Earticle

This study investigates students’ perceptions of a writing-to-think, critical thinking (CT)-infused learning-log approach designed to support L2 learners’ reasoning and participation in English-medium classroom interaction at a Japanese university. Across seven oral communication classes (N = 182), students engaged in a write–speak–write cycle in which learning logs mediated pre- and post-discussion thinking by prompting learners to generate, clarify, and evaluate ideas through three recurring CTinfused tasks: opinion expression (OE), text interpretation (TI), and evaluation of viewpoints (EV). Quantitative results from a post-intervention survey indicated overall positive perceptions, with students reporting that log-writing helped them organize ideas, prepare for speaking, and participate more actively in discussions, although confidence in spontaneous oral expression remained comparatively lower. Thematic analysis of open-ended responses revealed three themes: (1) OE was perceived as the most effective task for clarifying ideas and building confidence; (2) writing before speaking supported participation by increasing clarity and easing speaking initiation; and (3) linguistic constraints, particularly limited vocabulary and difficulty verbalizing ideas, constrained fuller CT expression. Overall, CT-infused learning logs made students’ reasoning more visible while underscoring the need to strengthen L2 learners’ content-building and meaning-making skills to support fuller CT expression in interaction-focused learning environments.

本研究は、日本の大学英語授業において、L2学習者の思考形成および英語によるディスカ ッション参加を支援することを目的とした、「writing-to-think」に基づくクリティカルシン キング（CT）統合型ラーニングログに対する学生の認識を調査したものである。7つのオー ラル・コミュニケーション授業（N = 182）において、学生は write–speak–write サイクルに基 づく活動に取り組み、ラーニングログを通してディスカッション前後の考えを整理・発展さ せた。ログには、意見表明（opinion expression: OE）、テキスト解釈（text interpretation: TI）、 視点評価（evaluation of viewpoints: EV）の3つのCT統合型タスクが組み込まれ、学習者が自 らの考えを深め、表現することを促した。事後アンケートの量的分析の結果、学生は全体と して肯定的な認識を示し、ログライティングが考えの整理、発話準備、およびディスカッシ ョンへの積極的参加に役立ったと報告した。一方で、即興的な口頭表現への自信は比較的低 い傾向がみられた。自由記述のテーマ分析からは、（1）OEタスクが考えの明確化と自信形 成に最も効果的であると認識されていたこと、（2）話す前に書く活動が、思考を整理し、 発話のきっかけを与えることで参加を支援していたこと、（3）語彙不足や考えを言語化する難しさといった言語的制約がCTの十分な表現を妨げていたこと、の3点が明らかとなった。 以上より、CT統合型ラーニングログは学生の思考過程を可視化し、ディスカッションへの 参加を支援する一方で、より深いCT表現を実現するためには、L2学習者の内容構築力や意 味形成能力をさらに育成する必要があることが示唆された。

This exploratory classroom-based study examined changes in Japanese STEM undergraduates’ pragmatic awareness and production following a two-week intervention involving explicit pragmatics instruction on English suggestions. The instruction was integrated into a first-year English course incorporating project-based learning (PBL). Thirteen first-year pre-intermediate learners completed written discourse completion tasks (DCTs) before and after the instruction. Additional qualitative data were collected from a recorded peer-feedback interaction embedded in the project and from learners’ post-task reflections. The instructional sequence, adapted from Roever (2022), included noticing of model expressions, metapragmatic discussion of directness and mitigation, and guided practice leading to a semi-authentic peer feedback task. DCT scores increased from pre- to post-instruction. Qualitative analyses further revealed that learners broadened their repertoire beyond formulaic direct expressions to include conventionally indirect and mitigated forms (e.g., “How about …” and “Maybe you could …”), as well as supportive strategies such as reason-giving and collaborative repair. However, learners reported difficulty using indirect forms in real time due to limited formulaic resources and cognitive load. Overall, the findings suggest that integrating explicit pragmatics instruction into a context-rich PBL course can support learners’ production of suggestions and expand their repertoire of pragmatically appropriate forms.

