The One-step Tonal-range Leveling of Mandarin among Waishengren in Taiwan

Hui-ju Hsu

Chung Yuan Christian University, Taiwan


Hsu (2006) investigated the leveling of Taiwan Mandarin between Waishengren and Holo in two generations – those born between 1951-1960 and between 1981-1990. The former is the second contact generation (henceforth “second generation”) and the latter the third (henceforth “third generation”). The results showed that the leveling started in the second generation. Among the four examined variables, tonal range was the only one that was not leveled until the third generation.

However, the Waishengren in Hsu’s study only included those from Mandarin areas, both standard and non-standard (Figure 1).The current study aims to investigate the speech of second generation non-Mandarin speaking Waishengren (hereafter “non-Mandarin Waishengren”) to explore whether a leveled Waishengren Mandarin had ever formed in Taiwan before the Mandarin Waishengren - Holo leveling (Figure 2).

36 second generation Waishengren (Mandarin and non-Mandarin), including those from Waishengren - Holo intermarriage families, were recruited. The informants were subcategorized into 5 subgroups by their parents’ ethnicities (Table 1); their tonal ranges were examined. Tonal system is one salient distinguishing feature between Mandarin and non-Mandarin Chinese dialects. The former usually have four tones while the latter six to eight. Furthermore, in Hsu (2006), tonal range was the only unleveled variable in the second generation.

Each informant read 83 sentences in Mandarin, each consisting of eight or nine syllables with at least one tone4 (T4) syllable at either the initial or middle position, with 101 T4 syllables in total. Tonal ranges of these syllables were measured. It was assumed that T4, the high-falling tone in Mandarin, covers the majority, if not complete, of one’s tonal range; the highest pitch point is realized at onset and the lowest at offset (Figure 3). Lin (1989) also showed that T4 covered the full pitch range of a Mandarin speaker.

A one-way ANOVA on tonal range revealed no significant differences among the five Waishengren subgroups, ( F (4, 3198)= 2.107, p >.05). The tonal heterogeneity of the first generation Waishengren’s native languages did not seem to pass on to their children’s tonal range realization.

Three factors are proposed to interpret this one-step Waishengren tonal range leveling: (1) Mandarin being manipulated as an identifier of Chinese, the identity the demographically minor but politically predominant Waishengren intended to impose on non-Waishengren Taiwanese people; (2) the long-term Mandarin-only policy, which established standard Mandarin as a target for children to model after; (3) the veteran villages, or Juan4cun1, the self-contained residential areas exclusively for the families of Waishengren veterans.


Hsu, Hui-ju. (2006). Some Aspects of Phonological Leveling in Taiwan Mandarin. Doctoral dissertation, National Taiwan Normal University.

Lin, Mao-can.(1989). Shengdiao. InWu, Zong-ji and Mao-can Lin (Eds.) Shiyan Yüyingxue Gaiyao. Beijing: Higher Education Press.

Session: Paper session
Levelling 1
Saturday, April 5, 2008, 09:00-10:30
room: 15