|
The Japanese journal of neuropsychology
|
|
Full Text of this Article
in Japanese PDF (76K)
|
ArticleTitle
|
Surface dyslexia in Japanese -A computational account from simulation study- |
Language |
J |
AuthorList |
Mutsuo Ijuin, Takao Fushimi, Itaru F. Tatsumi |
Affiliation |
Language, Cognition and Brain Science Research Group, Tokyo Metropolitan Institute of Gerontology |
Publication |
Japanese Journal of Neuropsychology: 18 (2), 101-110, 2002 |
Received |
Sep 21, 2001 |
Accepted |
Dec 25, 2001 |
Abstract |
The Japanese script comprises morphographic Kanji and phonographic Kana. Whereas each Kana character invariably represents a single pronunciation, the majority of Kanji characters have two or more legitimate pronunciations, with one appropriate to the character in any given word determined by the word-specific knowledge. Surface dyslexia in Japanese-speaking cases has been traditionally characterized as selective disorder of reading aloud Kanji words, coupled with preserved ability to read aloud Kana words and Kana nonwords. Recent research, however, revealed that deficit for Kanji words is remarkable only for less common words with atypical character-sound correspondences in which patients assign incorrect but typical pronunciations to component characters whereas the performance for common words and/or words with typical correspondences is relatively spared. Because surface dyslexia often accompanies semantic dementia or progressive fluent aphasia, it is suggested that deterioration of word meaning and/or reduced phonological activation from semantics is responsible for surface dyslexia.
In the present study, surface dyslexic reading was simulated by the triangle model as, in part, implemented on a computer. The model was trained to compute the pronunciations of Kanji and Kana words directly from their written forms, supplemented by an additional input intended to stand for phonological activation arising from word meaning. This model successfully simulated skilled performance of Japanese readers: frequency-by-consistency interaction in computing pronunciation of Kanji words as well as ability to yield correct pronunciation for Kanji nonwords. After withdrawal of additional input from putative word meaning, the model reproduced surface dyslexic pattern, with its accuracy much better for Kana words and nonwords, high-frequency Kanji words, and Kanji words with typical correspondences than low-frequency Kanji words with atypical character-sound correspondences, to whose component characters the model often erroneously produced the typical pronunciations. Mechanisms and properties of this model are discussed. |
Keywords |
surface dyslexia, the triangle model, simulation, Kanji, semantics |
|