By | October 19, 2017 | 总浏览:14,827


???? 近期在蒙古国图拉河旁的慧苏图鲁盖(Hüis Tolgoi)山发现了疑似七世纪的蒙古语石碑。经法国社会科学高等学院(EHESS)的蒙古学家Mehmet ?lmez, étienne de la Vaissière, Dieter Maue, Alexander Vovin等专家的初步破解研究,宣布为使用婆罗米文(Brāhmī)刻写的蒙古语石碑,是属首次发现。

???? 之前学术界普遍认为在公元五到七世纪,蒙古族主要生活在蒙古高原东部和东北地区,但是这个考古发现证明了那个时代的蒙古高原中西部地带蒙古族也广泛繁衍生息。

???? 所以这碑文是一个划时代性的重大发现。蒙古国考古学会称“这是往后五十年内蒙古学领域的最伟大的发现”。

?????? 该碑文的刻写时代比成吉思汗石碑还要早600年,如发现和对其甄别鉴定无误,蒙古文的历史还可推前了600年。



Presented August 31, 2017 at 60th PIAC, Székesfehérvár, HUNGARY


?Alexander Vovin, 2017




(Draft version)


Alexander VOVIN



I am going to present below a tentative reading and interpretation of the Hüis Tolgoi inscription (hereafter HT inscription). This work would not be possible without the pioneering work by Dieter Maue on the decipherment of its writing system, which was also greatly added by the efforts of the team of specialists in 3-D photography that accompanied us on our 2014 expedition to Mongolia.

Although at this stage it is not possible to provide a complete reading and interpretation of the HT inscription, slightly going ahead it seems beyond reasonable doubt that it is in some form of Mongolic. It also appears that the language of the inscription, although it can be conditionally termed as a variety of Para-Mongolic, is much closer to the mainstream Mongolic languages, such as Middle Mongolian and modern extant Mongolic languages than to Serbi-Khitan,[1] although there are some features that the language of the HT inscription shares with the latter. It is necessary to note that the HT inscription is monolingual, which certainly opens the gates for the guess-work and speculations, some of which certainly will be corrected by the following generations of scholars.

As was already mentioned by my colleagues, the HT inscription (or inscriptions?) consists of two stones, one preserved in the basement of the National Archeological Institute, and the other placed near the entrance to the building. Certain circumstances prevented us from reading and photographing the text on it.


LINE 2-1[2]


b??tī?-?a?r? kagan ?dig?n ?ī????-n? bodi-satva to?ro?-x


Tentative transcription:

biti[g]-ńar qa?an? digin ??ińe-n? bodi-satva? t?r?-k


Morphemic analysis:

inscription-PLUR qa?an? tegin? new-GEN ?Bodhi-sattva ?be.born-NOM.FUT



  1. b??tī?, kagan, dig?n, and bodi-satva seem to be for the most part unproblematic, but they also provide no information on the genetic affiliation of the language.
  2. -ńar is probably the Mongolic plural marker -nAr, but there is a functional problem with this interpretation. In MM it is always used with either kinship terms of with deities. Khitan plural marker -ń?r[3] ~ -ń?ń is also used for nouns denting humans (Shimunek 2017: 264). Thus, in both Mongolic and Khitan these suffixes are limited in usage, being reserved only for animate nouns. However, crosslinguistically the functional changes in plural markers are not unknown. Thus, for example, WOJ -ra originally used to be a neutral plural marker, used both with animate and inanimate nouns, but virtually acquired pejorative usage with animate nouns in Middle (Classical) Japanese (Vovin 2005: 93). In the similar fashion, WOJ plural marker -tati for animate nouns gradually became highly restricted honorific plural marker in MJ (Vovin 2005: 100). The plural marker nd?m? was restricted to animate nouns designating people in WOJ (Vovin 2005: 98), but in MJ its descendant form -domo is used with both animate and animate nouns (Vovin 2003: 41).
  3. ?ińe-n ‘new’ is the second word that is highly diagnostic, clearly pointing to the Mongolic direction. Brāhmī <??> probably stands for [i] or [?], as noted by Maue, so in this case it is probably safe to transcribe <??> as [i], although [e] cannot be completely ruled out. Even if the form is to be read ?ini, cf. EMM ?ini 失你 ‘new’ (MNT §265), although the majority of attestations indicate ?ine, cf. ?ine ???? (Qub IV: 43), WMM ?ine ???? (Mu 334). Cf. Khitan *??n (Langjun 7: 6), thus phonetically HT form is closer to mainstream Mongolic. The final -n is likely to be a genitive.

As far as I can tell, there are no clear-cut cases of the adnominal usage of genitive in MM. Thus, e.g.:


Tümet irgen-ü oki-t

Tümet people-GEN girl-PLUR (MNT §241)


can be interpreted as ‘girls who are Tümet persons’ or ‘girls of Tümet people’. Meanwhile, in Khitan, as noted by Shimunek, a genitive case marker can be used as an adnominal marker, albeit this function is restricted to the cases when it is preceded by a numeral (2017: 260), e.g.


