中文摘要：

采用边界范式,在副中央凹呈现一个双字词或两个单字词,操纵字N＋2的预视（目标预视和假字预视）,来考察副中央凹中词汇特征是否影响读者随后的眼跳长度及注视位置。结果发现,读者能够获得字N＋2的预视;双字词与两个单字词条件下注视位置没有显著地差异;当副中央凹中为双字词时,字N＋2为目标预视时随后的眼跳长度显著地长于假字预视条件,随后的注视位置距离词首更远;当副中央凹中为两个单字词时,两种预视条件下随后的眼跳长度及注视位置没有显著的差异。本研究结果提示：在汉语阅读中,读者能够对字N＋2进行预视加工,这种加工的影响只发生在当副中央凹中为双字词时。

英文摘要：

Although some researches have found that the word is the basic processing unit in Chinese reading, there is little if any evidence to show that Chinese readers select upcoming words as the next saccade target. Research has shown that parafoveal processing affects saccade targeting in reading of word-based alphabetic languages like English. However, it is less clear whether it is the case in reading of character-based languges like Chinese, in which word boundaries are not clearly demarcated and there is often ambiguity about which characters form a word. Opinions differ about whether Chinese readers select a specific position within a word to fixate during reading. Several studies have shown that Chinese readers do not target saccades to a specific position within a word. However, others have argued that saccade targeting is based on ongoing word segmentation that occurs in parafoveal vision, and only if segmentation is successful, do Chinese readers select the center of the word as the next fixation location. Conversely, if segmentation fails, readers will move their eyes to the beginning of the word. If Chinese readers select the next word as a target, there should be difference in the fixation location when characters in parafovea do, or do not, comprise a word. In this experiment, we selected a high-frequency character as character N＋1. In the ＂two-character word＂ condition, the characters N＋1 and N＋2 formed a word, but in the ＂two single character words＂ condition, the two characters did not form a word. Using the boundary pradigm（Rayner, 1975）, we also manipulated the preview of character N＋2, such that it was either an identical or a pseudocharacter preview. The results showed that fixation durations on character N＋2 were significantly shorter when the preview of character N＋2 was identical than when it was a pseudocharacter. There was no significant difference in the length of the outgoing saccade from the pretarget to the target region between the two-character word condition and