Gerhard Doerfer (1920–2003) was a German Turkologist. The first and only English translation of “Zur Sprache der Hunnen ― On the Language of the Huns” in its entirety and in an exponentially better format is coming up soon.
Gerhard Doerfer, “Zur Sprache der Hunnen ― On the Language of the Huns” in Central Asiatic Journal, 1973.
And see there’s nothing we can know!
It almost sets my heart burning.
[Faust Part I: Scene I: Night, 364–365.]
§ 1. The problem of the language of the Huns is extremely controversial. A wide variety of opinions have been expressed without even the possibility of a solution ever arising. Under these circumstances, it seems almost presumptuous to even deal with the subject, let alone to make a definitive statement about it. So, I cannot say anything more than my own opinion—in addition to which others are easily possible. However, I would ask you to consider that an overall view of the Hun material is offered here, which should only be refuted and replaced by a new overall view, but not by the criticism of details.
I ask the reader’s forbearance if I have been extremely brief (which can easily give the impression of a certain—unintentional—harshness). Basically, this would be a (much comprehensive) dissertation, if not a habilitation topic. However, the results of this work are critically negative—and who wants to write a negative dissertation?
§ 2. The origin of the Huns is historically and archaeologically controversial: Since Deguignes (1756–8) there have been repeated researchers who assumed their identity with the East Asian Hiung-nu, who moved westward in a process that lasted centuries (Hirth 3310 (1899), Schuster 3318 (1940), Jettmar 3302 (1951/2) — albeit cautiously and subtly —, Pritsak 3304 (1954), Takats 3331 (1959), Shafer 1966 and others); There have also been attempts to connect the European Huns with an East Asian Hun people (Hün) or Ḥūn (Xwn), which is not identical with the Hiung-nu (Altheim-Haussig 3293 (1958)). On the other hand, since Abel-Rémusat (1820) there has been the thesis that the Huns' connection to any East Asian people (especially the Hiung-nu) is completely unproven and should be viewed with the greatest skepticism (Ritter 1832, Maenchen-Helfen 3300 (1944–5) ff., Thompson 3287 (1948), Enoki 3301 (1955), Moór 1963; Altheim-Haussig 3293 (1958), p. 14 also basically approaches this view.
In this case, can linguistics step in to help? Is it able to prove or disprove the identity of the Huns with the Hun or Hiung-nu? Is it also able to classify the language of all these peoples into a certain already known language family (such as the Turkic one)?
§ 3. The language of the Hun (Xwn) is completely unknown (at least not worked on); no source seems to provide any information about it. Therefore, a comparison with the language of the Huns (or the Hiung-nu) is not possible. And even if the Huns were originally identical with the Hun, this would still say nothing about the people — and that means above all their language. The Huns, moving westward, could have been a small master race in a broad stratum of subjugated people, losing their language (as well as their culture) and trading it in favor of another.There are an infinite number of such examples in the history (especially of the Altaic and other nomadic peoples) (Bulgarians, Turkish Mamluks, Mongols in Iran, etc.). Let us also think of the Normans, who over time used three different languages (Scandinavian, French, English). The identity of Huns =? Hiung-nu =? Hun is not only an archaeological problem (Jettmar 3302 (1951/2), 178), but also a linguistic one.
§ 4. About 20 words and a short distich in Chinese writing (comprising 10 syllables) have so far been examined from the language of the Hiung-nu. The paucity of the material corresponds to the abundance of hypotheses. The following hypotheses can be found regarding the identity of the Hiung-nu (older literature, especially according to Inostrancev 3299 (1900), Shiratori 2873 (1902)):
A. They were Mongols (such as Pallas, Bergmann, Venelin, Ilovajskij, Hyacinth, Neumann; also Deguignes, but he hardly differentiates between Mongols and Turks).
B. They were Turks (Abel-Rémusat, Klaproth, Semenov, Ritter, Koskinen, Shiratori 2873 (1902), Pritsak 3304 (1954), Samolin 3309 (1956), Altheim passim; differences here, too: According to Altheim they practically spoke = Old Turkic, after Pritsak Bulgar Turkic).
C. They were Finns (Saint-Martin).
D. They were a mixture of Turks and Mongols (Parker, Cahun).
E. They were a mixture of Mongols and Tungus (Shiratori 2863 (1923)).
F. They were a mixture of Turks, Mongols, Tunguses and Finns (Castrians).
G. They were not a tribe at all, but a political association (which means practically indefinable mixedness and multilingualism, according to Lacouperie).
H. They were Kets (“Yenisei Ostyaks,” according to Ligeti 2574 (1950), Pulleyblank 1962).
I. They were most likely Iranians (Moor 1963, 65f.).
J. “And yet is impossible to affiliate the Hsiung-nu language with one of the great linguistic families of Eurasia,” (Maenchen-Helfen 3300 (1944–5), 224f., similarly the same 3307 (1959), 225 and already Ligeti 1941–3, also 2574 (1950), 142).
As can be seen, opinions are still controversial even in recent times. We certainly cannot follow Samolin’s argument (3309 (1956), 149) that, according to the Chinese sources, the Turks were descendants of the Hiung-nu, i.e., the Hiung-nu spoke Turkic. Anyone who knows about Chinese indifference towards “barbarian” cultures and languages will find this to be a completely irrelevant testimony.
§ 5. The Hiung-nu material breaks down into a wealth of glosses (Pulleyblank 1962, 240 claims to have collected 190 of them, Maenchen-Helfen 3300 (1944–5), 224f. even speaks of “hundreds of Hsiung-nu words” in the Chinese sources) and (apparently) into the couplet mentioned. As is well known, isolated glosses (and in the Chinese script, which, among other things, does not distinguish l and r, and does not recognize the Turkic-Mongolian vowels ö and ü) are often difficult and can only be interpreted without certainty. Here, the oracle’s saying — which represents a text with a Chinese translation (in Chin-shu) (e.g. “Lead the army out, seize the commander.”) — should ensure certainty. However, this is not the case for the following reasons:
a) This text is also written in unclear Chinese script. In fact, the original was almost always interpreted as Turkish by the various authors (and edited accordingly), but in detail it was read and interpreted very differently, see the literature in Ligeti 2574 (1950), 143, Pritsak 3333 (1954), 135, Doerfer I (1963), 96. The latter suggested (but only ironically and in order to reduce the previous interpretations to absurdity) that the text should be understood as Akkadian and showed that this was easily possible (he might as well have used the Eskimo as an example). Also, according to Pulleyblank 1962, 264f. the couplet cannot be explained from any known language (not even from Ket).
b) As Maenchen-Helfen 3307 (1959), 225 noted, even if the text were actually Turkic, it would need to be originating from the 4th century AD, and in no way it would be related to the language of the Hiung-nu Mo-tun’s of the 3rd century BC. be identical (“William the Conqueror's ancestors did not speak French”).
c) Finally, according to a finding by Ligeti in 1941–3 who checked the original source carefully, the couplet is not written in Hiung-nu at all, but in Ho language, and one can, in no way, claim that Ho = Hiung-nu.
