Thracian ( ) is an extinct and poorly attested language, spoken in ancient times in Southeast Europe by the Thracians. The linguistic affinities of the Thracian language are poorly understood, but it is generally agreed that it was an Indo-European language.

The modern understanding of Thracian is built on a massive corpus of personal names (onomastics), place names, and a few deeply obscure inscriptions that have defied clear interpretation.

Remnants of the Thracian language

Little is known for certain about the Thracian language, since no text has been satisfactorily deciphered. Some of the longer inscriptions may be Thracian in origin but they may simply reflect jumbles of names or magical formulas.Olteanu et al.

Enough Thracian lexical items have survived to show that Thracian was a member of the Indo-European language family.

Besides the aforementioned inscriptions, Thracian may be attested through personal names, toponyms, hydronyms, phytonyms, divine names, etc. and by a small number of words cited in Ancient Greek texts as being specifically Thracian.

There are 23 words mentioned by ancient sources considered explicitly of Thracian origin and known meaning.Georgiev, Vladimir I.. "Thrakisch und Dakisch". Band 29/2. Teilband Sprache und Literatur (Sprachen und Schriften [Forts.]), edited by Wolfgang Haase, Berlin, Boston: De Gruyter, 1983. pp. 1151–1153. Of the words that are preserved in ancient glossaries, in particular by Hesychius, only three dozen can be considered "Thracian". However, Indo-European scholars have pointed out that "even the notion that what the ancients called "Thracian" was a single entity is unproven." The table below lists potential cognates from Indo-European languages, but most of them have not found general acceptance within Indo-European scholarship. Not all lexical items in Thracian are assumed to be from the Proto-Indo-European language, some non-IE lexical items in Thracian are to be expected.

