Faculty Publications and Conference Presentations

This presentation focuses on a case study of senior Math undergraduates’ learning of popular science communication through scaffolded reflective practice. The case begins with the challenge of transitioning from academic writing (essays and mathematical proofs) to writing for the general public. To students of Math/Computer Science majors whose first language is not English (despite learning in an English Medium Instruction/EMI environment), transitioning between two different writing styles doubles up on their existing effort of language development, as these students need to not only acquire a style that is less practised in academia but present abstract/ advanced math ideas to audience outside of the silo of math. To enable Math students to practice the popular-science communication style, the teaching of written communication has resorted to a preset learning portfolio and scaffolded practice of written reflection. This alternative ‘process writing’, as evidenced in students’ practices and their final products, has enabled various degrees of transition from the research-focused, impersonal and formal academic style to a more audience-centred and engagement-conscious approach to written communication. The consistent practice of written reflection also intends to mitigate unproductive use of GenAI for language learning – instead of generating the most probable content for task completion, students are encouraged to explicate the development of ideas, personal voices and choices of styles. As conclusion, this talk will discuss what constitutes critical reflection and how (possible) to assess written reflection by comparing human reflection and statistical modelling of human reflection.

TEACHER PERCEPTION OF GENAI AS AN ASSESSMENT AID<br/>Bruce Ma and Martin Ma<br/>Center for Language Education, Hong Kong University of Science and Technology, Hong Kong<br/>lcmbruce@ust.hk; lcmartinma@ust.hk<br/>ABSTRACT<br/>The emergence of Generative AI (GenAI) in November 2022 has initiated a proactive integration of this technology into pedagogical practices <br/>at the HKUST Center for Language Education (CLE). Apart from encouraging and teaching students to use GenAI, HKUST has made <br/>assessment-related GenAI tools accessible to faculty. One such tool is Pregrade - a grammar and spelling grader with the capability to <br/>assess and comment on writings based on a detailed rubric (Pregrade, 2024). HKUST’s second year Technical Communication course, <br/>designed for Engineering students, spearheads the university's integration by incorporating various GenAI learning strategies. This course’s <br/>ESP nature presents difficulties for language teachers who have little to no engineering background as one of its assessments revolves <br/>around analyzing the ethics of engineering incidents. The large amount of information that is needed to grade this assessment poses a <br/>challenge to the efficiency and consistency in grading. Pregrade has limited application in this case as the assessment requires specific <br/>content-based grading which it does not currently offer (Pregrade, 2024). Therefore, a tailored, content-based solution was developed: our <br/>course-specific GenAI bot designed to provide feedback for teachers.<br/>While tertiary students have a strong inclination to embrace GenAI (Arowosegbe, 2024; Chan &amp; Hu, 2023; Kohnke, 2024; Li et al., 2024), <br/>our literature review highlights the transformative potential of GenAI tools in language assessment and teacher feedback practices as well. <br/>Studies indicate that AI can enhance grading efficiency and provide valuable, accurate feedback to students (Mohamed, 2023), while also <br/>streamlining the assessment process, ultimately saving teachers time and effort (Koraishi, 2023). Yet, there is a gap in research examining <br/>the pedagogical effects of these tools, particularly regarding student and teacher attitudes toward AI integration in the classroom (Har &amp; Ma, <br/>2023). While considerable research has focused on student attitudes, teachers' perceptions and experiences have received relatively less <br/>attention. This study aims to explore teachers' perceptions and experiences with our course-specific GenAI bot that serves as a marking <br/>assistant, emphasizing its usability, reliability, efficiency, and overall impact on teaching practices. By addressing this gap, the study <br/>contributes to the ongoing dialogue about the role of GenAI in educational assessments and its broader implications for teaching.<br/>According to Ma (2017), the integration of technology provides an innovative alternative for delivering written feedback on student writing <br/>more effectively and efficiently. This shift is particularly evident in the development of Automated Writing Evaluation (AWE) tools, which <br/>utilize computer-based educational technology to offer students constructive feedback on their writing. The emergence of GenAI has the <br/>potential to revolutionize the use of AWE tools in assessing writing as it can enhance traditional AWE systems by providing more contextaware feedback. Unlike conventional AWE tools, which often rely on rule-based algorithms, GenAI can understand and generate natural <br/>language, allowing it to offer suggestions that are more aligned with the specific contexts and needs.<br/>Building on this content-aware feature, our course-specific GenAI bot, based on Claude-3.5-Sonnet-200k, is pre-loaded with a <br/>comprehensive knowledge base of engineering incidents. It is designed to assist teachers in evaluating students’ engineering analytical <br/>reports by comparing the evidence presented in student writings against this knowledge base. The bot generates accessible summaries of <br/>the analysis in both text and tabular formats, highlighting any missing components to enable teachers to quickly identify gaps in the reports.<br/>An explanatory mixed methods approach is adopted for this study, combining quantitative and qualitative data collection techniques to <br/>provide a comprehensive understanding of teachers' experiences with the GenAI bot. Specifically, these experiences relate to the extent to <br/>which the bot supports teachers in assessing whether students can identify factors contributing to engineering disasters, highlight the <br/>interplay between these factors, analyze engineers’ accountability, apply ethical codes to hold relevant engineers responsible, and evaluate <br/>the overall quality of students’ analyses in connecting these elements. We are currently in the data collection phase of the study, which is <br/>scheduled to span two semesters, ending in May 2025.<br/>REFERENCES<br/>Arowosegbe, A. (2024). Students’ perception of generative ai use for academic purpose in UK higher education. <br/>https://doi.org/10.20944/preprints202405.1158.v1<br/>Chan, C., &amp; Hu, W. (2023). Students’ voices on generative AI: Perceptions, benefits, and challenges in higher education. International <br/>Journal of Educational Technology in Higher Education, 20(1). https://doi.org/10.1186/s41239-023-00411-8<br/>Har, F., &amp; Ma, B. W. L. (2023). The future of education utilizing an artificial intelligence robot in the Centre for Independent Language <br/>Learning: Teacher perceptions of the robot as a service. In C. Hong &amp; W. W. K. Ma (Eds.), Applied degree education and the <br/>shape of things to come. Springer, Singapore. https://doi.org/10.1007/978-981-19-9315-2_3<br/>Kohnke, L. (2024). Exploring EAP students' perceptions of GenAI and traditional grammar checking tools for language learning. Computers <br/>and Education: Artificial Intelligence, 7, 100279. https://doi.org/10.1016/j.caeai.2024.100279<br/>Koraishi, O. (2023). Teaching English in the age of AI: Embracing ChatGPT to optimize EFL materials and assessment. Language Education <br/>and Technology, 3(1), 55–72. https://langedutech.com/letjournal/index.php/let/article/download/48/37… to <br/>Day 1<br/>Back to <br/>Day 2<br/>Back to <br/>Day 3<br/>Li, Y., Deng, Y., Peng, B., He, Y., Luo, Y., &amp; Liu, Q. (2024). Generative artificial intelligence in Chinese higher education: Chinese <br/>undergraduates’ use, perception, and attitudes. Frontiers in Educational Research, 7(4). <br/>https://doi.org/10.25236/fer.2024.070401<br/>Ma, B. (2017). The study of teacher written feedback: The effectiveness of electronic feedback on student writing revisions (Doctoral thesis, <br/>Durham University).<br/>Mohamed, A. (2023). Exploring the potential of an AI-based chatbot (ChatGPT) in enhancing English as a foreign language (EFL) teaching: <br/>Perceptions of EFL faculty members. Education and Information Technologies. https://doi.org/10.1007/s10639-023-11917-z<br/>Pregrade. (2024). Pregrade user guide, Pregrade. AI, https://app.pregrade.ai/guide/index.html#<br/>KEYWORDS<br/>assessment, written corrective feedback, GenAI, automated writing evaluation, ESP<br/>BIODATA<br/>Dr. Bruce Ma is an educator and researcher with a focus on writing feedback, learner autonomy, and second language writing. His research <br/>interests also include the integration of generative AI in teaching and learning, as well as computer-aided language learning.<br/>Martin Ma specializes in English for Academic Purposes (EAP) and English for Specific Purposes (ESP). His research interest includes <br/>applying Systemic Functional Grammar (SFG) notions, such as Thematic Progression and Grammatical Metaphor, to teaching, as well as <br/>applying SFG concepts through GenAI

