Mental representation of classifier phrases

The constituency structure of a phrase consisting of a noun, a numeral, and a numeral classifier (henceforth, classifier phrase) has been widely debated. One view holds that there is a universal or a preferred underlying structure of classifier phrases, though theories differ on the internal structure of the phrase. For example, some argue that classifiers form a constituent with nouns (e.g. Dékány, 2022) and others argue classifiers form a constituent with numerals (e.g. Greenberg, 1972). Another view argues that the structure of classifier phrases vary across languages or even across different types of classifiers within a language (e.g. Gil, 2013, Little et al. 2022). In other words, there is no universal or preferred underlying structure of classifier phrases. The representation of classifier phrases in a specific language is therefore shaped by language experience. In our previous research, we conducted a series of artificial language learning experiments to investigate whether learners have a preference among word order patterns predicted by the two constituency structures described above. Our findings suggest that, in the absence of explicit evidence, learners were approximately equally split between consistently using the order predicted by one of the two structures (Wang, 2024). This bimodal distribution suggests that there may be some flexibility in the representations of classifier phrases across learners. Here we build upon this line of research and investigate whether learners’ representation of classifier phrases can be changed and how it is influenced by the language input they receive.