Acoustic and articulatory analysis on Mandarin Chinese vowels in emotional speech

Acoustic and articulatory cues of Mandarin Chinese vowels were analyzed for ‘Angry’, ‘Sad’, ‘Happy’ and ‘Neutral’ speech by using EMA recordings from a male speaker. The results suggest that: (1) The F0 range and register of four emotions can be grouped into 2 emotional dyads: ‘Angry’ vs. ‘Happy’, and ‘Sad’ vs. ‘Neutral’; (2) Consistent differences in intonation were found across different emotions. ‘Happy’ has a high rising boundary tone, ‘Angry’ has a high falling boundary tone, and ‘sad’ and ‘Neutral’ have a low boundary tone. (3) The articulation of the vowels also exhibits differences for different emotions. The vowel triangle shifts downwards with the increase of arousal. The speaker seems to employ different articulatory strategies for different emotions. For instance, a higher lingual configuration is detected in ‘Angry’ and ‘Sad’ speech than in ‘Neutral’ and ‘Happy’; the ‘Sad’ speech demonstrates greatest changes in tongue shape; lip protrusion are somehow reinforced for both ‘Angry’ and ‘Sad’ vowels; a narrower vocal tract is detected for ‘Sad’ vowels than for ‘Neutral’ ones.