Language specific listening of Japanese geminate consonants: a cross-linguistic study

Various aspects of linguistic experience influence the way we segment, represent, and process speech signals. The Japanese phonetic and orthographic systems represent geminate consonants (double consonants, e.g., /ss/, /kk/) in a unique way compared to other languages: one abstract representation is used to characterize the first part of geminate consonants despite the acoustic difference between two distinct realizations of geminate consonants (silence in the case of e.g., stop consonants and elongation in the case of fricative consonants). The current study tests whether this discrepancy between abstract representations and acoustic realizations influences how native speakers of Japanese perceive geminate consonants. The experiments used pseudo words containing either the geminate consonant /ss/ or a manipulated version in which the first part was replaced by silence /_s/. The sound /_s/ is acoustically similar to /ss/, yet does not occur in everyday speech. Japanese listeners demonstrated a bias to group these two types into the same category while Italian and Dutch listeners distinguished them. The results thus confirmed that distinguishing fricative geminate consonants with silence from those with sustained frication is not crucial for Japanese native listening. Based on this observation, we propose that native speakers of Japanese tend to segment geminated consonants into two parts and that the first portion of fricative geminates is perceptually similar to a silent duration. This representation is compatible with both Japanese orthography and phonology. Unlike previous studies that were inconclusive in how native speakers segment geminate consonants, our study demonstrated a relatively strong effect of Japanese specific listening. Thus the current experimental methods may open up new lines of investigation into the relationship between development of phonological representation, orthography and speech perception.