本研究は、教室実践に基づく探索的研究として、日本の理工系学部生を対象に、英語の提案 表現に関する明示的語用指導を含む2 週間の介入を通して、語用的気づきおよび産出の変化 を検討したものである。本指導は、プロジェクト型学習（Project-Based Learning: PBL）を取 り入れた1 年次英語科目の一部として実施された。参加者は初中級レベルの1 年生13 名であ り、指導の前後に筆記式談話完成課題（DCT）に取り組んだ。さらに、プロジェクト内で行 われたピア・フィードバック場面の録音データおよび学習者の課題後リフレクションから、 補足的な質的データを収集した。指導はRoever（2022）に基づいて設計され、モデル表現へ の気づき、直接性および緩和に関するメタ語用論的な話し合い、さらに擬似的なピア・フィ ードバック課題へとつながるガイド付き練習から構成された。DCTスコアは指導前から指導 後にかけて向上した。さらに質的分析から、学習者は定型的で直接的な表現にとどまらず、 慣習的に間接的な表現や緩和された提案形式（例：「How about …?」や「Maybe you could …」）を用いるようになり、理由づけや協働的修復といった支援的方略も取り入れているこ とが明らかになった。一方で、学習者は間接的な形式をリアルタイムで用いることに困難を 感じており、その要因として定型表現のレパートリーの不足および認知的負荷が挙げられた。 総じて、本研究の結果は、文脈に富んだPBL 型授業に明示的な語用論指導を組み込むこと が、学習者の提案表現の産出を支援し、語用的に適切な表現形式のレパートリーを広げる可 能性を示唆している。

As generative artificial intelligence (AI) become increasingly prominent in education, their use in English language teaching (ELT) has drawn attention to issues related to teacher perceptions, technological readiness, and pedagogical adjustment, especially in contexts with limited resources. Guided by the Technology Acceptance Model (TAM), this study explores Vietnamese teachers’ day-today use of AI, their views toward the technology, and the ways these perceptions shape their willingness and capacity to integrate AI into ELT at a regional university. Drawing on reflective journals and semistructured interviews, the findings uncover both enthusiasm and concern among participants. Although teachers acknowledged AI’s ability to support personalized learning and foster learner autonomy, they also raised concerns regarding its potential effects on critical thinking, ethical issues, students’ dependence on technology, and their own insufficient digital competence. The study further highlights the importance of institutional support and professional development in facilitating successful AI implementation. Overall, the findings emphasize the need for contextually appropriate approaches, localized policy initiatives, and continued research into the influence of AI on pedagogy and teacher agency.

Việc ứng dụng AI trong dạy học tiếng Anh đang đặt ra những vấn đề liên quan đến nhận thức, mức độ sẵn sàng của giảng viên và sự thích ứng sư phạm, đặc biệt trong các bối cảnh còn những hạn chế nhất định về nguồn lực. Dựa trên Mô hình Chấp nhận công nghệ (TAM), nghiên cứu này tìm hiểu thực tiễn sử dụng AI của giảng viên đại học tại Việt Nam, nhận thức của họ về AI và tác động của nhận thức đến khả năng ứng dụng AI trong dạy và học tiếng Anh. Kết quả nghiên cứu cho thấy giảng viên nhìn nhận AI hỗ trợ công tác giảng dạy, giúp cá nhân hóa việc học, tăng cường tính tự chủ của người học, nhưng cũng gây ra những lo ngại liên quan đến tư duy phản biện, đạo đức học thuật, sự phụ thuộc quá mức vào công nghệ và hạn chế trong năng lực số của giảng viên. Nghiên cứu cũng chỉ ra vai trò thiết yếu của nhà trường và tính cấp thiết của hoạt động bồi dưỡng chuyên môn nhằm giúp giảng viên ứng dụng AI hiệu quả trong giảng dạy.

이 이론 논문은 영어 학습 동기를 맥락과 무관한 심리적 성향이 아니라, 정당성, 공동체, 긍정적 미래상과 관련된 역사적 맥락 속의 사회언어학적 구인으로 재개념화한다. 본 논문은 Gardner 의 통합성을 비판적으로 재해석하면서, 영어 학습 동기가 원어민 중심주의, 인종화된 화자 이미지, 자격 인증 및 평가 체계에 의해 유지되는 상상된 규범적 중심을 향한 구심적 지향으로 조직되어 왔다고 주장한다. 이어 국제적 자세를 경계가 분명한 “원어민” 공동체를 향한 동화주의적 욕망을 완화하고, 영어를 다원적, 다극적, 협상 가능한 매체로 재구성하는 중립적 재지향으로 해석한다. 그러나 평가 체계와 노동시장 진입 과정에서 중심 규범이 여전히 작용하기 때문에 이러한 탈중심화는 부분적일 수밖에 없다. 이러한 긴장을 바탕으로 본 논문은 플랫폼 매개 영어사용과 다극적 공중에 의해 형성되는 21 세기 영어 학습 동기의 원심적 전환을 제안한다. 이 전환에서 학습자는 영어를 외부지향적 참여, 보이스, 해석적 권위를 위한 매체로 상상하며, 자신의 로컬 레퍼토리를 상징자본으로 활용한다. 한국을 예시로 삼아, 본 논문은 영어 학습의탈통합적 근거와 구심적–중립적–원심적 힘의 장 모델을 제시한다.

