Effects of Assisted Tools and Learning Conditions on L2 Vocabulary Learning: A Study based on Large Language Model

In this study, we investigated the effects of assisted input, namely gloss and dictionary, on L2 incidental vocabulary learning. Furthermore, intentional and incidental learning conditions were also compared while merely utilizing a dictionary. Additionally, the gloss material was provided by a large language model (LLM) ChatGPT through prompt engineering. Besides, learning gains were measured not only solely from knowledge breadth (form-meaning connection) but also from more dimensions regarding knowledge depth (synonym discrimination, derivation production, collocation production). Sixty-four English learners of grade 2 from a senior high school were divided into three treatment groups and one control group. Those two kinds of comparisons were made respectively between every three groups. Results indicated that the gloss provided by LLM showed efficiency in collocation retention while the dictionary brought better effects in derivations and synonym discrimination. Furthermore, intentional learning may exert a good role in the long-term retention of knowledge depth and enhanced synonym discrimination effectively. The results are discussed along with students’ feedback from the questionnaire.