Storytelling in EFL is usually defined as the practice of telling a story for language learning purposes, like grammar and vocabulary acquisition or for the development of listening and speaking skills (Bland, 2015). A subset, digital storytelling, is usually defined as the combination of traditional narrative with electronic multimedia tools like videos (Bull and Kajder 2005; Porter 2005; Rule, 2010). The purpose of this study was to examine Thai EFL students’ English speaking ability, attitudes, and speaking anxiety after the implementation of digital storytelling and storytelling activities in class. Sixty-four first-year non-English major students at Rajamangala University of Technology Krungthep, Thailand, participated in the study and were divided into two groups: digital storytelling and storytelling. A pre- and post-speaking test, attitude questionnaires, and speaking anxiety questionnaires were used as research instruments. The findings revealed that students’ English speaking scores significantly improved after the implementation of both storytelling activities. In addition, students demonstrated positive attitudes toward both instructional approaches. The results also showed that storytelling (x̄ = 2.09, SD = 0.34) and digital storytelling (x̄ = 3.35, SD = 0.30) helped reduce students’ speaking anxiety.

Multiword units (MWUs), recurrent word combinations frequently used in language, support fluent language production (Pawley & Syder, 1983) and play a crucial role in children’s language development (e.g., Arnon et al., 2017; Skarabela et al., 2021). Adopting a corpus-driven approach, this study investigates the distribution of high-frequency single words and MWUs in parental and child speech, drawing on the Warren Corpus (Warren-Leubecker, 1982, 1984) from CHILDES. The dataset comprises home interactions involving children (mean age of 64.7 months) and their parents. High-frequency single words and two-, three-, and four-word MWUs were extracted and compared. The results revealed that (1) in terms of the high-frequency single words, the strongest correspondence is found in pronoun use, and (2) a higher degree of overlap is observed in two-word MWUs, whereas little or no correspondence is found in three- and four-word MWUs. Overall, the correspondence decreases as the length of MWUs increases. From a usage-based perspective, children’s emerging MWU production is associated with shorter, high-frequency patterns in the input, while longer formulaic sequences develop more gradually. These findings underscore the importance of examining MWUs of different lengths and suggest the pedagogical value of shorter, high-frequency MWUs for early-stage learners in both first and additional language contexts.

Teaching is one of the busiest professions, with teachers averaging 15-20 hours of overtime a week. When you add the “publish or perish” culture of teaching in universities, it is easy to constantly work, yet slowly become out of touch with recent developments, especially concerning AI. This workshop is aimed for teachers who want to use AI for research, but are new to it or don’t know where to start. Two research focused AI websites (SciSpace and Ai2 Scholar QA / Asta) will be demonstrated. These are commercial AI but have free versions, which will be used. Demonstrations will include finding relevant research, asking research questions, statistical analysis, and making chart/graphs for results sections. Demonstrations will be followed by participants trying them out, with the presenter assisting. Please bring laptops or tablets. If you’ve been meaning to look into AI geared towards research, but have never found the time, feel intimidated, or don’t know where to start, please join us as we explore these 2 programs. The presenter has no personal benefit from or connection to these websites, they are strictly being utilized for educational and research purposes.

In My Life New Edition by Matthew Taylor and myself (Nellie’s Publishing) is a scrapbook-based presentation and oral communication textbook for student-centered teaching and learning. Based on motivation and engagement principles explained by John Dewey and Zoltán Dörnyei. The coursebook centers around students creating scrapbook pages about their life to give presentations or as something to spark conversations. Paper scrapbooks can be used, or digital visual aids with apps like PowerPoint/Keynote/Google Slides, digital photos stored on computers, tablets, or smartphones, or a hybrid paper/digital format such as creating the scrapbook page on paper scrapbooks for classroom small group work and then taking a digital photo or series of photos projected on a screen for presentations performed in front of the whole class. This project has also been used in conjunction with Zoom in Breakout Rooms. In addition to explaining the digital formats described above, the structure of a unit, a list of the activities in one unit, and a sample syllabus covering the whole year are described on the poster Dewey, J. (1938). Experience and education. New York: Macmillan. Dörnyei, Z. (2001) Teaching and Researching Motivation (Applied Linguistics in Action). London: Pearson Education ESL.