This study explores how generative AIs interpret metaphorical expressions in Japanese, comparing the performance of text-generative and image-generative models. Metaphors rooted in embodiment and cultural context present unique challenges for AI comprehension. Using a corpus-based approach, metaphorical expressions involving body parts were extracted from the Balanced Corpus of Contemporary Written Japanese (BCCWJ) and the Tsukuba Web Corpus (TWC), and their interpretations were evaluated by text AIs (Copilot, ChatGPT-4) and image AIs (DALL·E 3, Copilot).<br/>Text-generative AIs consistently demonstrated superior understanding, accurately translating metaphors such as データが飛ぶ (data flies; lose data) and 頭が切れる (sharp-witted), preserving figurative meanings in English. In contrast, image-generative AIs often produced literal or incongruous visuals—e.g., flying objects for データが飛ぶ—highlighting a gap in metaphorical comprehension.<br/>The metaphors analysed include: 口が重い (reticent), 耳が痛い (painfully true), 目がない (extremely fond of), 顔が広い (well-connected), 首を切る (to fire someone), 胸に穴が開く (to feel empty), 腹を割る (to speak frankly), 腰が低い (humble), 手を抜く (to slack off), 腕が鳴る (eager to show skill), and 肝を冷やす (frightened). Some expressions were rejected by image AIs due to ethical or safety filters, further underscoring their literal processing.<br/>This discrepancy raises questions about why image AIs do not leverage text-AI outputs for metaphor interpretation. Possible factors include architectural differences, contextual limitations, and training data divergence. While humans intuitively grasp embodied metaphors and imagine abstract visuals, current AIs remain constrained by literalism and data boundaries.<br/>These findings contribute to understanding embodiment in language processing and suggest pathways for improving multimodal AI systems. Bridging cognitive linguistics and AI research may enhance metaphor comprehension and cross-modal integration in future models.<br/>

