In Russian
Contents Huns
Contents Türkic languages
Literature Index
Ogur and Oguz
Alans and Ases
E.Pulleyblank Eastern Hun Language
O.Pritsak Onomasticon of Western Huns W.B.Henning Xiongnu are Huns
L.Gumilev Language of Huns
Russian Version needs a translation
Kisamov N. The Hunnic Oracle
Tekin T. Hsiung-Nu Language
Alan Dateline
Avar Dateline
Besenyo Dateline
Bulgar Dateline
Huns Dateline
Karluk Dateline
Khazar Dateline
Kimak Dateline
Kipchak Dateline
Kyrgyz Dateline
Sabir Dateline
Seyanto Dateline
Alexander Vovin
(University of Hawai'i at Manoa)
Did the Xiong-nu Speak a Yeniseian Language?

Central Asiatic Journal, Edited by Giovanni Stary, 44 (2000) 1, CAJ 44/1 (2000), Harrassowitz Verlag

Sequell to E.Pulleyblank 1962 opus
“The Consonantal System of Old Chinese”, Asia Major 9: 58-144, 206-265
Appendix “ The Hun (E.Pulleyblank: Hsiung-nu) Language”, pp. 239-265




Like in politics, the only way in science to find out the truth is to listen to the opposition. The opposition may not be right, but without it you would think that the washed story advocated by certain groups is the only true reflection of reality. That the story changed in the last 100 years, there is no doubt. The initial players, like Türkic Scythians, dissipated into a blue smoke, and in their stead came an army of Proto-Indo-European/Semitic speakers that in the books and encyclopedias replaced indigenous population, indigenous languages, and got barely stained with a smattering of the old deities and indigenous vocabulary. In case when Indo-Europeans stubbornly do not fit the bill, a little known speck is pulled from the hat, and is filled with hot air to reach gigantic proportions. This is a story of Kurgan people, a baby of long-abandoned theories reincarnated by M. Gimbutas, this is a story of Ossetian Scythians, a baby of Vs. Miller and V.I.Abaev, this is a story of Ket Huns, a sickly baby of E.Pulleyblank and now A.Vovin, and some prominent others. Since no trickery can substitute for the facts, a best medicine is fresh air, the light of day always bleaches. This posting is to popularize the concept, the methods, and conclusions on the “Ket Huns”, which, like the Tokharians before them, turned out to be a misnomer, and also need be renamed to “Pumpokol Huns”, “Proto-Yeniseian Huns”, or to a safer generic “Yeniseian” (aka Enisean) Huns. We've seen what is being served at the table, now is a chance to peek into the kitchen. As the saying goes, if you've seen the cooking, you won't do the eating.

The suggestion of E.Pulleyblank raised a slew of criticisms, and forced him to abandon the idea. Possibly in light of that, A.Vovin advances more gingerly, proceeding in stages of debunking, disbanding, embarking. The first part demonstrates scientific honesty and inspires confidence for the second part. After a review of phonological arguments, he soundly dismisses E.Pulleyblank's argumentation and phonology as a viable tool, disbands the rest by gingerly concluding that E.Pulleyblank's lexical fishing is insufficient to establish the same very phonology that is useless as a tool, and embarks on new reincarnation of the E.Pulleyblank's paradigm from the platform just disbanded, but with a morphological twist. Using little-known and obscure bunch of languages, with scant records and a mass of theoretical reconstructions based on unsure assumptions, A.Vovin manipulates phonetics into construction of 2 (two) words with presumed morphology. Having assumed a priori a particular academic phonological and developmental scheme, and ignoring the facts on the ground, A.Vovin proclaims the 2 reconstructions being converging, his reconstructions of the words consistent with the other reconstruction, and the 2 reconstructed words confirming “the genetic affiliation” of the Xiong-nu discovered by Pulleyblank, “that the Hsiung-nu spoke a language of the Yenissei family”. Even buying in to the mass of all five (5) words found by E.Pulleyblank that A.Vovin recites, plus the mass of all two (2) words invented by A.Vovin, the total aggregate number constitutes only 2.5% of the claimed lexicon of 278 Hunnic words from Shi ji, Han shu, Hou Han shu, and Qin-shu assessed by E.Pulleyblank, “some” of them with real explanations that do not match the Yenissei paradigm and consequently left undisturbed. A closer account of questionable words, without the A.Vovin's manipulations, would leave just 1 (one) word linked to Enisean, or 0.4% of the claimed Hunnic lexicon. It is incredible that either 2.5% or 0.4% of the words should have survived representing a common ancestor so remote that 97.5% or 99.6% of the vocabularies of these languages are now entirely different. Most likely, the E.Pulleyblank's and A.Vovin's cognates have nothing to do with the Hunnic language, and are chance, or artificial aberrations created during phonetical reconstructions. Or following S.Starostin/A.Dybo concise diagnosis, “no paradigmatic morphology has ever been reconstructed for language families older than Indo-European, no matter whether they are commonly accepted (like Afroasiatic) or debatable, like Altaic. What is occasionally dressed up as “paradigms” on close inspection turns out to be a subjectively arranged set of isolated and often questionable comparisons of select morphological markers that prove little, if anything.” I bet anybody a six-pack of decent beer with a free shipment that among 278 Hunnic words will be more then 1 phonetically similar word in any living language of the world, for the chances see Statistical Linguistics. Random Coincidences.

Philology starts with basic universally applied rules. One is that no language changes occur at exactly uniform rate during the whole period of its existence. The other is that change rate cannot safely be projected backwards to determine the date at which cognate languages parted from a common ancestor and became separate languages. A third is that written data preserved in the script of foreign languages inevitably reflect phonological modifications (recall how “Route du Roi” became “Rotten Row” when filtered through English ears). All of these were breached in the embarking stage of the A.Vovin study. Applying a uniform change from Archaic Chinese to Old Chinese to Modern Chinese is a fallacy. The change rate was presumed to be uniformly applicable backwards, another fallacy. The records of Old Chinese were presumed to perfectly reflect the foreign language, another fallacy. Each one is turning the study into a dead duck, even without the difficulties specific to the Chinese language, which does not have grammatical apparatus to express phonemes ö, ü, r, or forms inherent to agglutinative morphology, and requires to consider all possible allomorphs. The -ng ending, endemic to Chinese lexemes, are used to emulate foreign syllables with regular -n, the vocalism of Chinese is a peculiar and distinct property, it may not catch important vocalism of the original, or create false vocalism in the emulation. An accurately transmitted agglutinative phrase in English is 1.5-3 times longer, requires many more words, and in Chinese may not be transmittable at all, reducing a formed agglutinative phrase to a pidgin format. A.Vovin study was received by some with enthusiasm not expressed or warranted even by the author, but it is clear that the rumors about the end of the traditional story are greatly exaggerated.

* * *

The posting's notes and explanations, added to the text of the author and not noted specially, are shown in parentheses in (blue italics), or in blue boxes.

Alexander Vovin
Did the Xiong-nu Speak a Yeniseian Language? 1
Until the early 1960s it was generally assumed that the Xiong-nu were ancestors of either Turks or other Altaic peoples (such as Mongols or Tunguses).
The very first phrase hits a false note. There were no assumptions, it is a certified and evidenced fact, based on solid foundation of:
1. Chinese annalistic association of the Huns and the Türks and the Se/Saka;
2. European and Asian records associate Türks with Huns and Scythians
3. Byzantine chronologies identify Türks with Huns and Scythians (G. Moravcsik, Byzantinoturcica II, p.236-39, numbers are astronomical)
4. Türkic genealogical legends recorded in the Middle Ages identify Türkic forefather Oguz-khan with Hunnic Maodun
5. Ethnological identity of the Huns and Türks down to minor details, not matched by any other ethnicity
6. Archeological identity of the Huns and Türks uncovered during previous century, such as Kurgan burial tradition, culture and economy
7. Biological-related anthropological identity, such as identical racial mixture of Caucasoid and Mongoloid types, practiced cranial deformation, “robust” osteology
8. Biological-related identical fare of milk-and-meat diet, incompatible with lactose intolerance widespread in foot hunter-gatherer and grain-based societies
9. Similar Eurasian geopolitical spread
10. Identical socio-economical tradition based on horse husbandry, high mobility, traceable genetically connected weaponry and military tactics
11 Identical habit of reining over vast multi-ethnic sedentary peoples with agricultural and foot hunter-gatherer economies
and, in post-1960 period,
12. Genetical evidence on Huns and Türks genetic affiliation, corroborative, and not conflicting, with previous annalistic, legendary, and archeological knowledge
13. Improved philological understanding of direct Chinese records on Hunnic linguistic evidence
14. Philological progress on pre-Hunnic, Hunnic, and Türkic borrowings in the lexicons of the peoples bordering on Inner Eurasia
15. Philological genetical connection of Türkic and Hunnic lexicon and morphology

No annalistic record whatsoever identifies Türks with Tunguses, Koreans, Japanese or other supposedly Altaic people, with the exception of Türkic genealogical legends that include Mongols as a parallel branch with the Türks. To advance a credible alternate, a credible explorer must debunk all other evidence as a preamble. As A.Vovin taught us, “One should not forget that every language has a history, and that this history is intimately connected with the cultural and sociopolitical history of the people who speak a given language.” (A. Vovin, 2005, The End of the Altaic Controversy//Central Asiatic Journal 49.1, p. 75) Truly, one should not forget.

One also should not forget that Kets (or Eniseians) are remarkable not only for their language-isolate, totally incompatible with all other languages and with the historical impact by the Huns that generously sowed its traces across Eurasia, but also for their unique propensity to carry almost exclusively the Q haplogroup of the Y-DNA. The Q haplogroup links Kets with Americas, but not with the Eurasia, where it is an oddball marker found in limited populations and in trace proportions, totally incompatible with the genetic footprint of the mighty Huns. The confined genetic isolation is also absolutely uncongenial with the Huns' unforgettable historical reality of uncounted genetic mergers with all kinds of folks, from Scandinavians to Koreans and from Indians to Selkups.

Several scholars also came up with proposals that they represented a mixture of these peoples (for the detailed overview of the bibliography see Doerfer 1973).

This goes without a saying, by a scholar or a layman: never on this Earth existed a monolingual empire. All empires, political and religious, are polylingual by definition. Some, like Japan and USA and Russia, engaged in physical and cultural genocide, and ended up as basically monolingual, but they all started and existed as polylingual empires.

In 1962, however, Pulleyblank proposed a new theory, connecting the Xiong-nu language with the Yeniseian (aka Eniseian) family of Central Siberia (Pulleyblank 1962).2

The Yeniseian family consists of the Ket language, which is still spoken today, but is lingering on the brink of extinction (less then 2,000 total population, compared with ca. 130 mln Türkic people across Eurasia), and several extinct languages: Yug, (aka Yugh) Kott, Assan, Arin, and Pumpokol (there are a few more. Also, see Note 10: crossreferences between different families are perilous). Our knowledge of Asan, Arin, and Pumpokol is limited mostly to word lists, as these languages became extinct in 18th century. The Kott and Yug languages are somewhat better known, especially the latter, as Kott became extinct in the later part of 19th century, and Yug only recently. Pulleyblank's hypothesis was enthusiastically embraced by some scholars, e.g. Yakhontov 1986, and energetically rejected by others (Doerfer 1973).

Since Pulleyblank's idea, expressed in Appendix to his 1962 work “The Consonantal System of Old Chinese”, Asia Major 9: 58-144, 206-265, was not anywhere close to being productive, and scientifically offered only a few possible morsels of connections, the reasons for the enthusiastic embrace by some scholars need to be openly questioned: they were not exited by scientific enthusiasm, but were obviously racially-driven.

The Eniseian family once was thought to be associated with Burushaski, within a concept where Burusho Indo-European non-Indo-Iranian people migrated from Central Asia in a wave of Indo-European conquest of the Indus Valley, and thus with with the Indo-European Urheimat milieu, i.e with Eurocentric interests.

1. I would like to thank Edwin Pulleyblank (Vancouver) and Ralf Stefan Georg (Bonn) for their comments on the first draft of this paper.
2. The first suggestion that the Xiong-nu language may belong to Yeniseian family belongs to Ligeti who demonstrated that Xiong-nu (more exactly Hu) *sak-dak “shoe” is reminiscent of Ket sagdi “id.” (1950), however, it was Pulleyblank who presented the credible evidence first that also included examples from basic vocabulary (With the same success we can demonstrate that Portuguese and Slavic may belong to Yeniseian family, with their “zapata” and “sapog”, both ascending to Türkic for “boots”. Ditto for other credible evidence, like “caballero” and Russian “kobyla”).


The Xiong-nu language is known to us only through the about hundred fifty glosses and possibly one very short text consisting of just two lines in the Chinese transcription. All previous attempts to identify the Xiong-nu language with one or the other later languages of East or Central Asia were relying on modern readings of Chinese characters, or in a few limited cases, on the outdated reconstruction of Old Chinese (OC) by Karlgren. It was, of course, a doomed enterprise from the start, as the Chinese phonology underwent tremendous changes during last two thousand years. Thus, it could be compared to an attempt to phonologically interpret a language X, written two thousand years ago by means of the Greek alphabet, through the prism of modem Greek phonology. Therefore, Pulleyblank's hypothesis, whether it finally proves right or wrong, was a significant step forward methodologically, as it is based on author's reconstruction of Old Chinese phonology, that itself represents a considerable advance compared to Karlgren's reconstruction.

