24. Prosodic Systems: China and Siberia Jie Zhang, San Duanmu, and Yiya Chen Abstract This chapter provides a summary of the prosodic systems of varieties of Chinese spoken in China and Taiwan as well as languages in Siberia, in particular Ket. What the Chinese languages and Ket share is their tonal nature. Three unique aspects of the prosody of these languages are highlighted. First, the typologically complex patterns of tonal alternation known as ‘tone sandhi’ are surveyed and a summary is given of current experimental findings on the productivity of these patterns. Second, the patterns of lexical and phrasal stress and their interaction with tone are discussed, with a focus on the similar metrical principles that underlie tone languages and other languages. Third, the interaction between lexical tone and utterance-level intonation, particularly pitch variations for focus and interrogativity, is reviewed. These issues are first discussed in the context of Chinese languages, and then echoed in a brief summary of Ket prosody. Keywords: tone, contour tone, tone sandhi, stress, intonation 24.1. Introduction This chapter provides a summary of the prosodic systems of languages in Northern Asia, including varieties of Chinese spoken in China and Taiwan as well as languages in Siberia, in particular Ket. A common theme in the prosody of these languages is their ability to use pitch to cue lexical meaning differences; i.e., they are tone languages. The well-known quadruplet ma55/ma35/ma213/ma51 ‘mother/hemp/horse/to scold’1 in Standard Chinese is an exemplification of the tonal nature of the languages in this area. We start with a brief discussion of the typology of syllable and tonal inventories in Chinese languages (§24.2). These typological properties lead to three unique aspects of prosody in these languages: the prevalence of complex tonal alternations, also known as ‘tone sandhi’ (§24.3), the interaction between tone and word and phrase-level stress (§24.4), and the interaction between tone and intonation (§24.5). The prosodic properties of Ket are discussed briefly in §24.6. The last section provides a summary (§24.7). 24.2. The syllable and tone inventories of Chinese languages The maximal syllable structure of Chinese languages is CGVV or CGVC (G=glide, VV=long vowel or diphthong) (Duanmu 2008: 72). The syllabic position of the prenuclear glide is controversial, and it has been analyzed as part of the onset (Duanmu 2007, 2008, 2017), part of the rime (Wang and Chang 2001), occupying a position of its own (Weijer and Zhang 2008), or variably belonging to the onset or the rime depending 1 Tones are transcribed in Chao numbers (Chao 1948, 1968), where ‘5’ and ‘1’ indicate the highest and lowest pitches in the speaker’s pitch range, respectively. Juxtaposed numbers represent contour tones; e.g., ‘51’ indicates a falling tone from the highest pitch to the lowest pitch. 1 on the language, the phonotactic constraints within a language, and the speaker (Bao 1990, Wan 2002, Yip 2003). Yip (2003) specifically used the ambiguous status of the prenuclear glide as an argument against the subsyllabic onset-rime constituency. The coda inventory is reduced to different degrees, from Northern dialects in which only nasals and occasionally [ɻ] are legal to southern dialects (e.g., Wu, Min, Yue, Hakka) where stops [p, t, k, ʔ] may also appear in addition to the nasals. Syllables closed by a stop are often referred to as checked syllables (ru sheng) in Chinese phonology, and they are considerably shorter than non-checked (open or sonorant-closed) syllables. There are typically three to six contrastive tones on non-checked syllables in a Chinese dialect. On checked syllables, the tonal inventory is reduced — one or two tones are common, and three tones are occasionally attested. Table 1 illustrates the tonal inventories on non-checked and checked syllables in Shanghai (Wu), Fuzhou (Min), and Cantonese (Yue). Table 1. Tonal inventories in three dialects of Chinese: Non-checked syllables Checked syllables Cantonese (Matthews and Yip 1994) 55, 33, 22, 35, 21, 23 5, 3, 2 Shanghai (Zhu 2006) 52, 34, 14 4, 24 Fuzhou (Liang and Feng 1996) 44, 53, 32, 212, 242 5, 23 24.3. Tone sandhi in Chinese languages A prominent aspect of the prosody of Chinese languages is that they often have a complex system of ‘tone sandhi,’ whereby tones alternate depending on the adjacent tones or the prosodic/morphosyntactic environment in which they appear (Chen 2000, Zhang 2014). Two examples of tone sandhi from Standard Chinese and Xiamen (Min) are given in (1). In Standard Chinese, the third tone 214 becomes 35 before another third tone;2 in Xiamen, tones undergo regular changes whenever they appear in nonfinal positions of a syntactically defined tone sandhi domain (Chen 1987, Lin 1994). (1) Tone sandhi examples: a. Tonally induced tone sandhi in Standard Chinese: 214 → 35 / ___ 213 b. Positionally induced tone sandhi on non-checked syllables in Xiamen: 53 " 44 " 22 ! 