하명정 Myung-jeong Ha

DOI:10.15702/mall.2025.28.4.1 Vol.28(No.4) 1-24, 2025

This study explored how to design and apply a blended learning model for English presentation classes using two metaverse platforms, ZEP and CoSpaces. The research examined how students’ prior interest in the metaverse and their digital skills affected their learning outcomes. Twenty-seven university students took part in a 15-week blended course. ZEP was used for real-time interaction and group work, while CoSpaces supported 3D content creation and virtual presentations. Data were gathered through online surveys before and after the course, including both quantitative items and open-ended questions. Descriptive and correlational analyses were conducted with SPSS 20.0, and qualitative responses were analyzed using thematic analysis. Overall learning experiences were highly positive (M = 3.94, SD = 0.96), with course interest (M = 4.22, SD = 0.85) and learning comprehension via blended learning (M = 4.22, SD = 0.76) also highly positive. Prior interest in the metaverse among learners was positively correlated with aspects of interest in the course (r = .53, p < .05), motivation to learn (r = .41, p < .05), and presentation confidence (r = .39, p < .05). Digital device proficiency was highly positively correlated with all aspects of metaverse literacy, including proficiency with ZEP functionality (r = .61, p < .01), design ability in CoSpaces (r = .47, p < .05), and virtual presentation (r = .58, p < .01). Qualitative findings indicated three main positive outcomes: decreased anxiety and enhanced confidence (44.1%), innovative learning experiences (29.4%), and increased engagement and enjoyment (20.6%). Overall, the results suggest that prior metaverse interest relates more to affective outcomes related to interest, motivation, and confidence, while digital device proficiency has a stronger connection to cognitive outcomes of metaverse literacy. At the same time, there remains an opportunity to develop differentiated instructional support that accounts for individual differences among learners. Furthermore, it recommends pre-course orientation and mentoring to address digital disparities and ensure effective blended learning design.

metaverse-based learning, blended learning, English presentation, learner characteristics, metaverse literacy

Myunghwan Hwang , Hee-kyung Lee

DOI:10.15702/mall.2025.28.4.25 Vol.28(No.4) 25-52, 2025

This study investigates how adult EFL learners interact with ChatGPT during writing revision and examines how these interactions affect the linguistic features of their revised texts. Using a case study approach, the research collected selfintroductory essays written before and after revision from four adult EFL learners, along with reflection logs detailing their purposes for using ChatGPT and the prompts they used. Linguistic features were analyzed using text analysis tools within the Complexity, Accuracy, and Fluency framework. The Wilcoxon signed-rank test assessed differences, and matched-pairs rank biserial correlation measured effect sizes. Thematic analysis was applied to explore participants’ purposes and prompt usage patterns. Five major findings emerged: (1) learners mainly used ChatGPT for lexical and syntactic corrections, rarely for content, style, or structural development; (2) prompts were generally simple, lacking contextual depth; (3) the use of vague autocomplete prompts suggested a personified view of ChatGPT; (4) multi-turn interactions were limited, even when the responses were unsatisfactory; and (5) while ChatGPT supported micro-level improvements in grammar, lexical variety, and readability, its impact on content elaboration and the overall length of the text was minimal. These findings highlight both the instructional potential and the constraints of using ChatGPT as a revision aid in second language writing.

ChatGPT, learner-ChatGPT interaction, prompt engineering, linguistic features, ChatGPT usage purposes