As labour-doing technologies become more prevalent for general use in wider society (and educational institutions in particular), instructors may find the need for more nuanced methods with which to teach writing in English for General Academic Purposes (EGAP) contexts, to emphasise the components, skillsets, and steps necessary to create a deductive logical essay. Drawing from the scaffolding principles of Constructive Alignment (Biggs & Tang, 2011) and with consideration of GoogleSheet’s technological features (Kozma, 1991) this poster covers the development of a directly-shared GoogleSheet template to support (via formulas and hyperlinked resources) first-year university students’ initial foray into structure-and-citation focused essay-writing. In addition, the methods and formulas employed in its creation, as well as students’ positive (support-materials, sentence-level emphasis, citation guidance, portability, accessibility) and negative (digital literacy requirements, accountability, privacy) reflections on writing an essay with said GoogleSheet will be discussed. Though such an approach is unlikely to be a universal treatment for all instructors, students, and classrooms, it is hoped that this poster presentation will provide food for thought – along with some takeaway ideas and examples – for instructors considering creating similar online, hyperlinked instructional materials as bridge between paper-based and computer-based writing in the AI age.

Research on AI in education often remains anthropocentric, separating humans ‘us’ (the knowing and agentive actants) from AI/technology ‘it’ (the object, the liability, or output). This framing simplifies the complexity of human-AI intra-actions and shapes how we understand what it means to be ‘critical’ with AI. Adopting a posthumanist perspective (Barad, 2007; Jones; 2025), this paper rethinks criticality as part of our AI-mediated language learning.

This study draws on a novel digital enthnographic method using multimodal digital portraits with 50 Chinese, Thai and Dutch university students. Using elicitation prompts and follow-up interviews, these digital portraits trace how learners feel, understand and make sense of their intra-actions with AI. Data were analysed through iterative, reflexive coding informed by posthumanist theory, attending to affective, embodied, and discursive dimensions of human–AI intra-actions.

Findings positions criticality in AI intra-actions as the ways in which language learners reconfigure and reshape relations; what happens when learners and AI together reshape the boundaries of what counts as knowing, learning and understanding. For educators, this work reframes critical engagement with AI as an emergent, embodied practice and offers the digital portrait as a pedagogical space for exploring how learners make sense of AI in language learning contexts.

This study investigates how digital multimodal composing (DMC)—the creation of meaning through the integrated use of modes such as written text, images, audio, video, and animation—supports disciplinary content learning in a content and language integrated learning (CLIL) context. The study is situated in a legal English course for university-level English majors, where students engaged in DMC tasks such as scripting, filming, and editing short explanatory videos on legal topics. Although prior DMC research has largely focused on language instruction, its potential for facilitating subject knowledge development in content-based courses remains underexplored. Drawing on data from eleven student-produced videos created by 32 students, 32 accompanying reflective essays, and records of the composing process, the study examines when and how DMC contributes to content learning. Using thematic analysis and multimodal analysis, the findings show that content engagement varies across stages of the DMC process. Substantial disciplinary learning occurs during topic selection and planning, while scripting promotes the integration and reformulation of specialized knowledge for non-expert audiences. Later production stages emphasize multimodal design and involve less explicit content learning. Overall, DMC supports content learning through conceptual integration, expansion, and recontextualization, highlighting its value for higher education CLIL settings.