This is not entirely accurate: we have N.Bichurin's phonetization, which reflected 1800's Manchu's time phonetization, drastically different from the modern Mandarin; we have archaic OC preserved in Korean and Japanese; they provide additional tools for reconstruction. In author's example, the ancient form of the Hebrew Ieshua is Iesus in modern Greek, Iisus in modern Russian, Joshua in modern Hebrew, Jesus in English, etc., quite informative. On top of that, like in Chinese case, we have *reconstructed phonetization following different concepts and methods of *reconstruction. On top of that, the search is for semantics, not for a precise phonetic form that likely did not exist at all. Understanding the expected morphology is much more important then precise phonetic, as is abundantly demonstrated by the author.

The goal of the present article is to evaluate Pulleyblank's arguments from the point of view of further advances in the reconstruction of Old Chinese and in the historical-comparative and descriptive study of Yeniseian languages. Since 1962 there has been a considerable progress in both the reconstruction of Old Chinese and, more importantly, in the reconstruction and description of Yeniseian languages. Thus, many of the materials on the latter, unavailable to Pulleyblank at that time, can be utilized now. As far as Old Chinese phonology is concerned, I will rely on the version of Old Chinese reconstruction, recently proposed in Starostin 1989 and Baxter 1991, rather than on Pulleyblank's own reconstructions. For the sources on Yeniseian reconstruction, I will use Sergei Starostin's comparative phonology (1982) and comparative vocabulary (1995) of Yeniseian, and Georgii Starostin's reconstruction of Proto-Yeniseian verbal system (1995). Several important descriptive studies of Yeniseian languages have also been consulted or used: Dulson's Ket grammar (1968). and, Vail and Kanakin's phonology and morphology of Ket (1990), Kreinovich's monograph on Ket verb (1968a), sketch of Ket (1968b), and sketch of Yug (1979), Werner's grammar of Koft (1990) and sketches of Yeniseian family (1996a), Ket (1996b), Yug (1996c), and Kott (1996d). 4

3 It must be noted, however, that since 1962, Pulleyblank has changed his views considerably on the reconstruction of Old Chinese, sometimes taking even an agnostic position on prospects of Old Chinese reconstruction (Pulleyblank 1973, 1991, 1992, 1994). His new version of Old Chinese does not seem to enjoy the same kind of scholarly support as did his earlier version.
4 My sincere gratitude to Half Stefan Georg who not only elicited certain Ket words from his informant, but also sent me some his field materials on Ket that were also consulted in addition to the above-mentioned published materials on Yeniseian languages.

It should be also appreciated that the problem can't rest on swings in the opinions of experts like Pulleyblank, or a wide brush stroke picture disposing of productive expertise. Since in every particular case every reconstruction has a certain probabilistic value, relying on any single opinion reduces the overall confidence level for each particular case, the confidence in S.Starostin's *reconstruction is less than the confidence in S.Starostin's + B.Karlgren + E.Pulleyblank-1 + E.Pulleyblank-2 + Wang Li 王力 + Baxter + other available sources' *reconstructions. Accepting any single *reconstruction as an absolute truth, and not as a probably useful step toward overall objective, risks a binary outcome, either totally right, or totally wrong.


Pulleyblank proposed a number of arguments against alleged Altaic affiliation, both phonological and lexical. Let us start with his phonological arguments.

Foreword for the coming phonology: A.Vovin debunks all of arguments. Pulleyblank's phonological arguments have no merit. Pulleyblank's conclusion, based on his phonological arguments, has no merit. The conclusion is: None of the prior Türkic reconstructions can be considered successful because of the phonological violence done. And then A.Vovin makes an 180° turnaround, and declares that he agrees with the Pulleyblank's conclusion that has no merit.

a) Pulleyblank noted that many Xiong-nu words start with a liquid that was transcribed in Chinese with a consonant that he reconstructed in 1962 as OC *1-, and now reconstructs as OC *r- (personal communication). Pulleyblank noted that initial l- is rare in Altaic and initial r- is non-existent (1962: 240). However, this cannot be accepted as valid argument against Altaic affiliation of Xiongnu, since initial l- occurs rather frequently in some Tungusic languages and, more limited, in some Mongolic languages: cf. Ewenki la:m,u “sea”, le:mba “salmon”, lipki:- “to stick”; Manchu labdu “many”, laia “porridge”, laya- “wither”; Monguor lavci “leaf', lumu “bow.” Although initial r- is indeed atypical for Altaic languages, three major exceptions should be noted: a) secondary initial r- in Monguor resulting from elision of initial vowel (e.g. rguen “wide”, cf. Written Mongolian (WM) orgen “id.”; red “river”, cf. WM and “island”; re- “to come”, cf. WM ire- “id.” (Todaeva 1973: 355-365)), b) secondary initial r- in Japanese resulting from the large influx of Chinese and Western loanwords {rilcai “under standing”, rosoku “candle”, raitaa “lighter”, reji “register”), c) secondary initial r- in North Korean literary standard, “reincarnated” from etymological spelling: ryuri “glass”, rodong “labor”, ripyel, “separation” (cf. South Korean standard yuri, nodong, ipyel). Since we know nothing about history of the Xiong-nu language, initial r- in it (if it was initial r- and not initial l-), could possibly be attributed to similar cases.

This debunking example is posted in its entirety to demonstrate the logic of argumentation. There are a number of flaws on both sides:
1. Altaic is a disputed family, using it a priori as the only true working assumption brings about more clatter then either expert can handle.
2. Using a little knowledge outside of phonetics would limit analysis to the Türkic family as a main contender and Enisean family as a runner-upper. Using much more vague criteria of “Altaic” debases the whole analysis.
3. Initial l-/r- argument is baseless. Just the Turkish vocabulary has about 1.03% of initial l- and 1.37% of initial r-. Turkish is Oguz language, comparison with Ogur languages could produce higher proportions. That proportion depend mightily on the lexical borrowings, which could possibly be incompatibly higher in Hunnic, especially allowing some borrowings from Enisean languages.
4. The histories of Hunnic and Tungusic/Mongolic/Ewenki/Manchu/Monguor/Japanese/Chinese/South Korean are drastically different, in that Hunnic had active direct connections with the languages to the west, while the other languages were separated from the west by the very same Hunnic language. Unlike the Huns, none of the other nations ruled Central Asia during the Hunnic times.
5. Pulleyblank's *reconstructions follow Pulleyblank1 reconstruction method, and being based on unknown accuracy of Chinese renditions, offer little confidence in the accuracy of OC => Pulleyblank-1 => Hunnic *reconstructions to come to confident and drastic conclusions.
6. Unlike A.Vovin, we do know quite a few things about the history of the Huns, which allowed numerous scholars to conclude that they had an Ogur-type language. Still, the A.Vovin's debunking-1 conclusion holds, though for entirely different reasons.

b) Pulleyblank also suggested that initial consonant clusters represent a non-Altaic feature (1962: 240). Again, this is not necessarily true as secondary initial consonantal clusters (originated due to vowel reduction) are found in Middle Korean and in Monguor: Middle Korean pco- “to weave”, pstay “time”, skweng “pheasant”, psu- “to use”; Monguor ndige “egg”, ndurda- “to become high”, rgen “other”, sgil “thought”, sde:~ “to tear off”. There is no reason to believe that the Xiong-nu language could not undergo the same kind of phonological development.

Consonant clusters is a peculiar argument, since Hunnic is *reconstructed from the records of a language abundant in consonant clusters. Hsiongnu for Huns is a handy example, the difference is that in case of Hsiongnu and Xiongnu we know the alphabetical form was Hun, no clasters por favor. The A.Vovin's debunking-2 conclusion holds, though for entirely different reason.


Existence (within Pulleyblank's *reconstructions of 278 Hunnic words) of initial and intervocalic *t, *d, *k, *g, and lack of *th, *kh is also given as evidence against Altaic origin (Pulleyblank 1962: 242). While this picture surely contradicts current Proto-Altaic reconstruction (Starostin 1991), the same distribution is found in Mongolic and Tungusic.

*t, *d, *k, *g, *th, and *kh are all *reconstructions, supposedly they include all allomorphs, which makes each entry a cluster, a la *t => *t/*d*/*th/*dh/*θ/*γ, and the evidence is only as good as the *reconstructions. The Chinese would spell any words of any language only with the apparatus of their dialectal version of the language, and this observation of Pulleyblank would apply to any record in Chinese rendition of any foreign language. On top of that, the Türkic linguists in Russia complain that the modern Russian does not have apparatus to record Türkic words with th, and deforms Türkic words with kh, which makes Pulleyblank statement inaccurate. Comparison with alternate *reconstructions demonstrates the probabilistic nature of paleolinguistics. The A.Vovin's debunking-3 conclusion holds, irrespective of Mongolic and Tungusic

As Pulleyblank noted himself, a peculiar distribution of initial *b- and medial *-p-, in Xiong-nu but not vice versa is found in both Old Turkic and Mongolian as well (1962: 242), therefore, this phonological peculiarity of the Xiong-nu language cannot be really used against its Altaic affiliation.

Initial *n-, as Pulleyblank notes does not exist in Turkic, but is widely present in Mongolian (1962: 242), and, I should add, in all other Altaic languages, too.

Initial n constitutes 1.72% of Turkish vocabulary, and may be more in other 35+ Türkic languages. Approximately the same proportion is listed in Old Türkic Dictionary, 1969, Leningrad, Science. The A.Vovin's debunking-4 conclusion holds

Thus, I do not find to be valid any of Pulleyblank's phonological arguments against Altaic affiliation of the Xiong-nu language. It is necessary to add, from the methodological point of view, that even if they were valid, and Xiong-nu did look very un-Altaic, it still could not constitute “proof of its un-Altaic origins, as any argument based on phonological similarity is essentially an argument based on typology, that can be very misleading in the case of genetic affiliation. At the same time, the fact that all Xiong-nu phonological features listed by Pulleyblank, can be found in other Altaic languages, cannot constitute any evidence in favor of its Altaic origins. For this evidence, we must look elsewhere.

A.Vovin brushes away any phonological arguments for phonological *reconstructions, a sober position that can't be disagreed upon. However, later on A.Vovin holds phonological *reconstructions as reliable source for his own arguments.

Besides the phonological arguments, discussed above, Pulleyblank proposed a (vanishingly small) number of lexical comparisons connecting Xiong-nu with Yensiseian. It is necessary to note that the majority of glosses found in Chinese sources represent either titles or personal names, which are practically useless in determining the genetic affiliation of a language, although it might be tempting to do so. Pulleyblank did not quite avoid this temptation, and among the words he discusses there are some Xiong-nu titles. I will not evaluate them here, as I believe that this kind of evidence is not admissible. Nor will I deal with several names for milk products, that Pulleyblank also uses, except for the word “milk” itself, as they obviously constitute cultural vocabulary, that can be easily borrowed and re-borrowed, and, therefore, cannot be really used as evidence for establishing genetic affiliation. Instead, I will concentrate here on few words from the Xiong-nu language that belong to basic vocabulary (some of them only to a certain extent).

Apparently, A.Vovin takes exception exclusively for the Hunnic case, because the titles and personal names are most frequent deciding element in determining the unknown language, take for example Scythians, from whom we have a total of 9 words, plus a 300-long list of names and titles. The Scythian language was proclaimed to be “Northern Iranian” precisely on the analysis of the names and titles by the august V.Abaev, who managed to compose, though never publicized, a “Scythian Dictionary”. It was the real Scythian lexicon that was excluded from the analysis... And not only the work of V.Abaev was enthusiastically embraced, it became a scientific dogma after the WWII.

The problem is not with the names and titles per se, but with their phonetical *reconstruction. Where the phonetics remained stable, and *reconstruction can be avoided, the results are within reach. The Türkic names and titles were spread far and wide, and there is enough distinction, uniformity, and abundance to project them retrospectively to find some Hunnic equivalents even with Chinese coding. Could A.Vovin possibly voice his disapproval against many non-sensical Indo-European etymologies found for a myriad of non-Indo-European names and titles?


