BMCR 2001.10.14

Les relations ethno-linguistiques en Thrace et en Mésie pendant l’époque romaine

, Les relations ethno-linguistiques en Thrace et en Mesie pendant l'epoque romaine. Sofia: Presses Universitaires "St. Kliment Ohridski", 2000. 174 pages ; 25 cm. ISBN 9540709180

‘Alle diese Bücher müssen mit der grössten Vorsicht benützt werden’ warns Warren Cowgill in his brief review of earlier scholarship on the Thracian language(s).1 Dimitâr Boïadjiev’s (henceforth B.) recent book continues this lamentable tradition. The ambitious title promises a general investigation of the linguistic landscape in what is today Rumania and Bulgaria and raises the expectation that we may finally learn more about the number of languages spoken in this region and their interaction both among themselves and with Latin and Greek. These expectations are let down as the author fails to present convincing evidence for the two principal conclusions of this book, namely that Thracians and Dacians were either one single ethnic entity or ‘une ethnie mixte’ (p. 151) and that Latin could not become deeply rooted in the region south of the Danube (p. 152).

Apart from a well-written introduction outlining his method and the state of research and a brief conclusion, B.’s book consists of a commentary both on chosen literary testimonies (Ovid, Strabo, Cassius Dio, etc.) and on a selection of inscriptions, organized around linguistic phenomena. In my review I shall concentrate on the latter, central section of the book.

B.’s interpretation of literary sources (pp. 32-81) gives a good selection of ancient testimonia and is mostly characterized by sound scepticism. Apart from the repeated references to forced migration in the Danube region (e.g. Str. 7.3.10; Eutrop. 8.3.1-2, 9.9.5; Hist.Aug. Aurelian. 39.7, Prob. 18.1-3), few of the ancient testimonies are accepted as credible by the author: B. demonstrates that Ovid does not bother to give an accurate picture of the ethno-linguistic situation in Tomis, being primarily interested in the effect of his poetry on the Roman audience, and he is doubtful of Cassius Dio’s statement (51.22.6-7) that the Dacians are Scythians of some sort. Likewise, B. questions the accuracy of the figures given by Strabo and others for the implantation of trans-Danubian tribes in the provinces south of the Danube (Str. 7.3.10, CIL 14.3608.9).

Already in this section, however, B.’s argumentation and evidence are often deficient or unconvincing. A few examples shall illustrate this. On p. 33 B. tries to illustrate the rarity of Latin speakers in Tomis in Ovid’s day by a numeric comparison of two Latin epigrams against 39 Greek ones. A diachronic perspective would have been more accurate: the Tomitan inscriptions dated to the time of Ovid’s exile are exclusively Greek; the earliest Latin inscriptions in Tomis are dated to the 2nd half of the first century AD, and Latin inscriptions then become more common in the 2nd and 3rd centuries.2 On p. 34 B. himself regrets not having arguments by which to prove that Geticis (Tr. 3.14.48) is not intended as an accurate indication of ethnicity: the use of a concordance would have supplied plenty of parallels for Getes and other ethnica pointing to all the tribes allegedly threatening the exiled poet, cf. e.g. Pont. 1.2.92 with 1.2.77-78; cf. also Ovid’s deceptive use of Scythia for Scythia minor. On pp. 38 and 40 B. seriously considers that Ovid had to use gestures to communicate with some of the inhabitants of Tomis, arguing that the lower-class, who could not afford to be represented on inscriptions, may have been in part unable to speak either Latin or Greek. This is highly implausible as the Getic lower class inhabitants surely had to be able to communicate with the (predominantly Greek) authorities and business people in Tomis and thus spoke Greek. On p. 43 B. suspects that the poetry written by the Thracian prince Cotys was, like most Thracian epigrams, of poor quality. He seems to be unaware that Ovid’s praise of Cotys’ poetic talent is supported by AP 16.75, where Antipater of Thessalonica, too, praises Cotys; also, B.’s negative judgement on the Thracian epigrammatists should be backed up by evidence. On p. 68 B. discusses the meaning of dilata at Flor.Epit. 2.28 (‘postpone’ or ‘repel’), using as his only evidence Jal’s Budé translation and a quotation from the modern scholar Pippidi, which, as B. himself acknowledges, is not even about the matter in question. A glance at the material at TLL s.v. differo 1075.78-1076.63 could have easily resolved the matter and prevented B. from trying to have it both ways by inadequately rendering dilata as ‘mise à la porte’. Thus, on the whole, pp. 32-81 provide a useful collection of source material, but B.’s commentary is unreliable.