This systematic review synthesised findings from 40 studies on the pedagogical impacts of generative AI in multilingual and bilingual language classrooms. Evidence shows consistently positive effects on writing, with medium-to-large improvements in vocabulary use, organisational quality, and grammatical accuracy (Gao, 2023). Speaking outcomes demonstrated the strongest gains, with year-long interventions yielding learning improvements more than twice those of conventional instruction, especially when systems provided real-time, multimodal feedback (Zhao & Hua, 2025). Affective benefits were similarly robust, including reductions in anxiety and increases in self-efficacy and self-regulated learning (Zhao & Hua, 2025). Findings for reading comprehension and critical thinking were more mixed. While several studies reported clear improvements, others found minimal effects or highlighted risks of reduced independent analysis (He, 2025; Eun & Bae, 2024). Outcome variability was explained by six mechanisms: task complexity, scaffolding design, feedback timing and modality, cultural and linguistic contextualisation, learner proficiency, and intervention duration. AI was most effective when integrated into pedagogically structured tasks that preserved learner agency, provided adaptive and culturally relevant scaffolding, and offered timely multimodal feedback. Limitations emerged when AI was used passively, lacked cultural adaptation, relied on delayed feedback, or failed to sustain engagement over time. Overall, the review emphasises that AI’s impact is contingent on thoughtful, context-sensitive instructional design.

Recently, many educators have been using generative AI to create materials for their courses. However, they often report frustration with the inability of Large Language Models (LLMs) to reliably produce language appropriate to their students’ proficiency levels. This raises the question: Can LLMs be adapted to generate level-appropriate learning materials consistently? This research project aims to develop an LLM capable of producing output at each CEFR level by taking into account vocabulary, grammatical features, and lexical complexity. Three open-source LLMs were selected and fine-tuned using datasets of level-differentiated CEFR texts. This presentation explains the fine-tuning process and compares the effects of different datasets on the three models. It will also introduce tools and workflows that allow participants to fine-tune models to suit their own need. Model outputs at each level will be evaluated against CEFR benchmarks, and the output of each model compared and shared with the participants for discussion. The presentation will conclude by considering how level-controlled AI-generated texts such as these can be integrated into courses, and the ongoing implications for material development.

Digital productivity tools are increasingly promoted as cure-alls to heavy research and teaching workloads that educators face, yet practical integration often involves experimentation, setbacks, and adaptation. In this presentation, I introduce four digital tools used in my own academic workflow: Focusmate for structured accountability, Scrivener for managing long-form writing projects, Elicit.com for AI-assisted literature exploration, and Headspace as a supportive practice for sustaining focus and well-being. Drawing on reflective classroom and research experiences, I examine what has worked, what has failed, and what required rethinking and tinkering when incorporating these tools into daily professional practice. While some tools significantly improved writing consistency and research efficiency, others revealed limitations. Rather than promoting technology as a universal panacea, I emphasize thoughtful, deliberate, and critical adoption of digital tools. Participants will gain concrete strategies for experimenting with free productivity technologies, insights into common pitfalls, and practical suggestions for building sustainable digital workflows that support both teaching and research. I will encourage participants to reflect on their own productivity challenges and to approach technology integration as an evolving process of trial, reflection, and refinement.

This presentation reports on a semester-long study at a Japanese university examining the effectiveness of the pronunciation practice website English Accent Coach in improving English pronunciation among first-year students. Although pronunciation instruction is essential for spoken intelligibility (Levis, 2021), it is often underemphasized due to teachers’ limited confidence or perceptions that it is less important than other language skills (Mahmood et al., 2024). This study investigated whether regular, structured online pronunciation practice could lead to measurable improvement over one semester. Participants were divided into treatment and control groups. The treatment group used English Accent Coach as part of their weekly homework, engaging in guided pronunciation activities with automated feedback. The control group followed the same curriculum but did not use the platform regularly. Both groups completed identical pre- and post-tests measuring pronunciation accuracy and intelligibility. Preliminary results indicate greater improvement in the treatment group than in the control group. These findings suggest that online pronunciation tools may help address gaps in contexts where pronunciation instruction is limited, contributing to ongoing discussions in CALL research on effective technology use for developing spoken language skills.