Foreword for the coming lexicon: all *words are *reconstructions, with Chinese explications. The Chinese explications can't or shouldn't be questioned as a semantic field, but as a translation they should be taken with a grain of salt, because Chinese annals do not translate, they explain. Thus, for one of many appellations for “son” we get a generic “son”, and for one of many appellations for “horse” we get a generic “horse”, among other possible scenarios.

a) *kwala “son” (). Pulleyblank compares this word with Arin Mkjal “son” (with bi- as a prefix), but Arin word seems to be isolated in Yeniseian: other Yeniseian languages have words based on Proto-Yeniseian (PY) *pu'f-b “son” (Starostin 1995: 252): Ket hi?p, Yug fi?p, Kott fup, Assan pup. It seems that there are no alternative Altaic etymologies for Xiong-nu *kwala “child.” Old Japanese kwo “child” superficially seems plausible (kwo < *ku[C]a or *ka[C]u), but it is likely to be related to Tungusic *kurja- “child”, and, therefore, is not applicable here (Contrary to A.Vovin, in modern Türkic languages “male child” or “son” is ogul, yvăl, ool, og'ul, ulan, ul, ül, ůgil, ool, uol, oolax, uul, half of which have the cognates of k/g-V-l, and since following A.Vovin the phonological arguments are not applicable, there is no need to turn to exotic Old Japanese and Tungusic. The g is silent in half of the Türkic languages, so ool = oĝol, etc. No need to go to the exotic dead Arin or *Proto-Yeniseian, the simple English ulan will do fine, it is a kin of the Hunnic “son”, no pun intended).

b) *kuti “horse” (). Appears to be a good match: PY *ku?s “horse” > Ket ku?s, Yug ku?s, Kott hus, Arin kus/qus, Pumpokol kut (Starostin 1995: 240). One minor problem is that the Xiong- nu form has -t- like Pumpokol, but unlike other Yeniseian languages. The other problem is however, more serious: the word for “horse”, although it does belong to basic vocabulary, can be borrowed (cf. Old Japanese uma/muma “horse” < Old Chinese *mra “id.”) (That problem is insurmountable: taiga foot hunters did not use horses, they used to drag their tackle on their backs. The quantity of terms for “horse” among Türks, who lived and died on the horses, is compatible to the quantity of terms for “work” among agricultural peoples or “boat” for Greeks. One of them is “kuti” for something small or gentle or nice, but is it applicable depends on the accuracy of Chinese rendition and phonetic *reconstruction).

c) *ket, ethnonym (corrected from ethnonim) Jie = “stone?” (羯). Pulleyblank compares it with Arin kes and Pumpokol kit “stone”. However, nowadays we know that both forms are results of fortitions typical for these languages, and that they go back to PY *ci?s “stone”, along with Ket ti?s, Yug ci?s, Kott Sis, Assan sis, Koibal Sis (Starostin 1995: 217-218). Pulleyblank believed that there are two Yeniseian etymons for “stone”, one with initial k- and another with initial c- (1962: 246), but he was misled by the fact that there are two Pumpokol forms for stone, kit, and ds. The latter is actually a Yug form, mislabeled “Pumpokol” (Helimskii 1985). This fact was not known in 1960s, and Pulleyblank, of course, could not avoid this mistake. We are left with two distinct possibilities here: (1) to speculate that Xiong-nu had the same fortitions as Pumpokol, which is not impossible (cf. the word for “horse” above), (2), to treat the term not as a “stone”, but as an ethnonym. In this case it does resemble closely the PY term *ke?t “person”, “human being” > Ket ke?t, Yug ke?t, Kott hit/hett, Assan hit, Arin kit/qit, Pumpokol kit (Starostin 1995: 236) (The second guess is closer to the story, which derives the name of the Hun's Jie people from the the name of the Qeshi area [modern Yuyshe county in the Shanxi province, Jinzhong Shi, Yuci 榆次区 district]: Wei-shu says: “Ancestors of Shi Le belonged to a separate Huns' pastoral rout, dispersed in the Shendan district, in the Usyan county, in the Qeshi area, after the name of which they started to be called Qeshi Hus” [Ch. 95, p. 9-a]. Just forget about linking Jie with Kangars~kang~stone, a dream of mega-Indoeuropeanists, these naive Chinese were jotting everything on the paper).

d) *t(r)engri “sky” (). The word is not found in Yeniseian: PY *?es “god”, “sky” (Starostin 1995: 188). On the other hand, it is present in Altaic: Old Turkic taηri, Written Mongolian tngri. It is considered, though, to be a loanword in these languages from a third unknown source (Pelliot 1944) (Pelliot did not go far enough, in Sumerian “dingir” is “sky”, “god” , exactly like in Türkic. So, Huns learned that word from the Sumerians. Or vice-versa. What's wrong with that?).

e) *d(r)ung “milk” ()- Pulleyblank compared this word with Pumpokol den and Arin terpxl “milk” < ten 'nipple' + ul “water.” It is a reasonably good etymology, although not without some phonetic problems (-u- to -e- and -n to -n, although Pulleyblank is aware of at least the latter one (1962: 252)). Starostin reconstructs PY *de(?)n “nipple”, “milk” > Kott ten “nipple”, Arin teqal, Pumpokol den “milk” (Starostin 1995: 220) (“Drung” Hunnic for milk? In all Türkic languages it is süt. *d(r)ung is suspiciously close to Germanic “drink”, and if Chinese recorded a Hunnic “drung”, it may explain the “uncertain origin” of the Germanic word. Considering that taiga foot hunters did not keep domesticated animals to milk, this must be an ingenuous Hunnic word, all the *reconstructions notwithstanding. We should remember that milk was a drink of the nomads, not of the taiga foot hunters. On the other hand, we all were suckled by breast milk, that must be a primeval word we carried from Africa 200,000 years ago. There must be a better etymology then to teach Germanics to suckle the Ket taiga milk. But what about consonant clusters? If the Huns did not carry the consonant clusters to Germanics, who did? Kets?).

The evidence above suggests that Pulleyblank's hypothesis is likely (If this can be called evidence, Bible can be called science and a textbook on biology. The “likely” on the lips of philologist must mean 0.000001 probability). However the scarcity of the material prevents any definite decisions, because five words (actually, only one suspicious word, for milk , subject to multiple contingencies of explications, renderings and *reconstructions) are obviously not enough to establish a significant number of regular phonetic correspondences, and all conclusions are based on phonetic similarity. The only regular correspondence that can be detected is that Xiong-nu (*reconstructed) medial and final /t/ correspond, like Pumpokol /t/ to *s (*reconstructed) in other Yeniseian languages and in Proto-Yeniseian. This may be classified as a shared innovation, further supported by the fact, that Xiong-nu and Pumpokol also share another innovation: fortition of *c- to k-. Unfortunately, there is just one example to support this. (Only in linguistics you can draw a trend based on a single point) Moreover, it looks like that Xiong-nu, similarly to Pumpokol, Ket, and Yug, preserves initial PY *d- while other languages devoice it to t- (cf. PY *de-s “eye” > Ket de:s, Yug des, Kott ti:s, Assan teg, tis, Arm ti-eη, Pumpokol dat (Starostin 1995: 220) (This argumentation is broken down by the Chinese peculiaruty, where for example da and ta for “great” are readily interchangeable). There is, therefore, the possibility that the Xiong-nu language, if it is a Yeniseian language, is most closely related to Pumpokol.

Thus, lexical evidence suggests that the Xiong-nu language is a Yeniseian language (Wow! Pause. Wow! Wow! Wow! ), but speaking in mathematical terms this type of evidence is necessary, but not sufficient. For the sufficient evidence we must again look elsewhere. Fortunately, there is a short poem in Jie (< *ket (羯)) language, that is believed to be a variety of Xiong-nu (Again, Wei-shu says: “Ancestors of Shi Le [i.e. Jie] belonged to a separate Huns' pastoral rout, dispersed in the Shendan district, in the Usyan county, in the Qeshi area, after the name of which they started to be called Qeshi Hus”). There are several attempts to explain this poem on the basis of Turkic (Ramstedt 1922), (Bazin 1948), (Gabain 1949). I agree with Pulleyblank that none of them can be considered successful (Pulleyblank 1962: 264), but for the sake of clarity I will discuss Ramstedt and Bazin's attempts as examples below.

Before doing so, however, I believe it is necessary to provide my readers with the poem itself, the text that surrounds it, and with the phonetic interpretation of Jie text, based on the recent advances in Chinese historical phonology.

The two-line poem in question is recorded in the biography of Fotu Cheng in the volume 95 of the “Jin shu [晉書, History of Jin].” 5 The Jin dynasty ruled from 280 to 420 CE, but the text itself was compiled under Song dynasty. The episode connected with the poem is as follows:

Then [Liu] Yao himself attacked Loyang, and [Shi] Le was going to rescue it. All his retainers [tried] to persuade [Shi Le] that [he] would not be able to do it. [Shi] Le visited [Futo] Cheng with [this problem], and Cheng said: “It is said in Xianlun Lingyin: “S(l)u(s)-ke thij?-re(ts)-kang bok-kok Ko-thok-tang.” This is Jie language. S(l)u(s)-ke means “army”, thij?-re(ts)-kang is “to go out”. Bok-kok is the barbarian title of Liu Yao, Ko-thok-tang means “to catch.” These words mean that army/armies will go out and capture [Liu] Yao.” [Cheng] also ordered a young boy to purify himself and fast for seven days, and [then] he mixed hemp oil with rouge, studied [the mixture] himself on his palm, and then showed his hand to the boy. It was clearly shining. The boy said with surprise: “There are many war horses, and [I] see a man, tall and big, and clearly white (Liu Yao was an albino - A.V.). His elbows are tied by a red rope.” [Futo] Cheng said: “This is definitely [Liu] Yao.” [Shi] Le rejoiced, following [that conversation] went to Lo[yang] to resist [Liu] Yao, and captured him alive. (Jin shu 95: 2486). 6

5. I use an edition of the “Jin shu” published by Zhonghua shuju (Beijing, 1988). “Jin shu” was compiled by Fang Xuanling and other scholars in the beginning of Tang dynasty, that roughly corresponds to the beginning of the seventh century CE. The events described in this passage refer to the war between two Xiong-nu chieftains, Liu Yao and Shi Le for the dominance in North China, that occurred in 329 CE.
6. Translation of the passage is mine (- A.V.).


The Jie poem from this passage is: (秀支替戾剛 , 僕谷劬禿當 in Unicode).

Its Chinese word-to-word translation is: that can be rendered into English along with the appropriate transcription, based on Baxter-Starostin reconstruction like the following:

秀支 = *s(l)u(s)-ke, (軍) “army” or “armies” (there is no indication of plural or singular in Chinese). Consonants in parentheses in this word and the following one existed in Early Old Chinese (7th c. BC) but are likely to be gone by the time of Late Old Chinese of Jin dynasty one thousand years later.

(Transcribed by V.Taskin in Cyrillic Ñþ÷æè, corresponding to English Süji, in modern Türkic Süči, meaning “army” - sü + -či = common Türkic noun-derivational affix to form profession or occupation, i.e. “army-man, commander”, like “potter” from “pot” )

替戾剛 = *thij?-re(ts)-kang, (出) “goes out” or “go out”. Possibly also “went out” or “will go out.”

(Transcribed by V.Taskin in Cyrillic òèëèãàí, corresponding to English tiligan, in modern Türkic tilegan, meaning “to wish” - tilek + -gan = common Türkic past participle, 3rd person singular, perfect tense verbal affix, i.e. “has ordered [to move, to act]”)

僕谷 = *bok-kok, (劉曜胡位) barbaric title of Liu Yao.

(Transcribed by V.Taskin in Cyrillic Ïóãó, corresponding to English Pugu, with several alternatives in Türkic titles, one is historically attested Bull; a homophonic pug/buk is excrement, poop, shit)

= *Ko-thok-tang, (捉) “captures”, “will capture”, or “captured.”

(Transcribed by V.Taskin in Cyrillic öçþéòóäàí, corresponding to English qüitudan, in Türkic 'yu tutar, meaning “will be” - -'yu/-'gyu = future conditional verbal transitive affix in reference to Pugu [i.e. Pugu will be] + tutar = capture in 3rd person future tense [i.e. (he) will capture] + dan/-tan = locative directional verbal affix “from, out of”, with several alternatives in Türkic, with homophonic quitudan - scare out of, from quyut = to scare, to spook)

Let us first see how our predecessors dealt with with this poem. G. J. Ramstedt uses mostly the modern Mandarin readings of the characters above and reads: Siuk'i t'ilikang, puh-koh kü-tu-tang (Ramstedt 1922: 31).7 The only obvious non-Mandarin readings Ramstedt employs are /k'i/ for 支 instead of Mandarin /zhi1/, and /k'i/ for 劬 instead of Mandarin /qu2/. Ramstedt's source of these archaic-looking readings is unclear: in any case none of them can pretend to be a true reconstruction: while consonantism is correct, vocalism is wrong. Ramstedt further interprets this poem as “Wenn man in den Krieg auszieht, so kann man Liu-yao gefangen nehmen.” (If someone go out to war, so can catch Liu-yao caught) The first thing that attracts our attention here is that Ramstedt's translation has too many words compared to original: there is no indication of impersonal (“man”) anywhere in the original, nor there is any hint at modality (“kann”), or temporal link between two clauses (“wenn”).

7. Ramstedt's interpretation practically repeats the interpretation of Shiratori (1900: 11-12).

Second, the word for “army” turned into the word for “war” (“Krieg”), in order to explain -k'i in siu-k'i as dative case -kä, which would be feasible in the light of Baxter-Starostin *ke, but not Ramstedt k'i. However, Old Turkic means “army”, and not “war”, and the replacement is not justified. In addition, Xiong-nu *s(l)u(s)-ke has a [+back] vowel, while Old Turkic word “army” has a front vowel. Furthermore, if *-ke were a dative or directional case, it would almost certain be indicated in the Chinese translation by prepositions 於 or 于 “in”, “to”. However, the absence of these prepositions clearly indicates that “army” is to be treated as a subject if we are dealing with SVO language, or as subject or direct object, if Xiong-nu was SOV language. Finally, Chinese 出 “go out”, “to put out” is translated as “ausziehen” (move out) rather than “ausgehen” (go out) or “hinausfuhren” (went out). Ramstedt does not stop here at his more than free interpretation of both Chinese translation and phonology. His t'i-li-kang further becomes tai-li-kang, and then magically turns into tal'iqyŋ or lal'iqaŋ The latter then is compared with Old Turkic tašyqyŋ “ziehet aus” (draws from, pulls from). Needless to say, such ad hoc interpretation of both Chinese phonology and translation does not give too much credibility to Ramstedt's “reading”. Furthermore, the title *bok-kok is compared with Old Turkic bögü “der wise” (the wise) and Written Mongolian boge “schamann” (shaman), ignoring completely the [+back] vocalism of the word in Xiong-nu and the final -k in it, absent in both Turkic and Mongolian. The last word, *Ko-thok-tang “to capture” is left without any Turkic interpretation at all.