This study investigates the sociolinguistic and psycholinguistic implications of a non-inclusive Cantonese term 北姑 bak1gu1 (‘northern girl’), which refers to women from Mainland China in Hong Kong. Corpus analyses reveal that bakgu is predominantly used in informal, pejorative contexts, often with discriminatory and sexualized connotations. A thematic analysis of interview data highlights significant differences in perception between locals and non-locals, shaped by geographical origins, language identity, and sociocultural contexts. While locals associate the term with its second morpheme, gu1 (‘girl’), reflecting gendered and sexualized connotations, non-locals focus on the first morpheme, bak1 (‘north’), interpreting it as a marker of their outsider status. Furthermore, native Cantonese-speaking non-locals perceive the term to exclude them due to their linguistic affiliation, whereas non-native Cantonese speakers view it as a general reference to Mainland women, amplifying their sense of exclusion. This study underscores the role of non-inclusive language in reinforcing social hierarchies, regional identities, and marginalization, contributing to broader discussions on language, discrimination, and inclusivity. The findings have implications for fostering diversity, equity, and inclusion (DEI) in transcultural education and beyond.

過去3年間にわたり大学の日本語初級コースLang1210のコースデザインの実践に取り組んできた。3年間の実践で、まずは、Blended Learningデザインの設計に取りかかった。教室での授業、家庭でのe-Learning、それぞれの良さを生かすことを考えた。学生にとって生産性の高い効果的な学びにするためにはどのようなデザインの視点が大切なのか検討した。また、学び方を工夫しながら少しでも学びの楽しさや学習動機の向上につなげるためにはどのようなデザインの視点が大切なのか検討した。教室での学習も、家庭でのe-Learningでも学生が受け身にならないように、自律的に学べるようにすることを念頭に置いた。家庭学習を考えると、学生一人ひとりが使えるツールが必要であると感じた。そこで、人工知能テクノロジーの産物であるChatbotの有効利用に着手した。Chatbotには改善の余地があるが、利用価値が大きく存在した。また、教室では学生の協働的な学習環境の構築、よりよく刺激し合え、力を注ぎ合えるグループ活動の在り方を検討し開発した。学生同士の学びにはむしろ楽しさやよりよい刺激が見え隠れした。こうした3年を経て、見えてきたもの、また、この先、どのような方向性が大切なのかを示したい。<br/><br/>Over the past three years, efforts have been made in the practical application of the course design for the Japanese beginners course Lang1210 at the university. Initially, the focus was on designing a Blended Learning approach. This involved leveraging the strengths of both classroom instruction and e-Learning at home. The aim was to explore what design perspectives are crucial for making learning productive and effective for students. Additionally, considerations were made on what design perspectives are essential to enhance the enjoyment of learning and motivation by improving learning methods. The emphasis was on ensuring that students are not passive but rather autonomous learners, both in the classroom and during e-Learning at home. In terms of home study, it was felt that each student needed accessible tools. Therefore, the effective use of Chatbots, a product of artificial intelligence technology, was initiated. Although there is room for improvement with Chatbots, they have significant utility. In the classroom, the focus was on constructing a collaborative learning environment and developing group activities that encourage mutual stimulation and effort. The learning between students often revealed elements of enjoyment and positive stimulation. After these three years, the aim is to present the insights gained and the important future directions.<br/>

チャットボットの言語教育への利用は、個人化されたインタラクティブな学習体験を提供する可能性を示唆している。一方、チャットボットによる応答の正確性や文脈に即した情報提供などには課題が残り、その使用の効果を妨げる障壁となっている。初級日本語コースにおけるチャットボット使用を評価した先行研究を基に、本研究では、チャットボットの応答の正確性や文脈に即したフィードバックを提供する能力など、改善すべき重要な領域を特定し、文法データベース、語彙定義、誤用などの情報を参照させ、より先進的なAIモデルを活用することで、チャットボットの正確性を向上させるためのフレームワークを提案する。この提案は、実験的な環境で調査・テストされ、予備結果は、チャットボットの応答の正確性と文脈理解において改善を示している。本研究は、AIを活用した教育に関して実践的な解決策を示し、言語学習におけるチャットボットの有用性を高めることに貢献する。今後の研究課題としては、チャットボットの長期的な影響を評価するための縦断的研究や、異なる学習者レベル間での比較など、チャットボットの使用に関するより包括的な理解を提供することなどが挙げられる。