In a recent talk, Microsoft CTO Kevin Scott (2025) discussed "capability overhang," the gap between what AI models can accomplish and what users actually implement. This issue is present in applied linguistics, where frontier models have demonstrated remarkable potential, yet most researchers remain limited to basic chatbot interactions or structured platforms like SciSpace. This presentation argues that agentic tools like MCP (Model Context Protocol) servers, customizable skills, and adaptive agents enable researchers to develop workflows tailored to their needs, bridging this capability gap.

This presentation will demonstrate how to build agentic research workflows for applied linguistics using tools such as Claude Desktop, Notion, custom MCP configurations, and agentic platforms. Specifically, I will show how researchers can leverage these tools to: (1) discover and retrieve relevant literature across multiple databases, (2) annotate and synthesise findings with persistent memory systems, (3) store and curate research in structured knowledge bases, and (4) streamline data analysis pipelines.

Rather than replacing expertise, this presentation shows how agentic approaches amplify it by streamlining repetitive tasks while preserving human judgment. Attendees will leave with concrete strategies for building personalised AI research assistants that can evolve alongside their projects, ultimately simplifying the path from data collection to publication.

This presentation outlines a practical process for designing institution-specific listening tests by combining statistical analysis with qualitative instructor feedback. The presenters were tasked with creating a listening test to re-stream approximately 800 first-year students into appropriate second-year course tiers, accommodating a wide range of English proficiency levels.

The development process consisted of five stages: creating and recording initial test items, selecting a content creation and assessment platform, trialling draft materials with instructors, administering two pilot versions to current second-year students, and conducting a classical test theory (CTT) analysis of both pilots to guide item selection for the final test.

The presentation explains each stage in detail, including the rationale for adopting BookWidgets as the test creation platform, the role of AI in early item generation, instructor feedback on test clarity and difficulty, and the decision to use teacher voice actors rather than AI-generated voices to enhance authenticity. Results from the CTT analysis will be shared. Attendees will leave with practical guidelines and replicable steps for developing reliable listening assessments suited to their own contexts.

In academic English courses, learners are often required to research and cite their sources appropriately. Teaching how to cite sources can be challenging for teachers. This blended research and practice-based study explores how H5P, a plug-in, supports learning APA style website citations. While previous studies reported that H5P can enhance learners' engagement and motivation, evidence on learning outcomes and instructional design is limited. Addressing this gap, the present study aims to explore how H5P supports learners' engagement and understanding of citation rules. A mixed-method approach was adopted to collect data from quiz performances and learners' feedback surveys. Seventy university students participated in a flipped learning activity by watching an H5P interactive video. The findings suggest that H5P can support learners’ engagement and understanding of website citation rules, but pedagogical impact requires careful design choices. This presentation will introduce H5P interactive videos as a tool for teaching APA citations in the L2 classroom. Participants will learn how the instructional design facilitated or hindered learning, as well as the challenges encountered during implementation. This presentation will contribute to CALL by providing insights into H5P, an underexplored tool with some implications for future research.

The purpose of this study is to explore the perspectives of Japanese university EFL learners on the use of generative AI tools, such as ChatGPT. Focusing on learner voices, it examines the benefits, concerns, and attitudes toward integrating generative AI into university English curricula. The study employed a mixed-methods survey design. A total of 400 first- and second-year engineering students at a Japanese university participated in the study. Quantitative items were used to measure AI-mediated learner autonomy, AI self-efficacy, and psychological dependence. In addition, open-ended questions were used to investigate EFL learners’ perceptions of the benefits and challenges of using generative AI tools and their potential integration into EFL curricula. The findings show that students recognize benefits such as increased efficiency and support for idea generation, while also expressing concerns about over-reliance and reduced learner effort. Attitudes toward curriculum integration varied depending on learners’ autonomy and dependence profiles. These results highlight the importance of pedagogical guidance in shaping the educational value of generative AI use in EFL classrooms.

