Neural correlates of semantic and phonological processing revealed by functional connectivity patterns in the language network.

Semantics and phonology are fundamental components of language. Neuroimaging studies have identified a language network (LN) that is distributed through multiple regions and exhibits preferential responses to semantic and phonological information. However, it is unclear how these regions work collaboratively to support the processing of these components. In the present study, we first defined the LN as voxels that responded more to sentences than to strings of Chinese pseudo-characters. We subsequently used a voxel-based global brain connectivity method based on resting-state functional connectivity (FC) to characterize the neural correlates of semantic and phonological processing. We specifically correlated the within-network connectivity (WNC) of each voxel in the LN with the participants' scores on the semantic and phonological components extracted from a battery of reading tests via principal component analysis. We found that individuals with stronger WNC in the left posterior superior temporal gyrus (lpSTG) and anterior superior temporal gyrus (laSTG) were better at semantic and phonological processing, respectively. Furthermore, the FC of the lpSTG with the laSTG and bilateral fusiform gyrus mainly contributed to semantic processing, whereas the FC of the laSTG with the left posterior middle temporal gyrus and inferior frontal gyrus largely contributed to phonological processing. Importantly, the semantic and phonological subnetworks overlapped in the laSTG, the WNC of which correlated with the participants' performances during semantic-phonological interactions. Our study revealed the hub and subnetwork for semantic and phonological processing, respectively, and highlighted the role of the laSTG in semantic-phonological interactions.