Turkic kümüš ‘silver’ and the lambdaism vs sigmatism debate

Turkic kümüš ‘silver’ and the lambdacism vs sigmatism debate∗ Anton ANTONOV† & Guillaume JACQUES‡ December 24, 2011 Abstract: e goal of this article is to contribute to the debate on lambdacism vs sigmatism by re-examining the etymology of the Turkic word for ‘silver’. We propose that the PT etymon re ected in CT kümüš and Chuvash kӗmӗl is a Wanderwort also found in various ST and AA languages. Although the source and direction of borrowing remain uncertain, all languages except CT have either a nal lateral or a segment whi originates from a lateral in the proto-language(s). erefore, the data presented in this article support the idea that the correspondence -š: -l between CT and Chuvash should be reconstructed in PT as a lateral *ɬ rather than as a palato-alveolar fricative *š. 1 Introduction e goal of this article is to revisit the etymology of the word for ‘silver’ in Turkic. As there is no obvious internal etymology for this word, resear ers have tended to look for an external one and seem to have found it in Chinese. We intend to show that this etymology raises a number of problems whi We wish to thank Andrs Róna-Tas, Laurent Sagart, Alexander Vovin for useful comments on this article. We are responsible for any remaining error. † Institut National des Langues et Civilisations Orientales (INaLCO)/Centre de re er es linguistiques sur l’Asie Orientale (CRLAO), Paris, France ‡ Centre National de la Re er e Scienti que (CNRS)/Centre de re er es linguistiques sur l’Asie Orientale (CRLAO), Paris, France ∗ 1 we will discuss in some detail before suggesting a new way of ta ling the data in a broader perspective. e purpose of our paper is threefold. First, it rejects the current etymology deriving the word from Chinese whi we nd untenable. is etymology while possibly not universally accepted among Turkologists seems to be the only one cited in reference works on Turkic etymology. Second, it proposes a new source for the Turkic word by taking into account data from several other language families, including Sino-Tibetan and Austro-Asiatic whi turn out to share the word for ‘silver’ despite super cially similar names whi could imply accidental lookalikes. In doing so, several scenarios are presented as we do not think it possible to be categorical in this ma er. ird, our hypothesis contributes in a non-trivial way to the debate on lambdacism vs sigmatism. 2 Lambdacism, rhotacism and the Altaic debate ere are mainly two sets of correspondences between Turkic languages as far as medial and nal -l and -r are concerned. For the purpose of the present study, only the rst one of these will be presented in some detail as it has a direct bearing on the etymology of the Turkic word for ‘silver’. In the case of -l, there are words where both Chuvash and Common Turkic (CT, all Turkic languages except Chuvash) have an -l and then there are those where Chuvash mostly has -l (sometimes -ś)ffi while CT has (virtually) only -š. e -l :: -š correspondence was rst noticed by S [1841:14], [1898:92]. then B [1864:243-244] and Aˇ e term lambdacism refers to the hypothesis that Proto-Turkic */š/ evolved into Chuvash /l/ R [1882:§ 288], G [1913], and the term sigmatism to the opposite hypothesis: namely, that Chuvash (as well as Mongolian and Tungusic in certain (loanword) cases) has in a way preserved the PT state of a airs, whereas CT has innovated by anging some of its laterals into palato-alveolar fricatives (Ramstedt). Now since we know that ffiAnd sometimes /-š-/ in what are most probably CT loanwords. 2 sometimes Chuvash l corresponds to CT l, the la er hypothesis claims the existence of two types of l in PT, usually termed l₁ and l₂, the second one being the one on whose re exes Chuvash and CT disagree. Hypotheses on the exact nature of the distinction between these two types of l in PT vary, but it is usually assumed that l₂ (wri en ĺ, ľ or ɫ) was a palatal lateral *[ʎ], a lateral fricative *[ɬ] or even a lateral a ricate *[tl~tɬ] or *[tʃl] (cf. Poppe 1925a:33, 1925b:27).² 2.1 Correspondences between CT, Chuvash, Mongolic and Tungusic Here are rst some examples of the correspondence CT -l/-š :: Chuvash -l.³ CT road yol heart köŋül tongue til winter qïš stone tāš outside taš opposite side tuš silver kümüš Chuvash śul kămăl čӗlxe xel čul tul tӗl kӗmӗl PT (lambdacism) *yōl *köŋl *til *qïš *tāš *taš *tuš *kmš PT (sigmatism) *yōl₁ *köŋl₁ *til₁ *qïl₂ *t(i)āl₂ *tal₂ *tl₂ *kml₂ Table 1: CT -l/-š :: Chuvash -l ese examples show that Chuvash -l corresponds sometimes to CT l, and sometimes to CT -š. As mentioned in the previous section, this has ²e same is true, mutatis mutandis, of r (cf. J [1998b:104-5]): sometimes Chuvash r corresponds to CT r, and sometimes to CT z. e hypothesis of rhotacism then says that Chuvash has anged PT z into r while CT has preserved it, whereas the hypothesis of zetacism claims that it is CT whi has anged an earlier r into z. Under this last hypothesis, we need to posit the existence of two types of r, usually termed r₁ and r₂, the second one being the one on whose re exes Chuvash and CT disagree. Hypotheses on the exact nature of the distinction between these two types of r in PT vary, but it is o en assumed that r₂ was a palatalized r [rʲ] (o en wri en ŕ). ³e following presentation is based on P [1925a,b], J [1998a:104-5], R´ T [1998:71-72] and M [2002]. Reconstructions of P(roto-)T(urkic) with lambdacism, resp. sigmatism, are given for the sake of illustrating the two oices we face when reconstructing PT. 3 prompted two di erent types of reconstruction, here called PT (lambdacism) and PT (sigmatism), respectively. Now, in certain (loan)words, Mongolian and Hungarian show a similar correspondence of -l to CT -š and -l- :: -š-. CT fall tüš noon tüš whelp köšek Mongolian döl ‘slope’ (edyr) düli ‘mid[day]’ gölige Hungarian dől ‘to slant; to fall’ dél ‘south; noon’ kölyök Table 2: CT -š/-š- :: Mongolian -l/-l- :: Hungarian -l/-lOne particularly controversial but o en cited correspondence is the word for ‘stone’, in whi Tungusic seems to pa ern in a way similar to Chuvash and Mongolian: CT Chuvash Mongolian Evenki stone tāš čul čila(xun) ǰolo Table 3: CT -š :: Chuvash -l :: Mongolian -l- :: Evenki -lFurther examples with Tungusic include the following. CT Chuvash Manu outside taš tul tule dream tül/š tӗlӗk tolgin Table 4: CT -l/-š :: Chuvash -l :: Man u -lFurthermore, sometimes Chuvash has -ś (instead of -l) where CT has -š and some old Turkic loanwords in Hungarian have -lcs(-) [ltʃ] (cf. J [1998b:105]). Mongolian shows a similar picture in some words whi may re ect old borrowings with -lǰ - :: CT -š-. Now, this last word could be analyzed as a (non-a ested) Turkic compound involving the word for ‘head’ (R´ T [1998:72]), whi is attested in the Volga Bulgar inscriptions of the 13th–14th centuries as *balǰ~baǰ. cf. K [2010], the most recent contribution on this topic, and the references therein. 