Next I shall turn to B.’s treatment of selected inscriptions from the region (pp. 82-150). It begins with eight inscriptions which contain a confusion of the ablative and dative plural endings -is and -ibus. The examples chosen by B. feature natibus, suibus, avibus (instead of natis, suis, avis) and involve, in addition to the morphological confusion, an ambiguity in relation to the regular forms of natis, sus, and avis. B. rightly suspects that the development of the forms natibus, suibus and avibus in place of natis, suis, avis presupposes that the similar forms of natis, sus, and avis were not in use any more. B.’s main conclusion, however, that the morphological confusion reflects poor knowledge of Latin in the Danube region, is overstated: the fact that the same phenomenon is found also in the western half of the empire3 shows that the Latin of the Danube region follows a general trend and is, at least with regard to the errant usage of the -bus ending, by no means worse than the Latin in other regions of the Roman empire.

Furthermore, B.’s presentation is marred by philological inaccuracies. In discussing CIL 6.13176/7, B. states: ‘la traduction selon les “règles” de filibus suibus aurait donnè “pour leurs cochons d’enfants”‘, thus ignoring or not noticing that filibus, too, is an irregular form. B.’s commentary on the betacism in CIL 6.35381 (pp. 87-88) could have profited greatly from the material gathered by Mihaescu from late-antique authors coming from the Danube region,4 and his argumentation for the existence of *bibus (‘ivrogne’) is unconvincing as it is based only on the separation of seribibi at CIL 4.581 (without a serious discussion of the syntactical consequences).5 Finally, the quotation of the African inscription CIL 8.4669 (also showing the form avibus instead of avis), leads B. to an ill-balanced digression on Apul. Apol. 98: ignoring both the context and other evidence on African Latin,6 B. too credulously interprets Apuleius’ reference to his stepson’s poor knowledge of Greek and Latin as reliable evidence that ‘la rue des grandes villes romaines en Afrique était hostile au latin’.

The next set of inscriptions (pp. 93-121) deals with the origin, meaning and attestations of brutes. B. rightly refutes the meaning given in OLD s.v. brutes‘bride’, showing that for all attestations of the word the meanings ‘daughter-in-law’ or ‘young woman/wife’ are more appropriate. On the basis of the inscription Fiebiger n. 308, p. 150 (from Aquileia, dated to the first half of the third century AD where socrus (‘mother-in-law’) and brutes are opposed, B. argues for the meaning ‘daughter-in-law’. Furthermore, B. interprets the concentration of the attestations of brutes in the Danube valley and the fact that the word has not entered modern Rumanian as an indication of a closer link between Rumanian and Albanian, where, too, the word has no descendant (p. 117-18).

Two main objections must be raised here. Since the relationship between the two women in the inscription Fiebiger n. 308, p. 150 is already sufficiently clear because the one is called socrus (‘mother-in-law’), brutes does not necessarily have to be in exact correspondence and could just mean ‘young woman’. This latter, more general meaning would also suit the Greek attestations of βρούτις / βρουτ, where the word does not seem to point to a particular family relationship,7 and the etymology of the word, which probably derives from Indo-European bhreu- / breu-d-‘spriessen, schwellen’8 and is linked to German ‘Braut’, which today is both used of the bride proper and, colloquially, of any young woman. Furthermore, regardless of what brutes means, B.’s argumentation for the meaning ‘daughter-in-law’ and suggestion of a closer link between Rumanian and Albanian with regard to this word is incomplete, as it leaves one major question unanswered: if nurus is common in inscriptions from Africa, Gaul and Italy but (as far as I can see) unattested in Dacia and Moesia, and if brutes, according to B., regularly replaces nurus in these regions, why and how has nurus survived in modern Rumanian ( nora)?

B.’s discussion of brutes is followed by the treatment of ISM 5.3, 5.73 and 5.290b (dated to the 3rd and 4th centuries; pp. 121-126), all of which name a person who controlled or even executed the inscription ( (de)scripsit). B. points out the numerous morphological and syntactical errors in the first two, private, inscriptions and contrasts them with the ‘better’ Latin of the third, official, inscription, arguing that the inscriptions reflect the generally poor knowledge of Latin in the Danube region in the 3rd and 4th centuries AD. Unfortunately, B. does not compare the indications of a ‘décomposition de la déclinaison’ in ISM 5.3 and 5.73 with the Latin of inscriptions from Gaul, Africa and Italy. Isolated as they stand, B.’s observations only show that the Latin of ISM 5.3 and 5.73 falls short of the classical norm but does not indicate whether the knowledge of Latin in the Danube region was any worse than that of inscriptions in Africa, Gaul or Italy and does not allow any judgement on the spread of the Roman language or culture in this region: it might just be that the inhabitants of Moesia fluently spoke the same vulgar Latin as the inhabitants of Africa, Gaul and Italy.

