1. On the tonal asymmetry in the dialects of the group Wu- an input-based view Te-hsin LIU Department of Linguistics Paris 8 University liu.tehsin@gmail.com

2. Program of presentation 1. The phenomenon - tonal asymmetry in Wu dialects 2. Fundamental problem of current accounts 3. How does our model solve the problem? Disyllabic forms Trisyllabic forms 4. Discussion 5. Conclusion 6. Appendix: Issues in African languages-what is downstep? 7. References

3. The phenomenon Tonal asymmetry in Wu dialects: The dialects of the group Wu, such as Shanghai, Suchou, Wuxi, Tangxi, are remarkable as to their sandhi behavior. Unlike Mandarin or Tianjin where the stress falls invariably on the last syllable and the preceding syllables are subject to tonal changes, the sandhi in Wu interferes with the syntax: The tone of the pretonic syllable reduces to a level tone deductible from the register

4. The phenomenon This asymmetry conditioned by the syntax seems to be more complicated in Wuxi, another Wu dialect. Chan&Ren (1986) report that there are 6 lexical tones in open syllables, namely 52, 131, 313, 213, 34, 213, classified according to their historical categories. /13/ in Chao (1928) No correspondence between historical categories and citation tones Initial lower pitch

5. The phenomenon When the first syllable is stressed, there are four tone patterns each of which is assigned one or two lexical tones. The following unstressed syllables lose their underlying tone (Chan&Ren 1986).

6. The phenomenon Monosyllabic disyllabic trisyllabic quadrasyllabic Pattern A T3 (313), T6 (213) Pattern B T1 (52), T4 (131) Pattern C T5 (34) Pattern D T2 ([2] 13) Tones in the same row share the same underlying tonal melody

7. Previous analysis -Yip (1989) Yip (1989) : contour tones = melodic unit Edge-in association Phonetic interpolation rules

8. Previous analysis - Duanmu (1994) Duanmu (1994): against contour tone units Initial syllables are all LH in patterns A, B and D, the final L tone being due to domain-final effect.

9. Problems of current analysis 1. There is no consensus as to the interpretation of these tone patterns: 2. They fail to explain: Why are there four tone patterns? What determines the relationship between a tone pattern and a lexical tone? Why is there always a final L tone in a multisyllabic domain? Where does it come from?

10. Our proposal 1.We attack the source of the problem, claiming that there are only three lexical tones in open syllables in Wuxi, which have a direct relation with the four tone patterns. 2.Common point between lexical tones and tone patterns: A limit of size: tone patterns never exceed HLHL whatever the number of the syllable is. 3.We posit that both lexical tones and tone patterns are constrained by a phonological input, a tonal template HLHL. 4.Following the assumption of a universal tonal periodic skeleton HLHL postulated by Carvalho (2002), analogous to the syllabic skeleton CVCV proposed by Lowenstamm (1996), tones in open syllables in Wuxi can be represented as follows:

11. Our proposal Given the allophonic variation of the register conditioned by the laryngeal state of the onset, there are only three phonological tones. The vertical line indicates the tonal head encoding the lexical register Why 3 lexical tones but four tone patterns?

12. Our proposal 1.The register of each tone being conditioned by the laryngeal features of the onset, we propose to analyze T3 (31[3])/T6 (21[3]) as an underlying falling contour with a low head. 2.The final H tone in T3/T6 is due to an epenthesis adjacent to the low head, which is also found in T2. 3.The derived character of complex contour tones is also supported by Mandarin, where the falling-rising tone 213 can be analyzed as an underlying low tone (Yip 2002)

13. Our proposal As opposed to the falling contour with a low head, T1 (52)/T4 ([1]31) will be analyzed as un underlying falling contour with a high head, the initial epenthetic L tone in T4 being triggered by the lowering effect of the voiced onset. T5 (34) and T2 ([2]13) will be analyzed as an underlying rising tone. T5/T2 have no lexical head, but contextual heads induced by the allophonic register.

14. Our analysis - disyllabic forms 1.Identity preservation: the head (i.e. the register) should be preserved in tone patterns. 2.We propose two general principles to account for the derivation of tone patterns: 1.σ2 associates to the last plain position of the tonal template of σ1 and to the following first empty position; 2. Being unstressed, σ2 cannot have its own tonal head.

15. Our analysis - disyllabic forms

16. Local conclusions 1. Why are there 3 lexical tones but 4 tone patterns ? => The rising contour doesn’t have an underlying head. Contextual heads, conditioned by the laryngeal state of onset, will serve as head in tone patterns. Hence pattern C and pattern D. 2.In the case of two falling tones with lexical head, i.e. pattern A and pattern B, the allophony of register induced by the onset is too fine to be perceived as a distinct pattern. Just like the lengthening of a short vowel before a voiced consonant in English: it will always be perceived as a short one, since there are lexical long vowels in English.

17. Our analysis - trisyllabic forms right-headed tones 1.Association from left to right; 2.Constrained by the size of the template, the median syllable will be “jammed” between σ1 and adjunction of the last syllable; 3.Consequently, σ2 will associate to the head so that the identity of the head can be preserved.

