Beschreibung:
<jats:p>A set of 118 synthetic stimuli was presented twice to three groups of listeners to be labeled as American English (AE) vowels. The stimuli were generated through a Klatt cascade formant synthesizer, where first (F1) and second (F2) formant frequencies were varied factorially. F1 ranged from 200–700 Hz in 50-Hz steps. F2 ranged from 600–2600 Hz in 200-Hz steps. The labeling of 11 AE vowels was compared for AE listeners, Japanese listeners living in the United States (immersed), and Japanese listeners living in Japan (nonimmersed). In addition to labeling, all listeners judged the goodness of the stimuli on a 1–7 rating scale. The modal responses of the immersed Japanese listeners were significantly more similar to those of the AE listeners than were those of the nonimmersed Japanese. The immersed Japanese also had higher ratings than the nonimmersed Japanese listeners, showing greater confidence in their labeling of AE vowels. As an additional task, the nonimmersed Japanese also labeled the stimuli as the five Japanese vowels, and demonstrated that Japanese vowel identities largely differed from AE vowels on F1−F2. This difference may be a factor in Japanese listeners’ difficulty in perceiving and producing AE vowels.</jats:p>