?? ?

tau-un ???u.ur

five-GEN ?division

the Five Divisions (lit.: divisions that are five) (Yelü Xiang-wen 7: 15-16)


  1. Bodhi-sattva is either a given name of the Turkic qa?an from the First Khanate, or Buddhist Bodhi-Sattva could also be meant here, as the text may be a historical, but with the religious overtones.
  2. to?ro?-x is likely to be Mongolic t?r?-kü ‘to be born’, nomen futuri, but not the transcription of the tribal name Türük (later Türk). Both t?r?-k[ü] ‘will be born’ and türüg ‘Türük’ appear more than once in the HT inscription, the former on lines 2-1, 2-8, and 2-9, and the latter on lines 2-5 and 2-10 in their specific contexts. What is even more important, is the fact that they have not only different vocalism, but also the difference between -x for t?r?-k[ü], and -g for Türüg. Both these discrepancies cannot be explained as a simple free variation.
  3. There are three typically Mongolic morphological markers on this line: plural -ńAr, nomen futuri -x < *kü, and genitive -n after vowel stems.


LINE 2-2


kagan bu?d-a?? kagan-u ruka-x ruka-ju? xī?r?? ?a?akay


Tentative transcription:

qa?an buda? qa?an-u? uqa-qu ?uqa-?u? kera? Ańaqay


Morphemic analysis:

qa?an Buddha? qa?an-GEN? realize-NOM.FUT? realize-CONV? country? Ańaqay



  1. Unless Bud-a is simply ‘Buddha’, which is the simplest solution, the only other possible explanation I can think about is Khitan bud  ‘external clan’ (Yelü Jue 31: 55), (Hsiao Hui-lian.Can 2: 1), (Yelü Pusuli 7: 18).[4] But this leaves final -a stranded: it is unlikely to be a MM dative-locative case suffix -A. There are no other fitting attestations either in MM or Pre-Classical WM texts to the best of my knowledge. Buda qa?an is reminiscent of OT bur-qan ‘lord Buddha’.
  2. Maue suggested that ruka- may reflect either uruqa or uqa. The first one is improbable, as only EMM uruqa ‘relatives, descendants’ (MNT §11, §23, §24 etc.), a likely loanword from Turkic, comes to mind, and it neither fits the context, nor does agree with apparently verbal nature of the stem. EMM verb uqa- ‘to realize, to know’ (MNT §18, §78, etc.) fits much better. No Khitan cognates.
  3. xī?r?? is probably a borrowing from Tumshuquese χ?era ‘country’, as Maue has suggested. In this meaning it could fit here and also on line 2-6 below. Less probably it can be equated here (but not on line 2-6) with EMM kere- ‘to fight’ (MNT §194), which is unrelated to WM kira- ~ kiru- ‘to cut’, and there are no attested Khitan cognates.[5]
  4. Ańaqay is a name of the last Ruan-ruan qa?an, who refused to give his daughter as a wife to Turkic Bumin qa?an in 546 AD.
  5. The most important is that this line is full of typical Mongolic morphology: MM idiosyncratic genitive -u after stems in -n as in qa?an-u ‘qa?an-GEN’, nomen futuri -qu, converbum imperfecti -?U, and possibly dative-locative -A, if the alternative proposed etymology for bud-a is correct. Once again we can see that the language of the HT inscription is much closer to mainstream Mongolic than to Khitan: a) there is no Khitan genitive -u, as the Khitan words with final -n take -en instead (Shimunek 2017: 253), and the Khitan word qa ~ qa.?a ‘qa?an’ takes the genitive in -an: qa.?a-an (Kane 2009: 132); Khitan has converbum imperfecti -? corresponding to MM -?U, and no nomen futuri in -qu ~ -kü or dative-locative case suffix -A.


LINE 2-3


[…]ī? tī?-n ?ja?-x? bo?d??? bigiy? ?a?r? ba?y??? do?lu?? ja?ju? hügbü? +?


Tentative transcription:

[.?.]-? te-n? ?a-qu? bod-?? beg-ey-ńar? bayy?-?? dolu? ?a-?u? (h)ügbü? +?


Morphemic analysis:

[title]-ACC say-CONV? promise-NOM.FUT? tribe-ACC ?beg-?-PLUR ?be/stand-IMP ?seven? promise-CONV? ?? ?



