Abstract: The human brain consists of two anatomically different but functionally complementary hemispheres. The two hemispheres complement each other for most functions, including language. Nouns and verbs are fundamental and important members of word classes, existing in almost all human languages. The investigation on the hemispheric lateralization in the processing of noun and verb will improve the understanding of human brain and language. A number of studies have investigated the hemispheric asymmetry in the processing of nouns and verbs. However, since the results obtained so far were mainly from studies on languages with rich inflectional morphology, it is unclear whether the hemispheric differences in the processing of nouns and verbs are due to semantic-conceptual properties or morpho-syntactic properties associated with the two word classes. It is reasonable to examine the word class effect along the semantic dimension in a language with little or simple inflectional morphology, such as Chinese. Furthermore, the techniques applied in previous researches were low in temporal resolution which might limit the findings. Event-related potentials (ERPs), which are voltage fluctuations that are associated with some physical or mental occurrence, can measure the processing activities taking place in the human brain with millisecond accuracy. The combination of ERP technique with half visual field (VF) technique which was commonly employed in the study of language asymmetry provided important windows into language processing in the two hemispheres. Thus, the current study, by combining ERP and VF techniques, aims to explore the hemispheric lateralization in the semantic processing of Chinese noun, verb and verb-noun ambiguous words in the left hemisphere (LH) and the right hemisphere (RH). The experiment contained five sets of stimuli: disyllabic unambiguous nouns (UN), unambiguous verbs (UV), verb-biased (V-VN), noun-biased (N-VN) verb-noun ambiguous words and non-words. The stimuli of UN and UV were mainly typical members of their respective word class and the stimuli of V-VN and N-VN selected according to results of subjective ratings. There were 80 stimuli for each set of real word stimuli. Word frequency and number of strokes were matched across four sets of real word stimuli. The non-words, numbering 160 in total, were the juxtaposition of two real characters that do not form real words. The disyllabic compounds were vertically arranged in the left visual field (LVF) and right visual field (RVF) to ensure that each character in the compounds was equally distant from the fixation. The experiment task was a lexical decision task. 17 native Chinese speakers were instructed to judge whether the stimulus, presented to the LVF or RVF randomly, was a real word or not as quickly and accurately as possible. Both behavioral and ERP data were collected and were analyzed with repeated measures ANOVAs. The results showed significant ERP differences between different word classes in intervals of 350-500ms (N400) and 500-650ms (LPC). N400 could index semantic difference between nouns, verbs and verb-noun ambiguous words. In the RVF/LH, the amplitude values of N400 were greater for UN and UV than for V-VN and N-VN and no significant difference was found between UN and UV, V-VN and N-VN respectively. In the LVF/RH, UN elicited more negative N400 than UV, V-VN and N-VN. UV elicited more negative N400 than V-VN and N-VN. However, there is still no significant difference between V-VN and N-VN in the LH. Furthermore, previous studies reported that the N400 could be influenced by the variables of semantic feature and concreteness. To further clarify the roles of semantic feature and concreteness in the word processing, the results of N400 elicited by UV and N-VN were further analyzed with respect to words of comparable levels of concreteness. Results showed that there is a significant difference in N400 between UV and N-VN in the RVF/LH but the difference was reduced in the LVF/RH when stimuli had matched concreteness, indicating that the N400 in the RVF/LH was sensitive to semantic feature and LVF/RH was influenced by concreteness. However, this did not mean that the effect of semantic feature was totally absent for the processing in the LVF/RH. This was because that in the LVF/RH, there were significant differences in N400 between UV and V-VN whose concreteness values were equalled. Therefore, the N400 in RVF/LH was mainly effected by semantic features while the N400 in LVF/RH was effected by both semantic features and concreteness. In addition, the LPC is a component that could reflect the confidence in the decision-making process; the higher the confidence, the more positive the LPC. In the lexical decision process, V-VN and N-VN activated more senses than UN and UV, so they would receive more semantic feedback which would lead to higher decision confidence. N-VN contained more senses than V-VN and this would result in higher confidence for N-VN. Thus, V-VN and N-VN should elicit more positive LPC than UN and UV and N-VN should elicit more positive LPC than V-VN. Results showed the predicted LPC effect in the LVF/RH but not in the RVF/LH. This was because that semantic activation was focal and strong in the LH but broader and weaker in the RH (Beeman 1998; Jung-Beeman 2005). In the RVF/LH, since only the dominant sense of each word was activated, there was no difference on LPC. While in the LVF/RH, multiple senses were activated and therefore LPC effect appeared. The results suggested different roles of left and right hemispheres in the word processing. LH mainly processes semantic features and RH processes both the semantic features and concreteness associated with different word classes. Besides, the difference between UN and UV in Chinese mainly lies in concreteness. UN and UV differ from N-VN and V-VN in semantic features, concreteness and the amount of semantic activation, which reflects different neural mechanisms for processing unambiguous and ambiguous words.

Key words: Key words unambiguous noun, unambiguous verb, verb-noun ambiguous words, lateralization, ERP

摘要： 将ERP技术和半视野技术相结合，采用词汇判断任务，对汉语名词、动词和动名兼类词在左脑和右脑中的加工机制进行了考察。实验结果显示，名词和动词的N400仅在左视野/右脑存在差异，名词和动词的N400在左视野/右脑和右视野/左脑中都比偏（动）和偏（名）更负。不同词类的LPC在右视野/左脑中没有显著差异；偏（名）和偏（动）的LPC在左视野/右脑中比名词和动词更正。实验结果表明，在没有语境条件下，汉语名词和动词的差异主要在具体性上，动名兼类词体现出不同于名词、动词的加工机制。

关键词: 名词, 动词, 动名兼类词, 偏侧化, ERP