§ 6. So much for the only “Hiung-nu” text. The glosses, in turn, consist largely of names; as is well known, these present great difficulties in interpretation, see § 15. It is a fact that so far (see § 4) 10 different theses have been put forward about the linguistic identity of these glosses, and that one and the same author (Shiratori)
once (1902) interpreted the same material as Turkic and once (1923) as Mongolian-Tungusian. By the way, almost all of Shiratori’s interpretations are disputed by almost all other researchers. It would be superfluous to refute all of the theses that have been put forward about Hiung-nu here; they are mostly based on some random word
similarity. Only two possibilities will be discussed here, for which there seems to be certain evidence:
The thesis that Hiung-nu is Turkic still largely prevails today. This is based on the undeniable fact that Hiung-nu and Turkic have some words in common, namely:
Hiung-nu ch'eng-li = Turkic täηri ‘Himmel’ (Shiratori 2873 (1902), Nr. l; Ligeti 2574 (1950), 143; Altheim 1951, 61, 95; Pulleyblank 1962, 240; Doerfer II (1965), 585),
hiep-ho, χiǝp-γǝu = yabγu ‘a title’ (Altheim 1951, 53; Altheim-Stiehl 3292 (1953), 49; Doerfer IV (1973), Stichwort 1825),
eu-ta, wo-lu-to, ao-t'ot = ordo ‘Army Camp’ (Shiratori 2873 (1902), No. 5, here interpreted differently; Doerfer II (1965), 35, 39).
Altheim (1951, 20, also 3291, I 7) wanted to conclude from this that the Hiung-nu must have been Turks (indeed, they would have spoken exactly Old Turkic). The following objection can be raised against this:
a) The vast majority of Hiung-nu words cannot be explained by Turkic.
b) There is also a circularity in Altheim’s thesis: it would only be conclusive if the words quoted above were first proven for Turkic and later for Hiung-nu. In fact, the Hiung-nu words are attested in the 2nd century BC, while the Turkic are documented only in the 8th century AD. Therefore, the assumption that these are Hunnic loanwords in Turkic is much more likely (see Ligeti 2574 (1950), 143 and Pelliot before him). The Hiung-nu word ordo, for example, has found its way into the European language (German “Horde”).
c) Furthermore, none of the words given (which are too long for a Turkic root) can be etymologized from Turkic (Altheim’s interpretation “Lord of the Archers” for yabγu, for example, is untenable; this would have to be *yačï bägi). Altheim-Stiehls 3292 (1953), 37 ff. Reference to the uncertain Proto-Bulgarian (Shumen) *ya-bäg does not hold water. A development of Turkic ya ‘bow’ + Eastern Iran baγu < bagam ‘the god’ > yabaγu > yabγu is quite unlikely.
d) Finally, täηri has a very strange structure: the sound combination -ηr- and the entire shape of the word are un-Turkic. cSince a variant *taηrï also appears in many Turkic languages (Turkish, Azerbaijani, tanrï, Turkmen, täηri, Yakut, taηara, Chuvash, tură), it makes sense to assume the original form as *taηri (which was later adapted in various directions of Turkic vowel harmony). However, this could not be originally Turkic because of the lack of vowel harmony.
The fact that there are words common to Turkic and Hiung-nu does not take us a step further, since these common words are actually primarily Hiung-nu, and only secondarily borrowed in Turkic. If we were to insist on Altheim’s assumption, it would be no different than concluding from Arabic haikal ‘temple’ and Sumerian é-gal (same meaning) that Sumerian was an Arabic dialect, although in fact the Arabic word ultimately comes from Sumerian.
§ 7. The thesis of the Yenisei-Ostjak origin of the Hiung-nu was first suggested by Maenchen-Helfen (3300 (1944–5), 224). It was then illustrated by Ligeti 2574 (1950) using the example of the Hiung-nu word so-to, older *sâkďâk = sagdaq ‘boot,’ which can only be related to Ket sāgdi, śāgdi. Finally, in 1962, Pulleyblank treated the subject monographically by identifying 12 of the 190 Hiung-nu words he had collected as Ket. The following objections can be made to Pulleyblank:
a) An arbitrary selection of 12 words out of 190 does not provide reliable and sufficient evidence. Simply based on the rules of probability, some such correspondences will always be found, no matter what language one takes as the basis. For example, one could relate ku-t’u ‘son,’ older *kwah-δah instead of Ket qalek, falla just as good with Khalaj Turkish qāl ‘child.’ (He who seeks will find.)
b) Therefore, many of Pulleyblank’s examples can also be interpreted differently. For *saγdaq = Ket sāgdi, the author himself points to the possibility of explaining the word using Middle Persian (mōčak) säxtak “boot of prepared leather.”
c) Ket is a language that has only been documented (in individual glosses) since the 17th century. It is separated from the Hiung-nu text by a period of almost two millennia. And yet the Hiung-nu text is written in Chinese script, which on the one hand (see above) is highly unsuitable for reproducing foreign sounds and on the other hand the sinologists themselves are still in complete disagreement about the correct reading in older times (Serruys, Haloun, Karlgren interpret the Chinese pronunciation quite differently from Pulleyblank and each sinologist differs from each other). Is an exact comparison even possible under these circumstances? Wouldn’t that be like comparing Javanese pat ‘four’ with (let’s assume we don’t know the Latin) Romanian patru (same meaning), which is palpably wrong? If we know neither the Chinese pronunciation of the 2nd century BC nor the Old Ket forms of that time, then we have an equation with two unknowns.
d) Finally, Pulleyblank has overlooked the possibility that—assuming some of his equations are correct—these could be Hiung-nu loanwords in Ket (just like German “Horde”). These words are only sporadic glosses, mostly cultural words, and such words migrate easily. It would be different if, for example, the Hiung-nu number words from 1 to 10 could be traced exactly in Ket. Here, we would have a systematically ordered group of basic words (and in this case, even the difficulty in c) could be overcome), but nothing like that can be found in the Hiung-nu material.
However, Ligeti-Pulleyblank’s thesis still seems the least improbable. Some words do seem convincing, such as Hiung-nu chieh < *ki̯at ‘stone’ = Ket khes, kit. I would also include here: Shiratori 2873, no. 11 t‘ieh-fah ‘iron’ < *tiet-bat. As Pulleyblank has shown in JAOS 85 (1965), 121–5, a consonantal pronunciation can be neglected in the reading even in very old texts, i.e., the actual pronunciation is *tieba. This form, however, could be Old Ket; according to Ligeti 2574 (1950), 151, we find ťip, ťep in today’s dialects.
Also revealing is the fact that there are a lot of words in Hiung-nu that start with l- (21 of 190), see Pulleyblank 240, 244. In Turkic and Mongolian, there are apparently no original words with l-. However, these are characteristic of Ket! (Incidentally, certain Old Turkish words such as lačïn ‘falcon,’ laγzïn ‘pig’ may be loanwords from Hiung-nu, but possibly also from Avar, see Doerfer IV, keyword 1728.)
§ 8. Let us summarize: Hiung-nu is certainly not a Turkic or Mongolian language. It is probably an isolated language that is now extinct (like Ugaritic or Sumerian). There is a faint possibility that Hiung-nu survives in present-day Yenisei Ostyak (possibly only as an adstratum), but the evidence for this is doubtful and insufficient.
