VII. Comparative phonetics of the Thracian language


Chapter VII. VERGLEICHENDE PHONETIK DER THRAKISCHEN SPRACHE (I. Duridanov, Die Sprache der Thraker, Neuried, 1985)

translated by Jordan Tchorbadzhijski



The concrete appearence of the words in every language is defined by the occurrence and sequence of a certain number of phonetic units (phonems) [1]. The Thracian language has developed its own phonetic system after its separation from the Proto-Indo-European (PIE). Some of the Indo-European (IE) phonems have remained unchanged, whereas others have experienced some change, especially regarding the consonants. The phonetic system of the Thracian could  be described purely descriptive, regardless of the origin of the particular phonems, or on a historical basis, in a diachronical perspective, that is comparing it with the phonetic systems of the other Indo-European languages. Following the second methodology would throw more light on the problem of the position of the Thracian in the group of the IE languages, which is why I have chosen it.


The preserved bits of the Thracian are chronologically erratic. The earliest records are from the end of the second millenium B.C. (some Thracian personal, tribal and geographical names attested in the Iliad and the Odyssey, whose oldest parts originate from the end of the Mycenaean period), the latest – from the 6. century AD. Certainly, during this long period of 1600 years the phonetics of the Thracian has been changing, a fact that can be observed on the basis of the evidence from the last centuries of the Thracian history.



1. A phoneme is a basic, theoretical unit of sound that can distinguish words.




In some cases the possibility of a certain Greek influence is not excluded, and after the arrival of the Romans – accordingly Latin. Those are problems, which have not been adequately researched yet. The important thing, however, is that on the basis of the etymological analysis of the material we are able to reconstruct to a large extent the phonetic system of the Thracian and to identify a number of very old sound changes, as well as some that must have taken place in the late period of the evolution of the language.



The vowels


The IE featured in its last period long and short vowels: a, e, i, o, u, ə and ā, ē, ī, ō, ū. In the first period of its history, the Thracian preserved that state, excluding the short o, which changed to a and the so called Schwa (ə), that was substituted in Thracian (as in the most IE languages) by a.


IE a > Thr. a. There are many examples of that change: Hydronym (HN) Asamus

IE *ak’mo- “stone“, Greek (Gk) ákmōn “anvil”, Old Indian (OI) aśman “stone, sky”; HN Angítēs – IE *ank-, *ang- “bend”, Gk ankn “bow, elbow”, ánkos “dale, ravine”

IE e > Thr. e: Sérmē – IE *sermā, IE stem *ser “flow“ in Latin (Lat) serum “the watery

part of the curd, whey”; Toponym (TN) Meldia – IE *meldh-, cf. Lithuanian (Lit) meldà, méldas “a kind of cane”. In the late period of the Thracian (it well may be the case only in some dialects) the e, followed by –nt(h)– developed to i, which can be concluded of the existence of the form –cinthus in the personal name (PN) Dias-cinthos alongside –centus, -kenthos (in Dias-centus and Dias-kenthos) < IE *kento-s “child, descendant”, Lat re-cens “fresh, young, new”. The unstressed e tended to i (reduction), maybe a vernacular feature, cf. the parallel forms Desa-(kenthos) and Disa-(centus), Diza-(kenthos) – IE dheso “god” in Gk theós.

 IE i > Thr. i: TN Idē, Idakos < IE *idhu “tree”, Old Irish (OIr) fid “tree, wood, forest”,

Gaulic tribal name (TrN) Uidu-casses, Old English (OE) widu “forest, wood, tree”; -diza, -dizos “castle, fortress” (in TN Tyro-diza, Burtu-dizos) < IE dhig’hā, Old Persian (OP) didā (from *dizā) “fortress”.

IE o > Thr. a: skálmē “knife, sword” < IE *skolmā, Old Icelandic (OIc) sklm “knife,

sword”, OE helma, English (En) helm (rudder); HN Pan(n)ysis,





Panisas < IE *pon-iso, stem *pon(i)- in Gothic (Got) fani “mud”, Middle Irish (MIr) HN On, Onach (swamp); Thr. TN Markellai < IE *morkā in Lit markà “a hole, used in the treatment of hemp”, Ukrainian morokva “mire” (from Proto-Slavic *mărky, -ъve) < IE *merk in Lit mekti (merkiù) “to put in water, to dip, to wet”. The IE ending in Nominative, singular of the –o-roots was accordingly –as in Thracian, which can be concluded of the exactly rendered HN Iuras, Panisas (alongside Panisos), PN Taruthinas (alongside –os). But in the most cases the ending was rendered as –os (in the Greek sources) or –us (in the Latin; sometimes under Greek influence also –os), being the result of an (morphological) adaptation of the Thracian nouns of masculine gender, ending with –as to the Greek –os and the Latin –us.

