「分類詞─名詞搭配」再探：分類詞特性與個人因素對於搭配關係之影響

Classifiers form an important but intricate system in Mandarin, and theoretical linguists have yet reached a consensus regarding whether classifiers are semantic or functional elements. The PI’s past research (Chan, 2019) has revealed that both semantic and grammatical processing are involved in processing classifier-noun agreement; however, why these two processes co-exist merits further exploration. This two-year project thus aimed to examine the co-existence through two experiments. The first experiment manipulated the existence/absence of the “class term” in a noun to see if it affected the agreement processing of classifier-noun combinations. The results revealed that the existence/absence of a class term did not differentially modulate the ERP response to the agreement relationship, suggesting that the co-existence of semantic and grammatical processing might be induced irrespective of the superficial structure of the noun in a classifier-noun pairing. The second experiment was to find out whether linguistic classifier-noun pairing could influence non-linguistic picture recognition in speakers with different native languages. The results showed that, although classifier-noun agreement was not required to identify an image, Mandarin speakers’ ERP response was still modulated by the object’s classifier-noun pairing, while English speakers failed to show such modulation. Taken together, both experiments suggest that the classifier system is no longer a system limited to language; instead, it is intertwined within the conceptual representations, along with other conceptual features, such as sensory and functional features. Therefore, when a concept is activated by linguistic information (as the classifier-pairing in Experiment 1) or sensory information (as the images in Experiment 2), the related features of classifiers are also activated. Since the features of classifiers include both semantic and grammatical ones (so that acceptable classifier-noun agreement can be achieved in everyday communication), we thus can observe both semantic and grammatical processing with the ERP technique in previous studies.