§ 9. Before we come to the language of the (European) Huns, a few words about the ethnic name itself, for it should be a particularly important element in determining Hunnic ethnogenesis. Its etymology is already controversial (see Moravcsik 236f., Maenchen-Helfen 3307 (1959), Altheim 3291, I 7f., Inostrancev 3299 (1900)). It is also disputed whether the name of the Huns (Oὗννoι, Hunni etc.) is identical with the name of the Hiung-nu, and how the latter name is to be etymologized. It is often assumed that Hun = Xwn (in the Sogdian letter, see Henning 3305 (1948)) is equal to chines, hun or hün, but not to Hiung-nu (see Bussagli 3303 (1950), Moravcsik 236 f., Maenchen-Helfen 3307 (1959) and 1961, Pritsak 3304 (1954) and 3306 (1959), Altheim-Haussig 3293 (1958), Altheim 3291, I 7f., Shafer 1966, Haussig 1969, Jettmar 3302 (1951–2), Pulleyblank (1962, 139). Furthermore, it is not necessarily clear whether the name Xoῦνoι (in Ptolemy, 2nd century) = Oὗννoι, Hunni (Priskos, Jordanes, 5th and 6th centuries), and whether these peoples were identical or not (Shafer 1966, Moor 1963 and others argue for identity, against Haussig 1969). Also, the name of the Huns appears in so many places and times that an identity of Attila’s Huns with all these peoples cannot be clear at all (Bussagli 3303 (1950), Moravcsik 236f., Maenchen-Helfen 3290 (1955) and 3307 (1959), Shafer 1966).
We will not delve into all these broad questions and will limit ourselves to the Huns of Attila's time and Attila’s empire, say, 5th-century Pannonia. As long as we cannot compare the language of the Ptolemy Huns with that of the Priscus Huns, for example, we know nothing about their identity, and the same applies to the Caucasian Huns of the 6th century.
§ 10. However, even if we limit ourselves to the European Huns, a number of questions arise. Yet, these questions can only be asked (and the problems can only be seen) if clear concepts are first established. I will, therefore, make the following distinctions:
a) Language names. Example: English, although encompassing a range of peoples (English, American , etc.), is a monolithic term (i.e., it denotes an essentially unified block, despite dialectal differences). Furthermore, this term is both diachronic and synchronic (there has been English before and there is English now as an existing system), and it is genuine (it is the actual term, English, used by the Anglo-Saxons themselves). Of course, “Anglo-Saxon” would also be a language name or speaker's name.
b) People names. Example: English. The same characteristics apply here as for language names (synchronic and diachronic, genuine, monolithic). However, it refers to a national, not a linguistic community.
c) State names. Example: Great Britain, Switzerland. Synchronic and diachronic, genuine, but not (necessarily) monolithic (can include a number of languages and peoples at a time). Also British, Swiss.
d) Migration names. Example: Prussians, French, Bulgarians. These names are the result of historical contacts, but are actually no longer applicable. (The Prussians are no longer Balts, the French no longer Franks, the Bulgarians no longer Bulgar Turks). Diachronic origin, genuine, not necessarily (but often) monolithic.
e) Collective names. Synchronic, inauthentic, non-monolithic. These names are “inauthentic” insofar as they are not reflected in the self-designations of the peoples (or language communities) concerned, but are based on ignorance or indifference of the sources. For example, all British people (including Scots, Valaisans, etc.) are often referred to as ”English” in Germany. Collective names often result from the fact that ethnic names of the decisive element in a state are wrongly used as state names (more precisely, as citizens’ names). This is also the case, for example, when Soviet citizens are referred to as “Russians” today. (In contrast, “Russian,” applied to a Ukrainian in pre-revolutionary Russia, for example, was a quasi-correct state name. And the situation is different with “Bulgarians.” Of course, the term is “wrong” insofar as the Bulgarians are not Bulgar Turks; however, it is a correct self-designation, hence the migratory name. This may also have been used as a collective name by foreigners in the past. The “decisive element” does not necessarily have to be the numerically strongest.
f) Transfer name. Diachronic, inauthentic. Such names result from the fact that names of peoples known from ancient times were transferred (out of ignorance or indifference or out of the intention of deliberate insult) to peoples who were similar to the previously known peoples, who emerged from the same area, etc. Such transfer names are frequently found in Byzantine literature. Thus (according to Moravcsik 279-283), Σχύθαι designates: Attila’s Huns; Kutrigurs, Utigurs, Onogurs; Turks (who should actually be called τοῦρχοι); Avars; Khazars; Bulgars; Hungarians; Pechenegs; Uzes; Cumans; Seljuks; Mongols; Ottomans; (according to ibid. 13) also Goths, Slavs, Russians (and of course the Scythians themselves). The term τοῦρχοι, türk, on the other hand, refers not only to Old Turks, but also Khazars, Hungarians, Ottomans, etc. (and according to Doerfer II 489 in Islamic sources also Mordovians, Ostyaks, Tibetans and Russians). One and the same name is, thus, used for different peoples; and one and the same people appears under different names; see especially Moravcsik 13–17. A disparaging transfer name: the designation of the Germans as “Huns” in the First World War. The line between e) and f) is often difficult to draw.
g) Foreign names. Many of the names that appear in the sources do not correspond at all with the self-designations of the peoples concerned. This is partly due to the fact that a whole people was named after the part that was first known to the foreign people (the Germans as allemands = Alemanni among the French, as salcsalaiset = (Lower) Saxons among the Finns). However, there are also names that have nothing to do with the peoples in question, either because they are disgusting names (i.e., contemptuous terms such as samojed, ‘self-eater, cannibal’ for the Samoyeds), or because they are taboo names (a rare case). I would also regard “Tatar,” used in Western sources for the Mongols of the 13th/14th century, as a taboo name. Although the Tatars were actually exterminated (see below), it nevertheless became the common name for the Mongols, their mortal enemies. The name could have been chosen by the Mongols themselves in order to escape the revenge of the national gods (after the slaughter of so many human lives), see E. Haenisch: Die Geheime Geschichte der Mongolen (The Secret History of the Mongols), Leipzig 1948, 137 f. By calling themselves Tatars to the Russians etc., the revenge of the national gods was transferred to the enemy tribe (see Uno Harva: Die religiösen Vorstellungen der altaischen Völker (The Religious Ideas of the Altaic Peoples), Helsinki 1938, 420; Hans Findeisen: Schamanentum (Shamanism), Stuttgart 1957, 24f.). Of course, transfer names are also foreign names, but the historical cause is quite different.
h) Homonymous names by chance. This case may not seem too frequent, but it must be taken into account. If two units appear under an identical or similar name, this does not necessarily have to be causal (i.e., due to historical contact), but it can be due to mere coincidence. For examples, see Maenchen-Helfen 3290 (1955): Wallonen and Walliser, Venedig and Venedi (Wends); the author also lists a number of cases which I would call wandering or transfer names. Examples of such coincidences can be found in any list of tribal names, for example in G. M. Vasilevic: Evenkijsko-russkij slovar’, Moskva 1958, 5576 Evenki tribal names such as: Alagir (cf. Alans), Bagdalid (cf. Baghdad), 579 Kantagir (cf. Cantabria). However, if such coincidences occur with quite obviously unrelated peoples’ names, i.e., of peoples who are geographically much too far apart, it can of course never be ruled out that they also occur with peoples who (chronologically one after the other) appeared in roughly the same area, e.g., the Xoῦνoι of the 2nd century and the Oὗννoι of the 5th century.
§ 11. As modern Europeans, we are generally accustomed to fixed terms and clear units (such as Italian, Greek, Swede). The categories a to c are well known to us, and we are all too inclined to consider them alone under all circumstances. In contrast, the more fluctuating categories d to f appear very often in older (not only oriental) sources.