  1. Ańakay is supposed to have a title, and since he cannot be equated with the Ruan-ruan qa?an Anagui as demonstrated by étienne de la Vaissière, it is natural to suppose that unreadable beginning of the line 2-3 corresponds to this title, but not to qa?an.
  2. te- ‘to say’ (quotational verb) was borrowed from OT t?- ‘id.’ into Khitan (Vovin 2013: 622-623. Although there is no similar evidence for mainstream Mongolic, it is possible that the language of the HT inscription acquired the same loan.
  3. ?a- ‘to promise’ is probably a cognate to EMM and WMM ?a’a- (HYYY 17b.2), (MNT §82, §84, etc.), (Mu 199)~ EMM ?i’a- (MNT §121, §145, etc.) ‘to report, to promise’.
  4. bod must be a loan from OT bod ‘tribe’, unless the direction of loan was opposite, cf. OT bodun ‘population, subjects’, apparently with a Mongolic plural -n.
  5. beg might be yet another OT loan (OT beg ‘bek’, ultimately from EMC p?k 伯 ‘elder’), but the -ey segment is unaccounted for. The vocalism is also problematic, since bigiy in HT seems to indicate [i], but there is at least one other case of e > i raising: *?ine > ?ińi ‘new’ in line 2-1.
  6. dolu must be a cognate to MM dolu’an ‘seven’, without the numerical suffix *-pan (> -’an in intervocalic position). Cf. Khitan dal? ‘id.’ with a different vocalism.
  7. hügbü is an interesting case. The language of the HT inscription has both p- and b-, so unlike MM, h- or h cannot be a reflex of proto-Mongolic *p-. I suspect that it either denotes initial g- or ?-. However, there are no apparent MM cognates. The last word on this line is unreadable.
  8. As in the previous two lines, there is plenty of Mongolic morphology in line 2-3: accusative -? (cf. MM accusative -i), converbum modale -n, nomen futuri -qu, plural suffix -ńAr (this time following an animate noun), and converbum imperfecti -?u. The last three has been already discussed above. As for the first two, note that there are no accusative -? or -i and converbum modale -n in Khitan.


LINE 2-4


b[??]tī? jī?lo?-na?r? kra-nya-gu-?? tu?wa?? pu?ro?-r ?c??cī?-ra?? pügt?g? ?a?la?n


Tentative transcription:

b[i]tī[g] jīlo-nar? q[a]ra-n[V]ya-?uń? tuwa? pu?ro?-r ??i?i-r?? pügt?g? ńele-n


Morphemic analysis:

inscription stone-PLUR? look-?-NML? ?-NML? stab-CONV? ?? join-CONV



  1. jī?lo must be related to MM ?ila’un ‘stone’ (with possible voicing of ?- in an intervocalic position), and -nar is a plural suffix. Note that this time we have n-, not ń-, possibly under influence of the final -n of the preceding word that itself elided (however, cf. also qato-ńar ‘qa?an’s wives’ < qatun-ńar in line 2-5). Note that the usage of the plural confirms that there were more than one stone in the inscription.
  2. Some remaining individual items on this line can be probably understood, but there is no coherent reading. kra-nya looks very ‘un-Altaic’, but on the basis of dro in the line 2-6 that is likely to stand for doro ‘law’, we can suspect the formation of secondary clusters due to the elision of the vowels in the first syllables here as well. I tentatively interpret this as q[a]ra- ‘to look’, cf. MM qara- ‘to look’ (MNT §5, §63, §77, §183), (Mu 292) + -n[V]ya-, obscure morphological marker without any obvious parallels in MM or Pre-Classical WM + ?uń, a deverbal nominalizer cf. MM -’Un (Godziński 1985: 53) and WM -?un ~ -gün (Poppe 1964: 46), which forms nouns designating qualities or abstract nouns. An alternative, but much more inferior explanation: q[a]ra n[o]ya ‘black (i.e., civilian, not military) officials’. Cf. MM qara ‘black’ (MNT §21, §25, etc.), (Mu 227), (IM 444) and noyan ‘official, lord’ (MNT §8, §51, etc.), (Mu 115), (IM 443). ?uń is difficult to explain. It certainly cannot be MM -wun or WM -?un, a deverbal nominalizer, because we have no verb preceding it. Since there is no -? in the language of the HT inscription, it could potentially be a loan from EMC ku? 公 ‘prince, duke’.
  3. tu?wa? occurs three times in the HT inscription: on the lines 2-4, 2-7, and 2-9. It is not clear what it can be, and the problem is further complicated by the fact that -w- occurs only in this word. The comparison with the tribal name Tab?a? ~ Ta?ba? should be discarded for phonological reasons. May be EMM to’a ‘number’ (MNT §229, §265, §278), (HYYY 22b.2), (KMQB I: 9b.2), but note that in other cases MM -’- ~ WM -?- is reflected as -?- in the language of the HT inscription. WMM tü’ü- ‘to gather’ (Mu 159) is more problematic for phonological and syntactical reasons. Maue suggested tribal name Dūbō ~ Dūbò (都波 ~ 都播EMC tuo pwa ~ tuo pwaC). Since the language of the HT inscription does not exhibit -p-, he is probably right.
  4. pu?ro?-r is obscure. -r can be a Mongolic deverbal nominalizer -r, as in WMM amu-r ‘peace, patience’ (Mu 102) < MM amu- ‘to rest’ (MNT §145, §201, etc.), (Mu 102), (IM 432), but the root is opaque.
  5. Maue suggested Mongolian ?i?i- ‘to stab’ + -rA, converbum finale. The form ?i?i- appears to be Post-Pre-Classical, as it is not attested in MM or Pre-Classical WM. Cf. EMM se?i- ‘to stab’ (MNT §127). Converbum finale -rA is well attested in both MM (Godziński 1985: 146) and pre-Classical WM (Poppe 1964: 98). Therefore, provisionally I accept Maue’s proposal.
  6. pügt?g is obscure. I do not have any possible etymologies for it.
  7. ńa?la?-n could potentially be EMM neyile- or WMM neile- ‘to join’, followed by the converbum finale -n.
  8. There is some Mongolic morphology in this line as well: plural suffix -nar, converbum modale -n, converbum finale -rA, and possibly a nominalizer -r. The first two has been already discussed above. As for the converbum finale -rA and nominalizer -r, I do not think they is attested in Khitan.


