Is phonological dyslexia in Japanese patients a selective impairment of reading aloud kana nonwords?

Abstract

  Phonological dyslexia in Japanese-speaking patients has been described as a selective impairment of reading kana nonwords in contrast to preserved ability to read both kana words and kanji words. In the framework of the classical dual-route model, this dyslexic pattern is interpreted as arising from a disruption of the non-lexical route in which a kana string is converted to phonology based on kana-mora correspondence rules, coupled with the preserved lexical route for whole-word translation from orthography to phonology. To understand the mechanisms underlying phonological dyslexia, we conducted a series of experiments involving both reading and non-reading tasks in a right-handed 64-year-old Japanese female patient, AC, with moderate non-fluent aphasia after cerebro-vascular injury in the left insula, the frontal operculum, the precentral gyrus, the postcentral gyrus, and the supramarginal gyrus. In reading aloud kana strings, AC demonstrated good performance with real words and pseudohomophones but moderate difficulty with non-homophonic nonwords. In reading aloud kanji strings, she again showed preserved performance with real words but a disruption with nonwords. Among a variety of phonological tasks in which written materials were not employed, she exhibited marked impairment in mora deletion (e.g., donaki → naki) and mora reversal (e.g., kodohi → hidoko) of words and nonwords. In lexical decision tasks with spoken and kana stimuli, her accuracy deteriorated for kana strings when nonwords were similar to their word counterparts. In comprehension tasks with kanji words when they are presented in kanji, kana(hence pseudohomophones), or spoken forms, her score was significantly lower with kana than with kanji or a spoken form. These results indicate that phonological dyslexia is neither a selective impairment in reading aloud kana nonwords nor a selective disorder of reading aloud. The symptoms observed in AC are assumed to derive from a disruption of the phonological system in the dual-route cascaded model, or from an impairment in phonology in the triangle model.