Robert C. Kerr

DOI:10.15702/mall.2025.28.4.53 Vol.28(No.4) 53-72, 2025

This study aims to understand how the use of ChatGPT for English language learning affects learners’ attention to language as measured through engagement with language (EWL). EWL is significant as it is attention to language and is understood to be the application of explicit language knowledge and the process that updates this knowledge. To accomplish this, a systematic review and qualitative analysis of research exploring ChatGPT and its effects on EWL were conducted. Recent empirical studies were selected, and the reported data were coded to identify evidence of EWL states and processes across its three subdimensions of affective, cognitive, and social EWL. The findings show that ChatGPT use does offer affordances for EWL. ChatGPT encouraged affective EWL as it provides motivation, reduced anxiety, and a sustained willingness to engage with language. Cognitive EWL was elicited through heightened attention to linguistic form, reflective problemsolving, and critical appraisal of AI-generated feedback. Social EWL was facilitated as learners initiate and maintain dialogic interactions with the model, negotiate meaning, and exercise agency in shaping the direction of exchanges. The study concludes with pedagogical implications and recommendations for future research.

ChatGPT, EWL, language learning, affordances, engagement with language

Mee-jee Kim

DOI:10.15702/mall.2025.28.4.73 Vol.28(No.4) 73-94, 2025

The integration of educational technology (EdTech) into language education has garnered increasing scholarly interest in recent years. Despite the central role of instructional materials in language education, systematic approaches to incorporating technology into these materials remain underexplored. In response to this need, the present study developed EdTech-integrated instructional materials for English education. First, the study analyzed current teaching practices and the integration of technology as reflected in Korea’s 2022 revised English curriculum. Based on this analysis, a framework for classifying EdTech application scenarios was proposed, along with a separate framework for categorizing existing EdTech products. Utilizing these frameworks, four instructional materials were designed, each consisting of a detailed lesson plan and practical examples of EdTech tool use. These materials were subsequently evaluated by 20 English education experts using a 5-point Likert-scale survey accompanied by qualitative feedback, which assessed subject relevance, curricular alignment, content validity, clarity, and usability. The findings offer practical insights into how EdTech can be effectively integrated into English teaching and learning, thereby contributing to the broader goal of fostering classroom innovation in the digital age.

Educational Technology, EdTech, teaching and learning materials, technology integration framework, classroom innovation, instructional material development

Sangmin-michelle Lee , Junseong Bang

DOI:10.15702/mall.2025.28.4.95 Vol.28(No.4) 95-114, 2025

Artificial intelligence (AI) has the potential to greatly facilitate student learning by providing personalized instruction tailored to each student’s individual needs and knowledge level. Although personalized learning is widely considered as one of the most effective teaching methods, it has been difficult to implement in traditional classrooms. In this context, AI offers a viable alternative. Various AI tools now perform key pedagogical functions, such as diagnosing learners’ knowledge levels, recommending content and learning paths, and providing feedback and personalized learning based on learners’ profiles. Since these tools differ in their capabilities and personalize learning in different ways, they should be applied differently according to pedagogical purposes and learner variables when implemented in educational settings. This study introduces five stages of AI-powered personalization grounded in learning theories. These stages take into account various factors, including technological capabilities, learner characteristics, and curricular structures at the elementary, secondary, and college levels. We further illustrate how these five stages can be applied to language education at different school levels. Finally, we present the pedagogical implications of using AI in education and address the challenges that must be considered for its effective and equitable implementation.

Artificial intelligence, personalized learning, affective computing, AI mentor, language learning