LINE 2-5


[+] k[a]ga[n-u? + ] ?kato-?a?r? düg?-d ?nī?r?? ?kagan ?türü?g ?kagan



Tentative transcription:

[+] q[a]?a[n-u? + ] ?qato-ńar? düge-d ?nī?r?? ?qa?an ?türük ?qa?an


Morphemic analysis:

[?] qa?an-GEN? [?]? qatun-PLUR? Niri? qa?an? türk? qa?an



  1. The first aks?ara is unreadable.
  2. In spite of the poor preservation, the next word must be qa?an-u ‘qa?an-GEN’. It is possible that the next aks?ara indicated just -n, forming therefore, an alternative genitive in -un after qa?an, see more on this in the notes to line 2-9.
  3. qato- can be equated with EMM qatu (MNT §54, §55) and MM qatun (MNT §64, §104, etc.), (IM 444), (Mu 203) ‘qa?an’s wife, lady, noble woman’. -ńar is a plural suffix already seen above.
  4. düge-d is ‘younger brothers’, metathesized form of MM de’ü ‘younger brother’ (MNT §18, §46, etc.), (Mu 151), followed by a plural suffix -d.
  5. Niri qa?an of the Western Turkic Khaganate (r. 579-603/604) is a grandson of Bumin qa?an.
  6. The rest of the line is quite straightforward: türü?g kagan is, of course, türük qa?an ‘Turkic qa?an’.
  7. We again see typically Mongolic morphology in this line: besides the genitive case suffix -u (or may be -un) and plural suffix -ńar (used this time after an animate noun), we also encounter for the first time Mongolic plural suffix -d.


LINE 2-6


ruc dro? ta?ya?-ju?? xīr??? h?rgin ?bar-go[l]? pa?lxī?r [+]xa?c?? h?gb?j


Tentative transcription:

u? d[?]r? tayaju? kera erkin ?bar-?o[l]? pa?lxī?r [+]xa?c?? h?gb?-j


Morphemic analysis:

? law? worship-CONV? country? title/people? take-NML? ? ?? ??-PST



  1. u? is not clear. A reduction from EMM u?ir ‘chance, reason, cause’ (KMQB I: 21a.2, 21b.1, 25b.4, 26a.3, 28a.3) does not seem very probable. Even less probable would be a connection with WMM ū?a ‘back’ (Mu 96, 154, 371), (Ist 71), u?a (L 1255).
  2. d?r? is ‘law, rule, principle’. Cf. EMM t?re (MNT §208), t?r? (MNT §216). For the discrepancy between HT d- and EMM t- cf. also HT digin vs. tegin on line 2-1 as well as Manchu doro ‘id.’, Jurchen doron ‘imperial seal’, borrowed from Mongolic with initial d-, not t-.
  3. taya- probably can be explained as EMM tayi- ‘to worship’ (MNT §189). -ju is converbum contemporale that we have already seen several times.
  4. For h?rgin Maue suggested OT erkin ~ irkin ‘title’ (p.c.), discussed by Clauson (1972: 225) and in other sources. It might be further etymologically connected to OT erk ‘authority, free-will, independence’ (Clauson 1972: 22). This may indicate a further connection with EMM erkin (sing.) ~ erkid (plur.) ‘important, excellent, masterly’ (de Rachewiltz 2004.1: 415), attested in MNT §105, §109, §153, etc., but not used as a title in Mongolic. It is not inconceivable, on the other hand, that we have here MM irgen [irg?n] ‘people’ (with a metathesis of vocalism) (MNT §5, §8, §11, etc.), (HYYY 14b.1), (Mu 119, 123, 136, etc.). Both explanations have their pluses and minuses. Maue’s theory does not involve metathesis, but leaves -g- instead of -k- unexplained. Also, in the HT inscription titles have consistent plural marking when the plurality is meant. May be only one erkin is meant here, it is difficult to say, even if the next word is explained as ‘taker, collector’ (see below). Vovin’s explanation has a metathesis, but does not create problem with the consonant -g-. Also, irgen is never marked with plural.
  5. bar-go[l] is probably a cognate of MM bari- ‘to take, to collect’ (MNT §19, §25, etc.), (Mu 199, 103, etc.) + MM -’Ul (Godziński 1985: 53-54), Pre-Classical WM -?ul ~ -gül, a deverbal nominalizer designating occupations. Alternatively, this form can be interpreted as bari- + a cognate of MM -’Ul (Godziński 1985: 53-54), Pre-Classical WM -?ul ~ -gül, a causative suffix, followed by a zero-marked imperative. étienne de la Vaissière believes that it might be Barkul, the place of Niri’s defeat.
  6. The rest of this line is opaque, although ? in h?gb?-j can probably be equated with MM past tense in *-?i found in distant past ?U’U(y) and its feminine form ?i’ay (Godziński 1985: 131-132).
  7. Again we have some clearly Mongolic morphology here: ?u, –? and -gol.


