Allograph priming is based on abstract letter identities: evidence from Japanese kana

It is well-established that allographs like the uppercase and lowercase forms of the Roman alphabet (e.g., a and A) map onto the same “abstract letter identity,” orthographic representations that are independent of the visual form. Consistent with this, in the allograph match task (“Are ‘a’ and ‘A’ the same letter?”), priming by a masked letter prime is equally robust for visually dissimilar prime-target pairs (e.g., d and D) and similar pairs (e.g., c and C). However, in principle this pattern of priming is also consistent with the possibility that allograph priming is purely phonological, based on the letter name. Because different allographic forms of the same letter, by definition, share a letter name, it is impossible to rule out this possibility a priori. In the present study, we investigated the influence of shared letter names by taking advantage of the fact that Japanese is written in two distinct writing systems, syllabic kana—that has two parallel forms, hiragana and katakana—and logographic kanji. Using the allograph match task, we tested whether a kanji prime with the same pronunciation as the target kana (e.g., special characters please refer to PDF, both pronounced /i/) produces the same amount of priming as a kana prime in the opposite kana form (e.g., special characters please refer to PDF). We found that the kana primes produced substantially greater priming than the phonologically identical kanji prime, which we take as evidence that allograph priming is based on abstract kana identity, not purely phonology.