ᠬᠦᠢᠰᠦᠲᠣᠯᠣᠭᠠᠢ ᠶᠢᠨ ᠪᠢᠴᠢᠭᠡᠰᠦ — ᠲᠡᠮᠳᠡᠭ ᠪᠣᠯᠣᠨ ᠠᠪᠢᠶ᠎ᠠ （ᠠᠩᠭ᠍ᠯᠢ ᠬᠡᠯᠡ ᠪᠠᠷ）

Prefatory remark The folowing article is a draft version which was presented on the ocasion of the PIAC 2017, Székesfehérvár (Hungary) within a panel on the Khüis/Hüis Tolgoi inscription. The panel included also contributions by – M. ÖLMEZ, On the discovery, the whereabouts, condition of the stones, and our expedition, – E. de la VAISIÈRE, The historical context to the Hüis Tolgoi inscription (uploaded on academia.edu), and – A. VOVIN, Interpretation of the Hüis Tolgoi inscription (discused sub: https:/ww.academia.edu/s/07da0d7bfb/interpretation-of-the-huis-tolgoi- inscription?source=news ). * Schwarzenborn 16 Sep 2017 © Dieter Maue 2017 The Khüis Tolgoi inscription - signs and sounds Dieter Maue 1 Generals 2 The sign inventory 3 The sound inventory 4 Apendices I Signs and sounds II Transliteration III Transcription 1 Generals The inscription is written in Brāhmī, the mother of al later Indian scripts and their derivatives , e. g. the Tibetan script, and the basis of five varieties which were used along the Silk roads for Khotanese, Tocharian A and B, Tumshukese, Sogdian and Uigur. The sixth variety was created for the Brāhmī inscriptions of the Mongolian steppe and their language or languages. As to the language of the KhT inscription, it appeared in the course of deciphering that we are most likely concerned with a (Para-)Mongolian language (: “(Para-)Mongolian hypothesis“) as will be shown in the following contribution by Prof. A. Vovin. The (Para-)Mongolian hypothesis exercised some controling efect on the deciphering in the narrow sense (i.e. on the identification of signs and definition of their value). To name two isues: a very special one: Initial r- should not occur in the inscription; a general one: The 2 underlying language should show front-back vowel harmony which might also be reflected in the sign system. col. 11 10 9 8 7 6 5 4 3 2 1 Fig. 1: Khüis Tolgoi inscription (detail) The Khüis Tolgoi inscription surprises with a couple of features which were unseen before in combination with the Brāhmī script: (1) The script is writen from top to botom 1 in columns which run from right to left. (Fig. 1) (2) The text is segmented by horizontal strokes mostly into words; ocasionaly morphemes are separated (regularly the plural morpheme); once a compound is di- vided into its constituents (bo dä| sa-t va). (3) The graphematic entity is the sylable, not the akṣara. 1 Traces of verticaly writen Brāhmī are known from coins of the late and post-Kushana period and from coins of the Gupta Empire on which the rulers are depicted in standing position accompanied by their names. While other parts of the legend are circularly aranged around the figure, the names are vertically written which was forced by the available space; for specimina s. under coinindia.com. 3 2 The sign inventory The KhT Brāhmī alphabet 2 is composed of a selection of signs taken from the basis alphabet and half a dozen additional signs. 2.1 The signs from the basis alphabet Conso- nants Occlusive Fricative Semi– vowel Nasal Liquid Vibrant unvoiced voiced un– voiced voiced un– asp asp. un– asp. asp. Labial p p h b b h v m Dental t t h d d h s n l r Retroflex ṭ ṭ h ḍ ḍ h ṣ ṇ Palatal c c h j j h ś y ñ Velar k k h g g h ṅ Laryn– geal h Vowels a ā i ī u ū e o Fig. 2: The KhT signs (high-lighted) taken from the basis Brāhmī alphabet 2.2 The consonant signs The table (Fig. 2) shows the basis alphabet; the elements ocuring in KhT are high- lighted (yellow). It is obvious that the KhT alphabet was deliberately and systematicaly organized. Unservicable signs of the basis alphabet such as retro- flexes and aspirates were dropped. The asymmetry of the velars (g but no k) is strange and most probably by chance as well as the lack of m and perhaps that of ŋ. 2 I fel fre to use the designations 'alphabet' or 'sylabary' for the list of basic signs though both expressions do not adequately describe the character of the Indian script: “Various terms have been suggested for this type of script There is as yet no comonly acepted term.“, R. Salomon, Indian epigraphy. New York, Oxford 1998: 15. 4 2.2.1 The special consonant signs On the other hand, there were obviously some sounds in the KhT language which could not be adequately represented by any of the basic signs. Therefore four special consonant signs were added, s. fig.3. No. 1 No. 2 No. 3 No.4 Fig. 3: The special consonant signs 2.2.1.1 Special signs nos. 1 and 2 ! Fig. 4 Special Fig. 5: Keregentas and Taraz (© Gogle Earth) signs nos.1 and 2 Signs nos. 1 and 2 form a sign group (Fig. 4) which ocurs 12 or 13 times in the inscription. The determination of their sound value was crucial for the decipherment. It suceded only through Brāhmī inscriptions which were discovered by the Kazakh scholar Eskander Baitenov: two inscriptions together with 5 a Bactrian fragment hail from