中文摘要：

采用眼动方法，通过两个实验考查了句子语境中汉语词汇形、音的作用及其作用的时间进程。结果发现：（1）在高限制性句子语境中，音同的凝视时间和总注视时间显著短于无关控制；（2）在低限制性句子语境中，音同、形似的首视时间均显著短于无关控制；（3）词频效应出现在低限制性句子语境年。该结果表明，句子语境影响词汇形、音的作用及作用的时间进程，汉语词汇的意义可以由字形直接通达，也可以由形和音两条路径得到通达。

英文摘要：

The issue of the role of orthography and phonology in activating the meanings of visual words has provoked a large amount of research for alphabetic writing systems, especially for English. A great deal of research has demonstrated that phonological codes are activated early in English silent reading. For Chinese, however, the role of orthography and phonology remains controversial. Specifically, some empirical evidence supports directly access hypothesis, some supports phonologically mediated access hypothesis, and some for dual-rout hypothesis. Unlike English, the basic unit of writing in Chinese is character which does not associated with phonemes but with meaning instead. Chinese provides an interesting case for testing the current models of visual word recognition that developed essentially for alphabetic writing systems. The current study focused the issue that how orthography and phonology is activated during Chinese silent reading when target words were presented in different constrained sentence contexts. Two experiments were conducted by manipulating the word frequency of homophone pairs and types of target words. All the target words were Chinese single character words and were matched for their stroke numbers across conditions. In addition, the homophone density （the number of homophones of a character has） was matched between the high- and low- frequency homophone pairs. Eye tracking method was used in this study and three types of eye movement measures such as first fixation duration, gaze duration, and total fixation duration were examined to determine how orthography and phonology is activated during Chinese sentence reading. The results showed that a reversed effect of word frequency occurred for orthographical targets in the fixation duration and total fixation duration, while both gaze duration and total fixation duration were significantly shorter for the homophonic targets than unrelated controls when the targets were presented in the highly constrained sentence context. However, when the