“…Although there are many studies on automatic test item generators for English language testing, the implications for testing Chinese as a second language are yet to be explored. Artificial Intelligence (AI) technologies, which are receiving increased attention in the field of automatic vocabulary item generation, can fill this gap for three reasons (e.g., Susanti et al, 2020; Ulum, 2020): firstly, both selected- and constructed- response formats can be generated with the application of NLP: (1) cloze items (Sakaguchi et al, 2013), (2) multiple-choice vocabulary items (e.g., Aldabe et al, 2006; Hoshino & Nakagawa, 2005), and (3) error correction items (e.g., Aldabe et al, 2006). Secondly, NLP technologies have the potential to create a larger number of distractors for multiple-choice questions (MCQs) in a short period of time using the four approaches as follows: (1) the corpus-based approach (e.g., Aldabe & Maritxalar, 2010), (2) the graph-based approach (e.g., Papasalouros et al, 2008), (3) Word2vec (e.g., Mikolov et al, 2013) and (4) visual similarity (e.g., Jiang & Lee, 2017).…”