4 CT Chuvash Hungarian cradle bēšek bölcső fruit yemiš śimӗś gyümölcs Table 5: CT -š/-š- :: Chuvash -ś :: Hungarian -lcs(-) CT Chuvash donkey ešgek ašak (<Tat. ?) sparrow hawk *taz bašı (‘bald-headed’) Mongolian elǰigen tarbalǰi(n) Table 6: CT -š/-š- :: Chuvash (-ś-) :: Mongolian -lǰ Given the absence of the sound and corresponding aracter for // in the Arabic alphabet whi these inscriptions use, this could point to the existence of a form ba(l)č in the language of the Volga Bulgars (Erdal 1993:107109, 121-122). Since it is known that Chuvash anged its * into ś sometimes a er that period, the Mongolian and Hungarian data have prompted Altaicists to reconstruct su words with a */l/ cluster as, for instance, in the above-mentioned word for ‘head’, whereas anti-Altaicists would see in su cases either a later borrowing from another Turkic language (usually Tatar), or an internal derivation by means of some kind of su x (possessive in the case of the word for ‘head’) (F [1996:452-3]). CT Chuvash PT (lambdacism) PT (sigmatism) head baš puś (not *pul) *baš *bal₂ sword qïlïč xӗś (not *xӗl) *qïlï *qïlï Table 7: CT -š/-š- :: Chuvash -ś Interestingly, the word for ‘sword’ seems to preserve this cluster thanks to the insertion of an anaptyctic vowel. Altaicists would thus claim that the various Mongolian, Tungusic and Hungarian correspondences, when they seem to agree with the Chuvash ones, are proof of a genetic relationship between the Turkic, Mongolic and Tungusic languages, whereas Anti-Altaicists would say that all these words are either loanwords from a Bulgar(Chuvash)-type Turkic language or else are loanwords in Chuvash or have an independent internal explanation in 5 that language. e l vs. š controversy is further complicated by the existence in CT of l-forms alongside š-forms in words su as tül ‘dream’ (a ested in Old Uighur and preserved in Yakut, cf. R´ T [2007:1]), whi has a more common variant tüš, and the corresponding verb tüšä- ‘to dream’, a ested in OT in the collocation tül tüšä- ‘to have a dream’. Some resear ers (cf. R´ T [1998:72], R´ T [2007:8], following Ligeti) take this to mean that the ange from l₂ (=/š/) to l had started already in the ancestor of CT and Chuvash, but shortly a er the ancestors of the Chuvash moved out, and so it came to a halt in CT but went to completion in Chuvash. Judging from the data, and following Poppe (1924:43-44, 1925a:32-34, 4142), it seems plausible to us to reconstruct two types of laterals, an ordinary one */l/ (with front and ba variants, as denoted by the runiform script, L1 and L² ) and a fricative one */ɬ/, both of whi could combine with *// in two types of clusters */l/ vs. */ɬ/. Ordinary /l/ was preserved in all varieties of Turkic, whereas the fricative lateral merged with */l/ in Chuvash, but with */ɬ/ in Common Turkic where it further developed into */š/. e */l/ cluster was preserved in Common Turkic by means of an anaptyctic vowel but merged with */ɬ/ in Chuvash to give */ś/, a er possibly losing its lateral component. 2.2 Borrowing vs. Inheritance e debate on lambdacism vs sigmatism together with the one on rhotacism vs zetacism is one of the most vexed issues in the eld of Turkology as it seems to be almost invariably associated with the Altaic debate, i.e. the claim that the Turkic, Mongolic, Tungusic, and according to S [2003] and R [2005] (to cite but the most recent literature on the subject), Korean and Japanese languages, are part of a larger language family called Altaic. Altaicists generally are also sigmatists (and zetacists), that is they claim that the above-mentioned correspondences show that Chuvash (and the Mongolian and Hungarian ‘cognates’ and loanwords, respectively) has merged the PT two types of l (and r), whereas the rest of Turkic has innovated by anging one type of l (and r) into š (and z, respectively). 6 We think that the two questions should be kept separate since even if it were proved that it is sigmatism and zetacism that really took place and thus Chuvash is really conservative in a way, ‘cognates’ in Mongolic languages could very well be old loanwords from Proto-Turkic and thus positing a genetic relationship does not seem to us to be the only logical consequence of this (hypothetical) fact. We now turn to the Turkic word for ‘silver’ whi we believe can contribute to this debate. 3 e history of the word ‘silver’ in Turkic 3.1 Ancient attestations e word kümüš is a ested since the 8th century. It occurs nine times in the following runiform inscriptions: Kl Tegin (3 examples) (1st side, line 11, 3rd side lines 5 and 14), Bilge Kagan (3 examples: 1st side, line 12, 2nd side lines 3 and 11), Begre (1 example), Tonyukuk (1 example: line 48), Golden vessel (1 example). e coda consonant is wri en with the runiform le er for Š in seven out of nine instances, the remaining two using the runiform le er for S (Bilge Kagan and Tonyukuk). It is noteworthy that there are six instances (1 in Kizil-çira II, E-44; 4 in Köjeelik-Hovu E-45 and one in El-Bazhy E-68) of what appears to be an ethnonym of the form Kümül (o en preceded by yüz ‘a hundred’) whi might be an earlier variant form of kümüš (cf. also R´ T [2007:9]). 