Animacy And The Processing Of Classifier-noun Combination In Chinese

The objects in the world can be generally divided into animate and inanimate ones, according to their animacy. And the information of animacy can be represented and stored in the mental lexicon of a noun, which makes the noun also carry the attribute of animacy. So far, great amount of evidence from cognitive neuropsychology and brain imaging has shown that the semantic memory network in the brain is sensitive to the distinction of animacy (for recent reviews, see Mahon & Caramazza, 2009; Martin, 2007).The present study aims to investigate how the information of animacy plays its role during sentence comprehension. Three theoretical views have been proposed: 1) the traditional view, which is based on the distinction of syntactic and semantic (e.g., animacy information); 2) the heuristic view, in which semantics like animacy has been assumed to play a role functionally prior to syntax; and 3) the interface view, which pronounces against the distinction between syntax and semantics. Instead, the concept of'prominence'is suggested (for very recent review, see Bornkessel-Schlesewsky & Schlesewsky, 2009). These disputes have been focused on how the information of animacy influences thematic role assignments, including role identification and role prototypicality evaluation.In the domain of sentence processing, the traditional psycholinguistic studies have been keeping animacy as one of the most important variation (e.g., Clifton et al., 2003; Feldman, 1986; Ferreira & Clifton, 1986; Frisch & Schlesewsky, 2001; Grewe et al., 2006, 2007; Hoeks et al., 2004; Jelinek & Demers, 1983; Kim & Osterhout, 2005; Kuperberg et al., 2003; Li et al., 1993, 2006; MacWhinney et al., 1984; Mak et al., 2002, 2006; Philipp et al., 2008; Roehm et al., 2004; Traxler et al., 2002, 2005; Trueswell et al., 1994; Weckerly & Kutas, 1999; for a review, see Bornkessel-Schlesewsky & Schlesewsky, 2009). Using eye-tracking, event-related brain potential, and other techniques, researchers have investigated the role of animacy both in processing syntactically complex or ambiguous sentences (e.g., Clifton et al., 2003; Ferreira & Clifton, 1986; Mak et al., 2002, 2006; Traxler et al., 2002, 2005; Trueswell et al., 1994; Weckerly & Kutas, 1999) and in processing syntactically simple and unambiguous ones (e.g., Frisch & Schlesewsky, 2001; Grewe et al., 2006, 2007; Hoeks et al., 2004; Kim & Osterhout, 2005; Kuperberg et al., 2003; Philipp et al., 2008; Roehm et al., 2004). However, it should be noted that these studies were all conducted on the background of noun-verb combination processing, with verb-argument-structure processing or thematic role assignments being involved. Crucially, the properties of animacy processing in the noun-verb combination, which have been revealed in numerous previous studies, do not necessarily suggest the same properties for the processing of other syntactic/semantic combinations, such as classifier-noun combination in some languages.In the present study, we investigated the role of animacy in processing classifier-noun combination in Chinese. For such a combination, no thematic role assignments are involved. We assume that although animacy information can be used rapidly during the processing of noun-verb combination, at least in the N400 window, as revealed by some earlier studies (e.g., Frisch & Schlesewsky, 2001; Li et al., 2006; Philipp et al., 2008; Roehm et al., 2004; Weckerly & Kutas, 1999), it may be not used very quickly in the processing of classifier-noun combination, due to its lower level of prominence for a combination without thematic role assignments being involved. Alternatively, one might argue that in Chinese, a language that lacks of explicit morphosyntactic devices, animacy information may have a general prominence during sentence processing. Thus, when processing the classifier-noun combination, whether the two components match each other in animacy or not can modulate the degree of difficulty in computing the combinatory relationship between them.To test the two alternative hypotheses mentioned above, we manipulated three types of sentences that had a non-canonical structure (object-subject-verb, OSV): (a) correct sentences in which the classifier was congruent with the sentence-initial object noun, such as the sentence"Car/ Qingfeng Zhao/ saw/ one liang [classifying ground vehicle]/ black"; (b) sentences containing within-category violations, in which although the classifier was incongruent with the object noun, both of them were inanimate, such as the sentence"Desk lamp/ Qingfeng Zhao/ saw/ one liang/ cheap"; (c) sentences containing between-category violations, in which the object noun was animate and was incongruent with the inanimate classifier, such as the sentence"Seal/ Qingfeng Zhao/ saw/ one liang/ clumsy". The numeral-classifiers served as the critical words and 30 participants participated in the ERP experiment.The results showed that for all 30 participants, both violation conditions elicited a broadly distributed negativity in the 300-550ms (N400) time window. Crucially, there was no difference in the N400 between the two violation conditions, suggesting that the congruency of animacy between the classifiers and object nouns were not considered during the processing of classifier-noun combination. Considering the fact that the sentence structure used in our critical items was a non-canonical structure and only half of the participants accepted such a structure, we divided the participants into two groups: the'accept'and'not accept'group, with 15 participants each. The results showed that for the"accept"group, there were still no N400 differences between the two violation conditions. In the 600-1000ms time window, the between-category violations elicited a P600 effect at the bilateral posterior sites, whereas the within-category violations did not. These results appear to suggest that once readers accept the non-canonical OSV structure, the word order functions as a prominent cue and the animacy cue plays no role in the initial computation of the classifier-noun combination, resulting in no animacy-N400 effects. The P600 effects most likely reflect the conflict between the word order and animacy cues in the between-category violation condition.In contrast, for the"not accept"group, compared with the correct condition, the between-category violations elicited a larger N400, while the within-category violations did not elicit any effects. These results suggest that when the OSV construction is not accepted, the word order is no longer an available parsing cue, such that animacy becomes the only prominent parsing cue, resulting in the presence of animacy-N400 effects in the animacy mismatch (between-category violation) condition and the absence of N400 effects for the within-category violations.In sum, the present study has revealed how the animacy information is processed during the computation of the classifier-noun combination in Chinese. The future studies should examine some other combinations that do not involve thematic role assignments, for example, the adjective–noun combinations, to gain an further insight into the functional nature of animacy information in language processing.