Bazin's explanation, also utilizing Turkic, does not fare much better. Bazin, however, must be given its due that his philological approach to the text is more solid than that of Ramstedt: he at least cites the poem in the original Chinese transcription and uses Karlgren's Chinese reconstruction (Kargren 1940) to interpret it (Bazin 1948: 208-209). However, even Bazin does not completely escape the temptation of a free interpretation of the poem that goes: “Si l'armée sort, Liu Yao sera capturé” (If the army leaves, Liu Yao will be captured). There is no conditional in the poem above. This is however, insignificant, compared to the major blunder that comes next. Citing Eberhard's authority, Bazin hypothesizes that Chinese usually divided words incorrectly in foreign languages (1948: 209-210). There is however, no reference to Eberhard's saying this anywhere in print, and one can assume that Bazin might misunderstood personal communication.

However, this assumption is simply not true. While most Chinese in ancient times were no more interested in the languages of surrounding people than they are interested in minority languages of China today, those who were interested left us samples of very careful and reliable transcriptions that can and must be utilized with application of proper knowledge of Chinese historical phonology. In addition, Bazin, in all probability, did not read attentively the preceding passage. The translation of Xiong-nu poem is done by Futo Cheng himself. And we do know that Futo Cheng, an Indian monk, was multilingual. He was of course, fluent in Chinese, and we can expect that being a close confident of Shi Le, he was quite familiar with the Xiong-nu language as well. Thus, the possibility of incorrect segmentation is almost non-existent. Bazin, however, does his own segmentation, which is solely justified by his Turkic interpretation.

First, Bazin believes that the word for army is just 秀, and not 秀支 as Futo Cheng says. He further analyses this 秀 as representing Old Turkic sü-g “army-acc” (army in accusative). Therefore, Bazin, like Ramstedt, tries to connect the first word with Old Turkic sü in spite of the difference in vocalism that I already mentioned above. In addition, his accusative case marker -g is based on Karlgren's reconstruction *siôg as a reading of this character in Archaic Chinese, but nowadays we know (corrected from now) that *-g simply was not there, but possibly final *-s was (Baxter 1991: 798). Bazin is more careful than Ramstedt with the vocalism of the title “bok-kok, but his proposal seems no less strange. He suggests that the word must be Old Turkic buquγ or boquγ, that is not attested per se, but reconstructed by Bazin on the basis of Old Turkic buqaγ found preceding Old Turkic tutuq “gouverneur militaire”, and Old Osman boγ “chef militaire, commandant d'une troupe”, that, as he believes, is a contraction from boquγ (Bazin 1948: 211.). There is, however, one difficulty on the Turkic side: although Old Turkic is known for using constructions consisting of synonym pairs, like ï ïγač “plants” (lit. plant+tree), it is by no means known that Old Turkic buqaγ in buqaγ tutuq is a title. “Old Turkic dictionary” glosses it as a personal name (Nadelieav et al 1969: 125). 8 Even if Bazin is right in this particular case, the title is useless for determining genetic affiliation of the Xiong-nu language: Turks could have just borrowed it from Xiong-nu.

8 No entry in Clauson 1972.72.

True to his principle of resegmenting Xiong-nu poem, Bazin gets rid of the first syllable in 劬禿當 = *Ko-thok-tang “to capture”, and interprets remaining *thok-tang as Old Turkic tut- “prendre, saisir” (take, capture) with the following suffix -qaŋ. This, of course, implies a metathesis of /k/ and /t/ in the Xiong-nu original, that Bazin attributes to Chinese notation (1948: 212). Why Chinese would be guilty of such misbehaviour remains unclear: in the fourth century CE the Chinese language had both final -t and -k. Another problem that faces Bazin here is that Old Turkic does not have second person plural imperative -qaŋ, but only -aŋ. Bazin tries to appeal to optative-subjunctive -qa or -qay and to possible elision of -q- to save his etymology, but this further makes his explanation even more speculative. The first syllabic *Ko, that was cut off to get Old Turkic *tut- “catch, take” is interpreted as Turkic accusative in -γ/-g, but note that it is transcribed in Chinese in a quite different way than after the alleged word “army.” To be more exact, as it was already pointed out above, there is no evidence for the accusative in the first line at all.

The most grave misanalysis occurs in the first line. Since Bazin already cut off the character 支 from the word 秀支 “army” in order to get Old Turkic sü-g “army-acc”, as was already discussed above, he has to explain somehow the remaining characters 支替戾剛. He reads the characters ;支替 as /t'iəg t-iei/, which he further interprets as Old Turkic tag- “to attack” + -ti, past tense marker (1948: 215). Characters 戾剛 are read as /liod kaŋ/, and interpreted as Old Turkic ïd- “send”, and -qaŋ (Bazin 1948: 215), a non-existent suffix of the second person plural imperative that has already been discussed above. Unfortunately for Bazin, both his readings and interpretations are plagued by grave problems. First, the initial of the character 支 was not /t'l at this time, but *k-, as amply evidenced by usage of 支 in Old Korean writing for /ki/ or /ke/ and in Old Japanese man'yôgana for /kyil. Second, it is completely incomprehensible how 戾 /re(ts)/ (Bazin's /liod/) can stand for Old Turkic ïd-: the final consonant there was probably lost in the Han period, and initial liquid cannot be simply ignored. Third, the Chinese translation given by Futo Cheng does not mention either “attacking” or “sending” as found in Bazin's interpretation, but plainly mentions an army (or armies) that will go out (One would be naive to read these types of blistering comments literally: the Chinese symbol 出 stands for 11 major English verbs - come out, go out, turn out, issue, arise, put forth, put up, happen, exceed, vent, expend - each of which has a cluster of English synonyms - come out has more then 30, so a translator has a choice of hundreds words to render the single 出 sign - all of them technically appropriate, beyond the pedantic criticism, and the semantic choice rests with the arbitrariness of the translator as much as the arbitrariness of criticism for that choice rests with the critic. To make matters worse, Chinese lexicon is built on idioms, and English has quite a few of them, like blow up has nothing to do with blowing, it stands for explode, detonate, and 37 other meanings).

Bazin concludes his “Turkicized reconstruction” of the Xiong-nu poem that reads as follows: 9

“süg tägti ïdqaŋ
boquγïγ tutqaŋ

with a new “translation”: “envoyez l'armée à l'attaque, capturez le commandant!” He then concludes that the Chinese translation is not bad, but just an approximation. Unfortunately, his own “translation” does not stand the slightest chance in the light of all the phonological, morphological, and philological mistakes and speculations discussed above. An attempt to rewrite Chinese translation on the premise that Xiong-nu language must be Turkic obviously does not pay off, nor can it be used as a solid foundation for analyzing the poem. The only credible solution will be to try to find matches which exactly follow the Chinese translation (These differences in approaches are more conceptual then technical: one sees a forest, the other only trees. It is a universally recognized fact that written data preserved in the script of foreign languages inevitably reflect phonological modifications, to expect a phonologically accurate transmission is more then naive, and to demand it as an absolute criteria for correctness is not within reason. Bazin is searching for a form that would be reasonably translated into Chinese and back into Türkic, preserving the semantics dictated by the circumstance, and to some degree ignoring the inevitable phonological modifications. Vovin is searching for phonological, morphological, and philological “mistakes”, which is a valuable criticism that in a long run will be productive, but raising the Chinese explanations to a status of perfect absolute has nothing to do with science, it is a matter of faith).

One of the four words in the poem, *bok-kok, the barbaric title of Liu Yao, is useless for our purposes: as was mentioned above, titles can tell us little about genetic origin of a language (This is an arbitrary nonsense, e.g. title Shanuy is associated with Huns, Raja with Indians, Shah with Iranian world, Czar with Slavic people, etc., as a very least they point in a certain direction, which in case of Türkic vs. Enisean is a valuable information). Another word, *s(l)u(s)-ke “army” does not have any parallels in Yeniseian: PY *kar1e “war” obviously does not fit. However, this is also a cultural word that is easy to borrow, so it cannot be diagnostic (Do we understand it correctly, if the text requires *s(l)u(s)-ke, and Enisean has banana, it is not diagnostic because the word banana came from Greenland? In a jury court case, the rule is: If it does not fit, you must acquit. In special linguistic cases, it is different: if the source says su-ke = army, and Türkic, a prime contender, has sü = army, we throw away the evidence because Enisean, a runner-upper, instead has banana = army, and because banana came from Greenland, that negates the evidence. We do not understand the logic correctly, do not see the logic at all, just a sleazy advocacy). The two remaining words are much more important, for two reasons: both of them represent basic vocabulary, and, being verbal forms, we can expect morphological elements in them that will give us some clue (Actually, we do not have any morphological elements, we are trying to *reconstruct morphological elements from the Chinese rendition that inevitably have unknown phonological modifications that we will ignore in order to pretend that Chinese rendition as an etalon of philological precision).

Pulleyblank noted that -ŋ is a frequent verbal ending in Yeniseian languages, especially in Kott (1962: 264), but did not go beyond this. I believe that it was due to the severe limitation of materials that he had at his disposal in early 1960s (Just by pure incidence, the Chinese also has -ŋ as a frequent ending that may inevitably show up in the Chinese renditions, producing inevitable phonological modifications). Thirty years later, however, the situation is much more favorable. Before going into detailed analysis of each verbal form, some general notes on the Yeniseian verbal system, are in order.

9. It is unlikely to be an Old Persian loanword contrary to (Starostin 1995: 235), since Old Persian kara- means “army” and not “war” (Starostin is no gospel, we can dispose with Starostin, Pulleyblank, Bazin, linguistical bananas etc. and other inconveniences at will. Actually, kara- is irrelevant, it can't go out, relevant is the banana which can go out).

Yeniseian verbal system is extremely complex, and best of all we know Ket system. There are two major types of verbs, verbs with interrupted root, and verbs with uninterrupted root (Kreinovich 1968a: 41, 97ff.). Two simple examples:

verbs with uninterrupted root
-uŋ “see”

ba-t-tuŋ “he sees me” daeŋ-t-uŋ “he sees us”
ku-t-uŋ “he sees thee” kaeŋ-t-uŋ “he sees you”
d-a-t-uŋ “he sees him” d-aŋ-t-uŋ “he sees them” (aŋ ~ ang)
d-i-t-uŋ “he sees her”

verbs with interrupted root
daγij “mock”

daγ-wa-γ-ij “he mocks me” daγ-daeg-γ-ij “he mocks us”
daγ-ku-γ-ij “he mocks thee” daγ-kaerj-γ-ij “he mocks you”
daγ-a-γ-ij “he mocks him” daγ-an-γ-ij “he mocks them”
daγ-i-γ-ij “he mocks her”
(actually, quite simple: he-me-mocks, etc. SOV)

Kott system seems to be simpler, but still quite complex. Simple example (Why jump the verbs in every example?):

hama?-a-th-ak-ŋ) “I love myself' (do-myself-love-I)
hama'r-an-th-ak-u “thou lovest me”
hamar-an-th-ak “s/he loves me”
hama?-an-th-ak-an-oŋ “you love me”
hama?-an-th-ak-an “they love me”
hama?-u-th-ak-ŋ “I love thee” (do-thee-love-I)
hama?-a-th-ak-u “thou lovest thyself' (do-thyself-love-thou)
hama?-u-th-ak “s/he loves thee”
hama?-u-th-ak-an-toŋ “we love thee”
hama?-u-th-ak-an “they love thee”
(actually, quite simple: do-thee-love-I, do-thyself-love-thou, etc. SVO. Where is SO agreement, where S and O change in concordance?)

The major difference between the Ket and Kott systems is that in Ket subject verbal agreement affixes are found as prefixes or infixes, while in Kott they are final suffixes concluding the verbal form. According to Georgii Starostin, it appears that Proto-Yeniseian system was much closer to Kott than to Ket (Starostin Georgii 1995: 158-160) (Totally irrelevant for Fotu Cheng phrase, but good information for general erudition).