3.2 Attestations in modern Turkic languages e word is a ested in all modern Turkic languages, and it is possible to reconstruct a proto-form of the type *kmš, or alternatively *kml₂. oted a er the electronic version of the texts available on Trik bitig, the website of the Language Commi ee of the Ministry of Culture and Information of the Republic of Kazakhstan at h p://irq.kaznpu.kz, last accessed in June 2011. O en in the collocation altun kümüš. is le er seems to be derived from the one used for the palatal variant of /l/, cf. S [1998] For a detailed list of Turkic forms see R [1994:211]. 7 Language Turkish, Azeri, Gagauz, Kumyk Turkmen Kara ay-balkar, Kirghiz Kazakh Khakas Tatar Bashkir Yakut Chuvash Form gümüš kümüš kümüš kümis kümüs kömeš kömöš kömüs kӗmӗl Table 8: Modern Turkic re exes Initial voicing in Oghuz languages is regular, though not universal (cf. Turkmen), as is lowering of ü in Tatar and Bashkir, but not in Yakut (!), where earlier ü and ö are usually preserved. Nevertheless, the Chuvash form has ӗ and so points to an earlier ü, as ö would have given ă as in the word for ‘heart; mind’ köŋül > kămăl (see Table 1). 3.3 Borrowings into Non-Turkic languages e Turkic word for ‘silver’ has been borrowed in some of the Yenisseian languages (Kot, Arin, Assan, Pumpokol), where it generally means ‘silver’, except in Pumpokol where it is used to refer to ‘gold’ (kümüč), the word used for ‘silver’ in this language being probably cognate with the Yenisseian word for ‘leaf’ (cf. W [2005:241; 267]). It has also been borrowed in some Caucasian languages, usually through Azerbaijani. On the other hand, Mongolic languages have not borrowed it as they have another (unrelated) word of similarly debatable origin (cf. R [1994]), mönggün, whi has itself been borrowed by Tungusic languages and even some Turkic languages whi have been heavily in uenced by Mongolic and/or Tungusic and whi now have two words for ‘silver’ (cf. Tuvan). 8 is is an important fact since Mongolic and/or Tungusic forms are o en cited in an e ort to prove a cognate relationship between them and their Turkic counterparts, and further press on the point of the primacy of l-forms, for instance, but this is simply not possible in the case of the word for ‘silver’. 4 Proposed etymologies 4.1 Internal etymologies R [1994] is the rst a empt at an internal etymology as far as we can tell. In view of the di culty to explain the Turkic form by the o en-cited Chinese donor form, R [1994:212] notes that, methodologically speaking, it could be a be er idea to look for an internal etymology before trying to explain the Turkic word by a foreign one. He then goes on to suggest a possible internal derivation of the word from a(n apparently) nominal root *km- and the noun-forming denominal su x -üš, for whi he gives two examples: bağïš ‘rope’ < bağ ‘tie; joint’ and bügüš ‘wisdom’ < bügü ‘(a) wise (person)’. Concerning the nominal root he reconstructs, he suggests it might be a variant form of kün ‘sun’ given that, according to him, terms denoting ‘silver’ o en have astral associations to them. However, since there seems to be no su variant of the word for ‘sun’ in any of the Turkic languages he is forced to reconsider the Chinese source as a possibility, although in a slightly modi ed form. First, even if it is problematic, as he himself points out, Rybatzki’s try at an internal etymology for kümüš is a welcome ange and is the rst a empt at nding an internal explanation for this word as far as we can see. We do agree with him that looking for an internal explanation must always precede the sear for an external one. e main problem with this etymology according to Rybatzki himself is the absence in Turkic of a form *km whi would be the missing link between the word for ‘sun’ kün and the word for ‘silver’ kümüš, under the assumption that su a link exists. is, however, is not entirely true since we do nd in the Kızıl dialect of 9 Khakas the form kum in the expression кум харағы (kum xarağı) instead of standard кӱн харағы (kün xarağı) ‘sunny spot’ (S [2006:211; 214]). Nevertheless, this form seems to be a hapax legomenon, since descriptions of this dialect do not mention a ange  > u and that the word for ‘sun’ in this dialect has a dental and not a bilabial nasal according to all extant historical a estations, whi nevertheless seem to con rm the non-front aracter of the vowel . A possible, though maybe not too probable, explanation for the form kum might be the following: In an 18th century dictionary giving Khakas dialect equivalents to Russian words the Kızıl dialect word for ‘sun’ (солнцe) appears as кун карагы (kun karagı, lit. ‘sun/day eye’) (B  [1973:125]). Now, given that the /-n/ of kün could easily assimilate to /-ŋ/ in front of the initial velar sound /k-/ of karagı, we could assume a further development along the lines of an Old Uyghur ange of /-ŋ/ into /-m/, exempli ed in the expression yürüŋ karak > yürüm karak ‘white eye’ (cf. E [2004]). All of this is of course highly hypothetical and does not come close to rescuing the internal etymology of Rybatzki. A second problem with this etymology is the link between the word for ‘silver’ and the word for ‘sun’. e word for ‘silver’ in those languages where we do know its etymology is usually derived from a root meaning ‘white ; shiny ; luminous’. is is the case of one of the two names of this metalffi⁰ in the Indo-European languages: PIE *h₂erĝ-n̥ t-om~*h₂reĝ-n̥ t-om (e.g. OIr argat, Lat argentum, Arm arcat‘, Av ǝrǝzatǝm, Skt rajatám, To B ñkante [with *r. . . n assimilated to *n . . . n]), whi is possibly derived from an adjective *h₂erĝ-n̥ t (genitive *h₂erĝ-n̥ t-ós), meaning ‘white’ (cf. M [2006:242]). Now, even if it is true that all of those could easily be used to speak of the sun as well, the name of the sun is usually derived in these same languages from a root with the meaning ‘to burn’. Indeed, since the Turkic word kün can also mean ‘day’ (and it is in this meaning that it is used in Modern Turkish, for instance, although not in the majority of Turkic languages where it can mean both ‘sun’ and ‘day’) we could compare it to the English word ‘day’ whi comes from ProtoAnd all of them do distinguish ba /u/ from front //. ffi⁰e other one, present in Germanic, Baltic and Slavic is best viewed as a non-Indo-European Wanderwort. 