This study employed electrophysiological measures to investigate the processes of expectancy fulfillment and semantic integration during the processing of binding and compound nouns in Chinese. Sequential expectancy and cloze probability between the two constituents of a two-character noun were manipulated in two experiments, resulting in four types of target characters (i.e., the final characters of two-character nouns): (i) high-cloze, binding character (HB); (ii) high-cloze, compound character (HC); (iii) low-cloze, binding character (LB); (iv) low-cloze, compound character (LC). Participants were asked to judge as quickly and accurately as possible whether the two characters, presented sequentially, formed a real word. The two experiments varied in stimulus onset asynchrony (SOA) between the initial and the target characters: 1000 ms and 300 ms. ERPs acquired at the target characters revealed a significant semantic integration effect in both experiments whereby low-cloze target characters elicited more negative-going activities than high-cloze target characters (L > H) with both long and short SOAs. Importantly, there was a graded N400 effect (LC = LB > HC > HB) at the central region: a clearly visible positive deflection for HB relative to HC in synchrony with an equivalence in negativity for LB and LC. This graded pattern was observed with an SOA of 1000 ms but not 300 ms. These results are discussed in terms of the possible coexistence of an expectancy fulfillment mechanism indexed by the P300 that monitors incoming information for realization of expectancies based on stored mental representations and a semantic integration mechanism indexed by the N400 that incorporates incoming information into the preceding context based on semantic-pragmatic knowledge. Manifestation of this coexistence appears sensitive to SOA-modulated attention orientation.

本研究采用电生理学方法研究汉语结合名词和复合名词加工过程中的期望实现和语义整合过程。在两个实验中操纵了一个两字名词的两个成分之间的顺序期望和完形填空概率，产生了四种类型的目标字符（即两字名词的最后一个字符）：（i）高完形、约束性字符(HB); (ii) 高完形、复合字符 (HC)；(iii) 低完形、具有约束力的特征 (LB)；(iv) 低完形、复合字符 (LC)。参与者被要求尽可能快速准确地判断这两个按顺序出现的字符是否构成了一个真实的单词。这两个实验在初始字符和目标字符之间的刺激起始异步（SOA）方面有所不同：1000 毫秒和 300 毫秒。在目标字符处获得的 ERP 在两个实验中都显示出显着的语义整合效应，即低完形目标字符比具有长 SOA 和短 SOA 的高完形目标字符 (L > H) 引发更多的负面活动。重要的是，在中心区域存在分级的 N400 效应（LC = LB > HC > HB）：HB 相对于 HC 的清晰可见正偏转与 LB 和 LC 的负性等价同步。这种分级模式在 SOA 为 1000 ms 但不是 300 ms 的情况下观察到。这些结果是根据 P300 索引的期望实现机制的可能共存来讨论的关于语义-语用知识。这种共存的表现似乎对 SOA 调制的注意力方向很敏感。