Phonemic effects in spoken-word recognition in Japanese

Abstract

Previous studies have shown that Japanese listeners are sensitive to the moraic structure of speech, and find it easier to manipulate or respond to morae than phonemes. We further examined moraic processing via two word reconstruction experiments, in which Japanese listeners heard three- or four-mora nonwords which could be changed into real words by substitution of a single mora. In experiment 1, listeners had to change the first mora of the nonword, in experiment 2 the final mora. We compared three types of substitution for CV morae: substitution preserving the C (e.g., kimera or kamere for, respectively, the first and last mora of the word kamera, which has three morae: ka-me-ra), substitution preserving V (namera, kamena), or substitution preserving neither (nimera, kamene). When C or V was preserved, responses were significantly faster and more accurate than when neither was preserved. In initial position, there was no difference between C- and V-preserving substitutions, but in final position, preservation of the C led to faster and more accurate responses than preservation of the V. These results confirm that spoken word recognition in Japanese is sensitive to vocabulary structure and similarity (inter alia at a submoraic level) between words.