Effects of two acoustic continua on the within-category perceptual structure of tones

Abstract

The present study investigated effects of two acoustic continua on the within-category perceptual structure of Putonghua Tone 2 and Tone 3. These two tones were simulated with tokens varying along two acoustic continua about F0 contour: the timing of F0 turning point and falling of F0. Three different syllable durations were tested with voice quality under control. Multidimensional scaling analyses were applied to investigate relative influence of phonetic identification and category goodness on the perceptual dissimilarity of synthesized tonal tokens. The result revealed that Tone 3 has later F0 turning point and greater F0 falling than Tone 2, which confirms former findings. The new finding is that perceptual representation of these two tones categories is different in their internal structures. Best tokens disperse within categories and are usually not unique. Perceptual space involving Tone 2 tokens shrink but that involving Tone 3 doesn't. Goodness rating contributes significantly to the dissimilarity scaling across Tone 2 tokens but not Tone 3 tokens.