the Keregenras valey and a third one from Taraz. (Fig. 5) © E Baitenov Fig, 6: The Keregentas Brāhmī inscriptions The two Keregentas inscriptions consist of four signs each (Fig. 6): two of them are the same in both inscriptions (in the read box), the other two are different from each other. The stones most probably served as balbals; thus the inscriptions should represent the name of the kiled enemy folowed by his title 3 as Baitenov sugested. All other posibilities eliminated, the best and only option was the reading “Kagan“. Thus the rough value of the two special signs was determined: no. 1 - transliterated as k with stroke above, k$ - stands for an unvoiced velar k and no. 2 - transliterated through g 1 – for its voiced partner. It appeared that the Keregentas sign group had direct counterparts in KhT and Bugut. The sign g 1 a-n is also visible on the Taraz stone. (Fig. 7). 3 Cf. the Ongi balbal with the inscription: išvara tarkan balbalı, cf. e. g. M. Ölmez 2012: 191. 6 ! ! (a)Keregentas (b) Khüis Tolgoi (c) Bugut (d) Taraz © E Baitenov © E Baitenov Fig. 7: “Kagan“ The heuristic value of the discovery is considerable: (1) It is relevant for the further deciphering that words preceding kagan are principaly candidates for names. (2) The KhT inscription is certainly a secular text and the Brāhmī part of the Bugut inscription very likely so; the latter is doubtlessly no Sanskrit sūtra as was generaly suspected. (3) One and the same Brāhmī variety was in use al over the territory of the First Turkic Kaganate. (4) The facts that (a) in the Bugut bilingual the main inscription was in Brāhmī script and non-Sogdian language and (b) that in Keregentas a Bactrian inscription existed beside the Brāhmī inscriptions - theses facts sugest that the Brāhmī script was used for writing the imperial or official language while the Bactrian and Sogdian scripts and languages had the status of “scripturae francae“ and “linguae francae“ in the Western and Eastern part of the First Kaganate resp. (5) The impact on the history of Brāhmī script neds a digression: (a) A closer lok to the signs of Fig. 8 reveals that k“a and g 1 a of KhT difer from those of Bugut and Keregentas. In KhT, the thre horizontal elements of g 1 a are not conected; k*a has the headstroke in common with the Bugut and Keregentas forms, but the body is shaped like KhT g 1 a. Clearly, the KhT signs show a systematic reshaping; they could continue the Bugut-Keregentas forms, but hardly vice versa. Thus the KhT script must be palaeographicaly later than that of Bugut-Keregentas. 7 ! (a) Khüis Tolgoi (b) Bugut (c) Keregentas ©E.Baitenov (d) Uigur ©BBAW ! (a)Khüis Tolgoi (b) Bugut (c) Keregentas ©E.Baitenov (d) Uigur ©BBAW Fig. 8: The signs (supra) and (infra) (b) Perhaps even more exciting is the second point. Hitherto it was believed that the coresponding Uigur Brāhmī signs (Fig. 8) were created after the Uigurs had left the steppe (A.D. 840 +) and moved to the o o area where they adopted the Brāhmī script under the influence of the Tocharian and Tumshukese varieties. However, the striking similarity to the Bugut-Keregentas signs teachs us that Uigur k“a and g 1 a were actualy inherited. In other words, the Stepe Brāhmī must have survived somewhere for centuries before it reapeared in the Uigur monasteries around the Turfan oasis. (c) From (a) we suggest that the KhT inscription postdates the Bugut in- scription (c. A. D. 582). From (b) we conclude that there is no terminus post quem non derivable from the palaegraphical data. For a more acurate dating of KhT one needs other evidence. We now return to the KhT velars proper. We have seen that the velars of the basic alphabet were adopted; at least g is attested. The question arises: what are the addi- tional velars for? The answer comes from other Mongolian and Turkic writing sys- tems: it is for the diferentiation of front and back velars which are/were distinct to such an extent that diferent signs are/were desirable to encode them apropriate- ly. The sensibility for this difference may be especialy high in speakers of languages with front-back vowel harmony. The basic signs were used for the front velars and the new ones for the back velars. Another question concerns the phonetic value, whether the back velars k and g were stops or fricatives. In Khüis Tolgoi, they occur in some titles: kagan (the g only there), kato(n) = katun(?), darkan with the plural darkad, and in añakay, possi- bly a name. As a proper name, Añakay is atested elsewhere in Chinese sources and once in Greek 4 , while the titles are frequently met with in various languages and 4 S. E. de la Vaisi re, The historical context to the Hüis Tolgoi inscription. Academia.edu. note 5. 