“The Frankish Name of the King of the Turks” — Christopher I. Beckwith (2006)

 

“The Frankish Name of the King of the Turks” — Christopher I. Beckwith

Archivum Eurasiae Medii Aevi, 2006

Abstract

The Chinese name of the Turks, T’u-chüeh 突厥, has been an unsolved problem for a very long time. In a recent paper (2005) on various early Chinese transcriptions of Central Eurasian ethnonyms a new solution is proposed. After that paper was in press the writer rediscovered a Frankish historical source, written in Latin, which contains transcriptions of a Turkic title that also clarify a hitherto problematic Turkic title in Greek transcription. Both actually transcribe the same name transcribed in Chinese as T’u-chüeh, confirming the writer’s earlier conclusion and further clarifying the meaning of the name.

For nearly a century the dominant interpretation of the Chinese name of the Turks, T’u-chüeh 突厥, has been that it represents the transcription of a plural form *Türküt ‘Turks,’ reconstructed as Türk plus a ‘Mongolian’ or ‘Sogdian’ plural suffix -t, with an epenthetic vowel, -ü-, determined by Turkic vowel harmony. A contribution by the present writer to the previous issue of this journal shows that the above analysis is untenable for a number of reasons, one of the most outstanding of which being the fact that the sixth century Bugut inscription does not actually contain a Sogdian transcription of a putative strictly Sogdian plural form *Türkīt, as has been claimed, and in fact an examination of all other occurrences of the name Türk in Sogdian texts shows it is always written the same, twrk *Türk, without a plural, even when the word has plural meaning, ‘Turks.’ The article concludes that the Early Middle Chinese period transcription of the name of the Turks, T’u-chüeh 突厥 MChi dial.^★türkwac, appears to represent a foreign compound *Türk-wač ‘lord(s) of the Türk’ or the like. Since that article was finished the writer has discovered other evidence which confirms and refines that conclusion.[1]

It has been established that in the early medieval period the Franks added an interesting detail to the legend they had been developing about their Trojan origins. According to the addition, the Franks belonged at first to a larger body. At the point in their migration when they crossed the Danube the group split into two, one of which became the Franci ‘Franks,’ who took their name from the name of their king, Francio.[2] The other group, the Torci, Turci, or Turchi ‘Turks,’ took their name from the name of their king, Torquotus, Turquotus, or Torcoth. There are two versions of the story in the anonymous Chronicle of Fredegar. The lengthier first version, in Book II, says:

Tercia ex eadem origine gentem Torcorum fuisse fama confirmat, ut, cum Franci Asiam pervacantis pluribus proeliis transissent, ingredientis Eurupam, super litore Danuviae fluminis inter Ocianum et Traciam una ex eis ibidem pars resedit. Electum a se utique regem nomen Torquoto, per quod gens Turquorum nomen accepit. Franci [...], quando ad Renum consederunt, dum a Turquoto menuati sunt, parva ex eis manus aderat.[3]

Tradition confirms that there was a third tribe from the same origin, the Turks, and that when the Franks in their travels and many battles crossed over and entered Europe, a group of them settled in that same place, above the bank of the river Danube between the ocean and Thrace. They elected from their midst a king named Torquotus[4] from whom they got their name, ‘Turks.’ The Franks, (...), since their numbers[5] were diminished by Turquotus, when they settled near the Rhine a small band[6] remained.[7]

 The second version, in Book III, is much shorter:

Residua eorum pars, que super litore Danuvii remanserat, elictum a se Torcoth nomen regem, per quem ibique vocati sunt Turchi; et per Francionem hii alii vocati sunt Franci.[8]

 

The remaining group of them, who stayed on the bank of the Danube, elected a king from among themselves named Torcoth,[9] after whom they were called ‘Turks’and after Francio the others were called ‘Franks.’[10]

The two names—the name of the Turks and the name of their king—mutually differ in each case. In its first occurrence in Book II the name of the Turks is Torc- [tork] or Turq(u)- [turk], while the king’s name is Torquot- [torkwot]; in the second occurrence the king is named Turquot- [turkwot].[11] In Book III the Turks are Turch- [tur ] or [turk] and the name of their king is Torcoth [torkoθ] or [torcot].[12] In all cases the second syllable has the vowel o; there are no examples with *u.[13]

