Wang, J., & Su, Q. (2019). Detection and Recovery Effects of Wrongly Written Characters in Context among Readers of Chinese as a Second Language. Journal of Technology and Chinese Language Teaching, 10(2), 1-60.
[王骏, & 苏茜. (2019). 汉语第二语言阅读者在上下文语境中的别字检测和复原效应. 科技与中文教学 (Journal of Technology and Chinese Language Teaching), 10(2), 1-60.]

Full paper

Abstract/摘要：

Reading processes are an important issue in psycholinguistics. Scholars have different opinions about the processes and cognitive models involved in reading. Moreover, whether wrongly written Chinese characters can be detected and restored in context is also the focus of much research (among which the relationship and role of orthographic layers, phonological layers, and semantic layers also lead to different views and divergent opinions). With the support of modern science and technology, eye movement experiments can be used to record and analyze the actual cognitive processes of readers. This study explores whether there is any effect on detecting and restoring wrongly written characters in context for readers of Chinese as a second language. How phonetics and orthography are linked to each other are used to further explore and discuss the above controversy. Therefore, based on the connectionist model this study uses an eye movement experiment to analyze the recognition speed of similar orthography, similar phonetics and orthography, and homophones in the correct context for Chinese learners from different language backgrounds (Sino-sphere and non-Sino-sphere) and different language levels. Through a comprehensive analysis of three variables, early eye movement indicators (first fixation duration and first run dwell time) are analyzed. Visual time and late eye movement indicators (total dwell time, regression path duration, and fixation count) are used to illustrate the recognition sequence and speed between orthography, phonology and semantics. In the process, a more specific connectionist model is established. Five noteworthy results are clear: (1) There is no significant difference in eye movement between sentences with and without wrongly written characters, so it is believed that the words can be read smoothly by learners, that is, the wrongly written character can be automatically repaired to complete the understanding of words and sentences. (2) Sino-sphere and non-Sino-sphere learners both show a strong connection between orthographic and semantics in the early recognition process of language, but the effect of recovery and detection of non-Sino-sphere learners is weaker than Sino-sphere learners. (3) Sino-sphere learners recognize better characters that have similar orthography, which means that the connection between orthography and semantics is stronger; the non-Sino-sphere learners recognize better the homophone. In other words, the connection between phonetics and semantics is stronger. (4) In the later stage of the cognitive process of character recognition, Sino-sphere learners rely more on the conversion of phonetics and semantics. Non-Sino-sphere learners have the same decoding speed for the phonetic, orthographic, and combination. (5) In the model of native speakers' recognition, the decoding speed of the combination of phonetic, orthographic, and combination are the same, and none of them is dominant. Therefore, its cognitive structure is a stable and mature structure in native readers. Overall, learners of Chinese as a second language perform differently in detecting and restoring wrongly written character in reading, and learners of different language backgrounds and language proficiencies have different performance in Chinese character recognition. This has important implications for teaching Chinese characters and words to speakers of other languages. Teachers can learn from this study how to improve teaching methods, and Chinese learners can adopt more scientific and effective learning methods.

阅读过程是心理语言学研究的一个重要问题，不同学者对阅读中的加工与认知模型有着不同的见解。并且，别字在上下文语境中是否能被检测和复原也是中外学者研究的焦点，其中学界对语音、字形和语义之间的关系和作用也产生了不同的观点和分歧。研究阅读过程可以采取眼动实验的方式记录和分析阅读者真实的的认知过程。本研究通过探究以汉语作为第二语言的阅读者在上下文语境中是否发生别字的检测和复原效应，以及其中语音、字形发生的联结作用，对以上争论做进一步的探究和讨论。 因此，本文立足于联结主义模型之上，采取眼动实验的方法，针对汉字圈和非汉字圈不同语言背景、不同语言水平的汉语学习者，分析同音异形、同音形似、异音形似别字在正确语境中的识别速度，通过三个变量的综合分析，运用早期眼动指标（首次注视时间、凝视时间）和晚期眼动指标（总注视时间、回视路径时间、注视次数）来说明字形、语音与语义之间的识别先后及识别速度，并建立更具体化的联结主义模型。 研究结果表明，（1）别字所在整句与无别字整句之间各项眼动指标都无显著差异，因此认为别字可以被学习者顺畅阅读，即别字可被自动修复，完成词语及语句的理解。（2）汉字圈学习者与非汉字圈学习者无论语言水平，在别字的早期识别过程中，都呈现出字形与语义的强联结作用，但非汉字圈的汉字复原与检测效应更弱于汉字圈学习者。（3）在汉语阅读文本加工难度上，汉字圈学习者更善于加工异音形似别字，即形码与语义之间的联结更强；而非汉字圈学习者则更善于加工同音异形别字，即音码与语义之间的联结更强。（4）在别字认知过程的后期阶段，汉字圈学习者更依赖于字音与语义的转化，非汉字圈学习者则对音、形、音形结合的解码速度一致，表示非汉字圈更多注重于早期字形的解码作用。（5）母语者对音、形、义的识别模型中，字音、字形与音形结合的解码都是相同速度的，无一者处于主导地位。因此其认知结构为成熟语言者对汉语阅读的稳定结构。 综上所述，汉语作为第二语言学习者在阅读中确实产生别字的检测和复原效应，并且不同语言背景，语言水平的学习者对汉字的识别有着不同表现，这对对外汉语教学中的汉字与词语教学有着重要的启示，教师可以从中汲取教学中可以改进的教学方法，汉语学习者可以据此采用更加科学有效的学习方法。