Phonological Involvement in the processing of Japanese at the lexical and sentence levels

The present study examined whether Japanese readers activate phonological information when reading kanji compound words and sentences and if so, how they do it. Experiment 1 used two-kanji compound words in a lexical decision task to study phonological processing at the lexical level. When nonwords were pseudo-homophones (朗費 /roR hi/ in place of the real word 浪費 /roR hi/), reaction times were longer and more errors occurred than with nonwords in the control group (削費 /saku hi/). Experiment 2 required participants to detect misspellings (i.e., incorrect kanji combinations) of two-kanji compound stimuli embedded in sentences. In the detection task of misspelled kanji, no homophonic effect was apparent. Experiment 3 used a semantic decision task. Included in this task were semantically similar but incorrect kanji compound words used as fillers in sentences (e.g., ‘向こうに見える大きな建物は、私の知人が設備したものだ' meaning ‘The building you can see over there was facilitated by my friend' instead of designed) as well as the sentences used in Experiment 2. Results from Experiment 3 indicated that participants could reject a sentence as incorrect more quickly when pseudo-homophones were embedded in the sentences rather than nonwords. These results suggest that readers activate phonological information of two-kanji compound words when reading for comprehension but not for simple proofreading