IE u > Thr. u: Suitulēnos, a sobriquet of the Heros, a development of the TN *Suitula,

cf. Lit švitulỹs “something glaring, light”; Muka- in PN Muka-boris, Muka-poris and others – IE *muk, cf. Iranian muka- in Ossethian mugæ “family”, muggag “seed, house”. Sometimes the Thracian u was represented by an o in Greek (and also in Latin), which could be explained by the openness of the Thracian u (so D. Detchev) or by a peculiarity of the Greek vowel development: the letter υ has already been used to denote ü or (in a later time) even i, so that the Thracian u could not be represented by an υ.

Compare for example the PN Moka-poris, Moca-por alongside  Muka-; TrN Odrysai < *Udrusai, from IE *udrā in avestan (Av) udra- “otter”.

IE ə [2] > Thr. a: PN Gaidrēs < IE *ghəidr- in Lit gaidrùs “light (colour), clear (weather)”,

Gk phaidrόs “clear, bright, blazing”, from IE *ghēi, *ghəi; TN Daton (-os) < IE *dhəto-m(-o-s), cf. Albanian (Alb) datё “place, locality”, from IE *dhē-, *dhə- “to sit, to put, to lay”.

IE ē > Thr. ē (represented as η in Gr): PN Rhēsos (ρσος, testified by Homer) < IE

*rēg’o-s, cf. Lat rēx, rēgis “king”, OI rāj- “the same”; TN Zrynthos, Zērýnthion, Zrinthon (in Gr Ζρυνθος and so on) < IE *g’hēr- “wild animal” in Lit žvėrìs “wild animal”, old bulgarian (OBul) zvĕrь “the same”, Gk thēr “animal”. In the late Thracian the ē could well have changed to an ī (cf. the word for wine, used by Hesychios zílai (ζίλαι) to the zēlas (ζήλας) and zelâs



2. This vowel has developed in the late IE from the composition of the so called second laryngeal and e, that is: H2e.




(ζελς) in other Greek sources; from IE *g’hēlā), if the first example does not reflect the Greek change of ē (η) to i. I see a sure example for that sound change in a Thracian loan word in Bulgarian: griv “gray (of a dove, falcon, eye)”, OBul *grivъ, borrowed in modern Greek as grivós, Romanian griv “colourful, spotted white and black” from the Thr. *grīvas < Old Thr. *grēvas < IE

*g()hrēo-s, cf. Old High German (OHG) grāo, genitive grāwes, German grau.

IE ō > Thr. ō (represented as ω in Gr): HN Strmn (in Greek Στρμών, so Hesiodus in the 7.

century B.C. and Herodotus in the 5. century B.C.) < IE *Srū-mōn, cf. Lit sraumuõ, -meñs “a quick current“, where the –uo traces back to IE ō; Mygdon (Μυγδών, a personal name, used by Homer) is in fact a name of a region, known later as Mygdonia, from IE *Mko-ghdhōm, the second part being connected to Gk chthōn “earth, ground”.

IE ū > Thr. ū (represented as υ, ου in Greek): HN Strmn (Στρμών) < IE *Srū-mōn, root:

IE *sreu- - sr - in Lit srúti (srūvù, srūnù) “to fill with water, to stream, to flow”.

The quantity of the Thracian vowels often was not reflected in Greek, particularly the ā and ī. On the basis of the above mentioned examples we can assume that long vowels were a part of the Thracian phonetic system, that is, the length of the vowels was a distinctive attribute; hence long a and i must have existed along ō, ē, and ī. For example the IE ā as an ending in the singular, nominative form of the feminine nouns of the ā-root would be preserved; cf. Thr. bria “town, city” < IE *riā, cf. Tocharian (To) A ri, B riye „town, a refuge”; -diza in TN Tyro-diza, Oru-disza < IE *dhig’hā “fortress”. One should also keep in mind that the Thracian ā was not rarely represented by an η in Greek (especially at the end of a word), accordingly to the sound laws of the Greek (Ionic-Attic), compare for example the Thr. TN Στρύμη, which in Thracian must have been *Str.