+ Word Meaning Attested by Cognates Notesἄσα (asa)colt's foot (Bessi)DioscoridesLithuanian dialectal asỹs 'horse-tail, Equisetum', Latvian aši, ašas 'horse-tail, sedge, rush' The etymology of both Baltic words is unclear and extra-Baltic cognates have yet to be established.βόλινθος (bólinthos)aurochs, European bisonAristotleProto-Slavic *volъ ("ox"). Per Beekes, Pre-Greek.See also Greek βοῦς 'cow', but Latv. govs ' id ' both < PIE *gwṓws. Proto-Slavic *vòlъ has no extra-Slavic cognates.βρία (bría)unfortified villageHesychius of Alexandria, compare the toponyms Πολτυμβρία, Σηλυ(μ)μβρία, and Βρέα in Thrace.Compared to Greek ῥίον (ríon; "peak, foothills") and Tocharian A ri, B riye ("town") as if < *urih₁-. Alternatively, compare Proto-Celtic *brix- ("hill").Greek ῥίον has no clear etymology. The Tocharian lemmata may be related.βρίζα (bríza)ryeGalenPerhaps of Eastern origin, compare Greek ὄρυζα, Sanskrit vrīhí- ("rice").The 'rice' words in Greek and Sanskrit are Wanderwörter. The Greek word may be borrowed from an Eastern Iranian language.βρυνχός (brynkhós)kitharaGeorgiev, Vladimir I. "Thrakisch und Dakisch". Band 29/2. Teilband Sprache und Literatur (Sprachen und Schriften [Forts.]), edited by Wolfgang Haase, Berlin, Boston: De Gruyter, 1983. p. 1152. Compared with Slavic *bręčati "to ring".The word may be onomatopoetic in nature. Furthermore, there is a grave issue with the inscription, as Gk /ŋ/ is written with a gamma before a velar, i.e., this word should be written βρυγχός, which it is not. βρῦτος (brŷtos)beer of barleymanySlavic "vriti" (to boil), Germanic *bruþa- ("broth"), Old Irish bruth ("glow"), Latin dē-frŭtum ("must boiled down").dinupula, si/nupylawild melonPseudo-ApuleiusLithuanian šùnobuolas, lit. ("dog's apple"), or with Slavic *dynja ("melon"). Per Vladimir Georgiev, derived from *kun-ābōlo- or *kun-ābulo- 'hound's apple'. Proto-Slavic *dyña (from earlier *kъdyña is most likely borrowed from Gk. κῠδώνῐον via Lat. cydōnia.γέντον (génton)meatHerodian, Suda, HesychiusPossibly descended from IE *gʷʰn̥tó- 'strike, kill', cf. Sanskrit hatá- 'hit, killed'The adjective *gʷʰn̥tós in the zero-grade has an *-s in the nom.sg., whereas in Thracian the word ends in a nasal, which is a serious issue that requires morphological remodelling in Thracian for it to be posited as the starting point for Thracian γέντον. Furthermore, the e-grade vowel of the Thracian potential avatar remains to be explained as well if from an original PIE *gʷʰn̥tós.καλαμίνδαρ (kalamíndar)plane-tree (Edoni)Hesychiusκη̃μος (kêmos)a kind of fruit with folliclePhotios I of Constantinople's Lexiconκτίσται (ktístai)CtistaeStrabomidne (in a Latin inscription, thus not written with Gk alphabet)villageinscription from RomeLatvian mītne 'a place of stay', Avestan 'dwelling'Πολτυμ(βρία) (poltym-bría)board fence, a board towerOld English speld 'wood, log'The OE lemma is poorly understood and extra-Germanic cognates are few and far between. OE speld may have descended from a PIE root *(s)pley- which is poorly attested and does not seem to be a formal match to the Thracian term.ῥομφαία (rhomphaía)broadswordCompared with Latin rumpō ("to rupture"), Bulgarian roféja, rufja (руфя), i.e. "thunderbolt"Предполага се, че освен названията на реки, селища, планини и други местни имена, като Пулпудева — Пловдив, Сердика — Средец, Месамбрия — Несебър, и др., в славянския език са били възприети и някои чисто тракийски думи: катеря се (качвам се), газя (тъпча), гудя (скътам, слагам), карпа (скала), руфя (мълния) и др., както и някои лични имена... В: Петър Петров, Образуване на българската държава (Издателство Наука и изкуство, София, 1981) стр. 67. and the Albanian rrufeja as derivatives of that word.L. A. Gindin (1999) Troja, Thrakien und die Völker Altkleinasiens. Versuch einer historisch-philologischen Untersuchung. Institut für Sprachwissenschaft der Universität Innsbruck. , p. 22.Ivan Duridanov (1985) Die Sprache der Thraker. , Hieronymus, p. 13. Slavic: Russian разрубать, Polish rąbać ("to hack", "to chop", "to slash"), Polish rębajło ("eager swordsman"), Serbo-Croatian rmpalija ("bruiser")The Slavic terms here must come from a medial *-bh-, whereas Lat. rumpō 'I break' must descend from a medial *-p- and therefore those words aren't even cognate with each other, let alone with the Thracian term.σκάλμη (skálmē)knife, swordSophocles, Julius Pollux, Marcus Aurelius, Hesychius, Photios' LexiconAlbanian shkallmë ("sword"), Old Norse skolmr 'cleft'The Albanian term is likely a secondary innovation. ON skolmr is unclear and has no extra-Germanic cognates; it is unlikely to be related to the Thracian term.σκάρκη (skárkē)a silver coinHesychius, Photios' Lexiconσπίνος (spínos)'a kind of stone, which blazes when water touches it' (i.e. 'lime')Aristotle PIE *k̑witn̥os 'white, whitish', Greek τίτανος (Attic) and κίττανος (Doric) 'gypsum, chalk, lime'.Although from the same PIE root, Albanian shpâ(ni) 'lime, tartar' and Greek σπίνος 'lime' derive from a secondary origin as they were probably borrowed from Thracian due to phonetic reasons pp. 159–161.τορέλλη (toréllē)a refrain of lament mourn songHesychiusζαλμός (zalmós)animal hidePorphyry of TyrePer Georgiev, derived from *kolmo-s. Related to Gothic hilms, German Helm and Old Iranian sárman 'protection'.Thracian initial ζ- can either be related to PIE *ḱ (as in these 'cognates' and several below) or to *ǵh-/*gh- as in the following entry, but not both. There does not exist an OIr word sárman, but a word śárman does exist in Sanskrit. However, Sanskrit ś- must go back to a PIE *ḱ-, not *k- as Georgiev states.ζειρά (zeira)long robe worn by Arabs and ThraciansHerodotus, Xenophon, HesychiusPer Georgiev, related to Greek χείρ (kheir) and Phrygian ζειρ (zeir) 'hand'.See above. The meaning of Phrygian ζειρα(ι) is unknown, not 'hand' as Georgiev believes.Klein et al., edd. (2018) HCHL:1820, chapter XVI.101 'Phrygian' by Ligorio and Lubotsky.ζελᾶ (zelâ), also ζῆλα (zêla), ζηλᾱς (zelās) winemanyCompared with Greek χάλις (khális; "unblended wine") and κάλιθος (kálithos; "wine")See above.ζετραία (zetraía)potJulius PolluxPer Georgiev, related to Greek χύτρα (khútra) 'pot'.See above.zibythidesthe noble, most holy oneHesychiusLith. žibùtė ("shining")

