视觉词汇识别（Visual Word Recognition）是人们通过视觉接受词形或语音信息，最终通达语义的短暂的过程。汉语拼音作为汉字语音的转写系统，既具有表音汉字的属性，又在视觉结构上与汉字完全不同。这种特性使得对比研究汉字与拼音两种脚本的加工机制，对全面了解汉字视觉识别过程具有重要意义。母语者和二语者在早期是如何加工汉字和拼音的，两种脚本的加工机制有何差异，两个群体之间的加工方式有何不同，是本研究关注的重点。

本研究的实验一和实验二分别招募母语被试25人，高水平汉语二语被试24人。采用1-back重复探测任务和词汇判断任务，以汉字真字、汉字假字、汉字非字、拼音真词、拼音非词、符号串为实验材料，通过事件相关电位（ERP）技术，考察母语者和二语者早期视觉识别过程中，从初级知觉加工到字形、字音加工的神经生理机制，结果发现：

（1）重复探测任务中母语者判断汉字真字的正确率最高，汉字假字与符号串的正确率最低。二语者对汉字非字的判断正确率最高，汉字假字和符号串的正确率最低。二语者汉语水平成绩越高，其在词汇判断任务中正确判断汉字假字、汉字非字、拼音真词、拼音非词的反应时越快。

（2）母语者汉字真字的P1成分平均波幅最低。在右半球，汉字真字和拼音真词的P1潜伏期分别比汉字假字和拼音假词短。二语者汉字真字的P1平均波幅最低。

（3）母语者汉字真字的N1平均波幅显著大于汉字非字，拼音真词的N1平均波幅显著大于拼音非词和符号串。仅在母语者中发现汉字非字的潜伏期最短，拼音真词的潜伏期显著长于拼音非词。二语者汉字真字的N1平均波幅最大，拼音真词N1峰值波幅显著大于拼音假词。

（4）母语者和二语者汉字与拼音在左半球上的N170平均波幅负于右半球。

（5）母语者拼音真词的P200峰值波幅显著最大。二语者汉字真字P200峰值波幅显著最大。

（6）母语者汉字真字的N250平均波幅最高。二语者N250成分的平均波幅在不同类型材料之间的差异边缘显著。

综合以上结果得出如下结论：

（1）相较于几何符号，母语者和二语者在识别加工语言材料时耗费认知资源更少。

（2）母语者和二语者在60ms～180ms进行了汉字和拼音的初级特征编码与知觉加工，P1、N1成分受文字材料熟悉度与视觉复杂度共同影响。

（3）母语者和二语者的正字法加工发生在115ms～210ms，两者在汉字和拼音上都出现了N170的左偏侧化。母语者N170平均波幅比二语者更大，这种效应是双侧的。

（4）母语者在120ms～200ms进行汉字的自动化亚词汇语音加工，随后在300ms前，语音加工过渡到整字水平。对拼音的自动化语音加工比整字更早，在120ms～200ms之间就已完成。二语者在120ms～200ms之间同样进行了汉字和拼音的语音加工，随后对汉字的语音加工结束，并没有从亚词汇水平转向整词水平，而对拼音的加工则进一步持续到了300ms。

研究比较了母语者和高水平二语者对汉字与汉语拼音的加工方式差异，深入探究了汉字早期视觉识别的脑电成分指标，为理解汉语单音节词早期加工提供了新参考。

Visual word recognition (VWR) is a short process by which people visually receive input in the form of words or speech information, and eventually make sense of semantics. Pinyin, as a kind of phonetic system of Chinese characters, has the properties of ideographic characters as well as completely differs from Chinese characters in visual structure. So it is important to study the processing mechanism of Chinese characters and pinyin to understand the process of visual recognition of Chinese characters. The event-related potential (ERP) opens up the possibility of revealing a very short visual word recognition process. This study focuses on how native speakers (L1 speakers) and Chinese-Second Language speakers (L2 speakers) process Chinese characters and pinyin in the early stage, how the processing mechanisms of the two scripts differ, and how the processing methods differ between the two groups.

In this study, 25 L1 speakers and 24 L2 speakers were recruited. The neurophysiological mechanism of primary perceptual processing to orthographic and phonological processing in the early visual recognition process of native and second speakers was investigated using ERP.

(1) For L1 speakers, real characters were judged most correctly in the one-back task. Pseudo-characters and symbols were judged least correctly. For L2 speakers, non-characters had the highest accuracy, pseudo-characters and symbols had the lowest accuracy. In addition, the higher score in Chinese test, the faster response time in LDT for correctly judging pseudo-characters, non-characters, real pinyin, and non-pinyin.

(2) For L1 speakers, the average amplitude of the P1 component was the significantly lowest for real characters. In the right hemisphere, P1 latency for real characters and real pinyin was shorter than that for pseudo-characters and pseudo-pinyin. For L2 speakers, the average P1 amplitude for real characters was the lowest.

(3) The average N1 amplitude of real characters was significantly larger than that of non-characters for L1 speakers. And the average N1 amplitude of real pinyin was significantly larger than that of non-pinyin and symbols. Only L1 speakers showed significant differences in N1 latency between different types of materials. For L2 speakers, real characters had the largest average amplitude of N1. N1 peak amplitude of real pinyin was significantly larger than that of non-pinyin.

(4) The N170 amplitude of L1 and L2 speakers on the left hemisphere of Chinese characters and pinyin is more negative than that on the right hemisphere.

(5) P200 peak wave amplitude of real pinyin for L1 speakers was significantly the largest. For L2 speakers, P200 peak amplitude of real characters was the largest.

(6) For L1 speakers, the average amplitude of the N250 for real characters is the most negative. For L2 speakers, the mean amplitude of the N250 component was significantly edged between different types of materials.

Taken together, the results are as follows:

(1) L1 speakers and L2 speakers use less cognitive resources to process languages. The ability of high level L2 speakers has little effect on the early processing of Chinese characters and pinyin.

(2) Primary feature coding and perception processing is performed by L1 speakers and L2 speakers between 60 and 180 milliseconds. The N1 component and P1 component are influenced by the familiarity and visual complexity of written materials.

(3) Both L1 speakers and L2 speakers process orthographically between 115 and 210 milliseconds. Both show the left-lateralization of N170 in Chinese characters and pinyin. N170 average amplitude of L1 speakers is larger than that of L2 speakers that is bilateral.

(4) L1 speakers perform automated sub-lexical phonetic processing of Chinese characters in 120 to 200 milliseconds, followed by a transition to the whole word level by 300 milliseconds. Automated speech processing of Pinyin predates the whole word and is completed in between 120 and 200 milliseconds. L2 speakers also process the phoneticist of Chinese characters and Pinyin between 120 and 200 milliseconds, and then the processing of Chinese characters do not change from sub-lexical to whole-word level, while the processing of Pinyin continue for a further 300 milliseconds.

This study innovatively compares the processing of Chinese characters and Pinyin between L1 speakers and L2 speakers, in order to explore the similarities and differences in recognition, as well as further explores the cognitive neural mechanism in early visual recognition of Chinese characters, providing evidence for the bi-modal interactive activation model.