Ewig (1999:845) notes that Krusch was wrong to say that the name Turquotus ~ Torquatus comes from the ancient name Torquatus.[14] He argues that it must instead be from “Tourxanthos/Türkshad, nom ou titre que portait le second frère de Tardu (575–603), second kaghan de Turquie occidentale.” While Ewig is right about the Turkic origin of these names for the king of the Turks, the direct equation of the Frankish and Old Turkic forms given by him is impossible phonetically. Also, the Frankish forms could derive from a Byzantine intermediary, but it could not have been the form found in the sources, Τούρξαθ Tourxath [turksaθ] or its variants,[15] because of the intrusive -s-.

The Byzantine form itself could not reflect a hypothetical Old Turkic *Türk-šad for two reasons. Firstly, the final consonants do not correspond: the equation of Byzantine Greek -θ as a transcription of Old Turkic -d is hypothetical and would actually be unprecedented.[16] Secondly, the putative title *Türk-šad, though widely cited as the Old Turkic equivalent of this Byzantine form, is unattested anywhere else. Thirdly, *Türk-šad does not make sense in the context of Old Turkic titles. The title šad was given to a ruler over one of the subdivisions of the Turkic realm, but in such titles, the designation of one of the subdivisions, X, is the first part and šad is the second, giving titles of the form X-šad. A title *Türk-šad would thus be incomprehensible. Fourthly, the independent Frankish and Chinese forms are nearly identical. Because the Franks learned the name from the Greeks, or possibly from the Turks themselves, during their stay in the Byzantine capital, it is manifestly clear that the -s- in the Greek spelling Τουρξάθ Tourxath [turksaθ] in the text of Menander Protector is wrong. The mistake was undoubtedly introduced by a copyist who misread the letters κο (-ko-) as κσ (-ks-) and transcribed them into ‘normal’ Greek spelling with the letter ξ (-ks-).[17] The name should therefore be emended to Τουρκοάθ Tourkoath [turkwaθ], representing the name or title *Türkwath.[18] However, the sources’ presentation of it as a title, not a name, is certainly correct.

Ewig dates the insertion of the Turkic episode into the legend in Fredegar on the chronology of the Frankish and Turkic embassies to Constantinople and on the Byzantine information on the Turk ruler known as *Türkwath. The information was acquired and added to the legend between 579 and 660.[19] However, the acquisition could not have been as late as Ewig’s terminus ad quem, 642–660. It is extremely unlikely that *Türkwath was remembered by anyone much into the seventh century, and he certainly would not have been referred to as ‘king of the Turks’ by anyone by that time. It is safe to say that the name must have been recorded by both Franks and Byzantines in the late sixth century, during the period of *Türkwath’s rule.

It is quite possible that the Franks and the Turks met in Constantinople. They surely saw each other there, and the Franks must have learned that the Turks had occupied the Western Steppe and impinged on Thrace. This is the locale for the common Franco-Turkic origin story, which the Franks subsequently developed and added to their Trojan migration legend. In any event, whether the Franks obtained their knowledge of the Turks directly or via the Byzantine Greeks, they must have gotten it orally, not from Greek written records, and wrote down what they heard.