In the Secret History of the Mongols (13th century), “Mongol” refers to a small tribe (from which Chinggis Khan, among others, sprang) and it is a people’s name. In later Islamic sources (Rašīd ad-Dīn, 14th century), it is also used as the name of peoples who used the same language as the Mongols (Oirat, Tatar, etc., i.e., language name), and on the other hand all these “Mongols” themselves are sometimes listed under the collective name “Turks” (Doerfer II 489). Through historical contacts, however, “Mongols” also became a migratory name. It appears as a tribal name among the Özbeks in Afghanistan (H. F. Schurmann: The Mongols of Afghanistan, ‘s-Gravenhage 1962, 99–101), as well as among Evenki tribes (Vasilevic 581: Momol, Mongo, Mongoli, Mungal—although this could also be a coincidence). Of course, “Mongols” also appears as a state name in contemporary sources of the 13th/14th century: the “Mongol” warriors of the Golden Horde, for example, were predominantly Turks (see Spuler: Die Goldene Horde, Wiesbaden 1965, 281–5), but subjects of the Mongol Empire (at the same time, this is a collective name). (So if today’s Turks around Kazan are called Tatars, this is not actually a language name, at least not historically, but it is, from various perspectives, a migratory name, a name of transmission and a foreign name). And finally, if at a certain time Red Army soldiers who belonged to the High Asian race were called ”Mongols” in Germany, this is a transfer name.
Eastern names are often ambiguous, for example “Khalaj” refers to a Turkic people in central Iran on the one hand, but also to a Kurdish tribe in northeastern Iran on the other; the author has collected linguistic material on both types of “Khalaj.” There is no evidence of historical contact between one Khalaj and the other. (It is, therefore, also quite unclear whether the Khalaj or Khalji of India, were Turks or Kurds or some third people). “Tatar” originally referred to a small Mongolian-speaking tribe that was exterminated as a national unit by Chinggis Khan (only women and smaller children were left alive and assimilated); the name, now applied to many Turkic tribes, is at the same time a migratory name, and when applied to the Mongols of the 13th century in Russian sources, for example, it is a taboo name (foreign name), see above. Often the slightest historical contact is enough to cause names to migrate; often the slightest historical connection is enough to cause the sources to transfer names or use them as collective names.
But this is exactly what happened with the name of the Huns, see e.g. Maenchen-Helfen 3290 (1955), 102ff. (in which some echoes may also be of purely coincidental nature). I consider this to be the most likely hypothesis: “Huns” was certainly once the name of speakers of a certain language (see § 20), and it was also once an ethnic name. But, it is also clear (see § 12) that the Hun Empire contained many peoples, races and languages, so “Hun” is necessarily also a state or collective name (not all citizens of the Hun Empire, simply referred to as “Huns,” were ethnically or linguistically Huns). However, the name of the Huns may also have migrated or been transferred (it is not clear whether the Caucasus “Huns” were Huns, see below). Looking at the occurrence of the name in Moravcsik 231–7, the following categories of the word Oὗννoι emerge:
1. The Huns of the Attila empire and its predecessors (sometimes certainly a linguistic or ethnic name, as when Attila is referred to as a Hun, but generally more of a state name; encompassing the citizens of this multiracial Hun empire; problematic is the extent to which Oὗννoι is a transfer name from Ptolemy Xoῦνoι),
2. ‘“Huns” in a general and summarizing sense’ (categories d–f),
3. “Unspecified Hunnic peoples” (hardly a linguistic name, more of a collective, sometimes transmitted name),
4. Kidarites,
5. Cadiseni, [Hunnic-Turkic note: an ancient Eastern Iranian or Hunnic-Hephthalite tribe that lived in Gharchistan before the appearance of the Chionites and mentioned by the Roman historian Procopius in History of the Wars, I, 14, 38.]
6. Hephthalites (i.e., peoples around Iran in 4–6, partly collective, partly transfer name),
7. Akatzirs,
8. Sabirs,
9. Onogurs,
10. Utigurs,
11. Kutrigurs (peoples in 7–11 collective or even more likely transfer name, see § 19),
12. Bulgars,
13. Avars,
14. Turks,
15. Hungary,
16. Uzes,
17. Cumans,
18. Seljuks,
19. Ottomans,
20. People in the neighborhood of the Alans in the 15th century (peoples in 12–20 transfer name).
We can now see the complexity of the problem. Under no circumstances should we, when people or tribes are referred to in the sources as”Hun,” immediately think of this as a linguistic or ethnic name, and use their names as Hun linguistic material. This would be no different than if we wanted to understand names of Soviet citizens such as Fazylov and Torganej as “Russian” (even though the former is Özbek and the latter Evenkitungusian—with Fazylov having Arabic roots!).
§ 12. Three appellatives have come down to us from the language of the Huns, as well as material containing around 100 tribal and personal names. As far as their identity is concerned, the following hypotheses can be found (older literature especially according to Inostrancev 3299 (1900)):
A. Deguignes (see § 4), Pallas, Bergmann et al.: They were Mongols.
B. Abel-Rémusat, Saint-Martin: Finns.
C. Venelin, Vel’tman, Zabělin, Ilovajskij, Florinskij: Slavs.
D. Klaproth: Hungarians, also Caucasians (Lezgen, Avars). They are the ancestors of the Hungarians. (Similarly Semenov.)
E. Inostrancev: A mixture of Turks and Finns.
F. Moór 1963: Caucasians.
G. Pritsak 3304 (1954), Barthold (see Maenchen-Helfen 3300 (1944–5), 225), Ašmarin 1971, 17, 192: Bulgar Türks.
H. Altheim-Stiehl 3392 (1953), 85 (the languages of the Hiung-nu and the Huns are both = Old Turkic, which means “things that have long been established”), also Altheim 3291, 1 (1959), 7. This is the most common interpretation, especially in older times held by many scholars: Zeuß, Kunik, F. Müller, Tomaschek, Sokolov, Korš, V. F. Miller, Vasil’evskij, Wietersheim (with the addition of Finnish and Tungusic elements), Vámbéry, Radloff, Aristov, also Németh (see Moór 1963, 65). Even Maenchen-Helfen, who otherwise tends to be skeptical, is convinced (see e.g. 1963) that the decisive component of the Huns was Turkish.
In the following, I will only deal with theories C, F, G and H, for which either certain evidence can be found (C), or which are widespread (H), or which have at least been advocated in more recent times (F, G).
First of all, it is clear that the Hun Empire was certainly just as multinational and multilingual as the Austro-Hungarian Monarchy that later occupied its space. It cannot, therefore, be a question of determining the language of a unified nation state, but of identifying the decisive and originally Hunnic component; this may (with Moravcsik I 5 f.) have been the language of a very small ruling class. Researchers from the most diverse directions agree on the colorful genetic and linguistic composition of the Hun Empire: Harmatta 3312 (1951), 144; Jettmar 3302 (1951–2), 178; Fettich 3328 (1953), 109f.; Werner 3329 (1956), 1; Maenchen-Helfen 3307 (1959), 237 and passim; Shafer 1966, 6.