LINE 2-7


tüg-jü ruka-ba-r-?a?r? kagan? xa?n??? ju?la?-ba? ?tu?n-u? ?tüg-ny?? tu?wa?


Tentative transcription:

tüg-jü uqa-ba-r-ńar? qa?an? qan-? ?ula-ba tün-ü ?tüg-n[V]y?? tuwa


Morphemic analysis:

be.enough-CONV realize-PST-NML-PLUR? qa?an? regnal.year-ACC? shine(?)-PST that-GEN? be.enough-??



  1. tüg- can be compared either with EMM tüge- ‘be enough’ (§92, §187) (cf. the similar elision of the second syllable vowel reflected by MM bari- : HT bar- on the previous line) or with MM tüge’e- ‘to give out, to distribute’ (MNT §213, §232), (Mu 358), possibly a derivative of tüge-. Converb -?U has previously appeared on lines 2-2, 2-3, and 2-6.
  2. uqa- ‘to realize, to know’ was discussed previously in conjunction with line 02. -bA is MM past tense suffix that appears twice on line 2-7. -r could be a nominalizer previously mentioned in the note 4 to line 2-4, although there might be a problem with this interpretation, since MM form in -bA is a final verbal form (Godziński 1985: 128-129), which has no adnominal function. But since the language of the HT inscription is not MM, I am going to propose this explanation, especially that adnominal forms in East Asia and Central Asia tend to acquire easily final predication function. Note also that Khitan past tense suffix -b?ń[6] can be used in the adnominal function (Shimunek 2017: 271). -ńar is plural suffix, which we have already seen on more than one occasion. Thus, uqa-ba-r-ńar is ‘realizations’ or ‘those who realized’.
  3. qa?an does not require any comments and xa?n- according to Maue is a ‘regnal year’ < Tumshuquese x?ana- ‘Zyklusjahr’ + accusative ?. Alternatively, it may be just qan ‘khan’ (Maue, p. c.).
  4. ?ula- can be possibly connected with MM ?ula ‘light, candle, torch’ (HYYY 10b.4), (HS 10.20), (Mu 169, 210, 325, 336), (IM 439), (Ist 39) (< WOT *?ula, cf. EOT yula ‘light, torch’),[7] which is a nominal root, but it is possible to suspect conversion here, so I tentatively interpret it as ?ula- ‘to shine’. Cf. MM to’o ~ to’a ‘number’ à to’o- ‘to count’.
  5. As for tün-ü, tün- must be cognate to EMM te’ün– (MNT §12, §116, etc.) ~ WMM tǖn- (Mu 94, 97, etc.) is the oblique stem of tere ‘that’. , is, of course, a genitive form after stems in -n (see the note 5 to line 2-2).
  6. -n[V]y? is a piece of morphology for which currently I have no explanation. The last word on this line is the enigmatic tu?wa? (see note 3 to line 2-4).
  7. As before, we see a number of typically Mongolic morphological markers: converb contemporale ?U, past tense suffix -bA, nominalizer -r, plural suffix -ńar, and genitive suffix -U after -n stems.


LINE 2-8


[+]xa?[] tu?[]? to[go?]-gun pügtig-ci? ?ī????-n bodi-satva ?to?ro?-x? kagan


Tentative transcription:

[ir]ge[n] tu?[]? to[?o]-?un pügtig-?i ?ińi-n? bodi-satva? t?r?-k? qa?an


Morphemic analysis:

people ?? count-NML ??-NA? new-GEN Bodhi-sattva ?be.born-NOM.FUT? qa?an



  1. The first two words on this line are unreadable, but the first one might be irgen ‘people’, judging from the context, if Maue’s identification of tu?wa? as tribal name Tupa is correct.
  2. to[?o]- can be compared with EMM to’o- ‘to count’ (MNT §161), WMM tō- (Mu 115). -?un must be cognate to MM deverbal nominalizer -’Un (Godziński 1985: 53), and Pre-Classical WM -?un ~ -gün (Poppe 1964: 46), which forms nouns designating qualities or abstract nouns.
  3. Enigmatic pügtig previously appeared on line 2-4 as pügt?g. Here it is probably followed by nomen actoris suffix -?i. The only information it provides us is that pügt?g ~ pügtig must be a noun.
  4. The rest of this line is identical to the second part of line 2-1 and the first word of line 2-2.
  5. Again we see several typically Mongolic morphological markers: nominalizer -?un, genitive case marker -n after vowel stems, possibly nomen futuri -k[ü], and nomen actoris -?i (which is, of course, a common Turco-Mongolic marker, but it remains unclear who borrowed it from whom).[8]


