How is semantic information in the mental lexicon accessed and selected during reading? Readers process information of both the foveal and parafoveal words. Recent eye-tracking studies hint at bi-phasic lexical activation dynamics, demonstrating that semantically related parafoveal previews can either facilitate, or interfere with lexical processing of target words in comparison to unrelated previews, with the size and direction of the effect depending on exposure time to parafoveal previews. However, evidence to date is only correlational, because exposure time was determined by participants' pre-target fixation durations. Here we experimentally controlled parafoveal preview exposure duration using a combination of the gaze-contingent fast-priming and boundary paradigms. We manipulated preview duration and examined the time course of parafoveal semantic activation during the oral reading of Chinese sentences in three experiments. Semantic previews led to faster lexical access of target words than unrelated previews only when the previews were presented briefly (80 ms in Experiments 1 and 3). Longer exposure time (100 ms or 150 ms) eliminated semantic preview effects, and full preview without duration limit resulted in preview cost, i.e., a reversal of preview benefit. Our results indicate that high-level semantic information can be obtained from parafoveal words and the size and direction of the parafoveal semantic effect depends on the level of lexical activation.

阅读过程中如何访问和选择心理词典中的语义信息？读者处理同凹词和副凹词的信息。最近的眼动追踪研究提示了双相词汇激活动态，表明与无关的预览相比，语义相关的副凹前预览可以促进或干扰目标词的词汇处理，效果的大小和方向取决于曝光时间中央凹预览。然而，迄今为止的证据只是相关的，因为暴露时间是由参与者的目标固定前持续时间决定的。在这里，我们结合视线快速启动和边界范例，实验性地控制了副凹的预告暴露持续时间。在三个实验中，我们操纵了预览的持续时间，并研究了在口头阅读中文句子时副凹窝语义激活的时间过程。仅当简短展示预览时（实验1和3中为80毫秒），语义预览才能比不相关的预览更快地对目标词进行词汇访问。较长的曝光时间（100毫秒或150毫秒）消除了语义预览效果，而没有持续时间限制的完全预览会导致预览成本，即预览收益的逆转。我们的结果表明，可以从副凹词获得高级语义信息，副凹语义效果的大小和方向取决于词汇激活的水平。仅在短暂展示预览时（实验1和3中为80毫秒），语义预览才能比不相关的预览更快地对目标词进行词汇访问。更长的曝光时间（100毫秒或150毫秒）消除了语义预览效果，而没有持续时间限制的完整预览会导致预览成本，即预览收益的逆转。我们的结果表明，可以从副凹词获得高级语义信息，副凹语义效果的大小和方向取决于词汇激活的水平。仅在短暂展示预览时（实验1和3中为80毫秒），语义预览才能比不相关的预览更快地对目标词进行词汇访问。更长的曝光时间（100毫秒或150毫秒）消除了语义预览效果，而没有持续时间限制的完整预览会导致预览成本，即预览收益的逆转。我们的结果表明，可以从副凹词获得高级语义信息，副凹语义效果的大小和方向取决于词汇激活的水平。