The fourth set of inscriptions (pp. 126-133) finally touches directly on the linguistic relations between Thracian, Greek and Latin in the region: the Greek inscriptions IGBulg. I(2).345, which gives the Thracian etymology of the toponym Mesembria, and IGBulg. 4.2083, which employs the Thracian gloss γέντη (‘flesh’), and the Latin inscription IMS 6.81, which quotes the Menandrian proverb ‘whom the gods love, dies young’. B.’s commentary on the three inscriptions is good, and he is obviously right in seeing the inscriptions as evidence of the interaction between speakers of the three languages.

The following pages (pp. 133-139) once again (cf. brutes, above) concentrate on a single word: burgus. It had been suspected before by v. Wartburg, Walde-Hofmann and others 9 that burgus may not be of Germanic origin but may stem from Greek πύργος. B. now adduces an inscription from Teteven (Beseliev n. 55, p. 33, dated to AD 151) which features burgi next to the Greek loanword phruri, and argues that a Greek origin of burgus is also more probable because the word is first attested in a region bordering on Greece. This sounds persuasive on first sight, but why should πύργος, upon entering the Latin military Fachsprache be changed to burgus, while φρυρός is adopted without change? Further doubts arise in view of the fact that πύργος itself is a loanword 10 and that toponyms such as πέργη or πέργαμος seem to be linked to Indo-European bheregh and German ‘Berg’ and ‘Burg’.11

The next two sections (pp. 139-141, 141-144) deal with the quality of the Latin of official inscriptions. B. quotes two previously unedited inscriptions from Deultum (dated to the middle of the 3rd century AD which feature matri castorum and matri casrorum instead of matri castrorum, and interprets these lapsus as an indication that ‘[les autorités] de cette colonie…étaient incapables d’assurer la correction des inscriptions officielles’ (p. 141) and of the poor state of the romanisation in this region. The third inscription, an official record on the rebuilding of the aqueduct of Serdica, dated to AD 580 (Besevliev, no. 3), is likewise taken as proof of the ‘latinité agonisante’ (p. 144) because of the use of a dativus adnominalis instead of a genitive in instantia dom(ino) v(iro) beatiss(imo) Leontio archiepiscopo and because of the phonetic spelling of -ts- for -t- and -e- for -ae- and the confusion of the numbers XIL and XIV.

Again, B.’s argumentation does not stand up to scrutiny: his reasoning concerning castorum/casrorum is problematic as, in a certain sense, the two errant spellings ‘add up’ to the right one and suggest that mere carelessness has made the lapicida forget the -t- and the -r- on the respective occasions (observe that on one occasion the lapicida has also omitted the second -e- of Severae). As for the errors detected in the inscription from Serdica, B.’s evaluation ignores basic linguistic facts: the confusion of -ae- and -e- was widespread and typical, for example, of the language of Italian peasants;12 the pronunciation of -t- before -i- is generally moving towards -ts- in the 2nd and 3rd centuries AD, even in Rome;13 the dativus adnominalis is also attested in the Augustan poets, Livy and Tacitus, and becomes increasingly common in late prose writers, such as Tertullian,14 which suggests that the chancellery of the archbishop of Serdica may, after all, have imitated the Christian prose of their day. Consequently, there is not much evidence of a ‘latinité agonisante’ in these three inscriptions.

The final subsection is devoted to the anthroponyms of the region (pp. 145-150). B.’s starting point is the inscription IGBulg. 3.2.1690 from Pizos, dated to AD 202, furnishing the names of 21 Thracian peasants who moved into a newly established emporion. In analysing the names, B. discerns 10 first members of compound names, and 4 second members of compound names. Claiming (rather than showing!) that this set of components constitutes ‘une partie essentielle de l’anthroponymie thrace’ (p. 147), B. reaches the conclusion that this set of components can be used to determine whether any given name of two roots is Thracian or not. This tool is then used to show that all names currently held to be Dacian (according to B. only Natoporus, Pieporus, Δεκέβαλος, and Scorylo) are in fact part of the Thracian anthroponymic system.