Rapport plays a central role in fostering engagement, participation, and interpersonal alignment in English as a Foreign Language (EFL) interaction, yet student-student rapport in online settings remains underexplored. Drawing on Spencer-Oatey’s (2000) Rapport Management Model (RMM), this study investigates how Japanese university EFL learners construct rapport during Zoom-based small-talk interactions, with a particular focus on the interactional strategies used across verbal, paralinguistic, and embodied nonverbal domains. Using video-recorded dyadic Zoom interactions, the study adopts a qualitative, micro-analytic approach to capture interactional strategies for rapport development across complete interactional episodes. A stratified purposive sample of dyads was analyzed to explore patterns of rapport construction and potential gender-based differences in interactional practices. The study offers pedagogical insights into how rapport can be systematically analyzed in online EFL interaction. By linking theoretical constructs to observable interactional practices, the findings inform task design, teacher mediation, and the development of interactional competence in digitally mediated language learning contexts.

Designing a coherent language course—whether for a semester, term, or short unit—requires time, expertise, and careful alignment between outcomes, materials, and assessment. This presentation demonstrates how generative AI tools such as ChatGPT can be used as a practical curriculum design assistant to build an entire language course from scratch.

Participants will see a flexible, start-to-finish workflow that works across educational contexts, including universities, secondary schools, and elementary classrooms. The session covers how to generate a course or unit outline; control and grade the language level of materials; design integrated activities for all four language skills; create active learning and critical-thinking tasks; and produce classroom-ready resources that are fully downloadable in Microsoft Word format.

Rather than treating AI as an automated shortcut, the presentation emphasizes explicit, teacher-driven prompting that embeds pedagogical intent, age appropriateness, and proficiency constraints. Attendees will leave with transferable prompting strategies, practical examples, and a clear understanding of how AI can support efficient, high-quality course design while maintaining professional judgment and instructional control.

Vocabulary knowledge is one of the strongest predictors of reading comprehension and overall language proficiency, yet many learners spend large amounts of time studying low-frequency or poorly sequenced words. Research in corpus linguistics, frequency-based word lists, and extensive reading suggests that learners benefit most when instruction is anchored in high-frequency vocabulary and supported by large amounts of comprehensible input.

This 60-minute hands-on workshop introduces a technology-supported framework for vocabulary learning that integrates the New General Service List (NGSL), extensive reading principles, and AI-enhanced corpus tools. Participants will explore how NGSL-aligned digital tools can be used to profile texts, evaluate lexical coverage, and generate level-appropriate reading and vocabulary materials. Through guided activities and live demonstrations, attendees will see how teachers can determine whether a text is suitable for a learner group, adapt authentic materials to different proficiency levels, and design vocabulary tasks that promote noticing, retrieval, and recycling.

The workshop is grounded in research on frequency effects, lexical coverage, and input-based learning, while also demonstrating how recent AI technologies can operationalize these principles in classroom practice. Participants will leave with practical experience using NGSL-based tools for vocabulary teaching and materials development.

Generative AI is increasingly embedded in language learning contexts as a language-producing system that interacts with learners, models, amplifies, and in some contexts actively reshapes linguistic norms, while influencing language across educational and digital spaces. This poster examines the sociocultural consequences of generative AI when it functions in the role of a participant rather than solely as a mediating tool. Here, participation is understood as a functional and relational role in interaction, not as evidence of intention, consciousness, or autonomous agency. From a sociocultural perspective, AI participation reshapes communities of practice and speech communities by contributing to norm-setting, authority, and legitimacy. In some contexts, AI stabilizes dominant registers and tones; in others, it amplifies, introduces, or accelerates emerging norms, including slang, euphemism, and ideologically charged language. When AI produces language alongside learners, assumptions about effort, authenticity, and developmental trajectories are reconfigured. Learner identity, investment, motivation, and affect are shaped through altered experiences of belonging, safety, anxiety, and recognition as legitimate speakers. By foregrounding AI participation, this poster isolates how sociocultural effects emerge through norm circulation and co-construction, offering a conceptual entry point for examining how participation itself is being redefined in AI-mediated language learning contexts.