10 Germanic *dagaz <*dʰógʷʰ-os, a noun of the type τόμος derived from the Indo-European root ‘to burn’ *dʰegʷʰ- (R [2001:133-4]), whi seems to be the most common IE verb for ‘burn’). erefore, we nd it quite reasonable to connect the Turkic word for ‘sun’ with the verb köń-/köj-/küj- ‘to burn’ (cf. Sevortjan (1997)), but would be more reluctant to do so in the case of the word for ‘silver’. 4.2 External etymologies e a empts at nding an external etymology for the Turkic word ‘silver’ have always, as far as we know, invoked a Chinese source. Despite certain minor variants, all of the external etymologies see the Chinese word 金 jīn (today usually ‘gold’, but it can also be, and certainly was, used as a generic term for ‘metal’). R [1949:116] suggests that kümüš is related to Sino-Korean kim ‘gold; metal’. Taking into account the Chuvash form of the word, kěmĕl, he reconstructs a proto-form *kml whi he in turn derives from *km+li, implying that a er vowel harmony has taken care of rounding the su x vowel a metathesis occurred. J [1952:210], R¨ ¨ [1969:308b], C  [1979:25-6] and D [2007:67] all derive the Turkic word from a nona ested Chinese compound *金鐐 jīnliào[lio] (MC kimlew OC *kəm *k(r)[ə]m*rˤew *[r]ˤewffiffi). Joki and Rsnen translate this compound literally ‘gold; metal’ and ‘argent’. C [1979] translate it as ‘pure silver’. Menges (1983: 120) says it means ‘(pure, precious) metal’. R [1994:212] notes that if su a compound had ever existed in Chinese its meaning would have been ‘wealth’, the true meaning, according to him, of the o -encountered Turkic compound altun kümüš whi is usually translated as ‘gold and silver’. He thinks that this compound could well be a calque of Chinese 金銀 jīnyín. Since he is unable to nd a satisfactory internal etymology, Rybatzki accepts the Chinese origin of the word, but rather than looking for it in a non-a ested compound, he proposes that it is derived from the Chinese word ffiffie MC and OC forms are cited a er B [1992] and B 11 [2011]. 金 jīn meaning ‘gold’ but also, originally, any kind of ‘metal’, to whi a rare noun-forming denominal Turkic su x -üš was added (cf. previous section). is means that the word originally meant ‘(precious) metal’, as this is the meaning in whi Turkic borrowed it from Chinese. Rybatzki cites as proof of this original meaning of the word data from Yakut where the meaning of this word varies according to the preceding adjective: with ‘white’ it does mean ‘silver’, but in the case of ‘red’ it means ‘gold’. e Yakut data is also used by Levitskaja’s article on kümüš in L [1997] to hint at the same possibility with no reference to R [1994]. Apart from the fact that the Chinese compound *金鐐 jīnliào is not attested, this etymology is also problematic from a semantic and a phonetic point of view. First, syntactically the only possibility for a N1N2 compound in Chinese is that N1 modi es N2, whi in this case would be completely meaningless. e only other possibility is that it was a kind of binomial (or paired) as is the case of Chinese 金銀 jīnyín. Even more importantly, phonetically the hypothetic Chinese compound should show up in Turkic as *kimle or *kimli, and even maybe *kml but the following metathesis one has to posit in order for this etymology to work is completely unwarranted. What complicates ma ers a li le bit more for this etymology is that the reverse order in this binomial pair, i. e. 鐐金 liào jīn, is a ested, in 11th century Chinese with the meaning ‘re ned silver’ (精美的银子, cf. L [1986-1993]) in the New Book of the Tang (新唐书·宣宗十一女传) (1060) where we read the following: 舊制：車輿以鐐金扣飾。帝曰：我以儉率天下，宜自近始，易以銅。 According to the old system, (a imperial princess’s) ariot has to be adorned with silver. e Emperor said: ‘In order to set the example of frugality to the whole world, it is be er to start with my close relatives: we will use bronze instead.’ is is of course quite late, and is really (another) hapax legomenon but we nd it important to mention its existence. It is important to note that except for Rybatzki, all the other resear ers adhere to the Altaic hypothesis whi forces them to consider the Chuvash 12 form more conservative in a way, at least as far as its /l/ is concerned, and so to look for external sources whi would have an /l/ sound. As we mentioned earlier, the problem of whether l₁ and l₂ were really two laterals or number one was a lateral and number two a palatal sibilant is almost invariably associated with the debate on the existence of an Altaic language family. is means that an anti-Altaicist would never resort to a foreign form whi has an /l/ and claim that it is the source of a word whi in Turkic participates in the CT :: Chuvash /l/, /š/:: /l/ correspondence. Indeed, in this case CT /š/ would have to re ect PT l₂ and the logical consequence of this is assumed to be that the Altaic languages must hark ba to a common ancestor. To sum up, none of the etymologies we have found in the literature, internal or external, seem convincing to us. In the next section, we are going to propose another one whi we think is be er as it not only takes into account data from Turkic but also from several other language families where the word for ‘silver’ seems to be derived from the same root. 5 Turkic ‘silver’ in a broader context T [1923:452], R´ T [1970:507-8] and S [1999:203], among others, have pointed out that a series of forms reminiscent of Old Turkic kümüš and Chuvash kӗmӗl are found in various languages of the SinoTibetan and Austro-Asiatic families. 5.1 Austroastic (Palaungic *kmuul) In Austroasiatic, we nd two groups of languages where the word for ‘silver’ is strikingly similar to the Turkic form: Palaungic and Khmuic. ese two bran es are not believed to be particulary close in the Stammbaum of the Austroasiatic family. Palaungic is a very diverse bran , comprising over thirty languages spoken in Yunnan and Burma. Fortunately, the historical phonology of these languages is relatively well known thanks to the work of D 13 [1980] and S [2010]. D [1980:#19-1] reconstructs *kmɨl ‘silver, money’ for proto-Waic, a subbran of Palaungic, and S [2010:#537] proposes proto Palaungic *kmuul ‘silver’. e Khmuic languages, spoken in northern Laos and neighbouring ailand and Vietnam, are relatively well described, though no complete reconstruction has been published yet. S [2002:#106] reports the form kmuːl ‘silver’ in Khmu. In both bran es, the noun ‘silver’ goes ba to a form *kmuul whi looks like the Turkic word with syncope of the rst vowel. 