汉语阅读时的眼动控制：眼跳目标的选择

PSYCH OpenIR > 认知与发展心理学研究室

	汉语阅读时的眼动控制：眼跳目标的选择
其他题名	Eye Movement Control during Chinese Reading: How to Select the Saccade Target
	刘萍萍
	2013-05
摘要	本论文围绕汉语读者的眼跳目标选择展开了系列研究，这是阅读中的眼动控制关注的两个重要问题之一。眼动控制主要包括眼睛什么时候（when）开始移动和眼睛注视在什么位置（where）上。拼音文字的研究认为，读者的眼跳目标在词的中央附近，在该位置上词的识别效率较高。但是，汉语读者的眼跳目标并没有指向词汇中的某个特别位置（e.g., Li, Liu & Rayner, 2011; Tsai & McConkie, 2003; H. Yang & McConkie, 1999）。由此，我们提出两个主要的问题：（1）为什么汉语读者的眼跳目标没有指向词汇中的某个特定位置？（2）汉语读者是采用了怎样的落点选择策略，从而选择眼跳目标的？我们通过研究一和二来回答第一个问题。第一种假设是，汉语读者的眼跳目标未指向词汇的某个特别位置，有可能是同汉语词汇和句子的特性有关。汉语词长较短，信息密度较大，眼跳落点在词汇上的不同位置上对词汇识别的影响无显著性差异。实验1表明，孤立汉语词的加工中存在着最佳注视位置效应（OVP）。不管是词还是非词，当被试注视在2-字材料的第一个字上、3-字和4-字材料的第二个字上，词汇判断的时间是最短的。也就是说，在汉语孤立词加工中，词上不同的落点位置对词识别产生了显著性影响。那么在句子阅读中，注视时间是否受到注视位置的影响呢？实验2的结果表明，在正常句子阅读中，同注视在其他位置相比，当首次注视在词的第一个位置时，词的整体被注视时间是最长的。研究一的实验表明，在汉语阅读中，不同的注视位置会对词识别或注视时间产生显著影响。因而，我们提出的第一种假设被否定。第二种假设是，由于汉语句子中词与词之间没有清楚的物理边界，汉语读者无法以词边界作为线索从而选择注视在词的某个特别位置上。研究二中，我们分别在短语边界（实验3）、词边界（实验4）、非词边界（实验5）后插入空格，考察词边界对眼跳目标选择的影响。结果发现，不管是词还是非词，被试总是倾向于注视在远离空格的位置上，且在词和非词条件下，被试的首次注视位置没有显著性差异。因而，可能是空格，而不是词边界引起了眼跳目标选择的变化。还有一种可能性是空格对眼跳目标选择的影响比较显著，从而掩蔽了词边界对眼跳目标选择的作用。因而，研究二的实验结果表明，汉语读者可能无法利用词边界信息选择注视到某个特别位置上。既然汉语读者无法选择某个特定位置作为眼跳目标，那么是采用怎样的落点策略，选择眼跳目标的呢？根据以往研究，我们认为汉语读者可能通过旁视野从注视点右侧提取了一些信息，来引导眼跳目标选择。比如在研究二中，被试可以通过旁视野很容易觉察到低水平的信息比如空格，从而引导眼跳，远离空格，注视到信息丰富，显著性高，且有利于加工的位置上。另外，Wei，Li 和Pollatsek (2013)提出了一种基于信息加工的眼跳模式。他们的研究表明，假如被注视词越容易被加工，注视时间越短，有更多的认知资源来加工旁视野的信息，则被试的眼睛跳出该词的长度越长。为了更深入考察汉语读者的落点选择策略，我们在研究三中设计了四个实验深入考察注视点和旁视野信息对眼跳目标选择的影响。在这四个实验中，我们分别考察了低水平（字间距和对比度）和高水平（语义冲突和词频）信息。在字间距实验6 中，我们发现，同正常条件相比，假如字与字之间的距离加大，则被试在旁视野提取的有效信息比较少，则被试的相对眼跳长度（以字为单位）比较小，但是绝对眼跳长度（实际眼跳长度）会比较大，句子的整体阅读时间也会增加。但是，在对比度实验7 中，我们并没有发现对比度和词频会对眼跳目标选择产生显著影响。而在语义冲突实验8中，我们选择了一些错别字作为实验材料，发现被试更少地注视别字信息，而更多地注视别字周围正确的字，且同正常条件相比，被试从别字上跳出的眼跳长度更短些。该实验表明，被试的眼跳目标选择有一定的灵活性。在实验9 中，我们通过边界启动范式考察了旁视野内信息加工程度和落点位置之间的关系。请被试来阅读句子，在句子的某个位置上确定边界，当被试跨过这个边界后，句子消失，会出现两个字，请被试来做词汇判断任务。结果发现，起跳位置影响着旁视野内信息的加工情况。起跳位置距离边界越近，则越容易通过旁视野加工边界后的信息。当起跳位置在边界前的第一个字以内的位置上时，落点位置在边界后的第二个字上时的词汇判断时间边缘显著地短于落点位置在边界后的第一个字上。因而，起跳位置影响着旁视野内信息被加工的程度，旁视野内信息被加工的程度影响了落点位置的选择。本实验只是初步的探讨基于信息加工效率的眼跳目标选择，这个研究假设仍需要可操作化的定义和更多实验验证。总之，汉语读者的眼跳目标选择受到旁视野内信息的影响。我们发现，汉语读者的眼跳落点选择模式具有较大的灵活性。汉语读者的眼跳目标选择同起跳位置有很大的相关性，当起跳位置确定的时候，读者可以通过旁视野加工部分信息，受到信息加工效率的驱动，从而选择眼跳目标。这些研究表明眼跳目标选择反映了实时的认知加工过程。
其他摘要	The main purpose of this thesis was to examine the saccade target selection strategy of Chinese readers when they read Chinese. The question we focus on is among one of two important questions that were asked by the researchers to study eye movement control during reading: what determines when to move the eyes, and what determines where to move. For English and other alphabetic languages, readers’ eyes tend to target the center of words where the words are identified most quickly. Nevertheless, due to saccade errors, they land halfway between the middle of a word and the beginning of that word in sentence reading. Readers of alphabetic writing system could target their eyes to the center of a word because there are spaces between words in those writing system so that reader could perceive word boundary with parafoveal vision. However, there are no explicit cues such as spaces between words in Chinese reading, and Chinese readers cannot rely on inter-word spaces to perceive word boundary at parafoveal vision. Without spaces to mark word boundary, how do Chinese readers target their eyes? Previous studies showed that Chinese readers do not target any specific position within a word (e.g., Li, Liu & Rayner, 2011; Tsai & McConkie, 2003; H. Yang & McConkie, 1999). Thus, two questions have been raised: (a) why do not Chinese readers target any specific position within a word? (b) how do Chinese readers select saccade target? Study Ⅰ and Ⅱ in this thesis were designed to address the first question. Study Ⅲ was designed to address the second question. Why do not Chinese readers target at any specific position within a word? One possibility was that word recognition efficiency may not be affected by landing position, since most Chinese words are very short and information density is very high. Study Ⅰwas designed to rule out this possibility. However, the results showed that fixation duration was influenced by landing position for both isolated word processing and sentence reading. Thus, the