A verbal form in Proto-Yeniseian was built in the following way (Starostin Georgii 1995: 170):

preverb *k-, *t-, *p- + object marker *-w-, *-t-, *-k- + conjugation marker *-a-, *-i-, *-o- + aspect/tense marker *-r-, *-n-, + stem + plural marker *-n- (for lpp and 2pp) + subject marker

Starostin also reconstructs the following system of subject verbal agreement suffixes (Starostin Georgii 1995: 162):

lps *-ŋ lpp*-zəŋ (Kott -toŋ)
2ps *-[k]u 2pp *-[k]oŋ
3psm *-a 3pp *-aŋ (aŋ ~ ang)
3psf *-i

On the basis of these reconstructions by Georgii Starostin I believe it is possible to provide the following analysis of the two verbal forms in the Jie poem:

替戾剛 = *thij?-re(ts)-kang (出 “goes out” or “go out”) = *t-i-r-ek-ang “went out”
preverb *t- + *-i- conjugation marker + perfective *-r- + *-ek-, stem + aŋ (aŋ ~ ang) 3pp
Cf. also stem *-ek- and Proto-Yeniseian *-jaq- “come/go”

劬禿當 = *Ko-thok-tang, (捉 “captures”, “will capture”, or “captured”) = *k-o-t-o-kt-ang
preverb *k- + o? + object marker *-t- + conjugation marker *-o- + *-kt- “to catch” + aŋ (aŋ ~ ang) 3pp

Foreword for the coming section: The upcoming explanations cover two distinct subjects, lexicon and morphology, interspersed or blended together. From the blend, we need to extract understanding of the provenance of lexicon, morphological schemes or laws used, and the morphology of *reconstructions, connecting every effect with its cause, and vice-versa. Like in any construction, the limits of applied materials and methods are important ingredients of the process and result.

Morphology seems to fit almost perfectly with Proto-Yeniseian morphology reconstructed by Georgii Starostin. Moreover, at least one of the verbal roots, *-kt- “to catch” can be paralleled by Ket -kas- “to take”, “to catch” attested in literature in the following forms: dkasaγus “he takes him”, dkasanem “he took him” (Vall & Kanakin 1990: 51), dkasanem “I caught” (Vall & Kanakin 1990: 67). Ralf Stefan Georg also kindly elicited for me numerous forms from his Ket informant, Zoia Maksiunova that are all based on the stem -qos'- “to take”, e.g. qos'-abᅀt (abᅀt apparently stands for ablaut [abΛt], a root vowel permutation, or ᅀ stands for [nj]) “I take (all the time).” It was mentioned earlier that Xiong-nu, like Pumpokol, has -t- corresponding to sibilants in other Yeniseian languages, so we should expect something like *-kat- in Xiong-nu. Less obvious, but still possible, is the equation of the stem of the first verb *-ek- “to go out” with the Ket stem -(y)aq- “id”, cf. di Gaq, diyaq “to go out”, ᅀlya “I go out.” Unfortunately, the verbs “to take”, “to catch” or “to go out” are not attested, in scarce Pumpokol materials (Helimskii 1986). Fortunately, however, a partial paradigm of one verb, “to stand”, is recorded in Pumpokol materials (Helimskii 1986: 211):

itscha-dingdi “I stand”
ue-itscha-dingdu “you (sing.) stand”
adu-itscha-dingdu “s/he stands”
adyng-it-schading-dun “we stand” (maybe adyng-itscha-ding-dun, for consistency?)
anjang-it-schadingan “you (plur.) stand” (maybe anjang-itscha-ding-an, for consistency?)
(unfortunately, the assembly of the morphemes is indirect and therefore speculative, the only common lexeme is itscha-ding, which must stand for “stand”, or rather “do stand/stand do”, the rest can be speculated to your soul's agenda, I suggest -di = I, ue-xx-du = you (sing.), adu-xx-du = s/he, adyng-xx-dun = we, anjang-xx-an = you (plur.) in Pumpokol. The “ang” is met once, in the composite “anjang-xx-an”, as part of “you” (plur.) or 2pp in A.Vovin's nomenclature. How A.Vovin jumps from “anjang-xx-an” ~“you” (plur.) or 2pp to the third person plural pronoun or 3pp, without any knowledge or parallels, is the A.Vovin's hocus-pocus, but it is a giant step toward *reconstructed composites -ek-ang and -kt-ang = -go out-3pp and -catch-3pp)

It is very difficult to offer a proper morphological analysis of this paradigm due to the scarcity of Pumpokol materials, but several possibilities can be suggested. The subject agreement markers are probably -di, 1ps, -du 2ps & 3ps, -dun 1pp, and -an 2pp. Among those, only plural suffixes have parallels in the Kott and Proto-Yeniseian system: lpp -dun, PY *-zəŋ (Kott -toŋ) and 2pp -an, 2pp *-[k]oŋ (Kott -oŋ). The verbal root is probably -ding- (Starostin reconstructs PY *d[i]k- “to stand” (1995: 221), cf. such Arin forms as at'aŋ æt'aŋ, aj æt'aŋ (aŋ ~ ang) “I stand”; æt'aŋtaŋ, aiŋ (itaij) æt'aŋtaŋ (aŋ ~ ang) “we (they) stand”; axorent'a:ŋ, aŋ (aŋ ~ ang) axorent'a:ŋ “you (pl.) stand” (Toporov 1968: 286-287). The initial ue- in ue-itscha-dingdu “you (sing.) stand” is probably just the independent second person singular pronoun, cf. Ket u, uk “you” (sing.) (Kreinovich 1968b: 461), Yug u, uk “you” (sing.) (Kreinovich 1979: 336), possibly also Arin au “you” (Toporov 1968: 287). The initial adu- in adu-itscha-dingdu “s/he stands” is also probably the independent third person singular pronoun, cf. Arin xatu “he” (Toporov 1968: 292). The same can be said about adyng- in adyng-it-schading-dun “we stand”, which be compared to Ket ᅀt'n' and Yug ᅀt'n “we” (Kreinovich 1968b: 461; 1979: 336). The initial anjang- in anjang-it-schadingan “you (plur.) stand” is likely to be the third person plural pronoun, too, but there are no obvious external parallels. The rest of the form, namely -itscha-, is unclear, at least to me, who is not a Yeniseian specialist.

In spite of an obvious poor quality of the recording, it is possible to see that Pumpokol, like Kott and unlike Ket, has subject verbal agreement suffixes at the end of the verbal forms. That, alongside with the fact that the Arin language has a system similar to Kott and Pumpokol, is favorable evidence for Georgii Starostin's reconstruction of Proto-Yeniseian subject verbal agreement of the basis of Kott system rather than Ket (Starostin Georgii 1995: 158-159).10 Therefore, it provides us further justification to believe that matches between Xiong-nu and Proto-Yeniseian morphology are not chance resemblances.

I, therefore, come to the following interpretation and translation of the Xiong-nu poem that closely follows the Chinese translation:

(The Jie poem:)
(in Unicode 秀支替戾剛  , 僕谷劬禿當 )
  *suke t-i-r-ek-ang          bok-kok k-o-t-o-kt-ang
  armies PV-CM-PERF-go out-3pp    bok-kok PV-?-OBJ-CM-catch-3pp
(PV - preverb, CM - conjugation marker, OBJ - object marker, PERF - perfective)

Its Chinese word-to-word translation is:

in Unicode 秀支 , 軍也 o 替戾剛 , 出也 o 僕谷 , 劉曜胡位也 o 劬禿當 , 捉也
Romanized Mandarin Xiù zhī , jūn yě o Tì lì gāng , chū yě o Pú gǔ , Púgǔ Liúyào hú wèi yě o Qú tū dāng zhuō yě
Translated Mandarin Xiù zhī , Army have o Tì lì gāng , Out of have o Pú gǔ , Púgǔ Liúyào Hu rank have o Qú tū dāng catch have
A.Vovin phonetization *suke   *thij?-re(ts)-kang   *bok-kok   *ko-thok-tang  
A.Vovin Proto-Eniseian N/A   t-i-r-ek-ang   bok-kok   k-o-t-o-kt-ang  

that can be rendered into English along with the appropriate transcription, based on Baxter-Starostin reconstruction like the following:

  Armies   have gone out   bok-kok   [They] will catch  

Note that 3pp subject agreement marker -ang on both verbs requires interpreting the word *suke as plural “armies” rather than singular “army.” As was mentioned above, the Chinese translation does not provide any clue as to whether Shi Le had one army or several armies involved in his campaign against Liu Yao. However, the small passage immediately preceding the passage quoted above does:

Liu Yao sent his second younger brother [Liu] Yue to attack [Shi] Le, and [Shi] Le sent Shi Jilong to deter him. [Liu] Yue defeated [Shi Jilong], and [the latter] retreated to the protection of Shi Liang's stockade. [Shi] Jilong strongly defended the fence. Cheng was [at that time] in Xiang country and said with a sigh: “I pity [Liu] Yue!” When his disciple Pazuo asked him about the reason of this, Cheng replied: “Yesterday, at the hour of the boar, Yue was already defeated and captured by [Shi Jilong].” It happened [exactly] as he said. 11

10 Verner believes that the Arin and Pumpokol languages constitute a subgroup within Yeniseian (1997a: 171). On the other hand, Kott, according to Verner, is in a different subgroup with the Assan language (1997a: 171). Thus, we have two independent pieces of evidence for the archaic nature of Kott subject agreement.
11. Translation of the passage is mine (- A.V.).


It becomes apparent that both Liu Yao and Shi Le moved their armies as chess pieces during the fateful campaign of 329 C.E. Therefore, there seems to be little room for a doubt that Xiong-nu *suke as well as Chinese 軍 means “armies” rather than “army,” and the usage of plural subject agreement markers in the Xiong-nu poem is justified.

Therefore, I believe that Pulleyblank must be praised for his important discovery of the genetic affiliation of Xiong-nu that he has made more than thirty years ago on the basis of much more limited materials than the ones we have at, our disposal today, which further support his ideas.

Comments on A.Vovin Reconstruction. Attention: these comments are not a part of A.Vovin article
Language type

Independent of the phonetization, the structure of the Hunnic sentence indisputably is of SOV or OSV language, the verb is at the end of the phrase: Army go out, Liu Yao captured SV, OV. Türkic is rigidly SOV language, while the examples given by A.Vovin indicate both SVO in Kott system and SOV in Ket system. A.Vovin treats with dead silence the expected language type, a major property of any language, and would rule out any Enisean language, including “*Proto-Yeniseian” that does not comply with this straightforward requirement: e.g. English would be ruled out based on the structure of the sentences: [When, if, etc.] Army go out, [it] [would, will, etc] capture Liu Yao SV, [S]VO; another possible structure in passive voice also does not fit, Army go out, Liu Yao [would be, will be, etc] captured SV, S[V]Adjective (in this case verbal adjective captured, but could be red or sad). Without addressing the type of examined language, any conclusion is inconclusive.

Number of words

A.Vovin scolded Ramstedt's translation for too many words compared to the original. That is a significant criteria. By their very nature, flexive languages differ from agglutinative languages in that they need more words to express the same meaning, the prepositions, postpositions, service words, articles, etc. add to the length of the sentence, and require greater number of words. A.Vovin himself in reconstructing the meaning of the sentence is committing the offence that he finds unacceptable at Ramstedt: Vovin “Armies have gone out. bok-kok [They] will catch” (8 words), Ramstedt “Wenn man in den Krieg auszieht, so kann man Liu-yao gefangen nehmen.” (If someone go out to war, so can catch Liu-yao caught) (11 Geman/10 English words, catch-caught = 1). Instead of berating Ramstedt, A.Vovin should have addresses the reasons for such a phenomena, and openly recognize a need for agglutination in the suggested language to match the properties of the Hunnic language. Stealthily, without discussing the subject, A.Vovin does exactly that, providing an abundance of examples of Ket, Kott, Pumpokol, and “*Proto-Yeniseian” agglutination, allowing his model to structurally conform to the Hunnic phrases.

Understanding the agglutinative nature of the verbs is a necessary step on the road to reconstruction, it allows to parse the words into their constituent elements, and search for suitable sources. Agglutination allows to recover additional morphemes imbedded in the compound words, providing an evidence about affiliation of the language: in English I zzz-ed the door, or I xx zzz-ing the door allows to recover a plausible meaning for zzz, and if in Chinese phonetics it was rendered hopung, that would strongly indicate “I hopung-ed the door”means “I opened the door”. Likewise, “I hopung-yet the door” and “I emg hopung-yeng the door” would strongly indicate that -yet corresponds to the past tense ending -ed, and -yeng corresponds to the gerund ending -ing, independent of the closing, closing, or swinging translation of the verb zzz. These kind of grammar guessing exercises are given at the prep stage of the middle school grammar courses, and probably do not need to be dwelled upon in the scientific paper on philology, except for the fact that these elements are vitally important for the linguistic reconstruction conducted in A.Vovin paper, and in fact constitute the greatest component of the theory and practice of his exercise. The Hunnic phrase has morphemes -zhī/-ke, -gāng/kang, -dāng/-t-ang, which absolutely must be established in the prior literature, and make morphological and grammatical sense for a successful reconstruction. In the Türkic grammar, they correspond to -či = noun-derivational affix to form profession or occupation, -gan = past participle, 3rd person singular, perfect tense verbal affix, and -dan/-tan = locative directional verbal affix “from, out of” respectively, all positively attested either in the Old Turkic Dictionary, or Mahmud Kashgari Devon lugotit turk, or both. They are also attested as active word-forming morphemes in the modern Türkic languages, used for daily language of the newsprint. In fact, a scan of a random daily newspaper would likely find all three of them, if allowed for specific dialectal modification and sufficient volume of contents.

