phase toward an integrated research environment that links technology, education, ethics, and service. In particular, since 2023, the emergence of systems such as ChatGPT and DeepSeek has broadened “AI translation” beyond NMT-centered technical debates to encompass collaboration, education, evaluation, and ethics; moreover, generative AI is transforming not merely translation quality but the verymeaning of translational practice. Accordingly, translator agency, evaluation standards, and pedagogical orientations are being reconfigured in tandem.