Articulatory–Acoustical Relationship in Cantonese Vowels *

This paper investigates the articulatory–acoustical relationship in the production of the Cantonese long point vowels [iː uː aː] in CV syllables and shortened equivalents [i u a] in checked CVC syllables, through an analysis of the tongue positions and corresponding formant frequencies for the vowels. The articulatory and acoustical data from four Cantonese speakers, two male and two female, show that: (i) the shortened high vowels [i] and [u], relative to the long [iː] and [uː], undergo formant undershoot in the acoustical vowel space and [u] also undergoes target reduction in the articulatory vowel space, and (ii) the effect of shortened vowel duration is not observed in the low [a]. The data on the articulatory–acoustical relationship show that: (i) the levels of sensitivity of F₁ and F₂ to the variations in tongue displacement are low in each of the vowels [iː uː aː] and [i u a], suggesting that the Cantonese point vowels are quantal, and (ii) the sensitivity of the formant frequencies is lower to the variation in tongue constriction in the x-dimension than in the y-dimension in the long [iː uː], but not the long [aː] and shortened [i u a]. This suggests that: (a) the low vowels are less quantal than the high vowels and (b) the shortened vowels are less quantal than the long equivalents. The results are discussed in connection with the theory of vowel target reduction, the quantal theory, and the articulatory–acoustical relationship in the English point vowels.