The Representations of Chinese Characters: Evidence from Sublexical Components
Xiaodong Liu,David Wisniewski,Luc Vermeylen,Ana F Palenciano,Wenjie Liu,Marc Brysbaert
DOI: https://doi.org/10.1523/JNEUROSCI.1057-21.2021
2022-01-05
Abstract:Little research has been done about the neural substrate of the sublexical level of Chinese word recognition. In particular, it is unclear how radicals participate in Chinese word processing. We compared two measures of radical combinability, position-general radical combinability (GRC) and position-specific radical combinability (SRC) depending on whether the position of the radical is taken into account. We selected characters with embedded target radicals that had different GRC and SRC measures. These measures were used as predictors in a parametric modulation analysis and a multivariate representational similarity analysis. Human participants with native Mandarin speakers (17 males and 24 females) were asked to read words in search of animal words. Results showed that SRC is a better predictor than GRC in decoding the neural patterns. Whole-brain analysis indicated that SRC is encoded bilaterally in the inferior frontal gyrus (IFG, pars opercularis, and pars triangularis), the middle frontal gyrus (MFG), and a region on the border of the superior parietal lobule and the inferior parietal lobule (SPL/IPL). Region-of-interest-based RSA confirmed the results of the whole-brain analysis. Furthermore, we observed a correlation of another sublexical variable, logographeme composition, with bilateral activity in SPL. Logographemes refer to the basic stroke combinations that form radicals and characters. Finally, we observed involvement of bilateral cerebellum activity in Chinese word recognition. Our findings confirm the importance of sublexical components (SRC and logographeme composition) in Chinese word recognition and also confirm that Chinese word recognition involves more bilateral processing than word recognition in alphabetical languages.SIGNIFICANCE STATEMENT Chinese is a logographic language. However, characters contain informative subword components (radicals, logographemes, and strokes). We investigated whether the position of the radical is important. We presented carefully selected words and looked where brain activity correlated with subword information. Results indicate that position-dependent radicals predict brain encoding in a network of regions associated with Chinese word recognition, including higher order regions such as bilateral IFG, MFG, and SPL/IPL. Logographeme composition had an effect as well. Our findings provide strong evidence (1) for the importance of position-specific radical information and logographemes in Chinese word recognition, (2) that current brain imaging techniques are best suited to study these, and (3) that confirms the interactive nature of Chinese character recognition.