On the patterning of voiced stops in loanwords in Japanese

TLDR: The authors argue that the instability of Japanese voiced stops under gemination may also be attributed to the phonological system of Japanese and propose a phonetic account grounded in perception for the variable patterning of voiced stops.

Abstract: In loanwords in Japanese, a consonant gemination process is frequently found. While voiceless stops consistently geminate and are voiceless in the appropriate environment, voiced stops may geminate as voiced stops, may fail to geminate, or may geminate as voiceless stops. A phonetic account grounded in perception has been proposed for the variable patterning of voiced stops under gemination conditions. I argue that while this is a possible explanation, the instability of voiced stops under gemination may also be attributable to the phonological system of Japanese. Thus, the role of perception in the variable patterning of voiced stops remains an open question.