이혜진 Hyejin Lee , 이도원 Dowon Lee

DOI:10.15702/mall.2025.28.4.115 Vol.28(No.4) 115-142, 2025

This study develops ACME, a comprehensive AI-based career support platform built through vibe coding. The ACME platform integrates four interconnected modules: Assistant AI, which supports preparation for the Secondary School English Teacher Recruitment Exams using a dataset of 806 English test items from 2002-2025 along with 365 documented success narratives systematically collected from Naver blogs (n = 167), online teacher certification communities (n = 101), university recruitment archives (n = 34), and YouTube interviews/vlogs (n = 63); Consultant AI, facilitating non-teaching career exploration through employment database integration and labor market intelligence; Mediator AI, coordinating institutional resources with individual career planning; and Educator AI, providing teaching methodology support via curated educational technology tools. Drawing on three primary data sources: (1) institutional records of career and extracurricular program participation (2022-2024), (2) a pre-survey of 54 English education majors, and (3) a post-pilot usability test involving 36 students―the study examined English education majors’ career identities, advising needs, and platform effectiveness. Findings from a preliminary needs assessment (n = 54) revealed significant career development challenges: while students demonstrated strong major satisfaction (55.6%) and interest (61.1%), only 44.4% expressed confidence in their career readiness confidence. Career aspirations fluctuated dramatically across academic years (first/second year: 76.9%/70.0%; third year: 33.3%; fourth year: 66.7%), reflecting students’ evolving awareness of teaching market realities. Demand for career counseling services tripled between 2022-2024 (20 to 61 annual participants), yet specialized programming for non-teaching careers markedly declined. A pilot evaluation (n = 36) confirmed platform utility, with highest ratings for examination preparation (M = 4.44) and teaching methodology resources (M = 4.36). Qualitative analysis revealed student requests for enhanced functionality including automated error-pattern detection, expanded non-teaching career outcome databases, community-based peer matching, and career-aligned activity recommendations. The study validates vibe coding as an effective method for enabling sustainable, low-cost platform development, demonstrating that regional universities can autonomously create department-level, discipline-specific AI systems supporting personalized, data-driven career guidance.

ACME, AI-based career support platform, vibe coding, English education majors

이정은 Jungeun Lee , 성은경 Eunkyung Sung

DOI:10.15702/mall.2025.28.4.143 Vol.28(No.4) 143-168, 2025

As artificial intelligence (AI) increasingly transforms education, the AI Digital Textbook (AIDT) has emerged as a key innovation for personalized and interactive learning. This study examined high school students’ perceptions of using the English AIDT according to their English achievement levels. A total of 149 first-year high school students who had used the AIDT in English classes participated in the study. Data were collected through surveys and open-ended responses from students, as well as in-depth interviews with both students and teachers. The survey measured four categories (i.e., selfdirected learning, level-based learning, functional aspects, and emotional stability). The findings showed that students generally held positive perceptions of the AIDT, particularly regarding its functional convenience and feedback system. However, self-directed learning received the lowest mean score, revealing that autonomy and self-regulated learning strategies have not yet been fully developed within the AIDT environment. The statistical results indicated that students showed significantly different perceptions across the four categories with self-directed learning scoring significantly lower than the other three categories. When comparing the two groups by achievement level, higher-achieving students generally showed more positive perceptions. However, significant differences were found only in emotional stability, suggesting that immediate feedback may enhance motivation for higher achievers but cause anxiety for lower achievers. Overall, while the AIDT supports personalized and interactive learning, its effective use also depends on instructional guidance that fosters learner autonomy and provides emotional support for students at different achievement levels.

AI digital textbook, English learning, perception, achievement level, self-directed learning, emotional stability

이동주 Dongju Lee

DOI:10.15702/mall.2025.28.4.169 Vol.28(No.4) 169-192, 2025

This study investigated the impact of using Padlet to support process-based English writing instruction on EFL university students’ writing motivation and writing strategy use. Forty-two students enrolled in an academic writing course at a Korean university completed scaffolded writing tasks, including planning, drafting, receiving peer feedback and revising, using Padlet. Data were collected via a background survey and two instruments: the Academic Writing Motivation Questionnaire (AWMQ) and the English Writing Strategy Test (EWST). These were administered at the beginning and end of the semester. Paired-samples t-tests revealed significant increases in the writing motivation subscales (enjoyment, self-efficacy and effort), while instrumentality and recognition remained stable, likely due to the students’ high initial scores in these areas. Writing strategy use improved significantly across all stages. Independent-samples t-tests showed no gender differences in writing motivation; however, male students demonstrated greater gains in pre-writing strategies. Overall, these findings suggest that integrating collaborative digital tools such as Padlet into process-oriented writing instruction can enhance the motivation and strategic engagement of EFL learners in writing.

