Earticle

This study examines the repair strategies employed by advanced Korean language learners in oral interview performances, focusing on their role in demonstrating interactional competence. Using conversation analysis, interactions from seven TOPIK Level 6 learners were analyzed, identifying 45 instances of repair. Repairs were classified into interaction-coordinated repair, feedback-responsive repair, and request-based repair, with interactioncoordinated repair being the most frequent. The findings indicate that learners actively monitor, interpret, and revise their utterances to sustain fluency and appropriateness during real-time communication. These results suggest that repair behaviors function not only as linguistic responses but also as indicators of adaptive interactional skills. The study highlights the importance of incorporating repair as an evaluative dimension in speaking assessments and offers pedagogical implications for fostering interactional competence in Korean language education.

This study examined the substitution errors occurring among Chinese-speaking Korean learners of Korean when using the conjunctive endings '-go(-고)' and '-aso/oso(-아서/어서)'. 2,855 instances of conjunctive ending errors selected from the National Institute of Korean Language learner corpora were analyzed. First, in written language, the substitution errors between '-go(- 고)' and '-aso/oso(-아서/어서)' was found to be more complex compared to those in spoken language. Second, errors where '-aso/oso(-아서/어서)' was substituted with '-go(-고)' occurred more frequently than the reverse. Finally, errors were observed across all proficiency levels, from beginners to advanced learners, particularly in contexts related to reasons/causes and chronological order of actions.

Under an unselectively binding analysis, it has been argued that island (in)sensitivity of wh-in situ falls under the so-called Noun versus Adverb Generalization (NAG), which states that an in-situ wh-phrase is island-free iff it is (or contains) a wh-nominal (Tsai, 1994ab 1999, 2008; Stepanov & Tsai, 2008; Fujii et al., 2014). In this study, however, we show that the NAG is not sufficient to capture the island behaviors of some (non-)standard in-situ wh-phrases in Korean. Borrowing insights from Murphy (2017), we instead suggest that the island-(in)sensitivity of in-situ wh-phrases in Korean may correlate not with their semantic or categorial status, but with their base-generated positions in syntax. Specifically, we suggest that an in-situ wh-phrase that is base-generated in the CP domain is island-sensitive, while an in-situ wh-phrase that is base-generated below CP or TP is island-insensitive.

This study explores the cultural and linguistic appropriateness of ChatGPT’s translations of culturemes from the Korean historical film The Forbidden Dream (천문). While ChatGPT demonstrates strong linguistic capabilities, its reliance on transliteration and literal translation often results in subtitles that are awkward, overly formal, or culturally insensitive. Such limitations are particularly detrimental in the context of subtitle translation—a medium that demands not only linguistic accuracy but also brevity, clarity, and cultural nuance due to spatial and temporal constraints. Through a comparative analysis of ChatGPT’s output and that of a professional translator, this study evaluates the strengths and limitations of ChatGPT’s subtitle translations in conveying culturally embedded meanings. The findings suggest that post-editing ChatGPT’s subtitles and employing well-crafted prompts during the translation process can significantly enhance cultural accuracy, preserve meaning, and improve the viewer’s experience.

This study aimed to empirically analyze the impact of English learners' mindsets, enjoyment, anxiety and grit on English proficiency, as well as the roles of Willingness to Communicate (WTC) and motivation in this process. While previous research has identified the contributions of these factors, the structural causal relationships among these variables when considered simultaneously have not yet been elucidated. To address this gap, the current study measured various factors related to English learning among university students and employed structural equation modeling to analyze the interrelationships and causal effects. The findings revealed that both mindset and WTC positively influenced English proficiency, whereas anxiety related to English learning had a negative impact. Additionally, the mediation analysis of WTC indicated its significant mediating role in the effects of enjoyment, anxiety, and mindset on English proficiency improvement. Finally, the analysis of the moderating effect of motivation showed that while motivation mitigated the negative impact of anxiety on English proficiency, it did not moderate the effects of mindset and enjoyment on English learning.

This survey compares machine and human translation of Qingwen's verdict from *Dream of the Red Chamber, focusing on fidelity, metaphorical expression, and linguistic style. Translations by Papago, ChatGPT, and Hong Sang-hoon were rated by 65 advanced Korean speakers. Human translation outperformed machine versions, especially in cultural metaphors, emotion, and rhythm. ChatGPT performed best among machines, but prompt tuning had little effect. The study shows AI still struggles with literary and cultural depth.