Hino & Lupker (1998) examined word frequency effects for Japanese Kanji and Katakana words using naming, lexical decision, and go/no-go naming tasks. Whereas word frequency effect sizes were identical for Kanji and Katakana words in the lexical decision task, frequency effects were larger for Kanji words than for Katakana words in the naming task. In the go/no-go naming task, in which participants were asked to name a stimulus aloud only if it was a word, frequency effects were larger for Kanji words than for Katakana words as in the naming task and these effect sizes were larger than those in the lexical decision task. These results were consistent with the predictions based on parallel distributed processing model in which identical phonological coding processes are assumed for Kanji and Katakana words. As such, these results appear to indicate that phonological coding for Kanji and Kana words is accomplished based on similar operations.

View full abstract

    We propose that the basic process of kanji and kana reading is captured in a triangle framework suggested by Seidenberg & McClelland (1989), which assumes that the orthography, phonology, and meaning of words are interactively computed on the basis of parallel distributed processing. Our proposal, that the identical architecture and algorithm apply to both kanji and kana strings, found support in normal readers performance : In reading aloud two-character kanji words, normal readers were slower at low-frequency words with atypical character-correspondences than at either high-frequency words or words with consistent/typical correspondences. They could read aloud two-character kanji nonwords fluently. Normal readers were slower at low-imageable words than at high-imageable words, but the imageability effect emerged only when the target was low-familiar kanji words with atypical character-sound correspondences. In reading aloud kana nonwords, normal readers were faster at pseudohomophones (viz. orthographic nonword homophonic with real words) than at non-homophonic nonwords at when the pseudohomophones were homophonic with high-imageable kanji words. These effects indicate that the phonology of kanji and kana strings is computed directly from orthography, with support from meaning when direct computation is inefficient.
    The nature of normal readers' performance suggested that Japanese surface dyslexic patients suffering from semantic impairment should show severe deficit in reading aloud low-familiar kanji words with atypical character-sound correspondences, but relatively preserved performance for high-frequency words or words with typical character-sound correspondences. Furthermore, Japanese phonological dyslexic patients suffering from phonological impairment should be expected to show severe deficit in reading aloud non-homophonic nonwords, but relatively better performance for pseudohomophone homophonic with high-frequency and/or high-imageable words, in whatever script the nonwords comprise. Our assumptions were confirmed by reports of surface and phonological dyslexic patients in the literature and by our own additional tests.

View full 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.

View full abstract

    We reported three cases of surface dyslexia in Japanese. Case 1 was a patient with Alzheimer-type dementia and Cases 2 and 3 were patients with semantic dementia. All subjects were able to read well both words composed of two to eight kana characters and nonwords composed of three to five kana characters, but could not read words composed of two or three kanji characters without a prevailing tendency to substitute other (more typical) pronunciations of the component characters. Cases 2 and 3 (typical surface dyslexia) exhibited prominent consistency and familiarity (frequency) effects in reading kanji-words, as English cases have shown. However, they demonstrated a significant number of “don't know” responses, which might reflect characteristics of Japanese orthography.

View full abstract

    The ability to comprehend concrete words was assessed in a case who probably had been suffering from Pick's disease with an MRI proven prominent atrophy of the left temporal lobe. When he was admitted for evaluation two months after the first sign of his abnormal language behavior, he showed a typical picture of Gogi (word meaning) aphasia.
    At the first evaluation, the patient showed a difficulty in understanding the meaning of a heard word. He was unable to complete a word when only the first few syllables were given. However he could repeat a sentence quite well and distinguish a real word from an unreal one easily. He was also able to point to a drawing corresponding to a heard word with about 60% accuracy. His ability to categorize drawings was normal.
    A year later, he still was able to distinguish a real word from unreal one although the response became slower. He could categorize drawings correctly just like the first evaluation. On the contrary his ability to point to a drawing corresponding to a heard word showed considerable deterioration with only 30% accuracy.
    We interpret that his preserved ability to distinguish between real and unreal words and categorize drawings reflect preservation of “phonological lexicons” and “non-linguistic” meaning throughout this observation period. Lack of spontaneous comprehension of heard words with relative preservation of pointing ability of drawings corresponding to heard words at the first evaluation suggest the preservation of connection between “non-linguistic” meaning and “phonological lexicons” although their “dictionary” or “linguistic” meaning was gone. Deterioration of this pointing ability a year after suggest the additional difficulty in connecting “non-linguistic” meaning with “phonological lexicons” . This dissociated affection of the “phonological lexicons” , “linguistic meaning” , and “non-linguistic meaning” over the course of the disease suggests a word is probably composed of these three relatively separable subsystems. Thus we suppose the progressively deteriorated processes of Gogi aphasia in the present patient as follows : first, the disruption of “linguistic” meaning, second, the disorder in association between “phonological lexicons” and “non-linguistic” meaning.

View full abstract

    We extracted consonants from syllables by computer, prepared well-controlled but easily perceivable syllable and consonant discrimination tests, and applied them to examine 21 aphasic patients and 18 normal geriatric subjects. The aphasic patients showed disturbances in the syllable discrimination test, but disturbance in the consonant discrimination test was milder. There was no significant difference between the scores of the aphasics and those of the normal controls when the consonants were presented to the healthy (left) ear. In the aphasics, there was a significant correlation between the results of the syllable discrimination test and auditory word-comprehension ability (auditory retention span). At the same time, there was a significant correlation between the consonant discrimination test and auditory resolution performance (click fusion threshold). Factor analysis of the responses of the aphasic patients in the consonant discrimination test showed that acoustic feature analysis of speech sounds in the human brain is performed based on the coordinate axes of the participation of the vocal code (voiced vs. unvoiced) , the participation of the velum palati (oral vs. nasal) , the point of articulation, and the manner of articulation. Thus the theory which classifies phonemes based on these phonetical axes is neuropsychologically valid. The findings suggest that in cases where comparison between the results of acoustic feature analysis and memory of phonemes stored in the Wernicke's area is disturbed, identification of speech sounds is impossible even though phonetical feature analysis may be valid.

View full abstract