Christopher Irvin

DOI:10.15702/mall.2025.28.3.1 Vol.28(No.3) 1-28, 2025

As blended learning (BL) becomes more common in English as a Foreign Language (EFL) contexts, educators must consider how digital tools such as discussion boards (DBs) influence collaboration and skill development. This qualitative study examines student experiences in a first-year BL EFL course at a Korean university across three DB conditions: 4- skill, 2-skill, and no-DB. A total of 81 students completed four collaborative tasks over one semester, with formats adjusted by instructional group. Data were collected through an end-of-semester survey combining descriptive statistics and thematic analysis of open-ended responses. Three research questions guided the study: (1) how students described their activity preferences and discussion experiences; (2) how they perceived language skill development in relation to DB format; and (3) how they viewed collaborative learning opportunities across conditions. Results showed that students valued discussion tasks that were engaging, personally relevant, and socially supportive. Perceived skill gains aligned with DB format: the 4- skill group emphasized speaking and listening development, while the 2-skill group noted reading and speaking gains. Findings suggest that multimodal, learner-generated DB tasks can enhance motivation and support language growth. Implications include the need for scaffolded design, flexible sequencing, and task formats that promote meaningful communication in BL environments.

blended learning, 4-skill DB, discussion boards, collaboration, EFL

Yiteng Jiang , Seongyong Lee , Jaeho Jeon

DOI:10.15702/mall.2025.28.3.29 Vol.28(No.3) 29-54, 2025

The growing role of Generative AI (GenAI) in language education highlights the need to understand how learners engage with it for writing. While GenAI can support second language (L2) writing, in-depth qualitative insights into learner interaction and perceptions are scarce. This multiple-case study investigated Chinese high school students' engagement with GenAI in English writing, using a sociocultural lens. Twelve L2 students participated in a four-week program involving seven training sessions on using GenAI for planning, drafting, revising, and metacognitive reflection. Four students were chosen for in-depth analysis due to their distinct engagement patterns. Multimodal data, including videos, screen captures, prompt logs, writing samples, and interviews, were analyzed. Three engagement trajectories emerged: (1) distant engagement (GenAI used only for phrasal translation); (2) weak hybrid intelligence (Passive adoption of AI outputs with minimal interaction); (3) robust hybrid intelligence (Reciprocal knowledge co-construction between learner and AI). Students with weak engagement viewed GenAI as a translation or ghostwriting tool, leading to one-way information transfer. In contrast, the student showing robust hybrid intelligence saw GenAI as a teacher, partner, or tool, engaging in meaning negotiation and iterative prompting. Findings suggest learners' perception of GenAI significantly impacts their engagement quality. Theoretically, this study extends sociocultural theory by showing how learner-AI interactions mediate cognitive and metacognitive processes in L2 writing. Pedagogically, it emphasizes the need for interventions to guide student perceptions and foster meaningful, reflective GenAI use as a collaborative learning partner, not just a technical tool.

Generative AI, L2 writing, Sociocultural theory, students’ perception, case study

Hea-suk Kim , Yoonjung Cha , Na-young Kim

DOI:10.15702/mall.2025.28.3.55 Vol.28(No.3) 55-80, 2025

This study investigated the effects of machine translators, ChatGPT, and their combined use on reading comprehension and vocabulary acquisition among 120 university students enrolled in English courses in 2024, who were randomly assigned to one of three tool-based groups. It also examined whether tool usage in EFL reading contexts led to varying outcomes and explored students’ perceptions of these tools. Results showed that all three groups improved in reading and vocabulary within groups, but no significant differences were found between groups. Additionally, learners’ perceptions regarding both the affective and motivational influence, as well as the educational impact of AI tools, showed no significant variation across experimental groups for the majority of survey items. Students who used machine translators found them helpful for understanding texts and reading speed but expressed concerns about inaccuracy, overuse, and unclear meanings. Those who used ChatGPT appreciated its usefulness and speed but also noted similar issues to those observed in the Machine Translation (MT) group. The group using both tools found them useful, clear, and fast, but still faced issues related to overreliance and occasional errors. Despite its limitations, this study provides an empirical example of integrating AI tools into foreign language education and offers meaningful pedagogical implications.

machine translators, AI, ChatGPT, reading comprehension, vocabulary, students’ perception

Sooyeon Kim , Eunseok Ro

DOI:10.15702/mall.2025.28.3.81 Vol.28(No.3) 81-108, 2025

This study offers a contribution to the growing body of research on informal language learning by demonstrating how second language (L2) pronunciation learning can emerge within media-mediated contexts. Drawing on unscripted interaction from a Korean reality TV show, the study expands the scope of L2 research beyond classroom instruction to include casual, socially embedded talk in travel-based environments. While previous conversation analysis for second language acquisition (CA-SLA) studies have primarily examined learning as a response to interactional trouble or within pedagogical settings, this study addresses a gap by investigating how learning can occur even in the absence of such trouble. Through conversation analysis, this study highlights how participants create learning opportunities by topicalizing others’ pronunciation or by requesting translations for phrases whose meanings are already known―not merely for the sake of learning, but as to develop resources for accomplishing other social actions, such as making a service request or offering a compliment. The analysis further shows that alignment between participants’ instructional orientations supports short-term development, while disalignment―often shaped by competing social goals―can hinder learning. By demonstrating how learning unfolds in a Korean reality TV show, this study argues that media interaction is a valuable yet underexplored site for revealing how participants co-construct learning opportunities―thereby deepening our understanding of how L2 learning emerges through interaction.

