Comparing Identification of Artificial and Natural Vowels

In this study, we investigate how listeners classify non-native vowels, comparing classification of synthetically produced stimuli to that of natural stimuli. A forced choice identification task reveals that synthetic vowels are labeled differently from natural vowels and take more processing time. Participants are also less concordant in labeling synthetic tokens as compared to natural tokens. Because listeners’ performance on and classification of synthetic and natural vowels is different, speech perception studies with synthetic stimuli should be cautiously interpreted in terms of how humans perceive the sounds of their natural language.