Differential roles of spatial frequency on reading processes for ideograms and phonograms: A high-density ERP study

Abstract

The neural substrate of the dissociation between reading Japanese ideograms (Kanji) and phonograms (Kana) is currently unclear. To test whether spatial frequency (SF) information is responsible for this phenomenon, we recorded high-density event-related potentials (ERPs) with unfiltered or spatially filtered word stimuli in Japanese-speaking subjects. Kanji (early-learned, late-learned), Kana (word, non-word), and scrambled characters served as stimuli. Fourier analysis revealed that Kanji and Kana were characterized by high-SF (HSF) and low-SF (LSF) information, respectively. In ERPs with unfiltered stimuli, bilateral occipital P100, left occipitotemporal N170 and fronto-central N400 were elicited. Scrambled characters did not evoke left-lateralized N170 or clear N400. Under the LSF condition, P100 and N170 latencies for Kanji were significantly longer than those for Kana. In the HSF condition, P100 and N170 latencies for late-learned Kanji were significantly longer than those for early-learned Kanji. There was no significant difference in the N400 between Kanji and Kana in both SF conditions. These results suggest that early visual responses, but not the semantic component, are influenced by SF. This indicates a close link between Kana and LSF information, and between Kanji and HSF information. The differential effects of SF could underlie the neural basis of the differences between Kanji and Kana reading.

Distinct role of spatial frequency in dissociative reading of ideograms and phonograms: An fMRI study

“…1Ab (right) and Ac (right), respectively. In a behavioral study using the same spatially filtered stimuli, no significant differences were found in their readability (subjects' mean reaction times or error rates) (Horie et al, 2012). Therefore, our spatially filtered Kanji and Kana stimuli were equally legible.…”

“…We found that early visual responses (P100 and N170), but not semantic processing (N400), are significantly influenced by SF, and that close links exist between Kana and LSF information and between Kanji and HSF information. Hence, we concluded that SF information contributes to the neural dissociation of Kanji and Kana reading (Horie et al, 2012). However, because the spatial resolution of ERPs is much inferior to that of functional magnetic resonance imaging (fMRI), the neural substrates involved remain unclear.…”

“…Each range of subtracted SF fit the cut-off frequencies for our stimuli (10.9 cycles/ 3 characters for LSF; 62.3 cycles/3 characters for HSF). In addition, Kana contained a significantly lower number of strokes than Kanji (p b 0.01) (Horie et al, 2012). Sugishita et al (1992) alleged that the dissociation between Kanji and Kana reading reported in previous studies was caused by methodological problems, such as failure to control for individual differences in facility for the two writing systems, or a lack of consistency in the level of difficulty across stimuli.…”

“…To create the HSF and LSF stimuli for Experiments 2 and 3, we manipulated the SF information in the unfiltered BSF stimuli using image-engineering techniques with 2D-FFTs (Horie et al, 2012). Eighty five-order Hamming window methods were applied for creating LSF images, while HSF images were created using 31-order Hamming window methods.…”

“…Recently, we investigated the role of SF in Kanji and Kana reading by recording high-density visual event-related potentials (ERPs) (Horie et al, 2012). We found that early visual responses (P100 and N170), but not semantic processing (N400), are significantly influenced by SF, and that close links exist between Kana and LSF information and between Kanji and HSF information.…”