8 scripts so that there was some hope to get relevant phonetic informations from there. In view of the limited time I have to make the long (and rather disappointing) story short: g 1 is almost certainly the representative of the voiced back velar frica- tive, while for the unvoiced counterpart k both options, stop or fricative, remain open without further evidence which can perhaps be provided in the next section. 2.2.1.2 Special sign no.3 (a) Special sign no. 3 (b) tpa (c) kṣa (d) TS no. 12: ša Fig. 8: Special sign o. 3 and (aleged) relatives The sign, s. Fig. 8a, is atested as syllable initial or syllable final (s. Fig. 9) and occurs in the begining, midle or end of words. syllable initial: (a) 7 3a ni (b) 2; 6 3ä 1 rä (c) 6 + 3a cä syllable final: (d) 3 ja-3 (e) 1; to ro-3 (f) 2 u qa-3 Fig. 9: Words with special sign o. 3 At the very begining of the decipherment it appeared to be simply the liga- ture tp-, s. Fig. 8b. But that resulted in monstrous forms like ?ukatp, ?tpani and was therefore son abandoned. Next, the reading kṣ or š was guesed, the later because there semed to be a genetic connection with the basis ligature kṣa (Fig. 8c) or the Tumšukese special sign no. 12 (Fig. 8d), which is, however, more developed and has got an aditional tail. With the equation no.3 = š two words of the inscription sugested themselves as foreign words: 9 (a) šani in the sense of 'regnal year', a wanderwort (wandering word) which is met with in several Central Asian languages 5 ; (b) ši 1 ri 'country', Tumš. šera-, Khot. kṣīra- id. Both concepts, 'regnal year' and 'country', being unfamiliar to the stepe people, might have ben adopted together with the foreign words. But ši 1 ri must be ques- tioned in as far as it is limited to the Sakan languages; the Sakans played no such do- minant role in the Tarim basin that their concept of 'country, kingdom' might have been attractive for being borrowed. When the (Para-)Mongolian hypothesis was accepted, the value of the sign had to be modified again, because the native words (Fig. 9d) ja-3 and (Fig.9 f) uka- 3 could hardly end on š. Tentatively, the fricative x was guessed instead: jax and ukax could then be analysed as deverbal derivatives on -k(u) which was developed to -x in final position. 6 It is questionable whether xani (instead of former šani) and xi 1 ri (instead of former ši 1 ri), dubious as it is (s. above), are still eligible as the above supposed for- eign words. A front allophone is probable though not demonstrable. (Here one might remind of the double realisation of the velar fricative in Standard High German in the words “ich“ and “ach“.) A consequence of the identification of no. 3 as velar fricative would be that the back velar k must have ben a stop. 5 KS kṣuṇa- 'section, period of time' (DKhS 68a), TS xšana- 'Zyklusjahr' (thus Konow 1935: 50), KrorPkt kṣuna-, KharInscr. kṣun/ṇa-, TochB kṣu 'regnal year' (Adams DTB 2 261: “It is likely that the three Tarim languages KS, TS, TochB,DM have borrowed this word, directly or indirectly, from Bactrian, either as a result of Kushan (10 BC to 20 AD) influence or influence of the later Hephthalites (first half of the sixth century AD)“) Bactr. (a) ono “'(calendar) year, (regnal) year', posibly loanword from Gk. (A. Thierfelder apud Humbach 1966, p. 24)“ (N. Sims-Wiliams, BactrDoc I ). 6 That would not touch its status as phoneme which is proved by the opposition k vs. x in the beginning of the words, e. g. x(ani) vs. k(agan). 10 2.2.1.3 Special sign no. 4 (Khüis Tolgoi inscr.) ( DTA TS 5 r1, © BAW ) ( DTA bs7 p.2 l.2 and “Eig. Zähl.1“ v3 © BAW) (a) Special sign no.4 (b) tu 4a (c) (d) v1a (e) v1e- Fig. 10: Special sign o.4 and its (aleged) relatives The sign under discussion (Fig.10a) is atested thre times in one and the same word tu 4a . Being without an exact counterpart in any Brāhmī variety, it was thought to be related to one of the two signs, (1) the long sylabic l (Fig. 10c) or (2) v 1 , the bilabial w (Fig. 10d-e). The first candidate, the long sylabic l, l̥̄, 7 is an invention of the Indian gram- marians. It ocurs in grammatical treatises or alphabet charts, but never in natural language texts. Excepted are some instances in the Uigur Brāhmī. There, it is used for lı or, exceptionaly, for li, both usually spelled as li, lī . Thus the long syllabic l, l̥̄,, of he Uigur texts turns out to be a superfluous and sophisticated orthographi- cal variant for which there was no other purpose than to show the scribe's intimate knowledge of the subtleties of the Indian script. In a perspicuously and economical- ly organized script like that of the KhT inscription, such extravagances are unthink- able. The second candidate is a non-Indian sign which ocurs in Tumshukese, Sog- dian and rarely in Uigur Brāhmī; transliterated by v 1 , it stands for the bilabial w. 8 The sign shows several variants 9 . Closest to the form of our inscription are two sam- ples from Sogdian mss which are reproduced in Fig. 10d and e. The lower part in KhT could be an attached -r. The reading of the word Fig.10b is therefore tuwa or tuwar. A second bilabial beside the basic si