Padlet, process-based writing, English writing motivation, English writing strategies, gender differences

이문정 Moonjung Lee , 이용상 Yongsang Lee , 홍익현 Ikhyeon Hong

DOI:10.15702/mall.2025.28.4.193 Vol.28(No.4) 193-212, 2025

This study aimed to explore the theoretical foundations of automatic item generation (AIG) methods and derive practical implications by examining the quality of English reading items automatically produced using generative language models, in order to support item writers seeking efficient development of high-quality items. To this end, open-source (Llama) and closed-source (ChatGPT) models were selected and test items were generated under four sets of conditions to compare the outcomes. These conditions specified eight variables ― standard performance, item type, genre, topic, keyword, number of options, language and text length ― to ensure maximal consistency between the two language models. Based on these conditions, two item types were pilot-generated for 7th- and 10th-grade students: identifying specific details and identifying mood/emotion changes. Subsequently, ten content experts evaluated the quality of the generated items. The results indicated that although test items produced by generative language models still require improvement and are not yet ready for immediate use in school settings, there was no meaningful difference in quality between the items generated by the opensource and closed-source models. These results suggest that institutions sensitive to data security may consider developing automatic generation systems based on open-source models.

Automatic Item Generation, generative language model, open-source model, closed-source model, education assessment

이대현 Daehyun Lee , 한혜민 Hae Min Han

DOI:10.15702/mall.2025.28.4.213 Vol.28(No.4) 213-233, 2025

This study explores the applicability of AIDT in the context of Korean language education for non-native speakers. As intelligent educational tools, AIDTs offer personalized feedback, learning diagnostics, adaptive content delivery, and learning history management, all of which support individualized learning paths. Drawing on recent implementation cases in Korean elementary and secondary education as well as government-issued guidelines, this paper examines the pedagogical potential of AIDTs beyond the boundaries of public curriculum constraints. Despite recent policy shifts that cast uncertainty on the official status of AIDTs within Korea’s public education system, their application in Korean as a foreign language (KFL) education remains viable due to the flexible use of diverse instructional resources in this field. The study categorizes core AIDT functions ―including AI tutors and teaching assistants ― and assesses their alignment with key competencies in Korean language instruction. By analyzing technological affordances and practical usage scenarios, the paper argues that AIDTs may serve as effective tools for differentiated instruction, formative assessment, and learnercentered pedagogy in KFL classrooms.

AIDT, Korean as a foreign language, personalized learning, intelligent tutoring, adaptive instruction, AI in education

오현주 Hyun-joo Oh

DOI:10.15702/mall.2025.28.4.234 Vol.28(No.4) 234-255, 2025

This study examined how ChatGPT’s evaluation and feedback patterns varied when identical text was framed as ‘student text’ or ‘AI revision.’ An exploratory mixed-methods design was employed, in which 66 argumentative essays written by university students were assessed under two framing conditions, and ChatGPT’s rubric-based scores and feedback were collected for comparison. The quantitative analysis indicated that ChatGPT assigned higher scores when an essay was presented as an ‘AI revision’ rather than as a ‘student text,’ with the largest differences observed in grammar and language use. The qualitative analysis further revealed distinct differences in tone, evaluative orientation, feedback function, and meta-evaluation. Under the ‘AI revision’ framing, ChatGPT provided more encouraging and reinforcement-oriented feedback, whereas under the ‘student text’ framing, the feedback was more critical and evaluative. These findings suggest that ChatGPT’s responses are influenced not only by textual features but also by contextual framing information, with its role shifting from an evaluator under the ‘student text’ framing to a facilitator under the ‘AI revision’ framing. Pedagogically, the study highlights the need for framing literacy, encouraging teachers and learners to recognize the contextual dependence and potential biases of AI feedback and to engage with it more critically.

ChatGPT, framing effect, AI feedback, assessment reliability, framing literacy