Andrew P. Lea
ATR Human Information Process. Res. Labs., 2-2 Hikaridai, Seicha-cho, Kyoto, 619-02 Japan
This paper presents new evidence that listeners can use the difference between voiced and whispered speech to segregate both voiced and whispered members of concurrent vowels consisting of voiced/whispered pairings. Previous experiments performed by Lea and Summerfield [Acoust. Soc. Jpn. Tech. Rep. H-92-31] required English listeners to identify the members of pairs of steady-state synthetic English vowels, called concurrent vowels. Accuracy of identification was higher when one vowel was voiced and the other whispered compared to control conditions. Surprisingly, the improvement in accuracy was restricted to the whispered member of the voiced/whispered pair, thus suggesting that the whispered vowel was segregated, but not the voiced vowel. Lea and Summerfield's study was replicated in this paper using Japanese vowels and listeners, however, the results suggest that both the voiced and whispered members of the voiced/whispered pair were segregated. These incompatible results either indicate a difference in segregational ability between English and Japanese listeners, or a difference in procedure between the experiments. Correspondingly, further experiments were performed using Lea and Summerfield's English vowels and English and Japanese listeners which show that the incompatible results are mainly due to the different methods of matching the levels of the voiced and whispered vowels between the two experiments. Thus the evidence suggests that the Japanese results indicate the true segregational ability of listeners.