On the perception and production of Japanese stops by native speakers of Korean

Abstract

In a series of experiments, native Korean (NK) speakers’ production and perception of Japanese stops was examined. It was predicted that Koreans would not produce Japanese voiced stops authentically given the fact that Korean does not contain these phonetic tokens and that this may also be due in part to their perception of these tokens [i.e., they perceive them differently than do native Japanese (NJ) speakers]. First, a production study revealed that NK produced /g/ VOT values that were significantly different than NJ subjects. Second, perceptual experiments where NK and NJ heard Japanese stops produced by NK and NJ speakers revealed that in general, the two groups were in agreement when deciding whether the tokens were voiced or voiceless. However, there was a disparity between the two in that some of the NK tokens of /g/ were perceived as /k/ by NJ but were perceived as intended /g/ tokens by NK subjects. Results of an additional perceptual experiment also demonstrated that NK subjects treat Japanese voiceless stops as equivalent to Korean aspirated stops and Japanese voiced stops as Korean plain unaspirated stops (even though the VOT values for the Japanese voiceless and Korean unaspirated plain stop are more similar).