24 in nonfinal positions of tone sandhi domain % ' 21 2 This is a vast simplification. While in identification tasks, T2 is indistinguishable from the sandhi tone for T3 (e.g., Wang and Li 1967, Peng 2000), recent phonetic, psycholinguistic, and neurolinguistic evidence indicates the sandhi tone for T3 is neither acoustically identical to T2 (e.g., Peng 2000; Yuan and Chen 2014), nor is it processed the same way as T2 in on-line spoken word processing (e.g., Li and Chen 2015, Nixon et al. 2015). 2 Tone sandhi patterns can be generally classified as ‘left-dominant’ or ‘right- dominant.’ Right-dominant sandhi, found in most Southern Wu, Min, and Northern dialects, preserves the base tone on the final syllable in a sandhi domain and changes the tones on nonfinal syllables; left-dominant sandhi, typified by Northern Wu dialects, preserves the tone on the initial syllable (Yue-Hashimoto 1987, Chen 2000, Zhang 2007, 2014). It has been argued that there is an asymmetry in how the sandhi behaves based on directionality, in that right-dominant sandhi tends to involve local or paradigmatic tone change, while left-dominant sandhi tends to involve the extension of the initial tone rightward (Yue-Hashimoto 1987, Duanmu 1993, Zhang 2007). We have seen in (1) that both the tone sandhi patterns in Standard Chinese and Xiamen are right-dominant and involve local paradigmatic tone change. In the left-dominant Shanghai tone sandhi pattern in (2), however, the tone on the first syllable is spread across the disyllabic word, neutralizing the tone on the second syllable (Zhu 2006). (2) Shanghai tone sandhi for non-checked tones: 52-X → 55-31 34-X → 33-44 14-X → 11-14 Zhang (2007) argued that the typological asymmetry is due to two phonetic effects. One is that the prominent positions in the two types of dialects have different phonetic properties: the final position in right-dominant systems has longer duration and can maintain the contrastive tonal contour locally; the initial position in left-dominant systems has shorter duration and therefore needs to allocate the tonal contour over a longer stretch in the sandhi domain. The other is the directionality effect of tonal coarticulation, which tends to be perseverative and assimilatory; the phonologization of this type of coarticulatory effect could then potentially lead to a directional asymmetry in tone sandhi. Duanmu (1993, 1994, 1999, 2007), on the other hand, argued that the difference stems from the syllable structure, and hence, stress pattern difference between the two types of languages, as discussed in §24.4. Despite these typological tendencies, phonetically arbitrary tone sandhi patterns abound in Chinese dialects. For instance, the circular chain shift in the Xiamen pattern (1b) has no phonotactic, and hence, phonetic motivation, as the base tone itself is not phonotactically illegal in the sandhi position. Left-dominant sandhi, likewise, often has phonetic changes that cannot be predicted by a straightforward tone mapping mechanism. In Wuxi (Wu), for example, the tone on the initial syllable of a word needs to be first replaced with another tone before it spreads rightward (Chan and Ren 1989), and Yan and Zhang (2016) argued that the tone substitution involves a circular chain shift, as in (3). (3) Wuxi tone sandhi for non-checked tones with voiceless initials: 53-X → 43-34 Falling " Dipping 323-X → 33-44 % ' 34-X → 55-31 Rising 3 The phonetic arbitrariness and complexity of the synchronic tone sandhi patterns raises the question whether all these patterns are equally productive and learnable for speakers. This question has been investigated using the ‘wug’ tests in a long series of work since the 1970s. For instance, Hsieh (1970, 1975, 1976), Wang (1993), and Zhang et al. (2011) have shown that the circular chain shift in Taiwanese Southern Min (a very similar pattern to Xiamen in (1b)) is not fully productive. Zhang and Meng (2016) and Yan and Zhang (2016) provided a comparison between Shanghai and Wuxi tone sandhi and showed that the Shanghai pattern is generally productive; for Wuxi, the spreading aspect of the tone sandhi is likewise productive, but the substitution aspect of the sandhi is unproductive due to its circular chain shift nature. The relevance of phonetic naturalness to tone sandhi productivity has also been investigated in non-chain-shift patterns. Zhang and Lai (2010), for instance, tested the productivity difference between the phonetically less natural third-tone sandhi and the more natural half-third sandhi in Standard Chinese and showed that, although both apply consistently to novel words, the former involves incomplete application of the sandhi phonetically and is thus less productive. In general, the productivity studies of tone sandhi demonstrate that, to understand how native speakers internalize the complex sandhi patterns in their language, we need to look beyond the sandhi patterns manifested in the lexicon and consider ways that more directly tap into the speakers’ tacit generalizations.