§ 13. The three Hun appellatives are most probably Slavic. I list them in an anticlimax according to the degree of certainty of their Slavic provenance:
(a) strava ‘drink of death, celebration of the dead (for Attila).’ Appears in Iordanes, Getica 49, 258: Postquam talibus lamentis est defletus, stravam super tumulum eius, quam appellant ipsi, ingenti comessatione concelebrant [He does not provide its translation: “When the Huns had mourned [Attila] with such lamentations, a strava, as they call it, was celebrated over his tomb with great revelling.”]. The attempt to etymologize the word in Gothic (see Walde-Hofmann: Lateinisches etymologisches Wörterbuch, Heidelberg 1938, 601 “in Gothic straujan ‘sprinkle’”; Mommsen 198; cf. also Samolin 3309 (1956), 144; Maenchen-Helfen 3300 (1944–5), 225; Moór 1963, 84) has been abandoned today. The word is not attested in Germanic. The attempt to derive the word from Turkic, from the root astra- (Karaim) ‘to conceal, to bury,’ is also untenable. This word in Turkic is in fact a Mongolian loanword that only penetrated in the 13th century, see Doerfer I 20f. (versus Arnim 3336 (1936) and Németh 3286 (1940), 217–226; cf. also Samolin op. cit., Maenchen-Helfen 3300 op. cit., Altheim 1948, 219f., same author 1951, 209, same author 3291, IV 335). The assertion in Moór 1963, 84 that the word cannot be etymologized at all is also erroneous. Nor can it be derived from Slavic trizna (against Ilovajskij in Inostrancev 554). The combination of strava with Bulgarian zdravica ‘toast’ in Altheim-Stiehl 3292 (1953), 48 is also erroneous (this is a derivation of Bulgarian zdrav ‘healthy’).
Rather, see Trautmann: “strava… is nothing other than the Slavic word which we later find in the Old Polish strawa ‘epulae, feralis [food, wild],’ and in the Old Czech strava ‘funeral feast.’” According to Niederle 1926, 51, 53, the word is also the Slavic term for the funeral feast, and its use in Jordanes indicates that “les sujets d’Attila, dans la Hongrie centrale, devaient être alors des Slaves [He does not provide its translation: “Attila’s subjects, in central Hungary, must then have been Slavs.”].” The word is attested in many Slavic languages, and relates (as sь-trava) to traviti ‘digest’ (which is why Mommsen’s assumption p. 198 that the Slavs adopted it from the Goths is invalid) and is of Indo-European origin, see Max Vasmer: Russisches etymologisches Wörterbuch, III, Heidelberg 1958, 21, 130f. (Russian, Ukrainian, Belorussian, Czech, Slovak, Polish). Many other Slavists have also explained Hunnic strava as a Slavic word, such as J. Jungmann: Slovník česko-německý [Czech-German Dictionary], Prague 1838 (Czech); Fr. Miklosich: Lexicon palaeoslovenico-graeco-latinum, Vienna 1862–5 (Old Slavic); A. Brückner: Slownik etymologiczny jezyka polskiego, Kraków 1927 (Polish). There cannot be the slightest doubt that strava is Slavic. Only a pronounced prejudice can decide otherwise. To be even skeptical here would be out of place. (Too much skepticism is uncritical.)
(b) medos, drink of the inhabitants of the Hun country, mead. In Priskos El 13112, see Moravcsik 186: instead of wine, there was what is commonly called mead (in the huts of simple residents of the Hun country, not at Attila’s court).
What is the origin of the word? It is certainly Indo-European, see Vasmer op. cit. II 110. However, similar forms appear in several Indo-European languages, see Pokorný: Indogermanisches etymologisches Wörterbuch, I, Bern and Munich 1959, 707: Slav. medŭ, but also Celtic (e.g. Cymr. medd), and Germanic (e.g. Anglo-Saxon meodo, Old High German metu—Goth. *medus?). The word could, of course, also have existed in Thracian. Except for Vasil’evskij (who regards it as Slavic or Thracian) and Moór 1963, 87 (who describes it vaguely as “Indo-European”), almost all researchers agree that the word is to be regarded as Slavic, and this is by far the most probable thesis. Cf. on this (apart from the literature cited by Moravcsik) Trautmann 1948, Ilovajskij (in Inostrancev 3299, 554), Niederle 1926, 37 (who describes medŭ as “la boisson principale des Slaves [He does not provide its translation: “the Slavs’ main drink”],” and proves it as such from many sources). Otherwise, we would have to claim that strava and medos are both Indo-European (which they obviously are), but each comes from a different Indo-Germanic language, and only coincidentally to be both in Slavic (and both in Slavic alone) in an exactly appropriate form.
(c) Finally, the most difficult and uncertain is kamon (rarely kamos), ‘drink of the inhabitants of the land of the Huns.’ In Priskos E[xcerpta de ]L[egationibus] 13114, see Moravcsik 148: τὸ ἐκ κριθῶν . . . πόμα· κάμον οἱ βάρβαροι καλοῦσιν αὐτό ‘a drink made from barley, “kamon” is what the barbarians call it.’ The explanation is difficult, because camum is already known in the Latin vocabulary in the 3rd/4th century as a drink of the Paions, i.e., in pre-Hunnish times. Could Priskos be referring to a learned reminiscence (a form of transference)? I hardly think that is possible (he describes it from his own experience, again in a barbarian hut). The Celtic, Thracian (or Illyrian) origin of the word is unprovable, and this thesis is hardly defended any more. On the other hand, it is often assumed to be of Turkic origin (qïmïz ‘Kumis’), see Moravcsik 148; it is often wrongly written “qumïz” (but this is only a Russian corruption, according to Németh 3286 (1940), Altheim 1951, 209, Altheim-Stiehl 3292 (1953), 85f.). One can object to the Turkic origin of the word:
(1) The meanings Hunnic ‘barley beer’ and Turkic ‘soured and fermented mare’s milk’ do not fit well together. However, this deficiency would not be insurmountable, as shown by Altheim-Stiehl. (Shifts in meaning can never be completely ruled out).
(2) Since kamon, camum is already attested for Pannonia before the Huns, the Huns (or “Huns”) could at most have adopted it from the original inhabitants, but they were by no means Turks. This is why Altheim 3291, IV 59f. abandoned the equation kamon = Turkic “qumïz.”
(3) Furthermore, the vocalism of Turkic qïmïz does not match kamon; it would be more likely Greek *ϰιμίς.
Perhaps this word can also be explained in Slavic terms (of course not with Ilovajskij in Inostrancev 554 from Russian kvas). I am thinking of the Slavic kom (see Vasmer, op. cit. I 606), which generally means ‘lump,’ but often in its basic form or with derivatives has taken on the meaning ‘draff, marc’ (Serbian kom, Bulgarian komina, Czech kominy); kom is used in Serbia to make komovica schnapps, see P. Budmani: Rječnik hrvatskoga ili srpskoga jezika, V, Zagreb 1898–1903, 231. However, the difference in meaning between ‘barley beer’ and ‘draff, marc’ can only be bridged by the bold assumption that kom ‘lump’ may also have been used to refer to the barley mass and, therefore, the beer itself brewed from it. Phonetically, of course, the equation works perfectly: Old Slavic o was actually spoken as a labial a and appears as a in Greek and Latin sources (see Hans H. Bielfeldt: Altslavische Grammar, Halle 1961, 47; Max Vasmer: “The Slavs in Greece,” APAW 1941, 238, 267).
Niederle 1926, 37, note 4 is interesting: The name of another drink made from barley, ϰάμος, appears among the Balkan Slavs since the 10th century (komina), but its Slavic origin is very doubtful. However, I was unable to find the meaning ‘barley drink’ for kom, komina in Slavic (even the Slovník jazyka staroslověnského [Dictionary of the Old Slavonic Language], 15, Prague 1967, 43 only mentions kominy ‘wine grape’ for Church Slavonic from Bohemia). It is possible, however, that Slavic kom(ina), ‘alcoholic fermentation, drink made from it,’ is taken from a pre-Slavic language, e.g. from Thracian or Celtic settlers (cf. English, whiskey < Celtic).