5.2 Sino-Tibetan 5.2.1 Western Tibetan ʂmul / χmul In Sino-Tibetan, lookalikes to the Turkic and Austroasiatic forms are found in ve distinct bran es: Western Tibetan, Tamang, Western Himalayish, Tani and Burmish. As Khmuic and Palaungic in Austroasiatic, these bran es do not form a coherent cluster within Sino-Tibetan. Tibetan and Tamang are relatively close to one another, and some authors believe that Western Himalayish and Tibetan form a ‘Tibeto-Kinnauri’ node, but Tani is not considered to be close to either Tibetan or Burmish by any author (see S [1993] for a detailed discussion). We do not present here an exhaustive review of all the primary data on these languages, as not all references are readily available. Whenever possible, we have osen the most reliable sources. Western Tibetan dialects, spoken in Ladakh (North-west India) and Baltistan (Northern Pakistan), have forms su as ʂ, x, χmul for ‘silver’ instead of common Tibetan dŋul :: Balti xmul ‘silver’ (B [1985:232]) or Purik ʂmul ‘rupee’ (Z [2006:79-80]). Tamangic, spoken in Nepal, is a group universally considered to be close to Tibetan. We nd mui in Risiangku Tamang and similar forms in other varieties (M [1994]). Note that in these languages, nal *–l generally anges to –i, and all initial clusters other than C{r, l, j, w} have been lost. West Himalayish is a group of languages spoken in Hima al Pradesh 14 and U arakhand (North-Western India), comprising Pa ani/Man ad, Byangsi, Darma, Chaudangsi and Kinnauri. We nd mul in Darma (W [2007:581]), and the STEDT online database cites the forms Pa ani mul and Kinnauri mölh (from notoriously unreliable sources). e only complex onsets in Darma (W [2007:61-2]) and other West Himalayish languages are of the C{w,j} type; a complex cluster su as /km/ is not permi ed by the phonotactics of these languages, as in Tamang. e Tani languages, spoken in Aruna al Pradesh (North-Eastern India) and neighbouring Tibet, are be er known than many bran es of SinoTibetan thanks to the comparative work of S [1993] and the grammar of Galo (citetPost08). Although some Tani languages have borrowed their word from ‘silver’ from Indic or Tibetan (S [1993:60,352]), we also nd `murkoo ‘silver’ in Galo (R . [2009]) and Bengni According to S [1993:213]’s sound laws, the syllable mur- in Galo and mɯr- in Bengni can come from protoTani *mul. Note the synonym *mul ‘amiss (verbal particule)’, Galo ‘mur ‘mistakenly’. We can therefore propose a proto-Tani *mul ‘silver’ based on Galo and Bengni. Only clusters of the type C{r, l, j, w} can be reconstructed for proto-Tani (S [1993:55-7]) and even these have been simpli ed in most languages.ffi² Lolo-Burmese is perhaps the best documented of all the bran es of Sino-Tibetan, spoken from Eastern Bangladesh to Vietnam, with the greatest diversity in Yunnan (China). e word usually reconstructed for ‘silver’ in proto-Lolo-Burmese is *C-ŋweffi (B [1979] #401b), but in several Loloish languages this etymon has become the autonym (B [1979]) and it was replaced by the word *pluffi ‘white’ (#501) in the meaning ‘silver’. However some Burmish languages su as Hpun (H [1986]), have forms that could go ba to proto-Lolo-Burmese *mwe and pre-protoLolo-Burmese *mul by regular sound laws. e actual Hpun form for ‘silver’ is myáiŋ. e rhyme Hpun –aiŋ has many distinct origins in proto-LB (quoted from B [1979]): Given the correspondence of ‘hair, feather’, a proto-form *mulffi is thereffi² We nd a similar form also in Sulong, a language of unclear a liation whi is probably unrelated to Sino-Tibetan: lə³¹mɯŋ⁵⁵ ‘silver’. is may be a borrowing from some Tani language. 15 PLB myàiŋ *ʔ-mwe³ #83 *m-rweffi #60 aíŋ *yimffi #341 ăʃaìŋ * (ʃ)-sin² #143 ămaíŋ *ʔ-m(y)iŋffi #419 PNC *ʰmʊl² #68 *rulffi #69 *ʔɪn² *tʰɪn³ #85 *ʰmɪŋffi #72 Tibetan meaning hair, feather sbrul snake khyim house mčhin-pa liver miŋ name Table 9: Proto-Lolo-Burmese origins of Hpun –aiŋ fore a possible reconstruction for myáiŋ ‘silver’. Incidentally, Hpun is spoken between Myitkyina and Bhamo on the Irrawady river in Burma, not far from the Wa-speaking area (see the preceding section). In all ve bran es of Sino-Tibetan, the word for ‘silver’ can go ba to a form su as #C-mUl (provisional reconstruction), where #C represents an undetermined consonant and #U a high ba rounded vowel. is form again is very similar to Proto-Palaungic *kmuul and Turkic kümüš. 5.2.2 Tibetan dŋul e Sino-Tibetan etymon for ‘silver’ presented in the previous section however is not independent from the most widespread form found across SinoTibetan (M [2003:415-6]). e following (non-exhaustive) data illustrate a estations of this etymon: Language Tibetan proto-Lolo-Burmese Tangut Old Chinese form dŋul *C-ŋweffi ŋwo² 銀 *ŋrən Reference a ested in the Zhol inscription (AD 763) #401b, B [1979] #3572, L [1997] Table 10: Words related to Tibetan dŋul across the Sino-Tibetan family. is word spread from Chinese to various Kra-Dai and Hmong-Mien languages, and through ai to some varieties of Austro-Asiatic. e correspondence between Chinese *-ən or *-un (whi cannot always be clearly distinguished in the reconstruction due to phonotactic constraints) and Ti16 betan –ul might seem counterintuitive, but many examples of this correspondence have been brought to light (Gong 1995 [2002:103]):ffi³ Chinese 分 *pən 貧 *brən 塵 *drən 鈍 *dˁun-s Meaning share poor dust blunt, dull (knife) Tibetan ‘phul dbul-po rdul rtul-po Meaning o er poor dust blunt Table 11: Examples of the correspondence between OC *-ən/-un and Tibetan -ul Although no reconstruction of Sino-Tibetan is possible at the present moment, the Tibetan form can come from a pre-Tibetan *C-ŋul, where C represents a dental or a velar stop: the contrast between pre xal *k/g- and *t/d- is neutralized even in Old Tibetan (L [1933]). In Western Tibetan dialects, thanks to the work of Marius Zemp (2006: 79-80), it is clear that the forms with m- presented in the previous section are secondary. /ŋ/ generally anges to /m/ before a rounded vowel: Kargil ʂmul ʂŋo ~ ʂmo ʂmultʃhu ~ ʂŋultʃhu ŋu Etymology dŋul rŋo rŋul-u ŋu Meaning rupee to fry sweat to cry Table 12: Examples of ŋ > m in Purik is sound ange does not apply to the simple onset ŋ-, and seem to be unstable to some extend in Tibetan, as both variants with ŋ and m are a ested. is suggests a rather recent ongoing sound ange across Western Tibetan. However, no su evidence is available for the four other bran es: there is no way to derive /m/ from /ŋ/ in either Tamangic, Tani or Hpun. For West ffi³ We cite here Baxter and Sagart’s (2011) reconstruction rather than Gong’s, but the comparisons are still valid. 