In A.Vovin paradigm, -zhī/-ke is not addressed at all, in fact A.Vovin clandestinely accepts the only suitable lexeme found in known languages, su/siu/sü = army in Türkic languages, only raising the question of back or front vowel in Türkic versus Chinese rendition, while completely ignoring a universally recognized fact that written data preserved in the script of foreign languages inevitably reflect phonological modifications, and without addressing the ability of the Old Chinese to understand the significance of the back or front vowels in the languages to which they never had or have any interest, their ability to adequately render the difference provided that they did appreciate the significance of the back or front vowels in the foreign languages, or that the character 秀 may have had more then one way of pronunciation across the time, space, and dialectal differences of Chinese history. In A.Vovin paradigm, his acceptance of the phonetic reconstruction 秀支= Xiù zhī/*suke passes without any historical, phonological, morphological, and philological explanation, except for a vague reference to a possible loan from unspecified language (Türkic by default?) that by simple sequence of events must precede the “*Proto-Yeniseian” borrowing, must replace the ingenuous “*Proto-Yeniseian” word xxx = army without a trace in any of its daughter languages, and indirectly stipulates that the Huns, these eternal warriors from the steppes, had their own Enisean word for the war, but the poor souls did not have a word for the army to run the war. To blast Ramstedt and Basin, who did not skip difficult problems, for their sins and then turn around and leave a gaping hole in his own paradigm does not inspire confidence in the research methodology.

I剛 -gāng/kang, 當 -dāng/-tang miraculously turn into -ek-ang and -kt-ang = go out-3pp and catch-3pp. Leaving aside for now the parsing that made the miracle possible, and a chain of dubious undocumented implications between different little-known vernaculars, a completely different miracle is the jump from you to they. The -ang started as a part, but not a morpheme, of the composite “anjang-xx-an” ~“you” (plur.) or 2pp in Pumpokol. The “anjang” is documented as “you” (plur.) or 2pp. The separation of “ang” from “anjang”, and its consecutive conversion from 2pp = second person plural pronoun to 3pp = third person plural pronoun are solely by a feat of urge. There is no justification for parsing morpheme “anjang” into sub-morphemes, by definition morphemes are minimal meaningful language units, they cannot be divided into smaller meaningful units. Neither is “ang” a separate morpheme, nor that fictitious morpheme is 3pp. And barring that triple-quadruple leap of the fantasy sends the whole A.Vovin paradigm into a deadly nosedive


It is well-known that the problem of parsing Chinese annals was not a problem for outsiders, for many centuries it was exclusively an inside problem. In the 1950's were re-published “authoritative” editions of a number of Chinese annals that resulted from a multi-year effort of numerous scientific collectives to parse the annals, which theretofore were published unparsed. A number of corrections made to N.Bichurin translations were caused by the parsing published in China during 1950's, N.Bichurin was parsing his sources himself before translating them. The parsing work is not over, many corrections were made after 1950's. To blame L.Bazin for a critical review of previous parsing is utterly unfair, no canonized parsing even existed at his time, and A.Vovin himself used a revised edition of 1988 that was a revision of the previous publications. The initial parsing of the phrase was done by K.Shiratori in his 1900 publication, he was a recognized expert on reading of the Chinese annals, and both L.Bazin 1948, and A.-M.Von Gabain 1949 benefited from the works of their predecessors. A significant handicap for all scholars was that Türkic languages were foreign for them, and their work was purely scholastic, without innate understanding of the language.

Unlike L.Bazin, who, like J.Ramstedt, tried to make sense out of 秀支 Pin. Xiù zhī, Wade-Giles Hsiu chih, for Türkic speakers that combination was not a puzzle, it was apparent that it stands for Süči, phonetical süchi, meaning army-man, army-trooper, army commander. The polysemantics exists in any language, and needs to be considered in any analysis. In the context, the army going, or army going commander going are synonymous, and the Chinese annals constantly use the terms as interchangeable. In today's lingo they are also synonymous and interchangeable: “Patton advanced” means “Patton's tank army advanced”, and “Patton ordered his tank army to advance”.This reconstruction is not an attempt to be mathematically precise in a situation that precludes precision, it is only a probabilistically better solution than the L.Bazin's alternate parsing. We can be positive only about the root word, that 秀 Xiù = 軍 “army”, an exclusively word for the army.

After berating L.Bazin for his daring to do a disparate parsing, A.Vovin advances his own paradigm, and his own parsing. He breaks down the phonemes of the two verbs, and re-constitutes them to his own purposes. The 替戾剛 Tì lì gāng is phonetically reconstructed to *thij?-re(ts)-kang, and then to t-i-r-ek-ang, the 劬禿當 Qú tū dāng is phonetically reconstructed to *ko-thok-tang, and then to k-o-t-o-kt-ang. Aside from phonetic liberties addressed below, in process the 戾剛 lì gāng ~ *re(ts)-kang becomes phonetically re-parsed into r-ek-ang, thus 戾 = *re(ts) is gaining -k from the 剛, turning into *r-ek, and 剛 = kang is losing the initial -k, turning into *ang. The 禿當 tū dāng ~ *thok-tang becomes phonetically re-parsed into *t-o-kt-ang, thus 禿 tū ~ *thok is gaining -t from 當 , turning into *t-o-kt, and 當 dāng ~ *tang is losing the initial -t, turning into *ang. Unlike the L.Bazin's attempt, this violence to Chinese morphemes is unprecedented, goes against every Chinese written and phonological practices, and is not justified by any most fanciful or flimsy explanations. Rejection of that methodological and phonological aberration negates the whole A.Vovin paradigm.

Reconstructed phonetics

The probability of a chance coincidence falls exponentially with the length of the potential candidates. For 3-phonem words the probability exceeds 100%, i.e. there will be more then a single chance coincidence; for 4-phoneme words the probability decreases by roughly an order of magnitude; for 5-phoneme words the probability decreases by roughly 2 orders of magnitude, and so on. For the 23-phoneme words the probability decreases by 23 orders of magnitude, or 1/1023 (give or take a couple of zeroes), that is that for any language, be it Türkic or English or Enisean, Classic Chinese or Modern Chinese, the probability of  chance coincidence P ≈ 0,000 000 000 000 000 000 000 01. And that value is purely mechanical, without any semantic contents. Adding the semantics for the 3 words (leaving aside the title), and allowing 10 semantically close words out of a conservative 2,000 word lexicon, adds another 6  orders of magnitude. The combined probability of a chance match would be much less than 1/1029 or P ≈ 0,000 000 000 000 000 000 000 000 000 01. The number is mind-boggling. Stringing letters in 1mm-wide small font into a continuous chain, a philologist would have to fly with the speed of light for 1 trillion years just to see the end of the string, and to survive 50 Big Bang events in between. Adding a single well-documented Türkic title Pugu to the calculations would send that poor philologist for another 1000 trillion years. And there will not be any formal grounds to dismiss the probability and save some flying time.

In contrast, finding two 5-phonem constructs, in this case Enisean, would be a child's play, 1/1014, including semantics. It would take 15 zeroes to display the ratio of probabilities:  Enisean/Türkic = 1/101 000 000 000 000 000. With or without formal grounds, statistically the Türkic version is 1 trillion times more probable. You can bet on it, it is a 100% bet.

The review of the Türkic reconstructions highlight their phonetic closeness to the Modern Chinese, and much less conformity with Old Chinese, even modified for the phonetical changes adjusted to the later period of 313 AD. As L.Bazin expressed, and was repeated over again by later scholars, the similarity is striking. The following is an example of Chinese phonetization and Türkic reconstruction, that at the same time differs from the previous attempts, and retains a continuity and phonetic closeness with them:

Chinese 秀支 替戾剛 , 僕谷劬 禿當
Romanized Mandarin Xiù zhī Tì lì gāng Púgǔ qú Tū dāng
English Phonetization Sü chi Ti li gang Pugu chu Tu dang
Chinese to English Army go out Pugu capture
Türkic Süči (Süchi) tiligan Pugu'yu tutar
Türkic to English Army-man would go Pugu (he) would capture
Comment -či (= chi/ji) - std. occupational affix -gan - past participle, 3ps, perfect tense verbal affix -'yu - future conditional verbal transitive affix capture in 3rd person future tense ablative ending

On the last word, tudan/tutar = capture, G. Doerfer made an observation: “It is interesting to note that the Uzbek (i.e. Karluk, i.e. Uigur) loanwords in Dari, or Afghan Persian, often show different forms from those in Tajiki and more often coincide with the standard Persian forms (cf. Kiseleva and Mikolaichik). Thus to Tajiki qapidan “to catch” corresponds Dari and literary Persian qapidan. The Uzbek loanwords in literary Persian, literary Tajiki, and literary Dari still need to be investigated; however, it appears that Dari contains fewer Turkish loanwords than Tajiki.” (Encyclopaedia Iranica, 1995, Volume 5, Sect.14, p. 231) Parsing the 劬禿當 as “chu-tu-dan” would bring us to the Persian/Dari/Tajiki qapidan “to catch”, and this Uigur loanword may excite somebody to claim an Iranian attribution of the Hunnic language. The ultimately Uigur loanword was borrowed in its entirety, with the Türkic ablative ending.

The parsing is different from previous attempts to accommodate a conditional compound Pugu'yu, the modern -'yu (y like York, u like mule) does not ring a q or *ko or ch, which could be a phonetical shift, or the Ogur initial y ringed like ch to the Chinese ear, or -'yu was really -'gyu, as may be attested in some modern dialects of some Türkic languages. L.Bazin must be given a credit, not sacking, for perceiving that tut- is the stem of the “capture”, and -qu belongs to the previous noun, the compound exactly corresponds to the English forms Pugu'll ~ Pugu will and Pugu'd ~ Pugu would. The -ng in the ending is peculiar to the Chinese phonetics, it is unlikely that it reflected Hunnic ending -ng, it is not attested in 35+ modern Türkic languages, but without additional evidence that possibility may be left open. Note that reference to other Türkic languages is conceptually identical to the A.Vovin's references to the dead languages, with an advantage of easy verification and no need to fantasize *reconstructions, and thus is debunking-proof. The offered form follows the attested languages without -ng: tiligan, not tiligang or tiligaŋ. The Chinese -ng in the ending of tutar, because of the absolute semantical id. and nearly absolute phonetical similarity, can be positively accepted as representing the final -r, which is one of the well-known methods of Chinese rendition of -r. With these relatively minor modifications, the modern Mandarin Romanization is extremely straightforward and likewise extremely close to the ancient phrase expressed in the modern Türkic. It is well known that the Chinese language underwent considerable modification during the 16 States period, departing from the archaic language and transitioning to the Old Chinese language. A.Vovin pointed out that the reconstructions of the archaic language must be modified for the Old Chinese of the 4th c. AD. Possibly, the few examples found in the Jie phrase belong to the layer of the language that accepted the modern form in the 4th c. AD. It is as well possible that the selected tiny sampling of the Chinese characters took their form during archaic period, and does not fall under *reconstruction rules numerously and variously deduced from unrelated examples. To claim otherwise is a gross misrepresentation of the extent of our knowledge, the *reconstructions are *reconstructions not because we know, but because we imply. Phonetics is not a religious belief where one climes a pulpit, flashes a sacred book, and claims a supreme knowledge based on exact phrasing written by a supernatural being. From the standpoint of the modern Türkic languages, a return to the archaic language would distance and distort the phonetics of the phrase. A.Vovin's accusation that incompetent linguists used modern Chinese instead of modified Archaic Chinese does not hold the water. In science, when a theory conflicts with practice, it is the theory that is in need of revisiting, and that frequently leads to new and improved knowledge, and even breakthroughs.

In case of -ng (), which E.Pulleyblank found frequent verbal ending in Enisean, and which is also endemic to archaic and modern Chinese, A.Vovin got a good mileage from it, in a fairly distorted way. In the A.Vovin *reconstruction *suke *t-i-r-ek-ang *bok-kok *k-o-t-o-kt-ang, the -ng () is present twice, both times in *reconstructed verbs, since only 2 verbs were examined, but in both cases they are not verbal endings, they are, at least in the A.Vovin interpretation, a part of agglutination denoting -3pp, i.e. “(go) them” and “(capture) them”, contrary to the construct of the type ba-t-tuŋ “he sees me” where -uŋ, with -ng () ending, is “see”. Besides the debacle with A.Vovin unable to draw -ang as a stand-alone morpheme, and his hocus-pokus with magic conversion of -2pp (you) into -3pp (them) under a cover of the E.Pulleyblank's frequent verbal ending in Enisean and G.Starostin *reconstruction, in the A.Vovin paradigm the -ng () gained a utility to mimic the Chinese -ng () to fit the Hunnic Tì lì gāng and Tū dāng in the Chinese transmission. This is a true art of phonytics.

Among the other artistic phonytics we learn that Xiong-nu, like Pumpokol, has -t- corresponding to sibilants in other Yeniseian languages. Accidentally, the Xiong-nu “is known to us only through the about hundred fifty (supposedly useless) glosses and possibly one very short text” under examination, and it is the very unknown that A.Vovin is venturing to attribute, likewise the Pumpokol is known only from “scarce Pumpokol materials”, but nevertheless, based on a single case it is possible to declare regular correspondences between these two unknowns, and even declare “the possibility that the Xiong-nu language, if it is a Yeniseian language, is most closely related to Pumpokol”. This is a classic example of circular logic. If that is a breakthrough, it is not in science.

