A phonetic study of the vowels in Ningbo Chinese


Student thesis: Doctoral Thesis

View graph of relations


  • Fang HU


Awarding Institution
  • Yun Yang ZEE (Supervisor)
Award date3 Oct 2005


This dissertation investigates the production of the vowels in Ningbo Chinese from the acoustical, articualtory and aerodynamic perspectives. It interprets the linguistic vowel features, such as vowel height, vowel backness, lip rounding, nasality, and duration by examining the acquired acoustic, articulatory, and aerodynamic data. The diphthongs in Ningbo are investigated by examining the diphthong targets and the movement between them in terms of their acoustical and articulatory characteristics. Twenty speakers, ten female and ten male, provided the speech data for acoustical analyses, seven speakers, three female and four male, provided articulatory data, and six speakers, three female and three male, provided aerodynamic data of the nasal vowels and vowel nasalization. The acoustic results show that vowel height and backness can be well interpreted in a perceptually scaled acoustic F1/F2 vowel plane. Ten normal-length vowels in Ningbo form a pattern of four levels of vowel height and three degrees of vowel backness. Results from the articulatory study show that there is no clear correlation between linguistic vowel height and tongue height based on the individual speaker’s data. Nevertheless, a PARAFAC modeling of the lingual articulation successfully decomposes the complicated tongue shapes into two underlying lingual movement mechanisms, namely “retraction and back raising” (retraction of the tongue body and raising of the tongue dorsum towards the velum) and “front raising” (raising of the front part of the tongue towards the hard palate). Using the extracted two tongue movement mechanisms, the linguistic vowel backness and vowel height can thus be defined in a purely tongue-based and speaker-independent representation of the vowels. Both the acoustic and articulatory data suggest that there is a three-way distinction of lip rounding in Ningbo: spread for [i], horizontal protrusion for [y], and vertical protrusion for [Ү]. The acoustic data show that the two apical vowels in Ningbo are in a semi-high central position in the F1/F2 plane, and the articulatory data suggest that the Ningbo apical vowels are best understood as doubly articulated vowels, i.e., being apex and dorsal simultaneously. The acoustic data also show that Ningbo vowels, in particular the high, mid-high, and low vowels, tend to be lowered when acquiring nasality. Meanwhile, the articulatory data reveal that for most speakers lip protrusion is involved during the production of the nasal vowels, possibly enhancing the difference in vowel quality between the nasal vowels and their oral counterparts. Results of formant frequency analysis of the diphthongs in Ningbo show that in general, both the falling and rising diphthongs begin in an onset frequency area close to their target vowels, while only the normal-length rising diphthongs end in the frequency region close to the offset targets, but the falling diphthongs and short rising diphthongs do not reach the offset targets. The articulatory data are generally consistent with the acoustic data. With respect to the dynamic aspects of diphthongs, the acoustic data show that the Ningbo diphthongs can be characterized by the formant (F2) rate of change if only the falling diphthongs are concerned. But the F2 rate of change fails to distinguish the rising diphthongs from the falling diphthongs. Meanwhile, the articulatory data of tongue kinematics show that the average velocity and peak velocity can be used to characterize Ningbo diphthongs, given that the falling and rising diphthongs are considered separately. The time to peak velocity serves as a better criterion in distinguishing the falling diphthongs from the rising diphthongs. In short, peak velocities usually occur late in the falling diphthongs, but early in the rising diphthongs. This is because the peak velocity of the lingual articulator occurs during the gestural change, i.e. during the spectral transition from an acoustical point of view, from diphthong onset to offset. And the falling diphthongs in Ningbo have steady states on both the onset and offset elements, while the rising diphthongs only have steady states on the offset elements. Both the acoustic and articulatory data suggest that the main difference between a short and normal-length vowel or diphthong in Ningbo is in duration, which differs from the tense/lax distinction in English, where lax (short) vowels are mainly correlated with lower tongue positions vis-à-vis their tense (long) counterparts. The aerodynamic data reveal that nasality can be successfully quantified using the acquired flow data. Results show that in Ningbo the nasal vowel [ ] has higher nasality than the other nasal vowel [ã]. Regarding the three nasalized vowels, the high vowel [i] has the highest nasality, to be followed by [õ] and [ә] in descending order. In this study, in addition to searching for the physical and/or physiological correlates for vowel features, attention was also paid to examining articulatory mechanisms and articulatory-to-acoustic relations in vowel production. The articulatory data show that jaw movements are usually coordinated with lingual gestures during the vowel and diphthong production. However, different speakers may employ different articulatory strategies to achieve similar acoustic goals. The data also suggest that the three Ningbo point vowels exhibit a quantal nature during production. It should be noted that the relationship between speech articulation and speech acoustics has been shown to be complex by the comparisons of the articulatory and acoustic data of the diphthong targets in Ningbo. Further studies are indeed necessary in order to have a better understanding of the nature of vowel production and the relationship between articulation and acoustics.

    Research areas

  • Chinese language, Vowels, Wu dialects, China, Ningbo Shi