Acoustic correlates of Tone 3 and Tone 4 in Mandarin.

Abstract

Perceptual similarity between Tone 2 and Tone 3 in Mandarin was widely discussed in previous studies [Moore and Jongman (1997), Huang (2004), Bent (2005)]. Other tonal contrasts are hardly addressed. However, recent findings of Mandarin tones show that Tone 3 and Tone 4 are confusing in terms of descent slope. A big difference between previous studies and current ones is kinds of Tone 3 stimuli: previous studies used isolated Tone 3, namely, 214, and current studies used derived Tone 3, namely, 21, which is coarticulated with other tones. Spectrographic analysis was conducted. Descent slope and height of starting pitch were found critical between Tone 4 and derived Tone 3. Then three continua with ten instances for each were synthesized in respect of three different factors, descent slope, height of starting pitch, and height of ending pitch, which makes 30 stimuli total in three experiments, an AXB discrimination test, an identification test, and a lexical decision test. Results show that starting pitch height is a salient acoustic factor for 30 Mandarin natives, although descent slope was predicted more important. In addition, lexicon has an influence on ambiguous instances between endpoints of Tone 3 and Tone 4. Current findings may make contributions to understanding of perception, articulation, and lexicon during speech processing.