This poster reflects on how generative AI has supported my work as a language educator with ADHD, and how these same supports may reduce learning barriers for a wide range of language learners. While generative AI is often discussed in terms of productivity and content generation, less attention has been paid to its role in easing cognitive friction such as task ambiguity, task initiation difficulty, working-memory overload, and attention regulation.

Drawing on autoethnographic reflection, the poster identifies recurring AI-mediated behaviors that emerged in daily academic and instructional work: task clarification and instruction reframing, task chunking and sequencing, cognitive warm-up through drafting and outlining, real-time idea externalization, attention recovery, and low-stakes feedback. Although these behaviors align with ADHD-related executive-function challenges, they also address difficulties that non-ADHD learners experience intermittently, particularly in complex, high-cognitive-load language tasks.

A key observation concerns availability: generative AI provides continuous, on-demand access to cognitive support, lowering help-seeking barriers and enabling frequent, context-sensitive use. Using a light Universal Design for Learning lens, the poster argues that generative AI can reduce barriers by externalizing cognitive processes in ways that benefit many learners, not only those with diagnosed learning differences.

Implications for CALL pedagogy and accessibility-oriented design are discussed.

Generative artificial intelligence (Gen AI) tools are increasingly utilized to support autonomous English language learning in higher education. However, research remains limited regarding how Gen AI use relates to student learning in the context of ESP, particularly among first- year non-English- major students transitioning from a high school environment to a university setting. The research therefore, aims to explore how first- year students use Gen AI tools to assist their ESP learning routines and their perceptions of the use of these tools in ESP class. Adopting a mixed-methods approach combining questionnaire data with semi-structured interviews, the study surveyed 158 first-year students across various academic disciplines at a university in Ho Chi Minh City, Vietnam. The results indicate that a majority of ESP students used Gen AI tools more frequently for tasks related to their assignments and exams. Furthermore, the study also highlights psychological shifts of some students feeling overwhelmed by the ubiquity of AI use compared to their high school experience. Finally, a change in teacher-student dynamics was observed, as students expressed a preference for AI assistance over seeking instructor intervention for tasks they perceived as manageable through automation.

This qualitative study aims to initiate a 5-month online yarning circle (relational indigenous methodology fostering mutual help through reflection and discussion [Shay, 2021]) in teacher training on enhancing English pre-service teachers’ (EPTs’) corpus and AI literacy in English oral teaching skills. Three doctoral students who had received applied linguistics training for over 4 years served as mentors and paired up with 3 undergraduate students who were EPTs. The three mentors facilitated weekly discussion with their mentees on key indicators that influence English oral performance and communication strategies that could be used in English oral communication and co-developed corpus and AI-integrated English speaking lesson plans.

Qualitative results identified the effectiveness of this online yarning circle. Both mentors and mentees highlighted the benefits gained from their partnerships in English oral teaching. The EPTs reported growth in their linguistic knowledge, corpus and AI literacy, and the ability to integrate linguistic knowledge with educational technology to design English speaking lessons. Mentors learned how to develop teacherfriendly resources from EPTs. For instance, mentors provided guidelines on key indicators that influence English speaking performance. The EPTs offered suggestions to make the indicators more suitable for secondary school contexts based on their teaching practice in local schools.