Of course, historically there is much that is unclear here: according to N. S. Deržavin: Die Slaven im Altertum [The Slavs in Antiquity], Weimar 1948, 1–9, we first find evidence of Slavs in the 1st/2nd century (in Pliny, Tacitus, Ptolemy) (although not for Pannonia), then again in the 6th century in Jordanes and Procopius. Niederle’s assumption (see Trautmann 1948) that Slavs had already settled scattered among the Illyrians and Thracians in the 1st century AD can hardly be proven. However, Jordanes (Mommsen 631-2) provides unambiguous evidence of Slavs directly in Pannonia in the 6th century: Sclaveni a civitate Novietunense et laco qui appellatur Mursiano usque ad Danastrum et in boream Viscia tenus commorantur [“The abode of the Sclaveni extends from the city of Noviodunum and the lake called Mursianus to the Danaster, and northward as far as the Vistula,” as translated by Charles Christopher Mierow in 1915; “The Sclaveni dwell from the town of Noviodunum and the lake called Mursianus to the Danaster and towards the north up to the Viscia,” as translated by Nuffelen and Van Hoof in 2020]. This is according to p. 163 = Noviodunum, Neviodunum (with Priscus Ηοβίδουνον) located on the Sava, about 45° 55’ N. It does not seem implausible that Slavs settled in Pannonia as early as the 5th century (the century of Attila). However, the 4th–6th centuries may have been a typical transitional period, which is perhaps already reflected in the city name in Jordanes: It is, namely, = Celtic “NeuBurg” (Noviodunum is also the name of various Gallic cities: today Nevers, Nyon, Soissons, etc.).
The fact that the main mass of Hunnish citizens did not speak Hunnish (and perhaps Slavic) seems to be supported by the passage in Excerpta de legationibus 13510–18 [This time the author and the editors choose subscripts when referring to the lines unlike above.], where Priscus says that the Scythians (transfer name, = Hunnish citizens, see Moravcsik 231) spoke Hunnish (as well as Gothic and Latin) in addition to their own barbaric language, see Moravcsik 6. “Scythians” are also called Goths, Slavs, and Russians in the Greek sources, see Moravcsik 13. It is, therefore, (in contrast to Τοῦρϰοι , Oὗννoι, see Moravcsik 16) not a word restricted to (originally) steppe Asian peoples. Cf. also Thompson 10f., Altheim 3291, IV 300, 302.
If we assume that the three appellatives discussed are all Slavic (which is uncertain in the case of the third), this would of course not mean that the Huns were Slavs (as opposed to the works of Inostrancev 103–9), but merely that for the Tisza-Danube area we can expect “Slavs subjugated by the Huns” (Trautmann 22), i.e., a narrow nomadic ruling class dominating a broad Slavic settler population, a familiar pattern from history (Bulgars in Bulgaria, Mongols of the Golden Horde in Russia, Ottomans in the Slavic Balkans, etc.). We do not find any Slavic names among the Huns (Venelin’s comparisons in Inostrancev 551–561, Balamber, Balamer = Vladimir, Attila = Tilan, Bleda = Vlad etc. are entirely unreliable). The name of the Hun king Balamber, for example, can be explained by Schhönfeld 3332 (1911), 275 as being of unknown origin; neither the Slavic etymology nor the Mongolian one (Inostrancev 371) is convincing; a Germanic etymology, such as *Bala-mērs, would also be daring (Muellenhoff in Mommsen 147: Balamber… nomen nemo nisi imperitus pro germanico vendet [Balamber… no one but the uninitiated will sell the name as Germanic]); Reynolds 3319 (1946), 50 is also hardly credible with the assumption that Balamber contains Turkic böri ‘wolf.’ This also rules out thesis C (Huns = Slavs).
Vernadsky in 1951 interpreted βάλαν (accusative) ‘black horse with blaze.’ (Procopius Hist. II 9012–13) as Turkic bulan. The word used to be considered Germanic, see Moravcsik 85 (Gothic bals). One could possibly think of Slavic bělánu ‘mold,’ too, for which see, among others, Linda Sadnik, Rudolf Aitzetmüller: Vergleichendes Wörterbuch der slavischen Sprachen [Comparative Dictionary of Slavic Languages], I, Wiesbaden 1963, 129, 131, also Daničic: Rječnik hrvatskoga ili srpskoga jezika [Croatian or Serbian Language Dictionary], I, Zagreb 1880–2, 379 and others (Slavic ě was spoken ä in ancient times and was often transcribed α in Greek, see Bielfeldt 25, Vasmer: Die Griechen…, 238, 269). However, the fact that the word is used on the occasion of the description of the Battle at Pons Milvius in [537] between Belisarius and the Ostrogoths, about a century after the destruction of the Hunnish Empire (which is why the Germanic etymology is to be preferred), speaks strongly against the Slavic or Hunnic origin of the word. Vernadsky’s assumption (383) that Belisarius’ horse “apparently came from Hun horse breeders” finds no support whatsoever. Also, bulan does not fit semantically, either, see Doerfer II 356-8.
§ 14. There are some names whose meaning seemed so clear to some authors that they were inclined to equate them with appellatives. Two of them are discussed here:
a) Jordanes talks about the Dnepr, referring to it as “quam lingua sua Hunni Var [Dnepr, which the Huns call Var in their language.]” According to Pritsak 3304, 3333 (both 1954), this is simply = Hunnic var (today still Chuvash = Neo-Bulgarian var, “ravine, valley”), which corresponds to the Turkic öz. This proves that the Huns spoke Bulgar Turkic. (Thus, by the way, already Ašmarin 1971, 17, 192, going back to an article from 1902, see also Benzing 1950). Altheim 1951, 209 lists Var among the Hunnic appellatives without giving an etymology. Certainly Klaproth’s compilation with Lezgian-Avar “or, hor, ouor” ‘river’ is unlikely (Inostrancev 534). It has not even been adopted by Moór 1963, who regards the Huns as Caucasians. However, the most probable origin of the name is Iranian (Markwart 1903, see Altheim 3291, IV 335, also Haussig 1969, 774). In A. Thumb, R. Hauschild: Handbuch des Sanskrit/Handbook of Sanskrit, I, Heidelberg 1958, 87, it says: “The Volga also once had an Indian name: Herodotus calls it O’aros (Ionian for Varos) = ancient Indian vār(i) ‘water’.” However the name may be explained, it is important to note Moór’s remark (1963, 86) that Var need not have been originally Hunnic, but was simply the name of the inhabitants of the Dnieper, which the Huns had naturally adopted. “Tiber” or “Volga” are also used in German, but they are not German words. Incidentally, it is highly controversial whether Common Turkic ö:z could have developed into Bulgar Turkic var as early as the 6th century, or whether this is not an anachronism (for -z = -r see Doerfer II 521–3, III 208–10, IV keyword 1921; for ö:- = va- see Doerfer: Tschuwaschisch -m < urtürkisch *-m (> gemeintürkisch -n) [Chuvash -m < Proto Turkic *-m ( > Common Turkic –cn]), UAJb 45 (1973)). This now also settles the thesis G (Huns = Bulgar Türks), because this was its only halfway useful support. (Altheim-Haussig 3293 (1958), 23f., according to which the Hunnic [?] tribal name Οὐλτίζουροιὐ is to be interpreted as ultï čur and was actually the name of the six highest dignitaries, and that this proves the Bulgar origin of the Huns, since only the Chuvash = New Bulgar has ulttă ‘six’ with u-, as opposed to the Common Turkic altï, with a-. Even in the 13th/14th century Volga-Bulgarian grave inscriptions, ‘six’ appears written as alṭï, see G. V. Jusupov: Vvedenie v bulgaro-tatarskuju épigrafiku [Introduction to Bulgaro-Tatar Epigraphy], Moscow-Leningrad 1960, Plates 5, 48; also in the 15th/16th century Chuvash loanwords in Mountain Cheremiss [today called Hill Mari], a still survives; see my article mentioned in UAJb 45).