At some point A.Vovin comes to a conclusion “I agree with Pulleyblank that none of them (prior Türkic reconstructions) can be considered successful” because of the violence done, in the Pulleyblank's and A.Vovin's eyes, to phonetics. And that despite his own words: “Thus, I do not find to be valid any of Pulleyblank's phonological arguments against Altaic affiliation of the Xiong-nu language. It is necessary to add, from the methodological point of view, that even if they were valid, and Xiong-nu did look very un-Altaic, it still could not constitute “proof of its un-Altaic origins, as any argument based on phonological similarity is essentially an argument based on typology, that can be very misleading in the case of genetic affiliation.” At the same time, both damning opinions are grounded on an implied absurd claim to absolute unrestricted knowledge of the 4th c. AD Chinese phonetics in the Hunnic “Later Zhao” state. In reality, the berating is not about the phonological violence, it is about the claim to the absolute truth, because A.Vovin likes the violence done to phonetics in constructing the A.Vovin paradigm.


Unlike Pulleyblank, who issued a general statement, A.Vovin went into great detail on J.Ramstedt's and L.Bazin's work, setting up a high standard, and incidentally exposing himself to that high standard. It is given that the material is not abundant, and every little bit is a throve of information. One noun, one title, two verbs, and associated possible morphemes. The noun is taciturnly accepted as basically Türkic, with no detail analysis on how, why, and any nitpicking phonetics, without even invoking the back and front vowel crucial distinction that could firmly propel *suke into the Enisean fold. This surrender does not bode well for the A.Vovin paradigm, it loses 25% of its ammunition, and gives credence to the paradigm he opposes.

The unaddressed underlying historical scenario is peculiar: the Enisei Hunnic people from 3rd c. BC to 4th c. AD had an army and a borrowed a term for the army *suke = army. With that army the Enisei Hunnic people conquered multiple nations, including Eniseian taiga foot hunters in the S.Siberia. There is no archeological evidence that the Eniseians in S.Siberia are migrants from elsewhere, but there is evidence that Eniseian taiga foot hunters in the S.Siberia are indigenous population. At some time before 3rd c. BC, the Enisei Hunnic people and their armies borrowed the term *suke = army from some unidentified quasi–Türkic elements. In S.Starostin glottochronology, the Türkic language did not exist before the beginning of our era, so the Türkic word for the army came from the linguistic elements that did not speak the language that did not exist yet. However, these glottochronologically unnamed elements came out winners in a long run: the Enisei Hunnic people lost the word *suke sometime between the 4th c. AD and 18th-19th c. AD, while glottochronologically unnamed elements not only donated their word to the Enisei Hunnic people, but also carried it to all 35+ Türkic languages of today. The losses for Enisei Hunnic people were much more severe than a loss of a single lexeme, from a nomadic powerhouse with 400,000 army and 3 mln population (vs. 50 mln in China) and 100,000,000 heads of cattle (at 30-35 heads of cattle per family of 5), they dwindled to a residue of 2,000 people with no cattle whatsoever, no nomadic traditions or culture, no Kurgan culture archeological remains, impoverished taiga foot hunters in the S.Siberia enserfed by Türkic Tele as suppliers of coke, while the more lucky Türks grew, after gaining their own language in the 1st c. AD, from non-existence to 35+ ethnically distinct nations with 130 mln population, and China grew 10 times in space and to 1,300,000,000 population. That's what is called a really bad luck, both linguistically and demographically.

The title is mistreated as irrelevant, it could be a borrowing from unidentified source. In fact, the noun is taciturnly accepted as basically Türkic, with no analysis suggesting alternate source, or a single phonetically reminiscent Enisean word from dead or alive languages, even with the widely open semantic that imposed no limits on the imagination. The J.Ramstedt suggestion of the title being a Türkic term Pugu for Bull, semantically consistent with attested Türkic habit of entitling their leaders with eponymic mythical royal animals, with the use of that title attested by the historical records, and phonetically matching the Chinese 僕谷 Púgǔ (likely a more generic Pugu), is not even blasted into powder based on the archaic *reconstruction *bok-kok, with two (2) extra consonants reprehensively ignored by J.Ramstedt. This second surrender does not bode well for the A.Vovin paradigm either, it loses another 25% of its ammunition, and gives credence to the paradigm he opposes.

The remaining two verbs is all that support the Enisean paradigm. On the attestation and provenance of the first word, 替戾剛 Tì lì gāng /*thij?-re(ts)-kang/*t-i-r-ek-ang = have gone out A.Vovin supplies the following information. The stem is -ek, we can guess it means “come/go” from the reference “Cf. also stem *-ek- and Proto-Yeniseian *-jaq- “come/go””. It is clear that the word is not attested anywhere, it is a *reconstruct *-ek from unidentified unnamed source. What language? Who did the *reconstruction? Where to go to “Compare also stem *-ek?” As A.Vovin stipulated, the word “come/go” belongs to the basic vocabulary, but apparently -ek does not belong to the only alive language of Kets, otherwise there would be no need to turn to the *reconstructed *-ek instead. Unfortunately, the attestation and provenance are not stated. It takes a doze of credulousness bordering on gullibility to accept *-ek for a full face value, but suppose it is accepted that in the 4th c. AD the Huns used *-ek for “come/go”, and passed it on to Chinese for recordkeeping. That would make the phantom *-ek the only extant record of the non-Ket Enisean “come/go”, a certifiably weak evidence for the Ket genesis of the Hunnic language, especially confronted by the mountain of evidence on Hunnic-Türkic genesis.

But taking on faith the reality of *-ek still does not pass the muster. The *-ek has to be obtained from the 替戾剛 Tì lì gāng /*thij?-re(ts)-kang, and that is done with incredulous transformation of *-re(ts)-kang into -ek-ang by a feat of unprecedented mischievous phonetic artistry that would be harder to accept then for a rich man to enter into the kingdom of God. Even with the fast and lose rules of phonetical *reconstructions, that kind of transformations needs to be justified, or addressed, or excused, or whatever that makes it less odious. The conversion of phantom *-ek into real -ek is detestable but may be palatable; the conversion of phantom *-ek into real -ek by the way of conversion the phantom *-re(ts)-kang into indigestible phantom *-ek-ang is not palatable, the conversion of phantom *-ek into real -ek by the way of conversion the phantom *-re(ts)-kang into indigestible phantom *-ek-ang with simultaneous conversion of 2pp into 3pp is a patented nonsense. Even though technically 2pp > 3pp debacle is unrelated to the stem -ek and is only used to fit the stepsister foot onto the Snow White shoe, this is not a lexicon, it does not even rise to a parody on lexicon. Brooklyn bridge on sale or a young mare from a proverbial Gypsy is a far better deal.

The state with the first verb is clear, one more to go. On the attestation and provenance of the second verb, 劬禿當 , Qú tū dāng/*ko-thok-tang/*k-o-t-o-kt-ang = [They] will catch A.Vovin supplies the following. The stem is -kt, it means “to catch”, stated directly in the explanation *k-o-t-o-kt-ang = preverb *k- + o? + object marker *-t- + conjugation marker *-o- + *-kt- “to catch” + aŋ (aŋ ~ ang) 3pp. The apparitional -kt is not attested, it is a *reconstructed *-kt based on a chain of generous allowances:
1. Phonological: “Xiong-nu, like Pumpokol, has -t- corresponding to sibilants in other Yeniseian languages”. This is invalid circularity that uses “Xiong-nu” both as an unknown that A.Vovin attempts to link with Yeniseian languages, and a known used as proof that “Xiong-nu” belongs to Yeniseian languages.
2. Phonological: attested in literature Ket stem -kas- “to take”, “to catch”, or alternate Ket form -qos'- “to take” are not attested in “scarce Pumpokol materials”, and A.Vovin assumes, based on suspected typological example (or more then 1 example?) that Pumpokol would have had a stem -kat- with identical meaning “to take”, “to catch”
3. Even if that was a true fact, there is a phonological transformation of -kat- to -kt-, not supported by any reasons except for a desire to use phonetical extracts from 禿當 *thok-tang > *tho *k-t *ang > *t-o-kt-ang > *toktang. Is Pumpokol famous for its verb stems consisting of 2 (two) consonants? What is the percentage of 2-consonant verb stems in Pumpokol? If 100 Pumpokol stems were known, and 10% of them were 2-consonant, the probability of -kat- losing a vovel to become -kt- would be 10%, or 0.1. If the probability of Pumpokol converting Ket -kas- to -kat- is 50%, the probability of -kas-to -kt- is 0.05. Times the invalid circularity, and you need a telescope to fit all leading zeroes.
4. And although “to catch” and “to*ko-thok-tang capture” belong to the same semantical field, they are not synonymous. 劬禿當 is 捉 “captures”, “will capture”, or “captured”, not “to take”, “to catch”. In the taiga foot hunter parlance, “taking, catching”, and trapping animals is a way of life, “capturing” prisoners is not, the semantical extension from “catching” to “capturing” is an unwarranted move of the goal posts. J Ramstedt is rebuked for bad semantics with impersonal case and modality, for confusing “army” and “war”, for more than free interpretation of “move out” and “go out” or “went out”. L.Bazin is scolded for free interpretation with semantically conditional case, for false segmentation, for using siu for , for reconstructing buquγ, “that is not attested per se”, and “is by no means known” as a semantical title. Even the exact match tut- used by L.Bazin for “capture”, reliably attested in Old Turkic and modern Turkish, is rebuked as speculative. With a small fraction of that same critical eye turned to the A.Vovin *-kt-, the viability of *-kt is nill. It is nill with most generous allowances also, this is not a lexicon.


And from another artistic phonetics we learn that matches between Xiong-nu and Proto-Yeniseian morphology are not chance resemblances, because “In spite of an obvious poor quality of the recording, it is possible to see that Pumpokol... has subject verbal agreement suffixes at the end of the verbal forms.” In the cited single Pumpokol example, following A.Vovin in segregating the verbal root -ding- “stand”, and stipulating the obvious that ever-present -itscha- does not change and therefore has nothing to do with subject verbal agreement

ue-itscha-ding-du “you (sing.) stand” (2ps-itscha-stand-2ps & 3ps) = (you (sing.)-itscha-stand-you (sing.) & s/he)
anjang-itscha-ding-an “you (plur.) stand” (2pp-itscha-stand-2pp) = (you (plur.)-itscha-stand-you (sing.))

ue-stand-du and anjang-stand-an are subject verbal agreement you (sing.)-stand-du and you (plur.)-stand-an, probably not. And “due to the scarcity of Pumpokol materials”, any speculation is iffy. The suspicious morphology is absent in Ket, so the only hints come from the elusive Pumpokol, or its genetic sibling Arin. The morphological line of argumentation starts on iffy ground, but suppose it is accepted.

The next stage compares the *reconstructed Chinese phonetics 替戾剛 Tì lì gāng /*thij?-re(ts)-kang = go out and 劬禿當 , Qú tū dāng/*ko-thok-tang = capture with morphologically adjusted A.Vovin's illusory verb stems *-ek and *-kt, using G.Starostin's 1995 morphological reconstructions of *Proto-Yeniseian. Morphology is independent of the verb stems it applies to, and the probability of chance morphological coincidence has an independent value. Can the *morphology A meet another *morphology B if the do not belong to the same language? G.Starostin's reconstruction has a number of *markers, which can be assessed individually and in combination.

Starting with the preverbs, which are defined as “a class of words in the Chinese language whose function is analogous to the cases, prepositions, and postpositions of other languages, a verbal prefix in the Algonquin languages whose function is analogous to the adjective of other languages”. In light of G.Starostin's *reconstructions, the definition might need to expand to include *Proto-Yeniseian, *Pumpokol, and *Arin. Not a good omen to be an oddball in the human family of languages, but a valid plus in demonstrating unique coincidencies. Putting Hunnic in the same group would necessarily put all those tribes that the Chinese annals identify as Hunnic kins into the same group, for example Dinlin, Chidi, Di, extending to the Di states "Former Qin" (351-394) and "Later Liang" (386-403), which is fraught with undesired paradigmatic consequences.

Without knowing where is the morphology and where is the lexicon in each word, the leeway for imagination is limitless. Suppose there are 20 consonants and 5 vowels, the chances to get the morphological portion of t-i-r-ek-ang and k-o-t-o-kt-ang, which are t-i-r and k-o-t-o, are 20-2 x 5-1 = and 20-2 x 5-2, or 1/2,000 and 1/4,000, or 1 in 2,000 and 1 in 4,000 respectively (assuming that the vowels and consonants are equiprobable, which of course they are not). Long chances, right? Not really, since there are no indications that Hunnic verbs have any particular morphology, any allophonic tir and koto in any language would do, once found, they can be re-interpreted as *Proto-Eniseian morphology. A quick look would find them in English (tire, tired and cotton), (tiraj, tire and kota, kotar), and probably in every other language. In case of tir, Greenberg & Ruhlen methods give an estimate for finding any combination of CVC in any language at 63%, and not finding 37% respectively, based on a 1000-word dictionary, 14C + 10V phonetics, and including all possible allomorphs. For 4-phonem koto, the dictionary must be expanded to ensure 4-phonem minimal lexicon for CVCV, and chances are 38% and 62% respectively. Accepting *Proto-Eniseian morphology of G.Starostin, where verbs start with preverbs *k-, *t-, *p-, the number of choices grows immensely, for each stem there are a *k-stem, *t-stem, and *k-stem forms, for each *k-stem there are *kw-stem, *kt-stem, and *kk-stem forms with object markers, for each *kw-stem there are *kwa-stem *kwi-stem *kwo-stem with conjugation markers, and so on. The numbers grow exponentially, the opportunities for wildcat exploration are vast, applying *Proto-Eniseian morphology guarantees success for any language. At the same time, the value of finding *Proto-Eniseian tir and koto in any unknown language, as far as proving their *Proto-Eniseian genesis, is nil. The available methods allow to calculate the odds with better precision, that wouldn't change the qualitative assessment. Qualitatively, A.Vovin himself preempted this conclusion noting that drawing on monophonemic morphemes “makes the possibility of chance resemblances rise dramatically high” (A. Vovin, 2005, The End of the Altaic Controversy//Central Asiatic Journal 49.1, p. 74). Very dramatically, the PV-CM-PERF- = tir and PV-?-OBJ-CM- = koto are equally applicable to English and Turkish when rendered in *reconstructed Old Chinese phonetics.