LINE 2-9


[+]l[][+]? []??yu? ?ruc bit?h?-? ??[ + + ] g[×-+] tu?wa? pa?r? kagan to?ro?-x? kagan-un


Tentative transcription:

[+]l[][+]? []??-yU u? bitig-iń? [ + + ] g[×-+] tu?wa? par? qa?an t?r?-k? qa?an-un


Morphemic analysis:

? ?-PRES? ?? inscription-GEN ?? ?? person? qa?an be.born-NOM.FUT? qa?an-GEN



  1. This line is poorly preserved. The first, second, fifth, and sixth words are unreadable, although -yU in the second word could be a suffix of deductive present MM -yU (Godziński 1985: 127), WM -yU (Poppe 1964: 92).
  2. On enigmatic u? see note 1 to line 2-6.
  3. bitig-iń is a genitive form of bitig ‘inscription’, which previously appeared as biti[g] on lines 2-1 and 2-4.
  4. On the enigmatic tu?wa? see note 3 to line 2-4. This word also appears on line 2-7.
  5. par is likely a cognate of EMM hara ‘person’ (MNT §246) ~ haran (MNT §39, §91, §100, etc.) ‘person’, ‘persons’, ‘population’ and WMM haran ‘id.’ (Mu 198), (Ist 31), (IM 431). Alternatively, it could be par ‘ten’ ‘cf. MM har-ban and Khitan par. Note that like in Khitan, and unlike in MM numerals in HT are not followed by the suffix -ban~ -’an < *-pan, cf. HT dolu ‘seven’ on line three
  6. qa?an does not require commentary and t?r?-k has been treated above in the note 5 to line 2-1. It also appears on line 2-8.
  7. The language of the HT inscription has alternative genitive -un for qa?an: qa?an-un, which is probably a result of secondary paradigmatic reanalysis, although it also could be a metathesis of a form similar to EMM alternative genitive qa?an-nu (MNT §1, §48, §51, etc.).
  8. Again several clearly Mongolic morphological markers are present: genitives -un and -iń, deductive present -yU, and possibly nomen futuri -k[ü].


LINE 2-10


pa?da?-?? n<ī?>r??? kagan? türüg? [ka]ga[n]? []ī?j??-n rubī-j? ja?lo?ba?-j? darka-d ?ja?y b??


Tentative transcription:

pada- Niri? qa?an? türüg? [qa]?a[n]? [k/g]ī?i-n ubī-? ?alo-ba-?? darqa-d ??ay bi


Morphemic analysis:

be.cut-(?) Niri ?qa?an? türk? qa?an ?approach/follow-CONV? ?-PST? direct-PST-PST? happy? be



  1. There is a lacuna after pada-. There is WM ada- ‘to be stubbed, to be cut’, but as far as I can tell the word is not attested most importantly either in MM (therefore, we cannot confirm the presence, or the absence of an initial h- < *p-) or in WM prior to seventeenth century. Therefore, this identification remains dubious.
  2. We have already met Turkic qa?an Niri in line 2-5, so Niri qa?an türüg [qa]?a[n] does not require any commentary.
  3. []ī?i-n looks like a verbal form ending in converbum modale -n, but establishing the identity of the root might be difficult, although MM [g]i?i- ‘to follow’ (MNT §253), [k]i?i- ‘to approach’ (MNT §195) seem to be the only candidates.
  4. The root ubī- does not have any parallels, but on distant past ? see the note 6 to the line 2-6.
  5. ?alo-ba-? may consist of the root ?alo- (cf. WM ?alu- ‘to direct’, ‘to make straight’[9]), past tense -bA, and distant past -?i.
  6. darqa-d is a plural form of darka-n ‘free man, official’.
  7. ?a?y looks like a cognate to EMM ?aya’an ‘happy, blessed’ (MNT §66, §194, §248), where -’an is likely to be a suffix.
  8. bi is a cognate to MM büi ‘to exist’ (MNT §18, §35, §44, etc.), (Mu 105, 110, 139, etc.).
  9. There are several typically Mongolic morphological markers on this line as well: converbum modale -n, distant past -?, past -bA, and plural suffix -d.


LINE 2-11


[?? ]run bitig? [+]s?g?? pag [ +? +? ] j[] [ +? +? ] darka-n? b[]?? t?? ba?? ka


Tentative transcription:

[e’]ün bitig? [+]s?g? pag ?[ +? +? ] ??[] [ +? +? ] ?darqa-n? b[i]ti-be? qa


Morphemic analysis:

this inscription? ?? ?? ?? ?? ?? official-SING? write-PST? ?



  1. The last line is in a very poor state of preservation, but it looks like a signature of a scribe.
  2. I substitute e’ for the first lacuna. The resulting form is e’ün-, an equivalent to EMM oblique stem e’ün- of the demonstrative pronoun ere ‘this’. Likely, this is a reduction of e’ün-ü, a genitive form of ere.
  3. bitig is ‘inscription’, see also lines 2-1, 2-4, and 2-9
  4. darqa-n is singular form of darqa- ‘official, free man’ that we have already seen as plural darqa-d on line 2-10.
  5. biti-be is the past tense form of biti- ‘to write’. The rest of the line is undecipherable at the present moment.
  6. Only two Mongolic morphological markers appear on this line: singular -n and past tense suffix -bA.



