| The aim of the current set of experiments was to understand the time-course of Japanese word processing. While numerous studies have looked at the time-course of word processing in Roman alphabetic languages, it is unclear whether the same processes apply to scripts that are completely dissimilar to the Roman alphabet. In the first two experiments, we examined within- (Experiment 1) and between-script (Experiment 2) masked repetition priming effects in two Japanese syllabaries -- Hiragana and Katakana. In both experiments native Japanese speakers were presented with repeated or unrelated pairs of words in the same script (Experiment 1) and different script (Experiment 2). In both experiments there was a significant N250 and N400 priming effect, but the N/P150 effect was only present in the within-script condition. N250 and N400 effects appeared later compared to prior work in alphabetic scripts, which suggests that syllabary scripts place more emphasis on phonological processing compared to languages that use alphabetic scripts. Experiment 3 investigated this possibility by conducting a masked cross-modal experiment. Using the same stimuli words as those used in the first two experiments, we exchanged the target stimuli to auditory words. Once again, we found robust priming effects on the N400 for both Hiragana and Katakana. Interestingly, however, we also found a priming effect on the P2 appearing only for Katakana but not Hiragana. We conclude that Katakana has closer connections with phonology, thus showing a strong mapping of phonological units onto phonological words during the early stages of word processing. Hiragana on the other hand seemed to influence priming predominantly through whole-word representations. Given these results, we proposed a modification of the bimodal interactive-activation model (BIAM) is to account for word recognition in syllabary scripts.;Experiments 4 and 5 focused on the processing of Japanese logograph (Kanji). In Experiment 4, we examined whether complex Kanji were decomposed into their constituent morpho-orthographic units (radicals) during processing similar to morphologically complex words in alphabetic languages. We achieved this by conducting an ERP masked repetition priming in prime-target pairs with semantically and orthographically transparent relationship, semantically opaque but orthographically transparent relationship, and semantically unrelated but orthographically transparent relationship. We found that semantically and orthographically transparent pairs were decomposed into its constituent radicals during processing similar to morphologically complex words, as reflected in priming effects in the N250 and N400 components. The other two conditions showed marginal priming effects in the N250, indicating a possibility of morpho-orthographic decomposition, however, the effects failed to reach significance. The semantically opaque but orthographically transparent pairs also failed to show priming effects on the N400, indicating a possibility of interference effects. On the other hand, we did see an N400 effect in the semantically unrelated but orthographically transparent pairs. However, this appeared to be anteriorly distributed, similar to those seen in object priming. Experiment 5 examined the effects of radical position and function on Kanji decomposition that was not covered in Experiment 4. This experiment used a lexical decision task and ERPs to examine the differences in processing of complex Kanji by breaking down the functional (semantic vs. non-semantic, free- vs. bound radical) and positional (left/right orientation, top/bottom orientation) information embedded in each radical. We found that in the early processes of Kanji processing, Kanji orientation and semantic radicals modulated the N250 component especially in Non-words, providing evidence that Kanji readers may be conducting a first scan of positional and semantic radical information. Non-semantic radicals modulated the later N400 component, again in Non-words than Words, and more so if characters contained free-semantic radicals rather than bound-semantic radicals. This indicated that non-semantic radicals are processed only after semantic radicals are activated, and more so if semantic radicals could only be processed when attached to a non-semantic unit. Overall, our results revealed that Japanese word processing can be understood using the same model of word processing as those used in Roman alphabets. While there are distinct differences in processing between Japanese and alphabetic scripts due to fundamental differences in form and function, the underlying processes seem to be similar. |
