Exploring generative AI in learning context generation for language learners: Friend or a foe for Wordhyve?
Submitted by:
Mohammad Nehal Hasnine
Abstract:
One year after the release of ChatGPT, there has been much discussion about how various types of Generative AI could be explored within the realm of language learning. According to Huang et al., Generative AI tools are significantly superior to their predecessors, heralding a new era in education and offering unparalleled potential to revolutionize learning experiences and outcomes (Huang et al., 2022). However, many language learning and teaching aspects are yet to be explored with Generative AI models. For example, learning context- refers to the learning environment, including the sociocultural environment where learning takes place (Gu, 2003). In language learning, many questions related to understanding and defining learning contexts are yet to be answered with Generative AI, although these models can produce human-like content. This research aims to explore the potential of Generative AI models, particularly Large Language Models (LLMs), on learning logs that are collected using the Wordhyve app (Hasnine et al., 2023) in order to produce learning contexts for each word a learner wishes to memorize. The Wordhyve app collects information on what kind of vocabulary a learner has learned. Hence, the app collects data on diverse modalities such as text, time, place, learner's vocabulary level, memos, and images associated with the vocabulary. Wordhyve can produce human-like learning contexts from the image the learner uploads to create a vocabulary learning log. However, the limitation of this approach is that when the learner uploads no image, Wordhyve fails to produce the learning context. Therefore, this study leverages the potential of large language models (LLMs) for producing human-like learning contexts using the learning logs collected by the Wordhyve app. This paper finds answers to the following RQs: RQ1: Can we use generative AI on Wordhyve collected mid-sized data to generate learning contexts? RQ2: What are the similarities and differences in learning contexts produced using generative AI, deep learning, and other data-driven approaches? RQ3: Can the generative AI generate appropriate images and learning contexts for words such as abstract nouns and adjectives? RQ4: Are the recommended learning contexts appropriate for the learner considering their cultural background and previous vocabulary level?
Huang, W., Hew, K. F., & Fryer, L. K. (2022). Chatbots for language learning—Are they really useful? A systematic review of chatbot‐supported language learning. Journal of Computer Assisted Learning, 38(1), 237-257. https://doi.org/10.1111/jcal.12610
Gu, M. M. (2010). Identities constructed in difference: English language learners in China. Journal of Pragmatics, 42(1), 139-152. https://doi.org/10.1016/j.pragma.2009.06.006
Hasnine, M. N., Wu, J., & Ueda, H. (2023). Wordhyve: A MALL Innovation to Support Incidental Vocabulary Learning. In International Conference on Human-Computer Interaction (pp. 246-250). Cham: Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-35998-9_34