ABSTRACT
In two eye-tracking experiments, we investigated the processing of information about phonological consistency of Chinese phonograms during sentence reading. In Experiment 1, we adopted the error disruption paradigm in silent reading and found significant effects of phonological consistency and homophony in the foveal vision, but only in a late processing stage. Adding oral reading to Experiment 2, we found both effects shifted to earlier indices of parafoveal processing. Specifically, low-consistency characters led to a better homophonic foveal recovery effect in Experiment 1 and stronger homophonic preview benefits in Experiment 2. These findings suggest that phonological consistency information can be obtained during sentence reading, and compared to the low-consistency previews the high-consistency previews are processed faster, which leads to greater interference to the recognition of target characters.
Acknowledgments
Data and R scripts used in the current study are available from the corresponding author on reasonable.
Disclosure statement
The authors declare that there is no conflict of interest.
Notes
1. An example is provided here to explain how type consistency and token consistency are calculated. There are five characters in a phonetic radical family, of which three (A, B and C) share the same pronunciation X. The other two (D and E) share a different pronunciation Y. Then the type consistency values for characters A, B and C are all 3/5 = 0.6. The token consistency value of a given character is the accumulative frequency of its pronunciation divided by the accumulative frequency of the whole family. If the frequencies of characters A, B, C, D, and E are 5, 10, 10, 5 and 20, respectively, the consistency value for character A (as well as for B and C) is (5+10+10)/(5+10+10+5+20) = 0.5. A higher consistency value indicates that the pronunciation of the character is more representative of its phonetic-radical family.
2. GD on pre-target words (i.e., preview duration) showed a longer processing time for identical previews (b = 0.017, SE = 0.005, t = 3.27, p = .001) and for homophone previews (b = 0.011, SE = 0.003, t = 3.10, p = .003), replicating the so-called parafoveal-on-foveal effects reported by Pan et al. (Citation2016). However, there was no statistically significant difference between high- and low-consistency conditions (p = .829), indicating that the consistency effect was not due to preview words having been processed with different amounts of time.
3. Low-consistency characters tend to have larger families (i.e., number of characters sharing a phonetic radical; M = 10.9, SD = 5.1) than high-consistency characters (M = 5.5, SD = 3.4; F (1, 207) = 84.41, p < .001). Similarly, low-consistency characters tend to have more phonological alternatives within their radical families (M = 6.4, SD = 2.4) than high-consistency characters (M = 2.5, SD = 1.9; F (1, 207) = 166.69, p < .001). These facts offer an alternative explanation to the consistency effect per se: lexical access is slower for low-consistency characters, not only because of their less-consistent orthography-to-phonology mapping, but also because these characters have more orthographic neighbors sharing the phonetic radicals serving as competitors. Nevertheless, both accounts are compatible with our explanation that lower-level lexical access of the low-consistency characters leads to less interference.