The Frankish name for the ‘king of the Turks,’ *Türkwot, is remarkably close to the Chinese name for the Turks, T’u-chüeh 突厥 MChi dial.^★türkwac, a transcription of the foreign form *Türkwač (= *Türkβač [tyrkβaʧ]) ~ *Türkwat^j (= *Türkβat^j [tyrkβat^j]) ‘ruler of the Türk,’ in which the second element ^★wac represents Gāndhārī Prakrit wati (vati) ~ wat ~ wac (from Old Indic pati) ‘ruler, lord,’ a compounding element in a number of early East Asian and Central Eurasian names and titles, where it appears as *βač ~ *vać ~ *bać (etc.) ‘ruler.’[20] The Frankish Latin text of Fredegar explicitly says that the foreign term is the name of the ‘king of the Turks,’ and that the name of the people is related to the name of their ruler, though the source has the direction of derivation wrong.[21] The name of the ruler is obviously derived from the simple ethnonym Turc ~ Torc (etc.) ‘Türk,’ attested already in Antiquity, which is regularly used throughout the text to refer without exception to the people. The king’s name is derived from the people’s name by affixation or compounding. The Byzantine Greek text of Menander also specifies that *Τουρκοάθ *Türkwath was “one of the leaders of the Turks” and the Greeks are equally consistent in always calling the people Τούρκοι ‘Turks’; clearly, this name too is constructed from Τουρκ- ‘Türk,’ not the other way around, as all in fact agree. It must be emphasized that in both the Frankish and the Greek cases the name *Türkwath is explicitly said to be the name of a ‘ruler of the Turks,’[22] not the name of the Türk people, the ‘Turks.’ The name of the ruler and the name of the people are always clearly distinguished in both Frankish and Greek sources.[23]

As noted above, the final -θ (th) in the title *Τουρκοάθ *Türkwath does not correspond to Old Turkic -d, but to something else. The Franks wrote the same phone as -t or -th. The Chinese transcription can be interpreted phonetically by reference to contingent transcriptional data in Old Turkic and other languages to yield the underlying transcriptional target, the foreign title *Türkwat^j ~ *Türkwač.[24] The agreement of the transcriptions in Chinese, Greek, and Latin, the clarity and consistency of the Frankish use of the title, and the agreement of the Greek use with the Frankish use, make it clear that there were two distinct forms: Türk, the name of a people; and *Türkwat^j ~ *Türkwač, the title of the—or a—‘ruler of the Türk.’ This meaning is established independently on the one hand from Old Turkic, Chinese, and other Eastern sources[25] and on the other hand from the Frankish Latin and Byzantine Greek sources.

Finally, it is necessary to comment on the previously dominant interpretation of the Chinese name of the Turks, T’u-chüeh 突厥, which goes back to Marquart (Harmatta 1972:263–264) and has been supported by Pelliot and many other scholars. The usual idea is that it represents *Türküt, a ‘Mongolian’plural with the epenthetic vowel -ü- determined by Mongolian or Turkic vowel harmony. However, the Frankish transcriptions, the Greek transcription, and the Chinese transcription unanimously confirm the lack of vowel harmony in the title in question.[26] The lack of vowel harmony in a Turkic linguistic context can only mean that the form is a foreign word, or a compound, or both.[27] More recently it has been argued that the Chinese name of the Turks derives from a purely Sogdian form *twrkyt, i.e., *turkīt ‘Turks,’ a Sogdian -t plural form of twrk ‘Türk,’ but this is hardly supportable.[28] In fact, as those who have written on the uncorrected Greek form of the name have argued, the name or title Tourxath seems to consist of two parts, *Tourk ‘Türk’ plus *sath, the latter being Old Turkic šad, a rank, or title of a rank, in the ruling hierarchy. Although the Frankish forms show that the hypothetical (and highly problematic) *sath ~ *šath in Tourxath is a mistake—the Greek form of the full title must be restored as *Tουρκοαθ *Tourkoath, as shown in this paper—all of the foreign transcriptions or analyses of the title reflect a form that is specifically not vowel-harmonized, and the lack of a Turkic possessive -i suffix further confirms that there are not two words,[29] but one, a compound. Such a compound in Old Turkic might seem to be alien, yet šad, yabghu, kaghan, and most, if not all, of the other major ‘Old Turkic’ titles and names are known to be non-Turkic, as is the word Türk itself.[30] It therefore is to be expected that this Old Turkic title should be linguistically non-Turkic. The reconstructed title, *Türkwat^j ~ *Türkwač, should thus be written more precisely as *Türk-wat^j ~ *Türk-wač.

