Weekes and Chen (1999) report the case of LJG, a Chinese-speaking patient with anomia and semantic impairment following a left hemisphere cerebrovascular accident. Despite severe difficulties with naming and comprehension, LJG retained relatively strong oral reading ability, especially for concrete Chinese characters. However, his performance showed classic signs of surface dyslexia—a pattern where reading irregular words is impaired due to disrupted semantic processing.
The study tested LJG’s ability to read monosyllabic Chinese characters aloud, which varied in regularity (predictability of pronunciation from components), frequency, and concreteness. Irregular characters—those whose pronunciation could not be reliably derived from their components—were especially problematic, particularly when they were also low-frequency and abstract. LJG often misread these words by assigning the pronunciation of a character component rather than the whole, known as a LARC (Legitimate Alternative Reading of a Component) error.
The authors argue that in normal reading, semantic support helps resolve response competition at the phonological output level, especially for irregular, low-frequency characters. When semantic systems are impaired, as in LJG’s case, the system defaults to more frequent or component-based pronunciations, resulting in surface dyslexia. This aligns with previous findings in English and Japanese, where semantic impairment led to difficulty reading irregular words.
Interestingly, LJG could still accurately read many concrete characters, even when he couldn’t name or comprehend them—suggesting the presence of a direct orthography-to-phonology route that bypasses meaning. However, this pathway is less reliable for abstract and irregular words, where semantic mediation becomes more critical.
In conclusion, this study extends the concept of surface dyslexia to Chinese. It supports the view that semantic impairment increases reliance on suboptimal decoding routes, particularly when reading irregular, abstract characters. This has implications for understanding reading mechanisms in logographic writing systems.