18. Our analysis - trisyllabic forms left-headed tones The adjunction of the last syllable will dissociate σ1 from the tail so that the median syllable can accede to the head of the pattern. loss of the tail

19. Local conclusions Relationship between tone pattern, stress, and tonal template 1. Just as lexical tones, tone patterns are constrained by a tonal template HLHL. 2.The tone patterns characterize un group where σ1 is stressed and the rest loses their tone and are subject to the spreading of the stressed lexical tone. 3. Isochrony between trochaic foot and tonal template. Wuxi: Emergence of a trochaic accentual pattern The tone patterns in Wuxi cannot be explained but by the inputs they derive from, viz. by lexical tones

20. Discussion What is a “tone pattern”? 1. Pasiego montañés (Tivoli 1998, 2006) 2. Koya (Tivoli 1998, 2006) Vowel harmony dependant on stress Tone pattern: tonal harmony dependant on stress  The tone of σ1, stressed, spreads to the following unstressed syllables.

21. Conclusions In Wuxi, lexical tones are not assigned to tone patterns arbitrarily, there being a direct relation between stress, lexical tones and tone patterns. Historical categories play no role in the modern dialect, and there is reorganization of lexical tones based on the phonological representations. The hypothesis of preservation of head in tone patterns predicts correctly the phonetic realization of multisyllables, and thus explains why there are 3 phonological tones but 4 tone patterns; The “domain final effect” of Duanmu follows directly from the hypothesis of a tonal skeleton whose order is HLHL, rather than *LHLH.

22. Appendix: Issues in African languages Hyman (1975, 2004) distinguishes “vertical assimilation” from “horizontal assimilation”: Vertical assimilation: lowering or raising of a tone by an adjacent tone Ex: HL + H => H.!H Horizontal assimilation: conservation or anticipation of a tone on a adjacent tone Ex: H + L => H.HL Hyman notices that, in vertical assimilation, progressive downstep and regressive upstep are natural phenomena, while regressive downstep and progressive upstep are not natural.

23. Appendix: Issues in African languages Why progressive upstep and regressive downstep are not frequent? A position available for H to settle down The chair is already taken by H. No progressive upstep, No regressive downstep Progressive downstep Regressive upstep

24. Thank you for your attention!

25. References Ballard, William. 1988. The History and Development of Tonal Systems and Tone alternation in South China. Tokyo. Institute for the studies of languages and culture of Asia and Africa: University of Foreign Studies. Bao, Zhiming 2000. The Structure of Tone, Cambridge, Cambridge University Press. Beckman, Jill 1998. Positional Faithfulness. Doctoral dissertation, University of Massachusetts, Amherst. Carvalho, J. Brandão de. 2002. De la syllabation en termes de contour CV, Habilitation thesis, Paris, EHESS. Chan, Marjorie&Ren, Hongmao 1986. Wuxi tone sandhi : from last to first dominance. UCLA Working Papers in Phonetics 63 :48-70 Duanmu, San. 1994. Against Contour Tone Units, Linguistic Inguiry, 25:4, 555- 608 Hayes, Bruce. 1985. Iambic and trochaic rhythm in stress rules. Niepokuj, N., M. VanClay, V. Nikiforidou, & D. Jeder (eds.), Proceedings of BLS 11: parasession on poetics, metrics, and prosody. BLS, Berkeley, pp.429-446 Hayes, Bruce. 1995. Metrical stress theory: principles and case studies. The University of Chicago Press, Chicago & London. Hyman, Larry M. 1975. Phonology : Theory and Analysis. New York, Holt, Rinehart&Winston Hyman, Larry M. 2004. Universals of Tone Rules : 30 Years Later. Paper presented at the international conference on tone and intonation, Santorini, September.

26. References Liu, Te-hsin. 2007.“A Fresh Look at the Paradoxical Nature of Chinese Contour Tones”, To appear in Online ConSOLE-proceedings. Liu, Te-hsin. 2007.“Reduplication as Evidence for the Skeletal Nature of Tones ”, ms. Lowenstamm, J. 1996. CV as the Only Syllable Type. In Jacques Durand&Bernard Laks (eds.), Current Trends in Phonology: Models and Methods, CNRS, Paris X : ESRI, vol. 2, 419-441 McCarthy, John & Prince, Alan. 1986. Prosodic morphology. Ms., University of Massachusetts and Brandeis University. Pierrehumbert, J. B. 1980. The Phonology and Phonetics of English Intonation. PhD dissertation, MIT. Smith, Jennifer 2002. Phonological Augmentation in Prominent Positions. Doctoral dissertation, University of Massachusetts, Amherst. Steriade, Donca 1994. Positional Neutralization and the expression of Contrast. Ms. UCLA. Tivoli, Majors 1998. Stress dependent harmony: phonetic origins and phonological analysis. Ph.D. dissertation, University of Texas at Austin Tivoli, Majors 2006. The development of stress-dependant harmony. in Southwest Journal of Linguistics. Yip, Moira (1989). Contour Tones, Phonology 6, 149-174 Yip, Moira (2002). Tone, Cambridge, Cambridge University Press.