This theoretical paper reconceptualizes English learning motivation as a historically situated sociolinguistic construct involving legitimacy, community, and positive future imaginaries. Through a critical rereading of Gardner’s integrativeness, it argues that English learning motivation has often been organized by a centripetal orientation toward an imagined normative center sustained by nativespeakerism, racialized speaker images, and credentialing and assessment infrastructures. The paper then reinterprets international posture as a neutralizing reorientation that weakens the assimilationist desire to target a bounded “native-speaker” community and reframes English as a plural, polycentric, and negotiable medium. However, such decentering remains partial because center-coded norms continue to operate in assessment and labor-market gatekeeping. Building on this tension, the paper proposes a centrifugal turn in twenty-first-century English learning motivation, shaped by platform-mediated English use and polycentric publics. In this turn, learners imagine English as a medium for outward participation, voice, and interpretive authority while mobilizing local repertoires as symbolic capital. Using South Korea as an illustrative case, the paper advances a post-integrative rationale for English learning and proposes a centripetal–neutral–centrifugal force-field model for future research.

This study explored the effects of using e-portfolios on English as a foreign language (EFL) learners’ speaking fluency, as well as their opinions regarding e-portfolio implementation in their program. Sixty EFL students participated in a mixed-methods sequential study and were assigned randomly to an experimental (e-portfolio) group and a control (non-e-portfolio) group. Whereas the experimental group kept portfolios on a social network (Eitaa), the control group presented their speaking activities in the classroom. The speaking fluency of both groups was assessed through pretest and posttest structured interviews, and four aspects of their fluency (speech rate, articulation rate, phonation time ratio, and silent pause rate) were analyzed using the PRAAT and SPSS software. Also, in the subsequent qualitative phase, 20 learners from the e-portfolio group took part in a semistructured interview. According to the quantitative data analysis, speaking fluency increased significantly in the e-portfolio group, and the gain was statistically higher than that of the non-e-portfolio group. The qualitative data analysis revealed 3 main themes of social-emotional and metacognitive benefits, social media portfolio challenges, and fluency in speaking. The findings encourage educators to consider using e-portfolios as a potential medium to improve second/foreign language (L2) learners’ speaking fluency.

این پژوهش به بررسی تأثیر استفاده از پوشههای الکترونیکی بر روانی گفتار زبانآموزان انگلیسی بهعنوان زبان خارجی و نیز دیدگاههای آنان درباره اجرا ی پوشههای الکترونیکی در برنامه آموزشی خود پرداخته است. شصت زبانآموز در یک پژوهش متوالی با روش آمیخته شرکت کردند و به دو گروه آزمایشی )پوش هی الکترونیکی( و کنتر ل )بدون پوشهی الکترونیکی( به طور تصادفی تقسیم شدند. گروه آزمایشی پوشههای خود را در یک شبکه اجتماعی )ایتا( ایجاد میکردند، ولی گروه کنترل فعالیتهای گفتاری خود را در کلاس ارائه م یدادند. روانی گفتار هر دو گروه ا ز طری ق مصاحبههای ساختاریافته پیشآزمون و پسآزمون سنجش گردید و چهار جنبه از روانی گفتار آنان )سرعت گفتار، میزان تولید گفتار، نسبت زمان گفتار/آواساز ی و میزان مک ث بیصدا( با استفاده ا ز تحلیل شد. همچنین، در مرحله کیفی ه م ۲۰ نف ر ا ز زبانآمو زان گروه (SPSS) و اس پی اس اس (PRAAT) نرمافزارهای پرت پوشهی الکترونیکی د ر یک مصاحبه نیم هساختاریافته شرک ت کردند. ب ر اسا س تحلی ل دادههای کمی، روانی گفتا ر در گروه پوشهی الکترونیکی ب ه طور معناداری افزای ش یافت و میزان پیشرف ت آن از نظر آمار ی بیشتر از گروه بدون پوشهی الکترونیکی بود. تحلیل دادههای کیفی ۳ مضمون اصلی شامل مزایای اجتماعی-عاطفی و فراشناختی، چال شهای پوشهی مبتنی بر شبکههای اجتماعی و روانی گفتار را آشکار ساخت. یافتهها مدرسان را ترغیب م یکند تا استفاده از پوشههای الکترونیکی را به عنوان ابزاری بالقوه جهت بهبود روانی گفتار زبانآموزان زبان دوم/خارجی مدنظر قرار دهند.