In conclusion, the Frankish name of ‘the king of the Turks’ is actually a title, the same title by which the Byzantine Greeks called the grandson of the Western Turkic kaghan Ištemi. This supports the present writer‘s earlier proposal that the foreign term used by the Chinese as the name of the Türk is actually an early title, but on the basis of the Frankish and Greek data it is now clear that it meant ‘ruler of the Türk’rather than ‘rulers of the Türk.’[31]

The Chinese sent their first known envoy to the Türk in 545, when their ruler, *Tumïn, was still a vassal of the Avars (Jou-jan);[32] that is, before the leader of the Türk overthrew his former overlord and adopted the Avar title kaghan ‘emperor’ in 552. His title was thus not kaghan, but what was it? Although *Tumïn did not yet have the imperial dignity, he was independent-minded and needed an impressive title to reflect his position. The name of his neighbors, the T’o-pa ~ *Ta -wat^j ~ *Taγ -wač,[33] whose name meant[34] ‘Ruler of the Earth,’ provided a convenient model.[35] It was a simple matter to come up with the title *Türk-wat^j ~ *Türk-wač ‘Ruler or Rulers of the Türk,’ on the same model.[36] Another possibility is that the grandson of Ištemi, *Türkwath —or more correctly, the *Türk-wath, i.e., the *Türk-wat^j, or the *Türk-wač—was as independent-minded as his grand uncle *Tumïn had been, and adopted the title himself. Unfortunately, his fate is unknown.

By the time the Chinese account of the Türk was written during the T’ang Dynasty, whatever the first known envoy from China to the Türk had understood of their language, state structure, and early titles was lost amidst later accounts of their origin myths, the story of the foundation of their empire, and the development of a much more complex state structure. Well before the beginning of the T’ang Dynasty the Chinese must have become habituated to the metonymous use of the title of the Türk ruler to refer to the Türk people and state.[37] Thus the old title *Türk-wat^j ~ *Türk-wač became established as the standard name for the Türk people and Turks in general in Chinese. As in many other similar cases in Chinese historical sources, it then remained unchanged ever after.

References

Beckwith, Christopher I. 2005. On the Chinese Names for Tibet, Tabghatch, and the Turks. Archivum Eurasiae Medii Aevi 14:5–20.

—— forthcoming. Empires of the Silk Road: A History of Central Eurasia.

Blockley, R.C. 1985. The History of Menander the Guardsman: Introductory Essay, Text, Translation, and Historiographical Note. Liverpool: Francis Cairns.

Boor, Carolus de, ed. 1903. Exerpta de legationibus, pars I. Exerpta de legationibus Romanorum ad gentes. Berlin: Weidmann.

Dobrovits, Mihály 2004. “They called themselves Avar” — Considering the pseudo-Avar question in the work of Theophylaktos. In: Matteo Compareti,

Paola Raffetta, Gianroberto Scarcia, eds., Ērān ud Anērān: Webfestschrift Marshak 2003. Studies presented to Boris Ilich Marshak on the Occasion of His 70th Birthday.

Ewig, Eugen 1997. Le myth troyen et l’histoire des Francs. In: Michel Rouche, Clovis: histoire & mémoire. Paris: Presses de l’Université de Paris-Sorbonne, 817–847.

Gerberding, Richard A. 1987. The Rise of the Carolingians and the Liber Historiae Francorum. Oxford: Oxford University Press.

Harmatta, J. 1972. Irano-Turcica. Acta Orientalia Academiae Scientiarum Hungaricae 25:263–273.

Krusch, Bruno, ed. 1888. Chronicarum quae dicuntur Fredegarii Scholastici libri IV cum continuationibus. In: Societas Aperiendis Fontibus Rerum Germanicarum Medii Aevi, ed., Monumenta Germaniae Historica, Scriptorum Rerum Merovingicarum, Tomus II. Hanover: Hahn, 1–193.

Ling-hu, Te-fen 1971. 周書 Chou shu. Peking: Chung-hua shu-chü.

Mackerras, Colin 1990. The Uighurs. In: Denis Sinor, ed. The Cambridge History of Early Inner Asia. Cambridge: Cambridge University Press, 317–342.

Sinor, Denis 1990. The Establishment and Dissolution of the Türk Empire. In: Denis Sinor, ed. The Cambridge History of Early Inner Asia. Cambridge:Cambridge University Press, 285–316.