first possibility cannot explain why Chinese readers do not target any specific position within a word. Study Ⅱ was designed to test the second possibility. Unlike in English, there are no explicit cues between printed word in Chinese. Thus, Chinese readers could not use physical boundary information to land on specific position during sentence reading. In this study, spaces were inserted between phrases, words and nonwords to examine how explicit word boundary information affect saccade target selection. Results showed that, the initial landing position was further away from the space that inserted into the text, whether it was before or after a word, a nonword, or a phrase. Thus, these results suggest that it was the inter-word spaces that guide the initial landing position on the word, not the word boundary.Nevertheless, we could not rule out the perspective: the effect of spaces on the saccade target selection may be so strong that the effect of word boundary could be masked or ignored. More studies are needed to explore the question. Moreover, Study Ⅱ confirmed that findings that Chinese readers do not target any specific position due to absence of explicit word boundaries. The purpose of Study Ⅲ was to explore how Chinese readers select saccade target. Wei, Li and Pollatsek (2013) suggested that Chinese readers saccade target selection is affected by the processing efficiency of the fixated word. According to this hypothesis, Chinese readers try to process as many characters as possible at a given fixation, and then saccade somewhere beyond these processed characters. When the fixated words are easy to process, more characters will be processed at that fixation, and hence saccade length is longer. This hypothesis also predicts that when the launch site is controlled, if the information to the right of fixation is easier to be processed, the landing position is further from the launch site. We tested these predictions in four experiments. For English and other alphabetic languages, many studies have found that: (a) where to move the eyes next is strongly influenced by low-level cues such as word length and space information; (b) saccade length is affected by the length of the fixated word and the word to the right of fixation. We explored how some low-level information (e.g.,inter-character spaces, contrast) affected where to move the eyes in Experiment 6 and 7.Meanwhile, we explored how some high-level information such as plausibility and word frequency affect where to move the eyes in Experiment 7 and 8. We used boundary priming paradigm to test the hypothesis directly in Experiment 9. We found that saccade length was influenced by the inter-character spaces and implausibility information (Experiment 6 and 8). But we did not found that target selection was influenced by contrast or word frequency (Experiment 7), which is consistent with some studies in alphabetic languages (e.g., Jordan, McGowan & Paterson, 2012). In Experiment 9, when the launch site was within one character long before the boundary, the information processing on the first and second character after the boundary was marginally significantly different. In summary, we have found some unique properties of saccade target selection strategies of Chinese readers. Studies Ⅰ and Ⅱ showed that fixation duration is influenced by landing position, and Chinese readers may not use word boundary information to select saccade target. Study Ⅲ showed that Chinese readers’ saccade target selection is flexible. They can adjust eye behavior according to the stimuli. Saccade length is influenced by both the current fixated character and the character to the right of the fixation. Information processing was influenced by launch site and landing positions.These studies suggest that saccade target selection data reflect moment-to-moment cognitive processes in the various tasks examined.
学科领域	语言认知（认知心理学）
关键词	汉语阅读眼动控制眼跳目标注视位置
学位类型	博士
语种	中文
学位专业	心理学
学位授予单位	中国科学院研究生院
学位授予地点	北京
文献类型	学位论文
条目标识符	http://ir.psych.ac.cn/handle/311026/19628
专题	认知与发展心理学研究室
作者单位	中国科学院心理研究所
推荐引用方式 GB/T 7714	刘萍萍. 汉语阅读时的眼动控制：眼跳目标的选择[D]. 北京. 中国科学院研究生院,2013.

条目包含的文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
刘萍萍-博士学位论文.pdf（3507KB）	学位论文		限制开放	CC BY-NC-SA	请求全文