Multiple-text reading (MTR) is an important skill nowadays; thus, an increasing number of language tests include items that require test takers to integrate information across texts. To clarify the effects of readers’ strategy use on the performance of intertextual questions while reading multiple texts in a second language, we conducted a small-scale think-aloud protocol study and a correlational study of Japanese learners of English. We created original texts, a combination of website announcement and an email, accompanied by intratextual and intertextual comprehension questions. The think-aloud protocol study revealed some differences in learners’ strategy use: (a) less proficient readers tended to have difficulty in word-level processing, while more proficient readers did not show such challenges; (b) less proficient readers tended to fix answers from local information without solving uncertainty. The major difference between proficient and less proficient readers was found in intratextual level, not intertextual level. However, the correlational study indicated that the strategy inventory, which was developed for MTR, did not show a significant correlation with reading performance. In sum, whether strategy use explains differences in learners’ processing depends on how strategy use is measured.

複数文章（テキスト）読解の重要性は近年ますます認識されており、言語テストでも2つ以上のテキ ストの情報を統合させる項目の出題が増えている。そこで本研究では、読み手がテキストを統合す る問題を解く際の方略使用を明らかにするため、日本人英語学習者を対象に、小規模な発話プロト コル調査と相関分析に基づく量的研究を行った。ウェブサイトの案内文とEメールを組み合わせた 独自のテキストを作成し、単一テキストで解答可能な「テキスト内問題」と、複数テキストを参照する 「テキスト間問題」を作成した。発話プロトコル調査により(a) 熟達していない読み手は語レベルの処 理に困難を感じる一方で、熟達した読み手にはそのような困難が見られないこと、(b) 熟達していな い読み手は不確定要素があってもそれを解決することなく、局所的な情報から解答を確定しがちで あること、が明らかになった。熟達度の高い読み手と低い読み手の差はテキスト間問題ではなくテ キスト内問題において見られた。しかし量的研究では、アンケートで測定された複数テキスト読解用の方略使用と読解得点に有意な相関関係は見られなかった。したがって、方略使用が学習者の処 理の違いを説明するか否かに関して、方略使用の測定方法に影響される可能性が示唆された。

이 코퍼스 기반 연구는 미국의 한 집중 영어 프로그램에서 수집한 아랍어권 ESL 학습자의 논증적 에세이 95 편을 Longman Spoken and Written English(LSWE) 코퍼스의 학술 산문과 비교하여, 학술적 영어 글쓰기에서 나타나는 형태적 시제(현재, 과거)와 상(단순, 완료, 진행)의 사용 양상을 분석하였다. 본 연구는 집단 간 설계와 비모수 통계를 활용하여, 학습자의 시제/상 분포를 기존 학술 산문의 사용 양상 및 학습자 숙련도와 관련지어 검토하였다. 연구 결과, 진행상을 제외한 모든 범주에서 아랍어권 학습자와 LSWE 필자 간에 뚜렷한 분포상 차이가 나타났으며, 학습자는 LSWE 비교 필자보다 현재 시제와 단순상을 더 자주 사용하고, 과거시제와 완료상은 덜 자주 사용하는 것으로 나타났다. 이러한 양상은 가능한 모국어 영향과 일치하지만, 본 연구 설계가 전이를 직접 입증하는 것은 아니다. 상위 수준 학습자는 일부시제/상 양상, 특히 상 사용에서 LSWE 비교 자료에 더 가까운 패턴을 보였으나, 중간 수준집단들은 유사한 양상을 보였다. 본 연구 결과는 학술 산문에서 숙련된 학술 영어 필자들이 유한동사를 어떻게 사용하는지를 보여 주며, 빈도 자료, 실제 예문, 명시적 대조 활동을 포함한 코퍼스 기반 문법 및 작문 지도의 교육적 가치를 시사한다.

This corpus-based study investigates Arabic ESL learners’ use of morphological tense (present, past) and aspect (simple, perfect, progressive) in academic English writing by comparing 95 argumentative essays from a U.S. intensive English program with matched academic prose from the Longman Spoken and Written English (LSWE) corpus. Using a between-groups design and nonparametric statistics, the study examined tense/aspect distributions in relation to established academic prose patterns and learner proficiency. Findings showed clear distributional differences between Arabic learners and LSWE writers in all categories except progressive aspect, with learners using present tense and simple aspect more frequently, and past tense and perfect aspect less frequently, than the LSWE comparison writers. These patterns are consistent with possible L1 influence, although the present design does not directly establish transfer. Higher-level learners showed some tense/aspect patterns, particularly in aspect use, closer to the LSWE comparison data, although mid-level groups showed similar patterns. The results show how established academic writers use finite verbs in academic prose and suggest pedagogical value for corpus-informed grammar and writing instruction, including descriptive frequency data, authentic examples, and explicit contrastive tasks.

