1 An earlier version of this paper was presented at the 40"' International Conference on Sino-Tibetan Languages and Linguistics, held from September 27-29, 2007 in Harbin, China, under the title "Reconstructing Old Chinese uvulars in the Baxter- Sagart system (ver. 0.97)". We are grateful to Chen Jian ROI], G. Jacques, M. Miyake, T. Pellard and the anonymous reviewers for their helpful comments. Cahiers de Linguistique - -Asie Orientale 38(2): 221-244 (2009) © CRLAO-EHESS 54, Bd Raspail 75006 Paris 0 153-3320/2009/038-221

2 The following abbreviations are used in this article: MC: Middle Chinese; OC: Old Chinese; PTB: Proto-Tibeto-Burman; TB: Tibeto-Burman; WB: Written Burmese; WT: Written Tibetan. 3 Our notation for Middle Chinese is not a reconstruction but a convenient representation, using ASCII characters only, of the traditional categories for Middle Chinese. It follows the transcription in Baxter (1992:45-85), except: ( 1 ) initial glottal stop ?- is replaced by an apostrophe '- (U+0027); (2) "ee", "e", and "i" are replaced by "ae", lea", and "+" respectively; and (3) the finals formerly written "-£i" and "-wet" are now written "-ea" and "-wea" respectively. Middle Chinese forms are not preceded by an asterisk. 4 ion our notation, initial h- sometimes represents MC S xiá and sometimes ^ -± yii sign. If the h- is followed by medial -j- or the main vowel -i-, then the syllable is in division 3 (sanddng j£3?), and the h- represents 1^ yù sun; otherwise, the syllable is in divisions 1, 2, or 4, and the h- represents the initial ? xid. When referring to Middle Chinese initials in isolation, we represent 14i xia as "h-" and "1(3 :c yu san as "hj='- even though not all cases of Ot- "" yù san are actually written with -j-.

5 The Old Chinese distinction between type A and type B does not coincide exactly with the Middle Chinese distinction between divisions I, 2, and 4 on the one hand and division 3 on the other; see Baxter (1992: 267-269, 580-581 ).

6 Our notation for Old Chinese uses square brackets in cases where more than one reconstruction is possible: *[X] means "either OC 'X, or something else with the same Middle Chinese reflex as *X". In the case of ��,Ltk *2'a[n]-s* we put brackets around the 0-n because in our system, the coda of the first syllable could be either *-n or 0-r. Following Starostin (1989: 338-343), we assume that Old Chinese had a coda *-r, contrasting with both *-j and *-n, which usually became MC -n, but in some dialects merged with *-j instead, and developed like original *-j. A reconstruction 94,ft *?lar-sok would fit the transcription of this name even better than *?*an-S3k, of course, but at present we are not prepared to make this choice without additional evidence or argument: a language without final Irl might well transcribe foreign irl as lnl, as has generally been assumed in the past in this case (e.g. Pulleyblank 1962: 228-230).

7 ion our Old Chinese reconstructions, we put round parentheses "()" around elements when they could have been present, but there is insufficient evidence to determine whether they were actually there or not.

8 For example, � shdo < syew 'burn' could reflect *4k(r)cw, but the reconstruction •ijew is also possible. 9 Here and below, we put square brackets around Mandarin readings that do not follow the usual correspondences with Middle Chinese.

10 On the notion of minor syllable, cf. Henderson (1952). Minor syllables are also referred to as 'presyllables' in the literatures; words composed of a minor syllable followed by a major syllable are called 'sesquisyllables' by Matisoff.

z- OC 0-ak and *-ak-s sometimes become MC -jak and -joH, sometimes (as here) -jek and -jaeH, under conditions that are not yet understood. In order to make the predicted Middle Chinese reflexes clear from the Old Chinese notation, we write *-ak and *-ak-s as'-Ak and *-Ak-s when they become MC -jek and -jaeH respectively. This is a purely conventional notation to mark this irregularity; we are not reconstructing *A as a vowel distinct from'a 12 Contacts like this between between MC -jek or -jiek and -it (like those between MC yeng or -jieng and -in) can be accounted for by assuming that in addition to *-ek and *-q there were also rhymes 0-ik and '-in, which usually merged with *-ek and *-et), but in some dialects merged with *-It and '-in instead; see Baxter (1992: 298-299, 422-425, 434-437).