In light of the phoneme -ŋ being endemic to Chinese, adding the putative broken off -aŋ (aŋ ~ ang) read as 3pp does not change the picture. Like in the case of the probabilities for the tir and koto, the weight of the Chinese endemic ending -ŋ only raises chances of the chance resemblances. The Proto-Yeniseian subject verbal agreement may be reasonably reconstructed on the basis of Pumpokol and Kott system rather than Ket, on which apparently G.Starostin has sufficient data for typological conclusions, but it has nothing to offer to Hunnic, where with the exception of circular logic any typological data is absent, and worse, the dismembered resemblance is purely phonological, while “any argument based on phonological similarity is essentially an argument based on typology, that can be very misleading in the case of genetic affiliation.” It seems that this is precisely the case.


The reviewed properties of the Hunnic phrase, the language type, number of words, parsing, reconstructed phonetics, lexicon, and morphology, each one separately, and in aggregate, demonstrate that the Enisean paradigm is inviable in any configuration, Ket, Pumpokol, or Proto-Yeniseian, and it can't rely on the only evidence offered, two (2) highly questionable verbs and unjustified morphological resemblance. The critique of the Türkic affiliation is superficial, inconsistent in application to pro and against paradigms, and in a twist of logics taciturnly accepts as many Türkic words as it tries to debunk. The enthusiasm afforded to the Enisean paradigm it totally unwarranted. A number of linguistic theories turned out to be anywhere between wrong and absurd, and the process of linguistic rectification of enthusiastic admiration is far from complete for many languages.



Baxter, William 1991. A Handbook of Old Chinese Phonology. Berlin & New York: Mouton de Gruyter.
Bazin, Louis 1948. “Un texte proto-turc de 4e siecle.” Orient! 1: 208-219.
Clauson, Gerald 1972. An etymological Dictionary of Pre-Thirteenth Century Turkish. Oxford: Clarendon Press.
Doerfer, Gerhardt 1973. “Zur Sprache der Hunnen.” Cenlral Asiatic Journal 17.1: 1-50.
Dulzon [Dulson], Andrei P. 1968. Ketskii iazyk. Tomsk: Izdatel'stvo Tomskogo universiteta.
Gabain, Anne-Marie 1949. Review of Louis Bazin. “Un text proto-turc due 4e siecle.” Der Islam 29: 244-246.
Helimsldi, Evgenii A. 1985. “Arkhivnyc materialy XVIII veka po eniseiskim iazykam.” Paleoaziatskie iazyki, ed. by Petr la. Skorik, pp. 179-213. Leningrad: Nauka.
Karlgren, Bernhard 1940. Grammata Serica Recensa. Stockholm: Museum of Far Eastern Antiquities.
Kreinovieh, Eruxim A. 1968a. Glagol ketskogo iazyka. Leningrad: Nauka.
Kreinovich, Eruxim A. 1968b. “Ketskii iazyk.” Tazyki narodov SSSR, vol. 5. ' Mongol'skie, tunguso-man'chzhurskie, ipaleoazialskie iazyki. pp. 453-473. Leningrad: Nauka.
Kreinovich, Eruxim A. 1979. “Ketskii iazyk” [sketch of Yug “dialect” - A.V.]. Iazyki Azii i AfriM, vol. 3. Iazyki Drevnei Parednei Azii, Iberiisko-Kavkazskie, Paleoazidtskie iazyki. pp. 330-847. Moscow: Nauka.
Ligeti, Lajos I960. “Mots de la civilization de Haute Asie an transcription Chinoise.” Ada Orienlalia Hungarica 1950.1: 141-188.
Nadeliaov, V. et al, ed. 1969. Drevne-tiurkskii slovar'. Leningrad: Nauka.
Pelliot, Paul 1944. “Tangrim > tarim.” T'oung-pao pp. 165-185.

Pulleyblank, Edwin G. 1962. “The Consonantal System of Old Chinese.” Asia Major 9: 58-144, 206-265.
Pulleyblank, Edwin G. 1973. “Some New Hypotheses Concerning Word Families in Chinese.” Journal of Chinese Linguistics 1: 111-125.
Pulleyblank, Edwin G. 1991. “The Ganzhi as Phonograms and their Application to the Calendar.” Early China 16: 39-80.
Pulleyblank, Edwin G. 1992. “How Do We Reconstruct Old Chinese?” Journal of American Oriental Society 112: 365-382.
Pulleyblank, Edwin G. 1994. “Reply to Baxter's Reply.” Journal of Chinese Linguistics 22.1: 161-169.
Ramstedt, John Gustav 1922. “Die Stellung des Tschuwassischen.” Journal de la Societe Finno-Ougrienne 38: 1-34.
Shiratori, Kurakichi 1900. Uber die Sprache des Hiung-nu Stammes und der Tung-hu-Stdmme. Tokyo [n.p.]
Starostin, Georgii S. 1995. “Morfologiia ketskogo glagola i rekonstruktsiia praeniseiskoi glagol'noi sistemy.” Ketskii sbornik. Lingvistika. pp. 122-175. Moscow: Vostochnaia literatura & Shkola “IazyM russkoi kul'tury.”
Starostin, Sergei A. 1982. “Praeniseiskaia rekonstruktsiia i vneshnie sviazi eniseiskikh iazykov.” Ketskii sbornik. Antropologiia, etnografiia, mifo-logiia, lingvistika. pp. 144-237. Moscow: Vostochnaia literatura & Shkola “Iazyki russkoi kul'tury”.
Starostin, Sergei A. 1989. Rekonstruktsiia drevnekitaiskoi fonologicheskoi sistemy. Moscow: Nauka.
Starostin, Sergei A. 1991. Altaiskaia problema i proiskhozhdenie iaponskogo iazyka. Moscow: Nauka.
Starostin, Sergei A. 1995, “Sravnitel'nyi slovar' eniseiskikh iazykov.” Ketskii sbornik. Lingvistika. pp. 176-315. Moscow: Vostochnaia literatura & Shkola “Iazyki russkoi kul'tury.”
Todaeva, Buliash Kh. 1973. Mongorskii iazyk. Moscow: Nauka.
Toporov, V. N. 1968. “Materialy k sravniterno-istoricheskoi fonetike eniseiskikh iazykov. 1. Arnisko-eniseiskie sootvetstviia.” Ketskii sbornik. Lingvistika, ed. by Viacheslav V. Ivanov, V. N. Toporov, B. A. Uspenskii. Moscow: Glavnaia redaktsiia vostochnoi literatury.
Vail, Margarita N. & Kanakin, Igor' A. 1990. Ocherk fonologii i grammatiki ketskogo iazyka. Novosibirsk: Nauka.
Verner, Genrikh [Werner, Heinrich] 1990, Kottskii iazyk. Rostov-na-Donu.
Verner, Genrikh [Werner, Heinrich] 1997a. “Eniseiskie iazyki.” Iazyki Mira. Paleoziatskie iazyki. pp. 169-177. Moscow: Indrik.
Verner, Genrikh [Werner, Heinrich] 1997b. “Ketskii iazyk. ” Iazyki Mira. Paleo-aziatskie iazyki. pp. 177-187. Moscow: Indrik.
Verner, Genrikh [Werner, Heinrich] 1997c. “Iugskii iazyk.” Iazyki Mira. Paleo-aziatskie iazyki. pp. 187-195. Moscow: Indrik.
Verner, Genrikh [Werner, Heinrich] 1997d. “Kottskii iazyk.” Iazyki Mira. Paleoaziatskie iazyki. pp. 195-203. Moscow: Indrik.
Yakhontov, Sergei E. 1986. “Iazyki sevemykh sosedei Kitaia v pervom tysi-acheletii do n. e.” Istoriko-kul'turnye kontakty ndrodov AUaiskoi iazyko-voi obshchnosti Tezisy dokladov XXIV sessii postoiannoi mezhdunarod-noi Altaisticheskoi konferentsii (PIAC). vol. 1: 123-134. Moscow: Institute of Oriental Studies.

Professor Alexander Vovin comments (the responses are embedded in the Prof. A. Vovin comments)

There are no reviews of my article in question, but it is favorably cited in various sources. Unlike books, articles are not reviewed - they might trigger academic discussion, though.

This "review" gave me a series of good chuckles. Its author needs to do some basic reading in historical linguistics, anthropology, and Eurasian history unless s/he wants to be simply entertaining.

Well, I am not going to rebute every single word or statement in this review, it is simply not worth my time, so just few major things:

1) It is a fallacy to use modern Chinese readings of characters: they were very different 2,000 years ago. (In the real word, no fallacy can bring productive results. If the results are real, corroborated by annalistic records, historical records, demographical records, ethnological records, genealogical records, genetical evidence, the fallacy lies elsewhere)

2) Oh, yes, DTS includes many words with initial n-. But before expressing his indignation at the well-known fact in Turkic historical linguistics that practically no native Turkic words could begin with m-, n-, n'-, ng-, l-, and r-, author should have looked very attentively: all of them with the exception of ne 'what' and its derivatives (and ne has a very complex phonological history) are loans from Persian, Sanskrit, Arabic, etc. (Supposedly, there was no limitation on the Huns to use exclusively Turkic lexicon. Neither was a limitation on Enisean to use exclusively Enisean lexicon, the Enisean reconstruction used Turkic “army” and Turkic title. Anyway, this comment is irrelevant since Prof. A. Vovin himself already debunked this Pulleyblank's thesis, the subject is not pra-Turkic-Turkic correspondences but Hunnic-Turkic correspondences, and none of the words in the phrase begin with m-, n-, n'-, ng-, l-, and r-, they begin with s-, t-, p-, and t-)

3) Ditto for the author's indignation that both Pulleyblank and myself use only tiny fraction of Xiong-nu words. This is no wonder, as most of these 200+ Xiong-nu words are personal names, which are useless for identifying the genetic affiliation of a language. Just to give one example, my first name does not make me a native speaker of Greek (There is so much truth in that, especially with the Chinese coding, replacement of the names and use of conditional names. But in case of "Iranian Scythians" that approach not only was not abandoned, it was singularly used by V.Abaev and enthusiastically accepted by many European scholars as valid evidence, and widely cited as an uncontroversial proof. Sounds like Do what I say, not what I do.)

4) It is well known that linguistic typology is meaningless for proving or disproving genetic relationships of languages. 75% of the languages of the world have SOV word-order like Turkic., e.g. Sepik in the New Guinea. On the other hand, Greek is SVO, and Sanskrit is SOV, and both are related. (The Turkic reconstruction of the phrase does not use the typology, while Prof. A.Vovin's Enisean reconstruction is the result of relying on typology, the premise is valid, its application for Enisean paradigm was not. The phrase's SOV order is documented in the Chinese record, and the Turkic reconstruction demonstrates the conformance to the SOV word-order as a consequence of reconstruction, not as its premise)

5) As I said above, the author has to do a lot of reading in different disciplines, unless s/he wants to strike his readers with "Turkic Scythians" (Skythians spoke an Eastern Iranian Language), his/her lack of understanding how deculturation process can occur (e.g. pastoralism > fishing/hunting/gathering), and lack of understanding how comparative linguistics and philology really work. (This all may be true, and pertinent in analyzing the messenger, but has nothing to do with the phrase on catching the Pugu in any language. Skythians spoke an Eastern Iranian Language because V.Abaev used personal names, which Prof. A.Vovin defined as “useless for identifying the genetic affiliation of a language”, V.Abaev did not dare to recite the real documented Scythian words. V.Abaev's conclusions are a fraud based on a fallacy acknowledged by Prof. A.Vovin. See G.Dremin's Scythian Lexicon, for example.
For how comparative linguistics and philology really work many a philologists have their own opinion on how it works and how it does not, they all must belong to the bunch that either does not understand or understands too well)


In Russian
Contents Huns
Contents Türkic languages
Literature Index
Ogur and Oguz
Alans and Ases
E.Pulleyblank Eastern Hun Language
O.Pritsak Onomasticon of Western Huns W.B.Henning Xiongnu are Huns
L.Gumilev Language of Huns
Kisamov N. The Hunnic Oracle
Tekin T. Hsiung-Nu Language
Alan Dateline
Avar Dateline
Besenyo Dateline
Bulgar Dateline
Huns Dateline
Karluk Dateline
Khazar Dateline
Kimak Dateline
Kipchak Dateline
Kyrgyz Dateline
Sabir Dateline
Seyanto Dateline
Рейтинг@Mail.ru “” Türkic A.Vovin Enisean “”