The rapid rise of generative artificial intelligence is challenging key assumptions in learner corpus research. As learners increasingly use AI to draft, translate, revise, and receive feedback, learner language can no longer be understood simply as unaided human production. This special session, Learner Corpus Research in the AI Era: Perspectives from Asia, explores how learner corpora are being reconceptualized across Asian contexts as evidence of interlanguage development, benchmarks for AI evaluation, and pedagogical resources for AI-mediated language learning. The contributions examine how national learner corpus traditions are responding to new technologies; how learner-produced and AI-generated texts differ in lexical, cohesive, phraseological, and discourse features; and how AI can support corpus compilation, annotation, transcription, feedback, and writing instruction while preserving human linguistic judgement. Drawing on studies from Japan, Korea, China, Hong Kong, Taiwan, Indonesia, and broader Asian learner datasets, the session highlights the continuing value of carefully designed corpora, transparent metadata, human-in-the-loop validation, and pedagogically grounded interpretation. It argues that Asian learner corpus research is entering a new phase in which authenticity, comparability, scalability, and ethical use must be reconsidered together.

While AI technologies have transformed the processes adopted by language learners to produce and revise academic writing, more approaches are needed to promote understanding of text patterns and linguistic features of GenAI texts so learners can engage with language skill development and maintain control as novice authors of academic texts. The current study presents a fusion of corpus methods (embedded in a free Windows/macOS tool), facilitating researchers, teachers and students in hands-on, interactive, comparative analysis of essays created by students and essays on the same topics created through GenAI tools. To demonstrate the feasibility of the approach, we analysed 2003 university level argumentative essays on 26 topics, alongside 546 AI-generated essays (21 essays per topic), focussing on four essays from one randomly selected topic for exemplification. The methods highlight notable strengths of AI-generated texts, including broad vocabulary profiles, balance between maintaining lexical focus and avoiding repetition, and rich collocational relations. By visually representing the roles of vocabulary items in text-level contexts, the tool provides a new approach for English for Academic Purposes teachers and students to engage with data patterns in the texts they have produced with and without GenAI, with potential to aid detection, identification, reflection and ultimately production.

This present study investigates the most frequently used lexical bundles in argumentative essays written by L1 English and L2 Taiwanese undergraduate students and compares the use of high-frequency lexical bundles between the two groups. The data were drawn from the ICNALE (International Corpus Network of Asian Learners of English) (Ishikawa, 2013), from which two sub-corpora were compiled: ICNALETW with 400 essays produced by L2 writers from Taiwan and ICNALE-EN as a comparable dataset with 400 essays contributed by native English speakers. Based on the structural and functional taxonomy proposed by Biber et al. (1999), the forms and functions of four-word lexical bundles used by L1 and L2 students were manually annotated and analyzed. The findings reveal that among the top 100 most frequent lexical bundles identified from the two sub-corpora, 16 lexical bundles are shared by L1 and L2 student writers and that through structural and functional analyses, both L1 and L2 writers’ lexical bundles exhibit a similar tendency, with a predominance of verb-based phrases and stance related expressions. The findings presented in this study suggest pedagogical implications for the instruction of formulaic language in tertiary-level academic writing.

Against the backdrop of the paradigm shift brought about by generative AI, this study conducted a diachronic analysis of Korean EFL Learner Corpus Research (KLCR) over the past two decades, examining thematic, methodological, and technological developments. Drawing on 211 KLCR-related articles published in the Korea Citation Index (2002–2024), the study quantitatively tracked changes in research topics, corpus types, and analytical tools. Three key findings emerged. First, research on Grammar and Vocabulary dominated the field, while Pronunciation studies remained scarce, revealing a persistent thematic imbalance. Second, although privately compiled corpora are still prevalent, the rise in the use of open-access corpora since 2017 signals a clear shift toward data openness and shared resources. Third, analytical tools have advanced from reliance on stand-alone software to the broader use of programming languages and online platforms, culminating in the recent emergence of generative AI tools. KLCR appears to be transitioning beyond its traditional paradigm toward more meaning- and function-oriented interdisciplinary inquiry, supported by expanding data accessibility and technological sophistication. To ensure sustainable development, this study emphasizes diversifying research topics, strengthening collaboration across academic institutions to build large-scale open-access corpora, and enhancing researcher training in programming and AI.

This article provides a historical overview of the evolution of learner corpus research (LCR) in Japan, tracing its development from its inception to recent advances in the 2020s. Initially centered on smallscale corpora, primarily in English, the field grew substantially in the 2000s with the emergence of large, systematically compiled corpora, including the NICT Japanese Learner English (JLE) Corpus and the Japanese EFL Learner Corpus (JEFLL). The 2010s brought further diversification through multinational and multilingual corpora, such as the International Corpus Network of Asian Learners of English (ICNALE), and extensive learner corpora for Japanese as a second language, including the International Corpus of Japanese as a Second Language (I-JAS). Methodologically, this period was characterized by greater use of standardized proficiency scales, the incorporation of large-scale data from online sources, and closer integration with second language acquisition (SLA) theories. Japanese LCR continues to mature through longitudinal data and increased attention to underrepresented learner groups and language pairs. More recently, the convergence of learner corpora with generative AI-based tools for annotation, feedback, and assessment has opened a new phase in the field. This article examines how these emerging practices are taking shape in Japan and considers their implications for learner corpus research and corpus-informed language education.