In this section B. is clearly out of touch with the state of research. He refers to C. Poghirc, ‘Thrace et daco-mésien: langues ou dialectes?’, Thraco-Daciaca, Bucarest 1976, and V.I. Georgiev, Trakite i tehnijat ezik, Sofia 1976, but does not (despite later quoting Georgiev’s article ‘Thrakisch und Dakisch’ in the same ANRW volume!) take notice of Georgiev’s contribution ‘Thrakische und Dakische Namenkunde’, ANRW 2.29.2, 1195-1213, in which his Bulgarian colleague had not only studied Thracian anthroponyms in greater detail but had also given a far longer list of Dacian anthroponyms, which are distinctly different from the first and second member components of B.’s ‘système anthroponymique’. Moreover, B.’s use of this system is methodologically unsound, as a glance at his argumentation in the case of Δεκέβαλος will show: B. links Δεκέβαλος on grounds of its 2nd member ‐βαλος with Dinibalis and then links Dinibalis on grounds of the first member Dini- / δινι to Δινίκενθος, which is attested on the Pizos-inscription, thus obviously ignoring the possibility that the Dacian and Thracian anthroponymic systems may have some similar, even common, first and second members, but still be distinct: in the present case, for example, Δεκέβαλος might theoretically just be the Dacian equivalent to Dinibalos. Furthermore, in the light of the extensive (forced) migration in the region, it is questionable whether all names attested in Teteven (‘au coeur de la Thrace’, p. 146) in AD 202 must be Thracian. Finally, B.’s conclusion of a continuous anthroponymic system covering Thracian and Dacian seems to contradict Georgiev’s demonstration 15 that the toponyms of Thracia and Dacia/Moesia are distinct: since B. cites and thus obviously knows this article by Georgiev, one would have expected him to engage critically with its results.

On the whole B.’s book shows that there is still a lot to be done on the linguistic situation in Thracia and Moesia. Unfortunately, B.’s own contribution to our knowledge is meagre. The author should have taken greater care in evaluating the evidence and presenting his results. Likewise, A. Tschaouchev, who (according to p. 4) prepared the book for publication, has overlooked a great number of most obvious typographical errors, e.g. ‘Istors’ instead of ‘Istros’, ‘agsse’ instead of ‘agisse’ (both p. 72; on average at least two errors per page), and has even had the hexameters indented instead of the pentameters in most of the quotations of elegiac verse.


1. W. Cowgill, Indogermanische Grammatik. Band I. 1. Halbband: Einleitung, Heidelberg 1986, 55. He is referring to D. Detschew, Die thrakischen Sprachreste, Wien 1957; V. Georgiev, Trakijskijat ezik, Sofia 1957; I.I. Russu, Limba Traco-Dacilor, 2nd ed. Bucuresti 1967, and G. Reichenkron, Das Dakische, Heidelberg 1966.

2. The first official Latin inscription from Tomis is dated to 2 December AD 76 (ISM 8); the Latin funeral inscription ISM 168 may be earlier; cf. also ISM 169, 170 (late I. AD 39,40,41 (I/II. AD).

3. Cf. e.g. the material gathered at F. Neue/C. Wagener, Formenlehre der lateinischen Sprache, 3rd ed. Leipzig 1902, 1.190-1; TLL s.v. filius 753.39-41, and J. Herman, ‘Le latin dans les provinces danubiennes de l’Empire romain’, ANRW 2.29.2, 1089-1106.

4. H. Mihaescu, ‘La langue latine dans le sud-est de l’Europe’, ANRW 2.29.2, 1107-1147, imprimis p. 1113.

5. The editor of CIL 4.581 comments: ‘punctum post seri non vidi, album enim hic illic delapsum erat.’ Consequently, no dot is to be placed between seri and bibi.

6. Cf. V. Hunink, Apuleius of Madauros. Pro se de magia, Amsterdam 1997, vol. 2, p. 241 ad loc.

7. Cf. Lydus de Magistr. 1.33, de Mensib. 4.24, Joh. Malalas Chronogr. 88, Suidas s.v. βρούτιδες, Zonaras s.v. βρούτιδες.

8. Cf. J. Pokorny, Indogermanisches etymologisches Wörterbuch, Bern/München 1959, vol. 1, 169-70 s.v.

9. Cf. W. v. Wartburg, Französisches etymologisches Wörterbuch, Bonn 1928ff, s.v.; A. Walde/J.B. Hofmann, Lateinisches etymologisches Wörterbuch, Heidelberg 1938/1954, s.v. (with further references).

10. Cf. P. Chantraine, Dictionnaire étymologique de la langue grecque, Paris 1968, s.v.

11. Cf. Pokorny (n. 8), vol. 1, 140-141 s.v.

12. Cf. Lucil. 1130 (Marx), Var.L. 5.97, and M. Leumann, Lateinische Laut- und Formenlehre, München 1977, 67-68 (with further evidence).

13. Cf. CIL 6.34635: nacione, and Leumann (n. 13), 154 (with further material).

14. Cf. E. Löfstedt, Syntactica, Lund 1928/1933, 1.164-8 and the material gathered by G. Landgraf, ‘Der Dativus commodi und der Dativus finalis mit ihren Abarten’, Archiv für lateinische Lexikographie und Grammatik 8, 1893, 39-76.

15. V.I. Georgiev, ‘Thrakisch und Dakisch’, ANRW 2.29.2, 1148-1194, pp. 1185-89.