Korean vocabulary approximately consists of 24.5 % of Chinese-derivative words and 69.32 % of pure Korean words. Previous studies suggested that the mental lexicon of Chinese-derivative words is segregated from that of pure Korean words. In this study, when aphasics with dyslexia read aloud, their ability was better in Chinese-derivative words than in pure Korean words. Further, whereas there were more errors at the syllabic level in Chinese-derivative words, we observed more errors at the phonemic level in pure Korean words. The most common errors were nonword errors by replacing a syllable in the former and nonword errors by substituting a phoneme in the latter. These results suggest that the storage and retrieval of Chinese-derivative words are processed in syllable units while those of pure Korean words are processed as whole words. Thus, this study supports the hypothesis for the segregation of these two kinds of Korean vocabulary in the mental lexicon.