Lexical and character-based sublexical processing in Japanese kanji reading: evidence from kanji nonword reading in...

This study investigated mechanisms underlying oral reading of two-character kanji nonwords in Japanese. Unlike alphabetic orthographies, kanji does not permit straightforward grapheme-to-phoneme conversion because individual characters often have multiple readings whose typicality varies across lexical and positional contexts. We compared a lexical-only account with an account in which lexical processing is supplemented by character-based sublexical processing. Twenty-one Japanese-speaking persons with aphasia and 24 healthy controls read aloud two-character kanji words, pseudohomophone nonwords (PSH), and non-pseudohomophone nonwords (NPSH). In persons with aphasia, PSH accuracy exceeded NPSH accuracy but remained below word reading accuracy. Errors included legitimate alternative reading of components (LARC)-type responses, lexicalizations, and phonological nonword errors. Model-informed classification identified phonological dyslexia, surface dyslexia, mixed patterns, and unclassifiable cases. The findings suggest that kanji nonword reading depends on interactions between whole-word lexical activation and character-based candidate reading processes, supporting a dual-route account adapted to the specific properties of Japanese kanji.