Faculty Publications and Conference Presentations

This study compares formative feedback from ChatGPT-3 and human teachers on student essays, evaluating language, coherence, and content/argumentation in writing. ChatGPT primarily addressed surface-level issues like typos, citations, and sentence length, even when prompts targeted deeper aspects such as thesis development, argumentation, and structural coherence. In contrast, experienced teachers provided higher-quality feedback, offering nuanced insights into metacognitive processes, such as synthesising information, extrapolating and inferring from evidence and stance-taking. While ChatGPT’s feedback remained conservative and unchanged despite revised prompts, teachers identified subtle areas for improvement to guide students toward more sophisticated writing. These findings suggest that as of now ChatGPT is useful for proofreading and addressing mechanical language accuracy issues but lacks the depth to support advanced writing development. Consequently, Generative AI can be used as a supplementary tool for feedback in conjunction with critical comments from skilled educators in fostering deeper writing competencies. <br/>Is this a duty-visit?<br/>

This paper evaluates the implementation of a chatbot in an elementary Japanese language course at a tertiary institution, examining its benefits and challenges through detailed student feedback. Utilising a mixed-methods approach that incorporates both qualitative and quantitative research methods, data was collected over several months from 40 students enrolled in a zero-beginner Japanese course at a university. The chatbot, designed to function as a private tutor, assisted students with vocabulary, grammar, and pronunciation and provided an interactive and personalised learning environment. Student feedback was systematically categorised and analysed to identify trends and patterns in usage and satisfaction. The results indicate that while chatbots offer personalised and interactive learning experiences, their effectiveness is contingent on the accuracy of responses and their ability to adapt to the evolving needs of students as their proficiency increases. The study underscores the need for enhanced features, such as improved contextual understanding and accuracy, to support students’ progress in language learning. The study identifies several challenges, such as ensuring the accuracy of chatbot responses by constantly monitoring the information provided in relation to learners’ proficiency levels. These challenges highlight areas for future improvement and research.

The implementation of English Across the Curriculum (EAC) in Hong Kong has evolved to meet the diverse needs of non-native English-speaking students. Over the past two years, EAC initiatives at three universities have embraced multimodal learning, enabling students to express ideas through various digital formats. A newly formed inter-university EAC project team, co-led by faculty in engineering, ESL, and educational development, has developed self-designed GenAI platforms to support writing and speaking skills. Some of these platforms provide self-paced learning and personalized feedback, making language learning more interactive and effective. Additionally, another university language centre, in collaboration with its educational innovation unit, is developing a Massive Open Online Course (MOOC) to enhance learners’ proficiency in using GenAI for text analysis and summarization while fostering their ability to evaluate arguments, detect biases, and produce high-quality written and visual content, all within an ethical and responsible AI framework. This panel explores how “traditional” EAC initiatives now incorporate common GenAI tools and presents updates on technology-driven EAC projects that leverage AI to create a more responsive learning environment. It also discusses the challenges EAC faces in the era of GenAI in Hong Kong and highlights issues that may resonate with the wider WAC community.