---

[1] In Beckwith (2005) I concluded that the name *türk-wač meant ‘Rulers of the Türk’ (or ‘Türk Rulers’). That view is revised here. I myself felt uncomfortable with the article’s conclusion at the time of writing, though it seemed to be inescapable based on the evidence. It was only some time after finishing my article that I ran into a reference to the connection made in Fredegar between the Franks and the Turks. Upon reading it I at once realized that the material there was important for the problem of the Chinese name of the Turks and have therefore written this addition to my previous article.

[2] Ewig (1999:821).

[3] Krusch (1888:46); cf. Ewig (1999:821), who however has silently made a few minor changes in the text.

[4] Gerberding (1987:15) makes this “Turquotus” without comment.

[5] i.e., the number of the original people, including the Franks.

[6] That is, the remnant people who became the Franks.

[7] Gerberding (1987:15).

[8] Krusch (1888:93).

[9] Gerberding (1987:16) has “Torcot” here; perhaps it is a typographical error.

[10] Gerberding (1987:15–16)

[11] See note 14 below.

[12] Ewig (1999:829).

[13] See below on the ‘Mongolian’ plural and vowel-harmony theory. In Old Turkic vowel harmony, a harmonizing rounded vowel in a suffix can be u or ü but not o or ö.

[14] Krusch may be partly right, insofar as the spelling Torquatus could well reflect the influence of the name of several famous early Roman figures bearing that name, especially Titus Manlius Torquatus, who actually appears in the Aeneid, the story of the Trojan origin of the Romans. Although the latter is not mentioned in the text in the immediate vicinity of the discussions of the Turks, it may be that an identification of Torquatus with the king of the Turks could have been intended. It is thus necessary to note Torquat- [torkwat] as yet another possible variant. See the discussion by Ewig (1999:824 et seq.). The Byzantine and Chinese forms suggest that a combination of the Frankish spellings Torquat- and Turquot-, i.e., *Turquat, would indeed be closest to *Türkwatj the ‘original’ Turkic form presumably heard by both Franks and Greeks in Constantinople.

[15] The most frequent manuscript form is Τουρξανθ- Tourxanth- (= ‘Turksanth’), with an intrusive -n- as well as the intrusive -s-, but the Escorial manuscript, which is at the head of de Boor’s stemma (1903:ix, xvi), actually has Τουρξάθου in one instance (rather than the Τουρξάνθου of the edited text, de Boor 1903:204 line 20), preserving the form found in the text before Τουρξάθ was ‘normalized’ by Greek scribes to Τούρξανθ everywhere else, as discussed below. Blockley (1985:172, 276 n. 221) adds other textual variants of the type Τουρξάθ from manuscripts unknown to de Boor.

[16] Although the final -th (= [-t] or [-θ]) of one of the Frankish Latin forms could well reflect Byzantine Greek -θ -th (= [-θ]), the latter never transcribes Old Turkic -d (Peter Golden, p.c. 2006) and the -s- in the Greek form—via the letter ξ x (phonetically [ks], as in Alex)—is clearly due to a scribal error, as shown below.

[17] The prevalence of the spelling Τούρξανθ is clearly due to the further Hellenization of the name by later scribes: there are no words that begin ξάθ- xath- or the like but many that begin ξάνθ- xanth-, including ξάνθορ xanthos ‘yellow,’ a word every Greek would know.

[18] This form will be used below without comment, but with an asterisk.

[19] Ewig (1999:847), according to whom the terminus a quo is 579–581 and the terminus ad quem is between 642 and 660.

[20] See Beckwith (2005) for details.

[21] Even this mistake is revealing. The normal direction of ethnonym-development among Central Eurasians is that the nation was indeed often named after a historical founding hero. See Beckwith (forthcoming).

[22] Menander even has the Greek envoy Valentinus address him as “o leader of the Turks” (Blockley 1985:176–177). The Frankish elevation of *Τουρκοάθ *Türkwath to the level of a ‘king’ is generally believed to be an exaggeration in view of his evidently subordinate position (according to the Greek sources), so perhaps ‘a ruler of the Turks’ would be more accurate. However, it should not be forgotten that he was the grandson of Ištemi Kaghan and thus one of the highest ranking men in the Turkic realm.

[23] The same is of course also true of Sogdian and Old Turkic texts.

