Processing of Chinese Classifier-Noun Agreement: An Event-Related Potential Study

• Main Authors: Shu-hua Tsai, 蔡淑華
• Other Authors: Jenn-yeu Chen
• Format: Others
• Language: en_US
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/36360281663474554388

碩士 === 國立成功大學 === 認知科學研究所 === 97 === The transparent semantic relationship in most of the numeral classifier-noun phrases led some linguists to the belief that classifier-noun agreement was semantic in nature, and some even raised the possibility of Chinese native speakers’ using classifiers as taxonomic category labels in classifying world objects. However, since classifier-noun agreement was not always explicable based on semantics, a pure semantic view was questioned. In the study, classifier-noun agreement was examined in word pairs and sentences by using the event-related potential (ERP) technique. In Experiment 1, the electroencephalogram was recorded while 22 Chinese native speakers read word pairs in which taxonomic category, classifier, or word category relationship was manipulated. Disagreement in word pairs formed by a category label-noun (e.g., animal-dictionary ‘動物-字典’), a classifier-noun (e.g., a-xiCL_clothing-like – dictionary’一襲-字典’ ), an intensifier-noun (e.g., quite-dictionary’相當-字典’) produced an N400 effect (250 ms-450 ms) with the intensifier-noun disagreement showing an additional sustained negativity in the frontal lobe (500 ms-800 ms). These ERP effects supported the idea that classifier-noun agreement was semantic in nature, similar to taxonomic category processing but different from grammatical agreement between the intensifier and the noun, especially in the later phase of processing. In Experiment 2, exactly the same classifier-noun pairs, both acceptable and anomalous, were set in sentences with embedded object-gapped relative clauses, thus creating a long-distance dependency of the classifier-noun agreement. The classifier-noun relation was manipulated in such a way that in one condition the classifier matched its adjacent noun but not the second noun in the distance (Match-Short), in another the classifier matched its head noun in the distance but not the adjacent noun (Match-Long), and in the third condition the classifier clashed with both the adjacent noun and the distant head noun (Mismatch). Thirty native Chinese speakers joined the experiment. A frontal negativity with a central focus was found corresponding to the mismatches between the classifier and the adjacent noun (Match-Long and Mismatch conditions), reflecting the attention and working memory load arising from readers’ keeping the classifier abreast while at the same time searching for a proper head noun to go with it. A relatively increased N400 component at the central lobe was observed when the critical long-distance head noun mismatched with the given classifier (Match-Short and Mismatch conditions). This implicated that classifier-noun agreement was semantic in nature, even when it occurred in complex sentences with object-gapped relative clauses. An additional P600 component was found at the long-distance head noun position in both Match-Long and Mismatch conditions, revealing readers’ efforts to integrate the classifier with the long-distance head noun even when the noun mismatched with the classifier. Also, P600 was as a clear indicator for sentence integration in the study. The overall results showed that regardless of the context, in which a classifier occurred, the processing of classifier-noun agreement was semantic in nature and the violation of the agreement served as a precursor of a relative clause.