1-3. qa?an [and] prince [of] the inscriptions.[10] When qa?an, who will be [re]born as a new Bodhisattva, knows lord Buddha knowledge, and promises to the country’s Ańaqay [title], begs, stand for the tribe, and promising seven [times] … 4. Looking at the inscription stones, Tupa [people], … in order to stab … joining … 5. … qa?an’s wives [and] younger brothers, [and] Niri qa?an, qa?an [of] Türks 6. worshiped the Law, and country’s erkins and collectors … 7. is enough and there were enough of those who realized that qa?an’s regnal years were shining. Tupa 8. people … counting … qa?an who will be [re]born as a new Bodhisattva 9-10. … of the inscription … Tupa persons being cut from the qa?an who will be [re]born as a qa?an, [they] followed Niri qa?an, qa?an of Türks and … [He] directed [them]. Free men were happy. 11. Official … wrote this inscription …



marker Hüis Tolgoi MM Pre-Classical WM Khitan
genitive after -n stems -U ~ -Un -U ~ -nU -U -en
genitive after consonantal stems -Un ~ -iń -Un -Un -un, -en,
genitive after vowel stems -n -n -n -n, -on, -un
accusative -? ~ -i -i ~ -yi -i ~ -yi -?
plural suffix -ńAr -nAr -nAr -ńer ~ -ńeń
plural suffix -d -d -d -d
singular suffix -n -n -n
nomen actoris -?i -?i -?i
converbum modale -n -n -n
converbum contemporale -?U -?U -?U -? ~ -?
converbum finale -rA -rA -rA
nomen futuri -x -qu ~ -kü -qu ~ -kü
past -bA -bA(i) -bA(i) -beń
distant past -? *-?i -?uqui ~ -?üküi
deductive present -yU -yU -yU
nominalizer -?uń ~ -?un -’Un -?un ~ -gün
nominalizer -r -r -r
nominalizer -?ol -’Ul -?ul ~ -gül
functionally unclear verbal suffix -n[V]yA-


I believe that the chart above clearly demonstrates that the language of the HT inscription shows much more features common with mainstream Mongolic than with Khitan. This also makes sense geographically, since the HT inscription comes from the territory much more to the North than the home territories of Serbi and Khitan were.

Certainly, we have sufficient lexical difference between the language of HT and the first monuments in MM. But this also should come as no surprise, given the fact that more than half of a millennium separates these texts, as well as the hard-core reality (sorely missed by Nostraticists and other long-rangers) that the lexicon represents the most unstable part of any language.



I totally agree with Maue that in spite being a bilingual text, this is a much harder nut to crack than the HT inscription. Although it is written in the same Brāhmī script, I see no similarities or no commonalities with the language of HT inscription, except may be for a converb contemporale -?U, but one similarity proves nothing. The inscription is also much shorter than the Sogdian text, so it is not a Rosetta stone by any standards. I would not be as pessimistic as Maue to call the decipherment completely hopeless. I largely suspect that the language of the non-Sogdian part of the inscription might be Ruan-ruan but I would not attempt its decipherment before I finish digging out bits and pieces of this enigmatic language from the OT sources and acquiring a good working knowledge of Sogdian that I currently do not have. The latter necessity is due to the fact that any readings of the Sogdian part of the inscription provide solely the translation, but not the morphemic analysis, which is the crucial helping part in the decipherment of any unknown bilingual portion.



I trust that the most innovative and revolutionary linguistic part of our presentation is that the first ‘Altaic’-type language attested in the steppe by a continuous text is not OT, but Mongolic. Apart from the fact that it now places Mongolic as a textually attested language in the late sixth or early seventh century and makes it a contender among ‘Altaic’-type languages with Korean that has the first continuous text attributed to 594 or 596 AD. For the sake of the comparison, the first Old Japanese text is dated only by 697 AD, and the first OT text by 713 AD. Tungusic Jurchen texts are not attested before the twelfth century.

This discovery can also potentially reverse, at least in some cases, the traditional point of view that the directionality of borrowing was always from Turkic to Mongolic, as some of them will have now more certain Mongolic origin.

Finally, it appears that the same language as in the HT inscription appears in the same script, and more importantly in the same language in much shorter Keregentas inscriptions from Eastern Kazakhstan. What polity or what tribe/nation might have possibly spoken this language is the question that should be answered by historians.