The emergence and rapid development of Artificial Intelligence (AI) have led to significant changes in how humans live and learn (Isharyanti et al., 2025). In the field of education, it has created a need to teach people how to use it skilfully, ethically, and responsibly (European Commission & OECD, 2025; Walter, 2024). Concurrently, its use is prevalent and inevitable in today's teaching and learning processes, as more applications of AI are explored and practiced in education. Teachers in this AI era thus need to add a new set of competencies to their professional identities to address this necessity and incorporate AI into their teaching practices. However, considering the novelty of AI, many topics pertinent to teachers' competences remain unexplored. The poster aims to present the outcomes of a systematic literature review on EFL teachers' AI literacy and AI pedagogy, among other findings, a general positive attitude toward AI for pedagogical purposes. However, teachers face challenges in ensuring criticality toward AI-generated content and ethical use of AI, and in closing the AI divides in technical, financial, and infrastructure capabilities. The poster offers audiences potential topics and research gaps in the field, as well as an opportunity to discuss and collaborate with the presenter.

Generative artificial intelligence (GenAI) has received considerable attention in second language (L2) research since the release of ChatGPT. A recent systematic review by Li et al. (2025) identified 144 peer-reviewed papers on GenAI in the L2 context within two years. While this review highlights the affordances and constraints of GenAI, most of the included studies focused on the L2 English context. Therefore, more research is needed to understand how learners of non-English languages perceive and experience GenAI in L2 learning settings. This presentation reports on a study that addresses this gap in the literature through a synthesis of GenAI studies on L2 learning in non-English contexts. Web of Science will be used to search for primary studies involving GenAI and the L2 learning of non-English languages between 2022 and 2026. Data will be reported following PRISMA guidelines. Titles, abstracts, and full texts will be screened using predefined inclusion criteria. Methodological features of the studies will be analyzed to identify trends. The findings from the included studies will be synthesized using thematic analysis to identify the benefits and challenges associated with the use of GenAI in the teaching and learning of non-English L2s. Implications of the study will also be discussed.

UNESCO’s AI competency framework for teachers (Cukurova & Miao, 2024) describes the teacher’s AI competency in five aspects: human-centred mindset, ethics of AI, AI foundations and applications, AI pedagogy, and AI for professional development. At the heart of teacher professionalism, AI pedagogy is arguably the most important aspect, as it equips teachers to integrate AI purposefully and effectively into their teaching. Several studies on the implementation of AI in teaching and learning have called for teacher training to ensure successful AI pedagogy. However, to develop competent AI pedagogical capacity, teachers need to first acquire the other four aspects outlined in the framework. A preliminary study of other aspects is necessary to develop AI pedagogy training that aligns with their current level of AI literacy. The study aimed to discover the state of Indonesian EFL teachers' AI literacy and their most pressing training needs. Through a wide-scale online survey of 578 English teachers in Indonesia, the study gained insights into the current level of teachers’ AI literacy and analysed their teacher professional development needs, which contributed to the development of research-based AI pedagogy training aligned with the field's realities.

This presentation explores a practical, classroom-tested approach to working with AI rather than against it in Oral English courses. Acknowledging that students already use AI tools, we argue that technological change reshapes learning demands rather than eliminating learning itself. A common classroom challenge is that students often focus on producing a “perfect” written script for presentations rather than developing spoken fluency. This emphasis encourages memorization, which can increase anxiety and reduce students’ confidence when speaking spontaneously. To address this issue, we redesigned a speaking task that prioritizes human cognition before AI involvement. Students first research authentic data, develop their own ideas and sentences, and organize speaking notes independently. Only after this stage is AI introduced through clearly bounded tools rather than a single solution. NotebookLM supports source-based research within transparent constraints, while ChatGPT provides prompt-driven language scaffolding focused on clarity, structure, and delivery rather than content generation. Together, these tools form an ethical workflow that reduces anxiety and improves speaking performance. The presentation concludes by discussing implications for assessment design, AI literacy, and teacher mediation in CALL, emphasizing that effective pedagogy lies not in banning AI, but in teaching students when, why, and how to use it responsibly and ethically.