[24] I have from this point forth reversed the presentation of these forms to reflect their relative likelihood, as indicated by the Frankish and Byzantine sources.

[25] See Beckwith (2005).

[26] See note 13 above.

[27] I owe my discussion of this point to a question Gisaburo N. Kiyose posed “about the name *Türkwat/*Türkwac. I wonder if a front vowel (‘u-umlaut’in this case) and a back vowel (‘a’here) could co-exist in a word” (Kiyose, p.c., 2007). I would like to thank him for this and other perceptive comments on an earlier version of this paper. The failure of the name or title to follow vowel harmony is decisive, in the negative, for the tenacious traditional theory of a ‘Mongolian‘-type plural in -t, and also for the ‘Turco-Sogdian’ variant. Both are also unsupported by any actual textual attestations. See the next note.

[28] For the ‘pure’ Sogdian theory, see Harmatta (1972); for detailed examination of the Sogdian data and criticism of the theory see Beckwith (2005).

[29] The native Old Turkic two-word title il-ügäsi (or el-ögäsi), for example, is composed of il ‘realm’ and ügä-si ‘advisor-possessive,’ meaning ‘advisor of the realm’ (Mackerras 1990:318–319).

[30] Several scholars have long argued that the Türk ruling clan—the Türk bodun or ‘Türk people’ proper—were not ‘Turks’ in origin but something else, possibly Iranians. In fact, the earliest Chinese source on the Turks suggests that the Turks were originally Sakas—one of its accounts of the origins of the Turks says they came from the ‘kingdom of the Sak’ (So kuo 索國 NMan suǒ < MChi *sak; Pul. 1990:298), i.e., the Sakas, to the north of the Hsiung-nu (Chou shu 50:907–908; cf. Sinor 1990b:287–288). Menander similarly remarks that the Turks “had formerly been called the Sakas” (Blockley 1985:116–117), or Sak(ai). This is significant because the Byzantine Greeks frequently refer to the Turks generically as ‘Scythians,’ but never as Sakas, and this is the only occurrence of the name Saka in the text of Menander. The name Turcae ‘Turks’ is mentioned in the first century ad by Pomponius Mela and Pliny the Elder as the name of a people living in the forest north of the Sea of Azov (Sinor 1990:285). That region was part of the classic Western Steppe zone and at the time Northern Iranian speaking except for the Greek colonies. It is also notable that typologically, and in some other respects, Turkic is a more ‘western’ language than Mongolic or Tungusic.

[31] Unlike Greek and Latin, Chinese has no grammatical distinction between singular and plural, which features can be determined only by context or by semantically explicit qualifiers.

[32] See Dobrovits (2004).

[33] The Old Turkic form Taβγač ~ Tawγač is metathesized; see Beckwith (2005).

[34] Or at any rate it was widely interpreted as having this meaning and was so recorded in Chinese records. Whether or not the name originally meant anything of the kind is unknown.

[35] Cf. a Chinese form of the name of the Sakas that follows the same structural model, in note 37. On the Sakas and the Turks see note 30.

[36] In neither Chinese nor Turkic is the marking of singularity or plurality obligatory or usual, so it is not possible to be more precise in English. However, the Byzantine Greek and Frankish Latin forms are both explicitly singular. It therefore seems clear that the Old Turkic title meant ‘Ruler of the Türk(s).’ Accordingly, it is highly likely that the T’o-pa title similarly meant ‘Ruler of the Earth,’ not ‘Rulers of the Earth,’ and that title too was applied metonymically, but unfortunately the T’o-pa language is very little known—there are no actual texts in the language—and the crucial glosses are in Chinese, a language which, like Turkic, does not have obligatory marking of singularity or plurality. In both cases, it is probable that it did not even occur to the speakers to make an explicit distinction.

[37] The usage is rather like the World War I British use of ‘the Kaiser’ for ‘Germany’ and ‘the Germans.’ There are other examples of the same type of usage in medieval Chinese sources, for example Sai-wang 塞王, literally ‘Saka king(s).’ Though this is often interpreted as referring to the king of the Sakas it sometimes is used instead of the simple name Sai 塞 *Sak (i.e., Saka), to mean ‘the Sakas.’