本稿は、日本における学習者コーパス研究（LCR）の発展を、黎明期から2020 年代の最新動 向まで歴史的に概観する。初期には英語を対象とした小規模コーパスが中心であったが、2000 年代にはNICT JLE Corpus やJEFLL Corpus など、大規模かつ体系的なコーパスの構築によっ て研究が大きく進展した。2010 年代には、ICNALE に代表される多国籍・多言語コーパスや、 日本語学習者を対象とするI-JAS などが整備され、対象がさらに広がった。方法論的にも、 標準化された熟達度尺度の活用、オンライン資料に基づく大規模データの導入、第二言語習 得理論との連携が進んだ。現在、日本のLCR は、縦断データや十分に研究されてこなかった 学習者集団・言語への注目を強めながら成熟を続けている。近年では、アノテーション、フ ィードバック、そして、評価に生成AI を活用する実践が学習者コーパスと結びつき、新たな 段階が始まりつつある。本稿は、こうしたLCR 研究の日本における展開を論じ、学習者コー パス研究とコーパスに基づいた言語教育への示唆を検討する。

Lexical bundles play an important role in supporting fluency, coherence, and rhetorical clarity in academic writing. Their use is particularly relevant in argumentative essays, where writers must present claims, organize ideas, and maintain cohesive argumentation. Although previous research has examined lexical bundles in learner writing, limited work has compared learner-produced bundles with those found in AI-generated texts, despite the increasing use of artificial intelligence in educational settings. This study investigates the lexical bundles used in argumentative essays written by Indonesian EFL learners and compares them with bundles appearing in AI-generated texts. The objective is to identify the most frequently used bundles in each corpus, examine their structural and functional patterns, and describe the differences observed. A corpus-based comparative method was employed, drawing on two datasets: a learner corpus of university student essays and an AI-generated corpus produced by DeepAI and SuperAI. Lexical bundles were identified and categorized by structure (NP-based, VP-based, PPbased, others) and function (research-oriented, text-oriented, and participant-oriented). Frequency and distribution measures were also calculated. The findings show that the two corpora differ in their most frequent bundles, structural compositions, and functional distributions, with each corpus displaying distinct patterns of lexical bundle use in argumentative writing.

English as a lingua franca plays a central role in international communication, yet it poses persistent challenges for non-native users, particularly in achieving textual cohesion, which is essential for clarity and communicative effectiveness. ESL (English as a second language) student translators often struggle to manage cohesion through the appropriate use of connectives when translating into English. The emergence of generative artificial intelligence (GenAI) has created new possibilities for supporting language learners and translators in producing cohesive texts. This study examines the potential and limitations of GenAI in generating cohesive Chinese-English translations by comparing its outputs with those produced by ESL learner translators, using sense-aware connective-based cohesion indices. The results show that, although the two groups display largely similar overall patterns in the use of discourse and non-discourse connectives, they differ substantially in specific connective categories. GenAIgenerated translations show a greater reliance on comparison and expansion connectives and a relative underuse of contingency connectives, whereas ESL learner translations tend to preserve source text discourse relations more closely. These findings provide insights into the cohesion characteristics of GenAI-generated translations and point to important considerations for language learning and translation pedagogy.