17 Himalayish, it could be argued that the forms mul ‘silver’ are borrowed from Ladakhi. W [2007:72] reports that Darma speakers are in contact with Tibetan, though she does not specify whether it is Central Tibetan (a dialect with velar nasal in ‘silver’) or Western Tibetan. 6 Several historical scenarios for the spread of the Wanderwort ‘silver’ We now have to explore all logical possibilities to account for the data reviewed so far. Ar aeologically ‘silver’ is not a ested in China before the 6th century BC and te niques of cupellation of galena smelt would only have become practised during the late Zhou and Early Han periods. (Barnard & Sato 1975, Needham et al. 1980, Behr 2008:516-524). In any case, the late appearance of silver in China and Southeast Asia (L [1985:336-7]) excludes the possibility that ‘silver’ could be reconstructed to the proto-Sino-Tibetan or proto-Austro-Asiatic levels. 6.1 Accidental lookalike M [2003:416] suggests than the #mul forms are loans from AustroAsiatic. In this view, the resemblance between the #ŋul and the #mul forms would be a coincidence in languages other than Western Tibetan and West Himalayish. #mul-like forms in Tamangic, Tani and Hpun would be borrowings from Austro-Asiatic. Since Hpun and Palaungic are spoken in neighbouring areas, and since the Austro-Asiatic language Khasi is spoken in Meghlaya, not far from Aruna al, this explanation could tentatively account for the Hpun and the Tani forms. e form mui in Tamangic however would be more di cult to explain away as borrowing from a Mon-Khmer language since the hypothesis of direct contact between Tamangic and Mon-Khmer is highly implausible, and even indirect contact is not evidenced by any other term observed so far. e resemblance between Turkic and Austro-Asiatic would also have to be ruled 18 out as coincidence, since no direct contact can be assumed between AustroAsiatic and Turkic speakers in pre-historic times. Most importantly, since silver te nology (cupellation) seems to have originated in Anatolia a later spread to East Asia from the north is far more plausible. 6.2 ST > Turkic or Turkic > ST e presence of both ŋ-forms and m-forms in Sino-Tibetan corresponding only to m-forms in Turkic and Austro-Asiatic can be accounted for by ea of the following three hypotheses, all of whi are compatible with both the ST > Turkic or Turkic > ST borrowing scenario. 1. Assimilation ŋ > m in Sino-Tibetan 2. Assimilation ŋ > m in Turkic 3. Dissimilation m > ŋ in Sino-Tibetan 6.2.1 Assimilation ŋ > m in ST In view of the assimilatory ange ŋ > m in Western Tibetan, we could either propose that this ange occurred independently in Tani, Tamangic and Hpun or that it took place in another Sino-Tibetan languages and was subsequently borrowed by the three groups. In this view, both the AustroAsiatic and Turkic forms would have to be borrowed from a Sino-Tibetan language that had undergone the assimilation. Under the Tibetan > Turkic loanword scenario, the WT form dŋul either re ects an earlier *gŋulffi whi was borrowed in Turkic from a (para-) Tibetan language in whi ŋ > m/C V where C is {r, d} and V a rounded vowel as in Purik (cf. above): WT dŋul ‘silver’ > Purik ʂmul, or else the assimilatory ange took place in Turkic (cf. 6.2.2). e drawba of this hypothesis is that Western Tibetan cannot be the source for all of the AA, ST and Turkic forms, and that we have to hypothesize the existence of an una ested Sino-Tibetan language (presumably a ffi Since, according to L [1933] preinitial d- and g- are in complementary distribution in Tibetan, we can posit a phonetic rule of the form *g- > d-/ velar. 19 close parent of Tibetan) whi languages. gave this word to all of the neighbouring 6.2.2 Assimilation ŋ > m in Turkic Alternatively, we could propose a similar assimilatory ange for Turkic. Indeed, the form kümüš could be derived from an earlier **kŋɬ with labialisation of the velar between ba vowels, a ange a ested in 10th century Khotanese Turkic (cf. H [1977:511]) or in Chuvash, for that ma er (cf. the word for ‘heart’ in Table 1). According to E [2004:117]) “In some words in some varieties of Old Turkic, ŋ > m beside rounded vowels: H [1977] discusses a.o. kömül < köŋül ‘heart’. OTWF [=Old Turkic Word Formation, E [1991]] 99 and 104 document the lexemes boymul < boyun+ and kömüldürük < köŋül+ (whi is also the source of Turkish gömlek ‘shirt’). Another instance is yürüm karak < yürüŋ karak ‘the white of the eye’ in the Turkic-Khotanese hippological glossary (Wordlist 40).” is would mean that either the ST (and AA) word were borrowed from a Turkic language whi had undergone this ange, or else the word was borrowed from ST in Turkic and then underwent a parallel ange independently from the donor language (cf. 6.2.1). 6.2.3 Dissimilation m > ŋ in ST Finally, we could propose that the X-mul forms are original and that the Xŋul forms found across Sino-Tibetan are due to a dissimilation of m > ŋ. is dissimilation does not need to have occurred independently in Chinese, Tibetan and LB. Rather, it could have taken place in one (non-speci ed) language and have then been borrowed into most of Sino-Tibetan. Ironically, the Western Tibetan dialect would have reversed this ange. e direction of borrowing could have been either from ST to both Turkic and AA, or from Turkic to ST to AA. e major weakness of this hypothesis is that this dissimilatory ange is not a ested anywhere in Sino-Tibetan. 