Nasal duration variability in Singapore Mandarin Chinese

Daming Xu & David Deterding

Fri. 2:00-4:05 A

Variationist studies of nasals have usually been binomial analyses of the phonetic realization or non-realization of nasal codas of a syllable (e.g. Poplack 1979, Wolfram 1989, Xu 1992). In a departure from that tradition, the present study focuses on the variable duration of nasal codas when they are realized. This analysis is not only prompted by the necessity of accounting for nasal duration variation phenomena in Singapore Mandarin Chinese, where the complete non-realization of a nasal coda is rare; it also addresses the issue of "free variation" revived by recent studies (Dorian 1994).

The contrast between nasal duration variation (NDV) and retroflexion variation in the same speech corpus illustrates the contrast between a marker and an indicator (Labov 1972, Trudgill 1974, Fasold 1991). Although NDV appears to be below the level of social consciousness and not subject to stylistic manipulation, it does not seem to be totally free but appears to be constrained by certain factors.

The database for the present investigations consists of a group of recordings made of 100 educated young Singaporeans performing two linguistic tasks in a test situation. The education level of the speakers range from high school to university graduates. They are mostly between 20 and 30 years old and come from many different social and geographic sectors of the country. The test was for a job-related screening purpose. The candidates were recorded under the same conditions in a language laboratory. First they read aloud a short passage of text, and then they talked freely about the passage. We can assume that the candidates were striving to approximate their speech towards the standard variety of Mandarin, and according to Labov's (1969) contextual style spectrum of attention, their speech in reading and talking would constitute two different styles of formality.

As a pilot study of NDV, the CSL speech analysis software was used to make acoustic measurements of the nasal durations of twenty syllables of each of twenty speakers, ten male and ten female, selected randomly from the database. In order to minimize the effect of general speaking rate and to normalize the individual measurements, the duration of the rhymes containing the nasals was also measured, and the ratio of the nasal duration and rhyme duration was taken to be the basic unit of variation.

The preliminary results of statistical analyses of the nasal/rhyme duration ratio variation (NRDRV) suggest that (1) this variation significantly differentiates the male from the female speakers, with means of the female speech closer to the measurements of NRDRV of the standard (Beijing-like) Mandarin; (2) no significant style-shifting was found for NRDRV while the retroflexion (a marker) style-shifting for the same speech was significantly apparent; (3) the scores awarded to the linguistic performances of the candidates by local language teachers were not correlated with NRDRV but significantly correlated with the retroflexion variation.

It is also found that the difference in NRDRV between the alveolar and velar nasals is not significant. However, other phonological conditions such as tones, preceding vowels, and following segments and pauses are being tested for effects of NRDRV.