[Preserved ability to read aloud kanji idioms in left handed alexia].

We report a 69-year-old left-handed man, who developed alexia after a right medial occipito-temporal lobe infarction. On admission to the rehabilitation department two months after the onset, neurological examination showed left hemianopia, left hemiparesis, decreased deep sensation on the left side, and alexia. A brain MRI demonstrated infarcts in the right medial occipito-temporal lobe and the splenium of the corpus callosum. Detailed neuropsychological examination was performed two months after the onset. The patient was alert and cooperative. His speech was fluent with some word-finding difficulty. Comprehension for spoken materials, repetition, and naming abilities were all preserved. Systematic examination for reading revealed that reading aloud was disturbed in both kanji and kana words. Reading comprehension was significantly better for kanji words than kana words. First, we examined the effects of number of characters in a word. The number of characters in a word didn't affect his reading performance. Second, his performance on reading aloud of usual kanji words was compared with that of kanji words representing idioms. A kanji idiom is different from usual kanji words, in which pronunciation of each character is selected from several options. Reading aloud kanji idioms was significantly better than usual kanji words. In addition, reaction time to complete reading a word was much shorter for kanji idioms than usual kanji. An analysis of qualitative features of errors revealed that most errors in kanji idiom reading were semantically similar to the correct answers, while many errors in usual kanji word reading were classified into "don't know" responses. These findings suggested that a kanji idiom was tightly connected to its pronunciation, which resulted in his much better performance for kanji idiom reading. Overlearning of a unique relationship between a kanji idiom and its pronunciation might modify neuronal organization for reading.