20 As far as the coda consonant is concerned, if the Turkic form is borrowed from ST we can explain why Turkic has *l₂ *[ɬ] corresponding to –l in other languages rather than *–l₁. Indeed, in many ST languages su as Japhug Rgyalrong, nal sonorants are devoiced in coda position (thus Japhug tamar ‘bu er’ is realized as [r̥]). Under the hypothesis that the borrowing of ‘silver’ occurred from ST to Turkic, the presence of *–l₂ rather than *-l in coda could be explained by supposing that the donor language had a devoicing rule similar to Japhug, and that the hypothetical form *kmul was realized with a devoiced lateral *kmuɬ. is devoiced lateral was phonetically closer to Turkic *-l₂ (perhaps *[ɬ] rather than a palatalized l) than to the normal *-l₁ (both in its velar and palatal variants). In the alternative hypothesis (i.e. Turkic > ST), the correspondence of Turkic *-l₂ to Tibetan -l is straightforward since Sino-Tibetan languages only have at most one /l/ sound in coda position. 6.2.4 Summary We have no way to determine whi of these three hypotheses is the correct one, though the rst one seems considerably less likely. e etymon for ‘silver’ is not derivable in a straightforward manner from any known verbal or nominal root in either Turkic, Sino-Tibetan or Austroasiatic. We have already seen this for Turkic. In Tibetan, dŋul could be a deverbal noun derived by the non-productive d-/g- nominalizing pre x,ffi but no independent root *ŋul is a ested either in Tibetan or in any other Sino-Tibetan language. Since no internal etymology for the word ‘silver’ is available in either ST, AA or Turkic, both the ST > Turkic and the Turkic > ST borrowing scenarios are equally possible, as is the possibility of both ST and Turkic having borrowed the word from an unknown language. Indeed, independently of the fact whether the original form had a velar or a labial nasal, and of the direction of borrowing (from Turkic to ST or the reverse), the similarity between the Turkic, ST and AA words for a te nical ffi As in ɴkhyil ‘to ow together, to whirl’ > dkyil ‘center’, nag(-po) ‘bla ’ > gnag ‘bla ox’. is prex is probably related to the velar nominalizing pre x found in Rgyalrongic, Kiranti, Kuki-Chin and other languages (Japhug Rgyalrong kɯ-, Limbu kɛ- etc). 21 concept su as ‘silver’, whi has no obvious etymology in any of these languages, strongly supports the hypothesis that all of these forms are related Wanderwörter. is is actually a not too uncommon situation as we have a very similar one in part of the IE family, since the word for ‘silver’ in Germanic and Balto-Slavic is most probably a Wanderwort. 7 Conclusion Independently of the direction of borrowing, the relatedness of Proto-Palaungic *kmuul, Tibetan dŋul and Turkic kümüš has an important implication for the reconstruction of the correspondence –š to –l between common Turkic and Chuvash. As explained above, this correspondence is generally reconstructed as *–š by non-Altaicists and as *–l₂ by proponents of the Altaic theory. e reconstruction of *–š in this word is clearly invalidated by the comparative ST and AA evidence. One would have to suppose a borrowing from Bolgar Turkic to ST and AA, but the ange *–š > –l hypothesized for the Bolgarian bran of Turkic is too recent to explain the presence of –l in all ST and AA languages, especially given the fact that the Chinese a estation of 銀 yín goes ba to the Han period. Besides, the words for ‘silver’ in LB, inasmu as they t in the correspondence sets with nal *-ul as illustrated above, must have been borrowed before the proto-LB unity since nal *-l was already lost in proto-LB. is detail of reconstruction however has li le incidence on the Altaic debate: It does not support in any way the hypothesis of a genetic relationship between Turkic and Mongolic. It disproves, however, the idea that the *–l₂ to –l correspondence between Turkic and Mongolic should necessarily be interpreted as a feature of words borrowed from Bolgar Turkic into Mongolic. 22 References Aˇ , N. I. 1898. Materialy dlja izslědovanija čuvašskago jazyka. Tipogra ja Imper. Universiteta, Kazan’. B B , Noel & Sato Tamotsu 1975. Metallurgical remains of ancient China. Ni iosha, Tokyo. , N. A. 1975. Grammatika xakasskogo yazyka. Nauka, Moskva. B , William 1992. A handbook of Old Chinese phonology. vol. 64 de Trends in Linguistics. Studies and monographs. Mouton de Gruyter, Berlin/New York. B , William & Sagart Laurent 2011. Baxter-Sagart Old Chinese reconstruction.Version 1.00. URL http://crlao.ehess.fr/document.php? id=1217. last accessed June 2011. B , Wolfgang 2008. Reimende Bronzeinsrien und die Entstehung der inesisen Endreimditung. Projekt Verlag, Bo um & Freiburg. B , Roland 1985. Das Mären von Prinzen Cobzang. Eine tibetise Erzählung aus Baltistan. VGH Wissens a sverlag, Sankt Augustin. B , M. I. 1973. Slovarnye materialy po xakasskim dialektam XIII. Dans P , D. F. (Éd.), Dialekty xakasskogo jazyka. Xakasskoe knizhnoe izdatel’stvo, Abakan. B , David 1979. Proto-Loloish. n 39 Dans Scandinavian Institute of Asian Studies Monograph Series. Curzon Press. B , József 1864. Némely látszólagos képtelenségek a csuvas-török hangviszonyokban. Dans Nyelvtudományi Közlemények. vol. 3: 234–248. B , Christopher 2009. A Reconstruction of Proto Northern Chin in Old Burmese and Old Chinese Perspective. èse de doctorat. University of London. 23 C , V. I. & Bugaeva T. G. 1979. K ètimologii nazvanij metallov i ix splavov v altajskix jazykax. Issledovanija v oblasti ètimologii altaiskix yazykov. Nauka, Leningrad. D , Gérard 1980. e Wa languages. Dans Linguistics of the TibetoBurman Area. vol. 5.2: 1–182. D , A. V. 2007. Lingvističeskie kontakty rannix tjurkov. Leksičeskij fond. Pratjurkskij period. RAN, Institut jazykoznanija, Moskva. E , Marcel 1991. Old Turkic Word Formation. Harrassowitz, Wiesbaden. E , Marcel 1993. Die Sprae der wolgabolgarisen Insrien. vol. 13 de Turcologica. Harrassowitz, Wiesbaden. E , Marcel 2004. A Grammar of Old Turkic. Handbook der Orientalistik. Brill, Leiden. F , M. R. 1996. Ètimologičeskij slovar’ čuvašskogo jazyka. Čuvašskij gosudarstvennyj institut gumanitarnyx nauk, Čeboksary. 