Abbreviations of languages and grammatical terms

CONV??????????????????????????? Converb

EMC????????????????????????????? Early Middle Chinese

EMM???????????????????????????? Eastern Middle Mongolian

EOT?????????????????????????????? Eastern Old Turkic

GEN????????????????????????????? Genitive

IMP?????????????????????????????? Imperative

NA???????????????????????????????? Nomen actoris

OT???????????????????????????????? Old Turkic

PLUR???????????????????????????? Plural

PRES???????????????????????????? Present

SING???????????????????????????? Singular

WMM??????????????????????????? Western Middle Mongolian

WOT????????????????????????????? Western Old Turkic



Primary sources


Hsiao Hui-lian.Can?????? Inscription on the canopy of Hsiao Hui-lian muzhi ming, 1080 AD

Langjun?????????????????????? Langjun inscription, 1134 AD

Yelü Jue?????????????????????? Yelü Jue muzhi ming inscription, 1071 AD

Yelü Pusuli????????????????? Yelü Pusuli muzhi beiming inscription, 1105 AD

Yelü Xiang-wen?????????? Yelü Xiangwen muzhi inscription, 1091 AD


Middle Mongolian

HS?????????????????????????????? The edict of Hai-shan, 1305 AD

HYYY???????????????????????? Hua-yi yi-yu, 1389 AD

IM?????????????????????????????? Ibn Muhannā vocabulary, 14th c. AD

Ist??????????????????????????????? Istanbul vocabulary, 14th or 15th century AD

KMQB???????????????????????? Kitad Mong?ol Qaril?a?an-u Bi?ig (Sino-Mongolian Documents), 14th c.

L???????????????????????????????? Leiden vocabulary, 1343 AD

MNT?????????????????????????? Mong?ol niu?a tob?a’an, ca. 1240 AD

Mu????????????????????????????? Mukaddimat ‘al-‘Adab, 14th century AD

Qub IV[11]????????????????????? Fourth edict of Qubilai qa?an, 1280 AD


Secondary sources

Clauson, sir Gerald 1972. An Etymological Dictionary of Pre-Thirteenth-Century Turkish. Oxford: Clarendon Press.

Godziński, Stanis?aw 1985. J?zyk ?redniomongolski. Warszawa: Wydawnictwa Uniwersytetu Warszawskiego.

Kane, Daniel 2009. The Kitan Language and Script. Leiden & Boston: Brill.

Poppe, Nicholas 1957. The Mongolian Documents in ?P‘ags-pa Script. Wiesbaden: Otto Harrassowitz.

Poppe, Nicholas 1964. Grammar of Written Mongolian. Wiesbaden: Otto Harrassowitz.

de Rachewiltz, Igor 2004-2013. The Secret History of Mongols. Translated with a Historical and Philological Commentary. Vols. 1 & 2 (2004), vol. 3 (2013). Leiden & Boston: Brill.

Róna-Tas, András & árpád Berta 2011. West Old Turkic. Turkic Loanwords in Hungarian. 2 vols. Wiesbaden: Harrassowitz.

Shimunek, Andrew 2017. Languages of Ancient South Mongolia and North China. Wiesbaden: Harrassowitz Verlag.

Tumurtogoo, D. 2006. Mongolian Monuments in Uighur-Mongolian Script (XIII-XVI Centuries). Taipei: Academia Sinica.

Tumurtogoo, D. 2010. Mongolian Monuments in ’Phags-pa Script. Taipei: Academia Sinica.

Vovin, Alexander 2003. A Reference Grammar of Classical Japanese Prose. London: Routledge/Curzon.

Vovin, Alexander 2005. A Descriptive and Comparative Grammar of Western Old Japanese. Part 1: Phonology, Script, Lexicon, and Nominals. Folkestone: Global Oriental.

Vovin, Alexander 2013. ‘Old Turkic Loanwords in the Khitan Language’. In: Yal?m Kaya Bitigi. Festschrift for Prof. Osman Sertkaya. Ed. by Hatice ?irin User and Bülent Gül, pp. 621-25.

Wu, Yingzhe & Juha Janhunen 2010. New Materials on the Khitan Small Script. Folkestone: Global Oriental.

[1] There was an unprecedented growth of Khitan studies in the West alone, with three significant monographs published in the recent years: Kane 2009, Wu & Janhunen 2010, and Shimunek 2017.

[2] The first number indicates the number of the stone, and the second the number of a line on it.

[3] Note that the onset ń- of this suffix is closer to Khitan, while the vocalism is identical with Mongolic.

[4] Shimunek glosses this word as ‘external clans’ (2017: 265), but the plural would probably work only if Khitan -d - is here a plural suffix.

[5] Shimunek’s hypothesis that there must be one based on various look-alikes in the neighboring languages (2017: 406-407) is, of course, nothing but a fantasy, unsubstantiated by evidence.

[6] Kane reconstructs -bo.ń ~ -b.uń ~ -bun (2009: 147).

[7] Possibly from Turkic *yul- ‘to catch fire, to be ignited, be kindled’ (Róna-Tas & Berta 2011: 409).

[8] Since the HT inscription is more than one century older than the earliest OT texts, it is likely that the directionality of borrowing might have been Mongolic à Turkic, and not the vice versa.

[9] Not attested in Pre-Classical WM (Tumurtogoo 2006) or in MM.

[10] This is likely to be a carry-over from the first stone.

[11] Numeration is given according to the edition by Tumurtogoo 2010, not Poppe’s edition of 1957, because many more documents in ’Phags-pa were discovered after Poppe’s edition.

文章出处(来源):   德力英蒙研究 微博
分类目录: 教学科研 总浏览:14,827