Speaker
Description
This research explores the value of generative AI as a tool for supporting multimodal reading and writing in the creation of crime stories. Currently, at this research site, the use of generative AI for expository essays is against program policy, but narrative creation is allowed as a no-stakes activity. To elucidate patterns in which language students explore and use generative AI as part of their language study, This study at a Japanese university, explored the tools and strategies that English students employ when given the task to work with a classmate to iteratively prompt ChatGPT (or equivalent LLM AI) in the creation of a crime narrative. The research questions were, 1) What strategies and tools do students use when crafting prompts to develop a crime narrative? 2) What concepts and skills are needed for students to generate comprehensible output that meets the criteria for a mock trial narrative? CEFR B2- English students (in pairs) chose a crime and recorded their iterative input and output from ChatGPT in developing a crime narrative. Data from three classes were analyzed. Pairs of students were observed in the process of using the AI and field notes were taken. Follow-up interviews and observations were used to validate prior data. Finally, a survey of language program students was conducted to elicit more general attitudes toward and patterns of use regarding generative AI tools. The results of this study will elucidate basic issues regarding L2 student prompt authoring and revision, the types of online tools students use, and strategies needed to create a crime narrative with the help of AI systems. Implications for L2 learner, human-AI hybrid authorship of narratives, and the normalization of tools like ChatGPT for language learning through hybrid authorship will be discussed.
| Keywords | Human-AI hybrid learning, multimodal CALL, hybrid authorship |
|---|