b) Maenchen-Helfen 3307 (1959), 231 f. lists Atakam among the names whose Turkic etymology is beyond any doubt (as already Vámbéry, see Inostrancev 558, and see also Moravcsik 76). Even Moór 1963, 95 says that this is “the only Turkic interpretation of a Hunnic personal name... that can be considered flawless at least in formal terms” (it is, however, merely “a game of chance”). Haussig 1953, 360 concluded from this: “The dignity of this shaman-chief priest required... a higher age. This is also expressed in the name of this man, who was referred to by the Huns as Άτακάμ, ata-qam, ‘shaman father’ (Priscus, Excerpta de Legationibus, p. 122, 15).” This example seems highly plausible at first glance—and that is precisely why I chose it (namely to show that even the names whose Turkic character seems convincing cannot be Turkic, not to mention the many names for which only very tortured Turkic etymologies have been given, and which I do not want to bother with refuting here). The objections can be raised:
i) Factual: There is no evidence of shamanism for the Huns in any historical source (Moór 1963, 95); Haussig himself concluded it just from the name (hysteron proteron [from the Greek: ὕστερον πρότερον, an inversion of the natural order: what should come last is put first]).
ii) The original text is unclear. It is said that Attila had some barbarians who had defected to the Romans [Greek: ἡ Βασιλεία τῶν Ῥωμαίων] crucified, ἐυ οἶς καὶ παῖδες Μάμα ακὶ Άτακὰμ τοῦ βασιλείου γέυος [ef oís kaí paídes Máma akí Átakám toú vasileíou géyos]. Here Μάμα ακὶ Άτακὰμ is to be understood as a genitive (thus Haussig, also Moravcsik 180; Altheim 3291, I 365, Moór 1963, 95, note 42 and Thompson 3287 (1948), 77 understand the passage appositionally, so that Mama and Atakam would be children, and this would of course completely exclude the interpretation “father-shaman”). Correct translation, therefore, is: “among them (were) also the children of Mama and Atakam of royal lineage.” However, the genitive could easily be interpreted as ‘the children of [he] Mama (or [she] Mama?) and Atakam,’ i.e., not even the gender of the person (otherwise not mentioned in Priscus) is certain. But then the interpretation ‘father’ is not certain.
iii) There is also an anachronism. The word ata ‘father’ was certainly originally a slang word in Turkic, as was baba, which later largely replaced ata. However, ata has only been documented with certainty since the 11th century (Qutadgu Bilig of 1069); in the Old Turkic Orkhon inscriptions of the 8th century, only the old word qaŋ appears for it. And even in the Dunhuang manuscripts of the 10th century, qaŋ alone prevails (see, among others, J. R. Hamilton: Le conte bouddhique du bon et du mauvais prince en version ouïgoure [The Buddhist tale of the good and the bad prince in Uighur version], Paris 1971, 4, 116). The Hunnic ata would then have been separated from the actual first record by six centuries. (An anachronism is no better in linguistics than in historiography).
§ 15. We now turn to the explanation of the Hunnic names. However, I would like to preface this with a general methodological explanation. All essential things in linguistics can be expressed in paradoxes. The paradox of onomastics is: “In the interpretation of a name, a possible determination of the meaning is certainly a pleasant side effect, but not essential or even a conditio sine qua non [indispensable condition]. On the other hand, certain formal criteria (quite inconspicuous to the layman) are of decisive importance.” The most important of these formal criteria are:
A. The phonetic structure of the names is relevant. For example, sounds such as st-, sk-, gl-, tr- generally refer to Indo-European names. (It is then generally inadmissible to explain a name with st- from a language that does not recognize this sound, e.g. by explaining st- < vowel + st- or zt-).
B. Certain characteristic suffixes are essential for the ethnic classification of naming materials. (For example, the suffix -ōn- is characteristic of Illyrian names for bodies of water. It is generally inadmissible to interpret a name from a root + unproductive suffix, unless it can be shown that the suffix in question is exceptionally common in names).
C. Attention should be drawn to recurring characteristic compositional elements. (E.g. Germanic names are usually bipartite, the second compositional element is often -wīg ‘fight,’ but never a neuter).
Of course, these formal criteria can only play a role and be applied effectively with a significant accumulation of material; you cannot do anything with one or two words. The process of first forming an indifferent mixture from an obviously heterogeneous name material and then picking out individual disparate elements (which happen to fit one’s own thesis) is unusual in onomastics.
The reason why the meaning does not play such an important role here is obvious: the creation of meaning is always possible and often arbitrary (whoever seeks will find), it is often not even controllable, for example in the case of poorly preserved old sources. Names also often contain obsolete elements, i.e., those whose meaning can no longer be determined (this is extremely characteristic of Mongolian personal names, for example). Especially in the case of controversial material (such as Hunnic), determining the meaning is completely unreliable. Everyone will simply inject their own specific prejudices into the original.
I would now like to give a methodologically instructive (albeit fictitious) example of this. I will take any two Eskimo names, let us say the first two names in E. S. Rubcova: Materialy po jazyku i fol’kloru éskimosov [Materials on the language and folklore of the Eskimos], I, Moscow-Leningrad 1954, 38. I want to prove that these can be interpreted easily and beautifully in Turkic (this fictitious game could just as well be played with Khoekhoe [nomadic pastoralist indigenous population of South Africa] or Botocudo [South American Indian people who used to live in what is now the Brazilian state of Minas Gerais] names. The names are: (male name) Qali̯ú. This is clearly Turkic, namely = qal ‘old’ (according to Drevnetjurkskij slovar’, Leningrad 1969, 410) or ‘great, strong’ (ibid., also Clauson: An etymological dictionary of pre-thirteenth-century Turkish, Oxford 1972, 614). It also contains iyü ‘the persecutor,’ from iy- ‘to pursue, to supress’ (in the same dictionaries p. 205 and 266, respectively), with the suffix -u/-ü (see A. v. Gabain: Alttürkische Grammatik, Leipzig 1950, § 106). Thus, ‘the old (or the great) persecutor.’ A perfectly fitting hunter’s name!.
And, there is the woman’s name: Arïpaki. This is, of course, also Turkic, namely = arïγ ‘pure, chaste; (an honorific epithet in Turkic that is common for women, see the dictionaries p. 51 f. and 213, respectively) + bäk ‘steady’ (or baqi (same meaning), see the dictionaries p. 92, 82, and 323, respectively). A name that beautifully adorns a woman! It is a shame that these Turkic explanations of Eskimo names are fictional. They are actually much nicer than some Turkic explanations of Hun names (which are also fictitious, which the authors just did not realize). The “beauty” of an etymology is not very impressive to the onomast; it proves nothing.