13 The cases where MC hj- occurs in kiiikðu 1m L=1 syllables (those without rounded vowels or medial -w-) are of two kinds: ( I ) cases where labialization in the initial was lost by dissimilation before the syllable codas *-w, *-m, or *-p, e.g. � ·[a]"(r)am > hjem > ydn 'brilliant'; and (2) the two final particles y *([?])a[n] >

hjen > ly4n] '(3p locative pronoun)' and � .(qp7> > hiX > yi '(final particle)', where what the Middle Chinese sources interpret as hj- (Ptq---) probably reflects a development (perhaps from *q- or •?-) peculiar to unstressed syllables.

14 It may also be possible to distinguish .oY- and ic3"'s- from *g*- and *g**- based on some modem dialects, but the correspondences are not yet clear to us.

15 We use a period to separate a minor syllable from a main syllable when the minor syllable plays no detectable morphological role and could be part of the root.

�6 Øj *1«D? is in the list because Shen Pei tt!;1 proposed that the character 'f1. on strip 24 of the Guodian 111m text "Yucong 2" sB3£_l should be understood as -9i (cited in Bai Yulan 2008:257). The original editors give no indication that they understood the character as anything but � (Jingmen Shi Bowuguan 1998: 204, 206).

17 As mentioned above (note 7), it is not clear whether :VA: shdo < syew 'bum' reflects *[a]ew or *ctcw. By parallelism we would expect to find labialized fi"-, '�'�-, *a"'-, and *4*t- also; but many forms previously reconstructed this way seem to have uvular connections (see examples below).

BAXTER William H. (1992). A handbook of Old Chinese phonology. Berlin : Mouton de Gruyter.

BENEDICT Paul K. (1972) Sino-Tibetan: a conspectus. Cambridge : University Printing House.

ECKERT Penelope and McCONNELL-GINET Sally (2003). Language and gender. Cambridge : Cambridge University Press.

HENDERSON, Eugenie J. A. (1952) The main features of Cambodian pronunciation. Bulletin of the School of Oriental and African studies, 14, pp. 149-174.

JAXONTOV Sergej Evgen'evič (1956). Review of B. Csongor (1952), Chinese in the Uighur script of the T'ang period (Acta Orientalia Hungarica vol. 2, no. 1, 73-121). Sovetskoe vostokovenedie 1956, no. 2, pp. 189-195.

KARLGREN Bernhard (1954). Compendium of phonetics in Ancient and Archaic Chinese. Bulletin of the Museum of Far Eastern Antiquities (Stockholm), vol. 26, pp. 211-367.

KARLGREN, Bernhard (1957). Grammata serica recensa. Bulletin of the Museum of Far Eastern Antiquities (Stockholm), vol. 29, pp. 1-332.

NORMAN Jerry (1994). Pharyngealization in early Chinese. Journal of the American Oriental Society, 114(3), pp. 397-408.

MATISOFF James A. (2003). Handbook of Proto-Tibeto-Burman. Berkeley, Los Angeles, London : University of California Press.

PULLEYBLANK Edwin G. (1962). The consonantal system of Old Chinese. Asia Major (n.s.), vol. 9, part 1, pp. 58-144; part 2, pp. 206-265.

PULLEYBLANK Edwin G. (1965). The transcription of Sanskrit k and kh in Chinese. Asia Major (n.s.), vol. 11, pp. 199-210.

PULLEYBLANK Edwin G. (1973). Some new hypotheses concerning word families in Chinese. Journal of Chinese Linguistics, vol. 1, pp. 111-125.

SAGART Laurent (1999). The roots of Old Chinese. Amsterdam : John Benjamins.

STAROSTIN Sergej Anatol'evič (1989). Rekonstrukcija drevnekitajskoj fonologičeskoj sistemy. Moscow: "Nauka".