중국인 학습자는 전 세계 영어 학습자 가운데 가장 큰 집단을 형성하고 있음에도 불구하고, 이들의 영어 작문을 대상으로 포괄적인 오류 주석을 제공하는 공개 코퍼스는 아직 충분히 구축되어 있지 않다. 이러한 자료적 공백은 중국인 영어 학습자의 오류 양상에 대한 체계적 연구를 제약할 뿐만 아니라, 실제 교육 현장에서 활용 가능한 피드백 자료의 개발 또한 제한한다. 본 연구는 이러한 한계를 보완하기 위해 신뢰도 높은 오류 주석을 산출하는 동시에 교육적 피드백을 생성할 수 있는 인간 참여형 AI 보조 프레임워크를 제안한다. 이 프레임워크는 문법, 어휘, 표기 및 형식의 세영역으로 구성된 간소화된 오류 분류 체계를 채택하며, 각 오류 주석은 오류 유형, 오류 범위, 수정안, 설명의 네 가지 항목으로 이루어진 최소 기록 구조에 따라 제시된다. 본 연구에서는 해당 프레임워크를 중국인 영어 학습자 10,000 편 작문 코퍼스(TECCL)에 적용하였으며, 주석 과정에는 중국인 영어 학습자 코퍼스(CLEC)에 기반한 문맥 내 학습 예시, 연쇄 사고 추론, 검색 증강 생성이 통합되었다. 100 편의 에세이를 대상으로 한 개념 증명 평가 결과, 주석자 간 일치도는 전체 Cohen’s κ = 0.78 에 도달하였고, 오류 범위의 정확 일치율은 85%였으며, 주석자 내 일치도는 90%를 초과하였다. 또한 164 개 문장을 분석한 결과, 총 187 개의 오류가 확인되었고, 그 분포는 문법 66.3%, 어휘 23.0%, 표기 및 형식 10.7%로 나타났다. 전문가 검토에서는 형태통사 및 어휘 오류에 대한 높은 정밀도가 확인되었으며, 주로 관용표현 또는 연어와 관련된 9 개의 누락 오류가 추가로 발견되었다(재현율 = 95.4%). 이러한 예비 분석 결과는 본 연구에서 제안한 프레임워크가 연구 간 비교 가능성을 유지하면서도 오류 주석 작업의 투명하고 확장 가능한 수행을 지원할 수 있음을 보여준다. 나아가 본 프레임워크는 교실 수업에 활용 가능한 오류 피드백 자료를 생성할 수 있다는 점에서 학습자 코퍼스 연구와 언어 교육 실천을 연결하는 방법론적 가능성을 지닌다.

Large-scale error annotation in learner corpora remains challenging, particularly for under-resourced L1 populations. Although Chinese learners represent the world’s largest group of English learners, publicly accessible corpora of their English writing with comprehensive error annotations are scarce, limiting research and pedagogical applications. This study introduces a human-in-the-loop, AI-assisted framework designed to deliver reliable annotations while generating pedagogical feedback. The framework employs a streamlined three-domain taxonomy (Grammar, Lexis, and Mechanics) and a four-field record structure (error type, span, correction, and explanation). Applied to the Ten-thousand English Compositions of Chinese Learners, the pipeline integrates in-context demonstrations from the Chinese Learner English Corpus, chain-of-thought reasoning, and retrieval-augmented generation. In a proof-of-concept evaluation of 100 essays, inter-annotator agreement reached Cohen’s κ = 0.78, with an exact-span match of 85%, and intra-annotator agreement exceeded 90%. Analysis of 164 sentences yielded 187 errors: Grammar 66.3%, Lexis 23.0%, and Mechanics 10.7%. Expert review confirmed high precision for morphosyntactic and lexical errors and identified nine omissions (recall = 95.4%), involving idiomaticity or collocation. These results suggest that the approach can support scaling while preserving research comparability and producing materials for classroom use. Although further validation with larger datasets is required, the framework shows potential for adaptation to other learner populations.

This study evaluates whether automatic speech recognition (ASR) systems can support scalable spoken learner corpus development without compromising the analytical rigor required in second language acquisition research. Using recordings from 60 Taiwanese learners of English at A2–B1 proficiency levels, this study compares three ASR platforms—Microsoft 365 Word, OpenAI Whisper, and Vocol.ai—across 1,200 elicited sentences and 60 weather-broadcast tasks. Transcriptions are assessed for lexical accuracy, sentence segmentation, and semantic coherence using multiple AI-based evaluators. Results show substantial variation across systems: Microsoft 365 Word frequently produced distorted or implausible lexical substitutions; Whisper maintained overall meaning but introduced systematic tense changes; and Vocol.ai provided the highest fidelity, preserving both lexical and structural details. No system generated error-free output, particularly with L2-specific pronunciation patterns. A threephase validation framework is proposed to ensure transcription reliability. The findings demonstrate that ASR can enhance corpus creation only when embedded within rigorous, multi-step validation procedures.

Producing concise and rhetorically effective prose is an essential skill in academic writing. This task, however, poses a particular challenge for EFL learners, as it involves various procedures for eliminating redundancy, ranging from simple lexical adjustments to the use of highly complex grammatical structures such as reduced relative clauses or elliptical concessive constructions (e.g., though not perfect). While a key objective is to compress information without sacrificing textual clarity, many EFL learners lack the requisite knowledge and skills, possibly because this area has received little attention in formal educational contexts. Drawing on 59 essays written by advanced Japanese EFL learners, this study identifies the major sources and patterns of redundant expressions and examines the potential of generative AI (GenAI) to support learners’ revision processes. The findings indicate that even advanced learners often fail to detect redundancy arising from seemingly simple repetition of lexical items and propositional content, and that these difficulties are closely associated with challenges in reconceptualizing sentence-level components as discourse-level elements. The study further demonstrates that carefully designed AI prompts, informed by linguistic research, can serve as an effective pedagogical tool to help learners identify and resolve redundancy-related issues.