Here are two examples of (1) an incorrect and (2) a correct interpretation of a name:
(1) Let us explain the name of the greatest ruler of the Ottomans, Süleyman. We take it for granted that it must be Turkish and find a suitable base: Old Turkic sülä- ‘to wage war.’ This results in the extremely beautiful etymology ‘the war leader.’ This interpretation is wonderfully convincing for two reasons: a) because it explains the name of a Turkish ruler in Turkish, b) because it fits so wonderfully for a ruler. It is just wrong: Süleyman is in fact borrowed from the Arabic sulaimān (this in turn from Hebrew, derived from the Semitic root šlm ‘peace’—the exact opposite of ‘war’). The formalist, on the other hand, would have stated: a) The word is not vowel-harmonic (the vowels e and a can never occur together in a genuine Turkish word), thus contradicting condition A, b) a suffix -yman is not attested in Turkish, thus the interpretation also contradicts condition B. It is, therefore, inadmissible, and the word must be interpreted differently, i.e., non-Turkish.
(2) Suppose we find the name Barulatai in a Central Asian or Eastern European source of the 13th/14th century. We want to find out which language the name belonged to. We know that in Mongolian (and only there) there is a common name type “tribal name + suffix -tai (or -dai).” Now, there is evidence of a Barulas tribe [The Barlas (Mongolian: Barulās) were a Mongol and later Turkicized nomadic confederation in Central Asia] in Mongolian. So, Barulatai must be Mongolian. And it does not matter that the Mongolian tribal name Barulas cannot actually be etymologized.
Even with completely unknown languages, it is still possible to operate with the formal criteria. Let us assume that we have a mixed name material consisting of (1) a layer in which many names begin with sk-, st-, and which often have the suffix -tun at the same time (and never a suffix -lak), (2) another layer, on the other hand, never has sk-, st-, never -tun, but often a suffix -lak. Then, we can easily separate two layers (1) and (2), although we cannot etymologize a single word. Let us approach the Hunnic names with this knowledge. But again, we must be forewarned:
§ 16. There are three alternatives for the explanation of Hunnic names, all of which are equally unpleasant:
a) We will limit ourselves to the names of those persons whose Hunnic origin is beyond doubt. Basically, only the names of Attila himself and some of his relatives remain. This material is far too small to allow a reliable statement (i.e., it does not fulfill the accumulation condition in §15). Moreover, as will become clear, it is of heterogeneous (at least controversial) origin.
In this strictest sense, only the following names would remain: the name of the Huns (Οὗννοι, Hunni) themselves, also from Greek literature (Moravcsik): Άτακάμ, Άττίλας, Βλήδας, Γιέσμος, Δεγγιζίχ (also Δινζίχ, Latin Dintzic, see Moravcsik 48, 117), Ἠρνάχ, Μάμας, Μουνδίουχος, Μοῦνδος, Οὔλδης, Οὔπταρος, ‘Ρούας, Χαράτων, Ώηβάρσιος. From Latin literature (Schönfeld, Mommsen) there are also: Balamber, Ellac, Laudaricus, Emnetzur, and Ultzindur. That makes a total of 20 names.
However, the boundary between b) and c) is sometimes difficult to draw (which makes the material even more uncertain). Δovάtoς, for example, probably does not belong here (not a Hun, see Schramm 144, note 45, Altheim 1951, 202f., note 26). In the case of Οὔλδης and Χαράτων, it is also quite unclear whether they were somehow related to Attila’s lineage or were Huns at all, see Schramm 143, note 43. But see b).
b) A middle alternative would be to also include names of Huns (“Huns”?) who are mentioned as courtiers of Attila, or in other more or less unclear contexts (which can still mean that they are of foreign origin). Then, from Greek literature (Moravcsik) would be added: Ἄδαμις, Βέριχος, Ἐδέκων, Ἐσκάμ, Ἤσλας, Κρέκα (or rather Ἡρέκα), Ὀνηγήσιος, Οὐλιμούθ (? Schönfeld 280, after Procopius), Σκόττας. From Latin literature (Schönfeld) there might also be Tuldila. The boundary between a) and b) is sometimes difficult to draw. Οὐάλιψ belongs to c) (presumably better non-Hunnic), see Thompson 217 f., also Σηγγίλαχος (Thompson 72). Βασίχ and Κονρσίχ, on the other hand, seem to belong to a). Moravcsik sees them as simple Hunnic military leaders (c. 440); however, according to Pritsak UAJb 26 (1954), 219 and Maenchen-Helfen JAOS 79 (1959), 298, they are ancient Hunnic rulers (c. 395). For a more detailed explanation, see Altheim 3291, I 12f., IV 319, who places them even older than Pritsak. Altheim’s version is probably correct.
c) The last, extreme possibility would be to include everything in the investigation that could possibly refer to the Huns. This would mean that we would also include the “unknown Huns” from Moravcsik, as well as, for example, Μοδάρης and other names that are recorded as “Byzantine military leader of Scythian (= Hunnic or Gothic?) origin.” (Cf. also Moravcsik 17, note 3: almost all “Huns” so designated have Iranian names). We would also have to use names such as Acatziri, although actually (against Thompson 3287 (1948), 95, Altheim 3291, IV 274-9) the statement that the Huns subjugated the Acatziri (“an obscure but valiant people”) would seem to indicate that they were not Huns. (Priscus’ statement that the Acatzirs were Huns—alongside the statement that they were a Scythian people—is, in my opinion, simply to be understood as a “collective name”: a people with a certain steppe-nomadic way of life, see § 16 [A typo? This is § 16]. On the many Caucasian “Huns,” see also § 19.
So, if we use everything that could possibly be Hunnish, we get a completely uncertain material, since we must certainly regard the Hunnish empire as a multi-ethnic state, and genuine Hunnish origin would first have to be proven in any case (hysteron proteron [= i.e., an inversion of the natural order]). It would, therefore, be the same as if we wanted to explain all the personal and ethnic names of the Soviet Union from Russian, see the end of § 11.
§ 17. In any case, it is certain that much of what is described as Hunnish can be excluded from the outset. Thus, the “Hunnic” name material in Altheim-Stiehl 3292 (1953) has been refuted by Harmatta-Pékary 1971 and already by Henning 1954 (in fact, these are Iranian official titles).
Haussig in Altheim-Haussig 3293 (1958), 9–29 has convincingly dismissed the idea that the Danube Bulgars were Huns, and that their list of princes from the 7th/8th century is, therefore, a monument to the Hunnish language (see Moravcsik 352–4).
And the “Hunnic runes” examined by Altheim in 1948, Chapter VIII (195–230) are certainly to be interpreted differently than Altheim did (if they can be interpreted at all). Let us take as an example the inscription on a drinking cup consisting of only four words. Altheim interpreted it as Old Turkic qadγu qoqunï qu(w)raγïn öy, “the sorrow reduces the time for sociability.” From the point of view of Old Turkic grammar and lexicon (as well as of the Turkic language type in general), this contains 12 errors. Let us remember that Altheim 3291, I 7 states that “It may be admitted that the Huns had an Altaic language that was closely related to Old Turkic, if not identical with it,” in Altheim-Stiehl 3292 (1953), 85 even more sharply that the languages of the Hiung-nu as well as the Huns are both Old Turkic, which are “things that have long been established.” So, let us apply Altheim’s strict standard “Hunnic = Old Turkic” to his own interpretations. Then, the following turns out to be wrong:
(1) “The sorrow” should be qadγuγ. Moreover, a preceding suffixless indefinite accusative qadγu ‘(any) sorrow’ would be stylistically unusual in Old Turkic (if only because of the possibility of confusion: qadγu in the first position would normally be expected as nominative = subject case), the object would have to come before the verb, and the verb at the end of the sentence.
(2)Even qoq- does not mean ‘to diminish (something),’ but ‘to diminish oneself.’ The meaning is, therefore, wrong.