CA-SLA, doing learning, Korean as a second language, L2 pronunciation, media interaction

Chae-young Mun

DOI:10.15702/mall.2025.28.3.109 Vol.28(No.3) 109-132, 2025

This study aims to investigate the impact of an artificial intelligence tool, ChatGPT, on English as a Foreign Language (EFL) learners’ speaking skills, including fluency, lexical range, grammatical range with accuracy, and content. Learners’ perceptions and attitudes toward using an AI-mediated speaking tool to enhance their English-speaking skills were also qualitatively explored. Fifty students from a university in Seoul were assigned to an experimental group, which engaged in communicative tasks using a generative AI-powered voice interaction tool, and to a control group, which practiced without the aid of the tool. Data were collected through (1) a pre- and post-test on speaking skills and (2) a post-survey on affective attitudes administered to the experimental group. A five-point Likert scale online survey assessed students' perceptions of the tool’s use. Results revealed that the post-test speaking scores of the experimental group were significantly higher than those of the control group, which interacted with peers. The group using the AI tool showed significant improvement in EFL learners’ speaking skills. Analyses of the survey responses indicated high levels of perceived usefulness and positive attitudes toward the AI tool, and the study underscores that such positive attitudes significantly influence learners’ intention to adopt ChatGPT for English-speaking practice.

Artificial intelligence, ChatGPT, English as a Foreign Language, English speaking skills, EFL students’ perception

서정임 Jung Im Seo , 김희진 Hijean Kim

DOI:10.15702/mall.2025.28.3.133 Vol.28(No.3) 133-156, 2025

This paper aims to analyze the effects of AI chatbots and artificial intelligence speakers (application) on elementary school students in English vocabulary classes. Thirty students in grades 4-6 were randomly divided into two groups, A and B. Vocabulary tests (pre-post) and surveys (pre-post) on affective domains were used to collect data. The experimental classes consisted of six sessions using AI chatbots and AI speakers for group A (n=15), and Classcard application for group B (n=15). The results of the research are as follows. First, there was a statistically significant difference in vocabulary achievement between the two groups. Second, the findings of the survey suggested that there were significant differences at the overall affective domains (interest, confidence). To be specific, the experiment using AI tools for learning English vocabulary proved effective in improving young learners' vocabulary skills and enhancing interest, confidence, and immersion. In conclusion, AI chatbots-based teaching and learning methods can improve students’ English vocabulary abilities and positive attitudes toward foreign languages. This study emphasizes the potential of AI-assisted classes to suggest an effective approach to English vocabulary learning. If more systematic and diverse AI-based teaching & learning models are designed and implemented to improve English vocabulary instruction, it is expected to have pedagogical effects on the future classes. Furthermore, AI-incorporated classes need to be expanded and applied to higher education.

AI chatbots, artificial intelligence speakers, English vocabulary achievement, affective domain

Min-chang Sung

DOI:10.15702/mall.2025.28.3.157 Vol.28(No.3) 157-173, 2025

This study evaluates the reading problem-solving abilities of four leading generative AI models: OpenAI’s GPT-4o and GPT-o3, and Google’s Gemini 2.5 Flash and Gemini 2.5 Pro. The models were assessed using the English reading sections from the College Scholastic Ability Test for the past three academic years. To ensure reliability, each model was tested across three separate sessions. The findings reveal high consistency in each model’s performance but significant accuracy differences among them. Gemini 2.5 Pro performed best with 97.12% accuracy, while GPT-4o was the lowest at 86.83%. All models excelled at tasks involving context inference and detailed comprehension; however, they demonstrated clear weaknesses in higher-order discourse skills, frequently producing incorrect answers for questions requiring logical construction and referential understanding. These results suggest that while generative AI’s reading capabilities have advanced significantly, critical vulnerabilities remain. The study also highlights the necessity of comparative analysis when selecting AI models for educational purposes.

Generative AI, ChatGPT, Gemini, English reading assessment, College Scholastic Ability Test

Jiyeop Yang , Moonyoung Park

DOI:10.15702/mall.2025.28.3.174 Vol.28(No.3) 174-193, 2025

This mixed-methods study investigated how metacognitive reflection amplifies ChatGPT-based feedback in processoriented English writing instruction. Three first-year Korean high school students participated in a four-session argumentative essay writing program over four months, involving writing essays, receiving structured ChatGPT feedback through specialized prompts, and completing self-reflection journals. Data collection included essay assessments by three raters, pre-post affective surveys, self-reflection journals, and interviews. Results demonstrated significant writing improvements: combined scores increased from Session 1 (M = 55.17, SD = 8.29) to Session 4 (M = 76.33, SD = 6.81), representing 38% improvement. Revised essays consistently scored higher than initial drafts. Affective factors showed positive changes: learning motivation increased (+0.7), self-efficacy improved (+0.3), and writing anxiety decreased (-0.7). Qualitative analysis revealed enhanced argument specificity, logical connectivity, and conclusion effectiveness. Selfreflection journals served as crucial amplifiers, transforming passive reception of algorithmic feedback into active strategy development and autonomous learning. The study demonstrates that metacognitive reflection moves beyond automated responses to create authentic learning that enhances both cognitive and affective dimensions of L2 writing development. These findings highlight practical implications for integrating AI tools into classroom-based L2 writing instruction through reflective pedagogy.

ChatGPT, AI feedback, process writing, self-reflection, English writing, affective factors, Korean EFL learners