研究目的：本研究的目的在於檢驗認字困難閱障生的基本組字規則知識及他們表徵文字的方式是否與一般學童有差異。在知覺的研究領域裡，倒立效果的研究典範已被視為是探討表徵能力的良好方式，也就是物體的倒立虧損愈大，代表該物體愈傾向以整體方式進行表徵。研究方法：本研究採用倒立的研究方式，要求參與者判斷各類正立及倒立的字（常見字、罕見字、假字、非字）是否像字，然後比較倒立虧損之大小。共有13名認字困難閱障生、12名無認字困難閱障生及13名一般生參與研究。研究結果：本研究發現認字困難閱障生比一般人及無認字困難閱障生有較多的倒立虧損，即當辨識漢字倒立時，所犯的錯誤較多。然而，認字困難閱障生對各類文字的區分程度，則與一般人及無認字困難閱障生沒有差異。研究結論：認字困難閱障生具基本的組字規則知識，但文字表徵卻異於常人。

Purpose: Dyslexia with Chinese character recognition disability characterizes significant word recognition difficulties that impacts on their comprehending of the text. Therefore, the purpose of this paper is to explore the character concepts of dyslexia with Chinese character recognition disability. It has been suggested that the face inversion effect is perhaps the best existing evidence for the holistic encoding of faces (Martelli et al., 2005); and further, some researchers have proposed that the inversion effect is associated with holistic pattern perception, where or not the pattern is a face (Farah et al., 1995; 1998; Freire et al, 2000; Murray et al., 2000). Thus, this paper uses the inversion paradigm to explore the representation of character in developmental dyslexia with Chinese character recognition disability. On the other hand, this study is also conducted to examine the basic orthographic knowledge of character structure in developmental dyslexia with Chinese character recognition disability. Methods: Thirteen dyslexic students with Chinese character recognition disability are compared with 12 dyslexic students without Chinese character recognition disability (reading-disabled group, RD) and 13 average readers of the same chronological age (CA controls) in the Lexical Decision task. The characters are presented upright or upside down, and then participants are required to decide whether the characters are like the real character or not. Results: The results show that dyslexic children with Chinese character recognition disability perform the most inverted cost among three groups, this is, their performance are disproportionately impaired by inversion. However, the percentages of looking like real characters for popular characters, rare characters, pseudo-characters, and non-characters in the CA controls are similar to the corresponding percentages in the dyslexia with or without Chinese character recognition disability. Conclusions: First, according to Farah et al. (1995, 1998) if orientation invariance in pattern recognition is achieved through a normalization process and that process operates on units of shape, be their parts or undecomposed wholes, then complex, holistically represented stimuli will be the most taxing for the normalization process. It would be possible that for dyslexia with Chinese character recognition disability, their characters are stored as undecomposed perceptual wholes in which individual parts (components) are not explicitly represented. And then they encounter the capacity limitations of the orientation normalization process. Further, according to Freire et al. (2000) and Murray et al. (2000) the encoding of spatial-relational information is disproportionately impaired when the face is inverted. The encoding ability of Chinese character of the dyslexia with Chinese character recognition disability is different from the other two groups. Second, for the performance profile of differentiating real Chinese characters or not on both upright and inverted characters, the dyslexia with Chinese character recognition disability is not different from the other two groups. The dyslexia with Chinese character recognition disability has the same basic knowledge of Chinese character structure as the other two groups. Finally, the inversion effect for Chinese non-characters is largest among that of all other characters. According to Farah et al. (1995, 1998), non-character is the most complex, holistically represented stimulus, and then it is the most taxing for the normalization process. Further, according to Freire et al. (2000) and Murray et al. (2000), the encoding of spatial-relational information is disproportionately impaired when the non-character is inverted. According to Bartlett and Searcy (1993), inversion reduces the perceived grotesqueness of faces in which spatial information has been altered, the non-character is constructed varying the legality of the radical positions, and then inversion could also reduce this perceived grotesqueness of non-character in which spatial information has been altered. At this time non-character looks like real characters, and then participants made more decision-making mistakes.