2 vols. G , Zoltán 1913. Zur Lautgesite der altaisen Spraen. Dans Keleti Szemle. vol. 13: 1–37. G , Hwang-Cherng 1995. e System of Finals in Proto-Sino-Tibetan. Dans W , W. S.-Y. (Éd.), e Ancestry of Chinese. n 8 Dans Journal of Chinese Linguistics Monograph Series. pp. 41–92. University of California, Berkeley. Reprinted in: Collected Papers on Sino-Tibetan Linguistics. Taipei: Zhong yang yan jiu yuan yuyanxue yanjiusuo oubei u, 2002: 79–124. H , James 1977. Nasales instables en turc khotanais du Xe siècle. Dans Bulletin of the Sool of Oriental and African Languages. vol. 40.3: 508–521. H , Eugenie J. A. 1986. Some hitherto unpublished material on Northern (Megyaw) Hpun. Dans L , John McCoy & Timothy (Éd.), Contributions to Sino-Tibetan studies. pp. 101–134. Brill, Leiden. 24 J 81–125. , Lars 1998a. e History of Turkic. Dans J [1998b]. pp. J , Lars & Csátó Éva Á. 1998b. e Turkic Languages. Routledge, London. J , Aulis J. 1952. Die Lehnwörter des Sajan-Samojedisen. n 103 Dans Mémoires de la Société Finno-Ougrienne. Suomalais-ugrilainen Seura, Helsinki. K , Béla 2010. Mongolian čilaɣun: Turkic tāš. Dans Turkic languages. vol. 14: 103–112. L , L. S. & Dybo A. V & Rassadin V. I. (Éd.) 1997. Ètimologičeskij Slovar’ Tjurkskix Jazykov. Obščetjurkskie i mežtjurkskie osnovy na K. RAN, Institut jazykoznanija. L , Daqin 2004. Sulongyu yanjiu. Minzu ubanshe, Beijing. L , Fang-Kuei 1933. Certain Phonetic In uences of the Tibetan Pre xes upon the Root Initials. Dans Bulletin of the Institute of History and Philology. vol. 6.2: 135–157. L , Fanwen 1997. Xiahan zidian [A Tangut-Chinese dictionary]. Zhongguo shehui kexue ubanshe, Beijing. L , Xueqin 1985. Eastern Zhou and Qin Civilizations. Yale University Press, New Haven. Translated by Kwang-Chih Chang. L , Zhufeng (Éd.) 1986-1993. Hanyu da cidian. Cishu ubanshe, Shanghai. M , J. P. & Adams D. Q. 2006. e Oxford Introduction to Proto-IndoEuropean and the Proto-Indo-European World. Oxford University Press, Oxford. M , James A. 2003. Handbook of Proto-Tibeto-Burman. University of California Press, Berkeley. 25 M , Martine 1994. Problèmes de comparatisme et de reconstruction dans quelques langues de la famille tibéto-birmane. èse de doctorat. Université Paris III. M , Oleg A. 2002. Bulgarskaja gruppa. Dans T ˇ , È. R. (Éd.), Sravnitel’no-istoričeskaja grammatika tjurkskix jazykov. Regional’nye rekonstrukcii. pp. 677–712. Nauka, Moskva. N , J. & Yu H. & Lu G.-D. & Sivin N. 1980. Alemy and Chemistry. Dans N , J. (Éd.), Science and Civilization in China. vol. 5. Cambridge University Press, Cambridge. P , N. N. 1924. K konsonantizmu altajskix jazykov. Dans Doklady Rossijskoj Akademii Nauk. vol. April-June: 43–44. P , N. N. 1925a. O rodstvennyx otnošenijax čuvašskogo i tjurko-tatarskix jazykov. Čuvašskoe oblastnoe izdatel’stvo, Čeboksary. P , N. N. 1925b. Čuvašskij yazyk i ego otnošenie k mongol’skomu i tjurkskim jazykam. Dans Izvestija Rossijskoj Akademii Nauk. vol. XIX(1-5): 23–42. P , Mark 2008. A Grammar of Galo. èse de doctorat. LaTrobe University. Melbourne. R R , Wilhelm 1882. Phonetik der nördlien Türkspraen. Weigel. , Gustaf J. 1949. Studies in Korean etymology. Suomalaisugrilainen Seura, Helsinki. R¨ ¨ , Mar i 1969. Versu eines etymologisen Wörterbus der Türkspraen. Suomalais-Ugrilainen Seura, Helsinki. R , Helmut & Kümmel Martin & Zehnder omas & Lipp Reiner & S irmer Brigi e 2001. Lexikon der indogermanisen Verben. Rei ert, Wiesbaden. R , Martine I. 2005. Is Japanese Related to Korean, Tungusic, Mongolic and Turkic?. vol. 64 de Turcologica. Harrassowitz, Wiesbaden. 26 R´ T , András 1970. Az altaji nyelvrokonság vizsgálatának alapjai. (A nyelvrokonság elmélete és a csuvas–mongol nyelvviszony). èse de doctorat. University of Budapest(?). R´ T , András 1998. e Reconstruction of Turkic and the Genetic estion. Dans J [1998b]. pp. 67–80. R´ T , András 2007. Nutshell Chuvash. Handout of a talk given at the Erasmus Mundus Intensive Program. Turkic Languages and cultures in Europe (TLCE). URL http://www2.lingfil.uu.se/afro/ turkiskasprak/IP2007/NUTSHELLCHUVASH.pdf. last accessed in June 2011. . 2009. Galo-English Dictionary. Galo R , Ìgoo & Mark W. Post Welfare Society, Itanagar. R , Volker 1994. Bemerkungen zur türkisen und mongolisen Metallterminologie. Dans Studia Orientalia. vol. 73: 193–251. S , Laurent 1999. e Roots of Old Chinese. John Benjamins, Amster- dam. S , Wilhelm 1841. De lingua Tsuwasorum dissertatio. Veit, Berlin. S` , È. V. (Éd.) 1980. Ètimologičeskij slovar’ tjurkskix jazykov. Obščetjurkskie i mežtjurkskie osnovy na V, G i D.. Akademija Nauk SSSR. Institut jazykoznanija. S , Paul 2010. Preliminary Notes on Proto Palaungic. URL http:// sealang.net/monkhmer/database/. last accessed in June 2011. S , Marek 1998. Zwei alürkise Konsonantenwesel (š ~ s, š ~ l), die Runik und die Altaistik. Dans L , J.P. & Ölmez M. (Éd.), Bahşı Ögdisi. Klaus Röhrborn Armağanı. pp. 391–399. Simurg, Freiburg/İstanbul. S , S. A. & Dybo A. V. & Mudrak O. A. 2003. An Etymological Dictionary of the Altaic Languages. vol. 8-1, 8-2, 8-3 de HdO. Brill, Leiden. 27 S , O. V. (Éd.) 2006. Xakassko-russkij slovar’. Nauka. S , Ja son T.-S. 1993. A Historical-Comparative Study of the Tani (Mirish) Bran in Tibeto-Burman. èse de doctorat. University of California. Berlin. S , Premsrirat 2002. esaurus of Khmu Dialects in Southeast Asia. vol. 1 de Mon-Khmer Studies. Mahidol University. T , Alfredo 1923. Elementi di gloologia. vol. 2 de Classe di scienze morali. Accademia delle Scienze dell’Instituto di Bologna.N. Zani elli, Bologna. W , Heinri 2005. Die Jenissej-Spraen des 18. Jahrhunderts. vol. 67 de Verö entliungen der Societas Uralo-Altaica. Harrassowitz, Wiesbaden. W , Christina 2007. A Descriptive Grammar of Darma: An Endangered Tibeto-Burman Language. èse de doctorat. University of Texas at Austin. Z , Marius 2006. Synronic and Diaronic Phonology of the Tibetan Dialect of Kargil. [Lizentiatsarbeit im Fa Historis Vergleiende Sprawissensa]. Mémoire de maîtrise. Universität Bern. 28