Abstract
The triangle model is a framework for lexical processing which computes word orthography, phonology and meaning using the architecture of a parallel distributed processing network. The computation takes the form of interactions among neuronlike processing units. The present research was designed to simulate normal and dyslexic reading performances of Japanese character strings using the triangle model as implemented on a computer. The Japanese triangle model computed phonology from orthography and meaning for both Kanji and Kana strings. This model successfully simulated certain reading effects seen in the reading performance of Japanese skilled readers. Moreover, different types of damage to the model reproduced data on both the surface and phonological forms of acquired dyslexia. In simulation of surface dyslexia, the model showed reading Kanji words with consistent character-sound correspondences, Kana words and Kana nonwords was much better than Kanji words with atypical character-sound correspondences after damage to meaning. In simulation of phonological dyslexia, the model showed reading of words (both Kanji and Kana) was much better than of Kana nonwords after damage to phonology itself. These results are basically comparable to those of previous models developed for English, and thus demonstrate that the same computational principles of the triangle model can be applied to alphabetic and non-alphabetic writing systems. Mechanisms and properties of the model for Japanese are discussed.