This study investigates the impact of different prompt design strategies on the performance of GPT-4 in assessing undergraduate reports within an English as a Foreign Language (EFL) context. As Large Language Models (LLMs) increasingly integrate into educational assessment, understanding how prompt engineering affects grading accuracy and alignment with human judgment is crucial. Three prompt design methods—TELeR Taxonomy, Six strategies framework, and Zero-shot prompting— were applied to evaluate 40 student business reports. The assessments were compared against human teacher scores using quantitative statistical analysis, specifically box-and-whisker plots and inter-rater agreement between teacher ratings and AI-generated scores using Cohen’s κ and quadratic weighted κ. Findings indicate that the choice of prompt design significantly influences the alignment between AI and human grading. The Six strategies method demonstrated strong alignment with teacher assessments, particularly in evaluating coherence and cohesion. The Zero-shot prompt demonstrated the strongest alignment with teacher ratings, yielding the most similar score distribution and the highest inter-rater agreement (quadratic weighted κ = 0.65). The results suggest that although this strategy most effectively approximates human scoring in AI-assisted assessment, human oversight remains essential for capturing the subjective and nuanced aspects of student writing.

This study investigates the integration of 21st-century skills and CEFR alignment in a pre-intermediate non-formal English course. Using classroom observations across 34 meetings, the research examined pedagogical alignment with CEFR standards, integration of six key skills (communication, collaboration, critical thinking, creativity, citizenship, and character education), and instructional challenges. Findings reveal incomplete CEFR alignment: while learning objectives were explicitly stated (100%), they lacked measurable rubrics, and assessment practices showed critical gaps, particularly in evaluating listening and speaking skills. Technology remained underutilized despite available infrastructure. Regarding 21st-century skills, communication and character education were perfectly implemented (100%), but collaboration (41.5%), critical thinking (26.7%), creativity (12.8%), and citizenship (10.9%) required urgent attention. Additional challenges included abrupt instructor swaps, fluctuating attendance, and summative assessments that excluded oral skills. The study underscores the need for SMART learning objectives, comprehensive assessment frameworks including listening and speaking, enhanced technology integration, and systematic embedding of higher-order skills. These findings highlight the critical role of non-formal education in complementing formal systems while revealing significant gaps requiring targeted intervention to better prepare Indonesian learners for global workforce demands.

Guided by the academic literacies framework (ALF), this study investigates how EMI (English-medium instruction) students’ English writing challenges are perceived by three groups of stakeholders (i.e. students, English instructors, and content instructors) over the course of a nine-month academic year. Six postgraduates, six English instructors, and six content instructors participated in two semi-structured interviews spaced nine months apart. Interview data were subject to thematic analysis in NVvio. The thematic analysis identified seven recurring challenges, with limited academic vocabulary and underdeveloped argumentation emerging as shared concerns across groups. However, group-specific emphases on English writing challenges varied among students, English instructors and content instructors. Over time, students reported increased awareness of grammatical issues and disciplinary expectations, although their recognition of other writing challenges might remain limited. These findings underline the need to bridge stakeholders’ perception gaps and account for students’ evolving awareness of writing challenges. The study extends the ALF by demonstrating how stakeholder perspectives and evolving student awareness interact across linguistic, cognitive, and disciplinary dimensions, offering both theoretical and pedagogical insights for EMI writing instruction.

This small-scale qualitative study explores how systematic spoken error analysis may illuminate second language (L2) development among Bangla-speaking learners of English. Drawing on Ellis and Barkhuizen’s (2005) performance analysis framework, the study examines grammatical, lexical, and pragmatic deviations in the spontaneous spoken output of two postgraduate learners during a task-based sociolinguistic interview. Adopting an interpretive analytic approach, analysis of the transcribed interaction identifies recurring patterns in interrogative formation, reliance on formulaic sequences, and L1-influenced pragmatic routines. While these features appear to support communicative fluency and interactional alignment, they also suggest partial proceduralisation of English morphosyntax and emerging limitations in register sensitivity. The findings indicate that spoken deviations may function not only as errors but also as contextually meaningful indicators of learners’ developing interlanguage under real-time processing constraints. Situated within an under-researched English as a foreign language (EFL) context such as Bangladesh, the study offers exploratory, context-bound insights into spoken L2 development. Pedagogically, it suggests that integrating focused spoken-error analysis into classroom practice, for example, through guided reformulation and contrastive awareness tasks, may support more responsive and developmentally attuned oral instruction in low-input South Asian ESL environments.