Brain networks associated with sublexical properties of Chinese characters

Cognitive models of reading all assume some division of labor among processing pathways in mapping among print, sound and meaning. Many studies of the neural basis of reading have used task manipulations such as rhyme or synonym judgment to tap these processes independently. Here we take advantage of specific properties of the Chinese writing system to test how differential availability of sublexical information about sound and meaning, as well as the orthographic structure of characters, pseudo-characters and “artificial” control stimuli influence brain activation in the context of the same one-back task. Analyses combine a data-driven approach that identifies temporally coherent patterns of activity over the course of the entire experiment with hypothesis-testing based on the correlation of these patterns with predictors for different stimulus classes. The results reveal a large network of task-related activity. Both the extent of this network and activity in regions commonly observed in studies of Chinese reading are apparently related to task difficulty. Other regions, including temporo-parietal cortex, were sensitive to particular sublexical functional units in mapping among print, sound, and meaning.

► Chinese readers show graded behavioral sensitivity to word-likeness, as defined by the inclusion of different types of sublexical information.
► A novel analysis approach to fMRI data identifies the typical ”reading network,” and shows evidence for stimulus selectivity in a much broader network.
► Throughout the reading network, activity is largely modulated by task difficulty, but some regions show selectivity for particular forms of sublexical information.
► The results are consistent with models in which sublexical processing is a natural part of reading in Chinese.