Inscriptions The following are the longest inscriptions preserved. The remaining ones are mostly single words or names on vessels and other artifacts. No translation has been accepted by the larger Indo-European community of scholars.: "There are as many interpretations of these as there are investigators; and as a result these monuments have not contributed anything to our knowledge of the language"

Ezerovo inscription

Only four Thracian inscriptions of any length have been found. The first is a gold ring found in 1912 in the village of Ezerovo (Plovdiv Province of Bulgaria); the ring was dated to the 5th century BC. The ring features an inscription in a Greek script consisting of 8 lines, the eighth of which is located on the rim of the rotating disk; it reads without any spaces between:

Dimitar Dechev (Germanised as D. Detschew) separates the words and proposes a translation as follows:

Kyolmen inscription

A second inscription, hitherto undeciphered, was found in 1965 near the village of , Varbitsa Municipality, dating to the sixth century BC. Written in a Greek alphabet variant, it is possibly a tomb stele inscription similar to the Phrygian ones; Peter A. Dimitrov's transcription thereof is:

ΙΛΑΣΝΛΕΤΕΔΝΛΕΔΝΕΝΙΔΑΚΑΤΡΟΣΟWritten from right to left. ΕΒΑ·ΡΟΖΕΣΑΣΝΗΝΕΤΕΣΑΙΓΕΚΟΑWritten from left to right. ΝΒΛΑΒΑΗΓΝ i.e. ilasnletednlednenidakatroso eba·rozesasnēnetesaigekoa nblabaēgn

Duvanlii inscription A third inscription is again on a ring, found in , Kaloyanovo Municipality, next to the left hand of a skeleton. It dates to the 5th century BC. The ring has the image of a horsemanPleket, H.W., and R.S. Stroud, eds. 1994. Supplementum Epigraphicum Graecum. 41–584. Amsterdam. with the inscription surrounding the image. Accessed 22 July 2024. It is only partly legible (16 out of the initial 21):

The word mezenai is interpreted to mean 'Horseman', and a cognate to Illyrian Menzanas (as in "Juppiter/Jove Menzanas" 'Juppiter of the foals' or 'Juppiter on a horse'); Accessed 22 July 2024. Albanian mëz 'foal'; Romanian mînz 'colt, foal'; Latin mannus 'small horse, pony';Kaluzkaja, Irina. "Thracian-Illyrian language parallels: Thrac. MEZENAI – Illyr. Menzanas". In: Thracian World at Crossroad of Civilizations – Proceedings of 7th International Congress of Thracology. Bucharest: 1996. pp. 372–373.Francisco Marcos-Marin. "Etymology and Semantics: Theoretical Considerations apropos of an Analysis of the Etymological Problem of Spanish mañero, mañeria." In: Historical Semantics—Historical Word-Formation. de Gruyter, 1985. p. 381. Gaulish manduos 'pony' (as in tribe name ViromanduiBalmori, C. Hernando. "Notes on the etymology of sp. 'perro'". In: Etudes Celtiques, vol. 4, fascicule 1, 1941. p. 49. DOI: https://doi.org/10.3406/ecelt.1941.1177 'men who own ponies').Georgiev, Vladimir I.. "Thrakisch und Dakisch". Band 29/2. Teilband Sprache und Literatur (Sprachen und Schriften [Forts.]), edited by Wolfgang Haase, Berlin, Boston: De Gruyter, 1983. p. 1161.

Stele from Samothrace

A damaged stele was found while excavating the Sanctuary of Great Gods in Samothrace. The stele carries an inscription dating to the late 5th or 4th centuries BC. The last two lines of the inscription could belong to a Greek postscript but the rest is clearly non-Greek. The preserved letters of the inscription read as follows:

Zone inscriptions