IE ī > Thr. ī: TN Ilion < IE * īl(u)- “mud, dirt” in Gk īlýs “mud, dirt”, Old Church Slavonic

(OCS) ilь “sllime, mud”.


In late Thracian several changes affected the long vowels. The long ū lost its dislabilized and evolved to a front vowel, i, possibly via ü. Such a change can be suggested on the basis of a comparison of the name Dinace (from a Latin inscription in Carnuntum, Austria), if it is correctly regiven, and




the Dunaka (Δούνακα, Accusative), mentioned by Strabo < IE *Dhūn-āk, root: IE *dhūn, cf. the OE dūn “elevation, mountain”. The same change of ū to ī is possibly reflected in the Thr. PN Bizo, Bizens, Bizias, Bizzos, Deo-Bizos (all of them known from inscriptions) alongside forms like Byzas, Byzēs, Byzos, which were represented in Greek by an υ in the stem; that Greek υ was used to represent Thr. ū is obvious, if we consider the etymology of the names: from IE *bhūg’o-s in Av būza “a he-goat”. However, the sound change ū > ī must have been typical just for some dialects and not for the whole Thracian language area. The rendering of the Thracian tribal name Bess as Βησσο (with η = ē) by Herodotus and as Βέσσοι by the late autors (like Strabo, Polybius etc.) cannot be seen as an evidence for a change ē > e: in the Greek language of the 1.-2. century AD η already denoted i corresponding to the Greek sound development, that is why the foreign ē could not be represented by that letter. It is well possible that the Greek autors had to use ε, although it didn’t represent a long vowel at that time.


Previous to ē and ū at the beginning of a word sometimes appears a prothetic j-, as in the following examples: HN Ieterus < IE *ētro-, OHG ātar “quick” (ā < IE ē); Iuras < IE *Ūro-s, cf. the Old Norse (ONo) ūr “drizzle”, Lat ūrina “urine”, Lit (also with a prothetic j-) jra (and plural jrios, -s) “sea”, Latvian jūra “the same”. However, this phonetic phenomenon was not common Thracian, as we can see from the examples of an ū at the beginning of words (sure evidence of ē does not exist): HN Ucasus, Utus (etymologically both had a long ū at the beginning, s. p. 48).


In Thracian (especially in its late period, beginning with the 3.-2. century BC) the unstressed vowels were apparently subject to a reduction, as a result of which they could sometimes even fully disappear. The PN Mygdōn (Homer) gives us the earliest example for this phenomenon (syncope): the name is connected to the land, which was later to be known as Mygdonia (Thucydides, 5. century BC), as per the offered etymology derived from *Mko-ghdōm “marsh, swampland” (s. p. 35). Another example for syncope is the HN Atlas (Herodotus, 5. century BC) < *Attulas, cf. the Latvian HN Adula and the German HN Attel < Attula (s. p. 23). Syncopized appear in other Thracian names too, e.g. Pyr-mērulas (a sobriquet of the Heros) alongside Pyru-mērulas and Pyry-mērulas; the deity name (DN) Zbel-thurdos, Zbel-thiurdos etc. <  older *Zibel-thurdos, cf. the Latvian zibele “twinkling”; TN Zburulus < older




 *Ziburulus, cf. Lit žiburỹs “fire, light”; Zym-drēnós, Zym-zdrēnós (a sobriquet of Asclepius) alongside Zym-ydrēnós, Zyml-yzdriēnós etc. Another example would be the word midne “village”, as in in its Latvian parallel *mĩtne “residence, dwelling, accommodation” < older *mītinē or *mītenē.


The diphthongs


The IE featured i- and u-diphthongs. They had their counterpart in Thracian as follows:

IE ai > Thr. ai: Aizikē (country), from the IE root *aig’- “a goat”, cf. Gk aix,

Genitive aigós, Armenian (Arm) aic “a goat”.

IE ei > Thr. ei: Seietovien(us) (a sobriquet of the Heros), derived from the TN *Seietovia

< IE *Seiətoā, cf. Lit sietuvà “a whirlpool, a deep spot in a river”.

IE oi > thr ai: Paisulēnos (a sobriquet of Zeus), derived from the TN *Paisula < IE

*Poik’ulā, cf. Lit paišaĩ “grime”, Gk poikílos “colourful”. In the TN Vevocasenus vicus, Vevo- (possibly from Thr. *Vaiva) < older *Vaevo is explained by the Latin mediation (s. p. 49).

IE əi (< H2ei) > Thr. ai: PN Gaidrēs < IE *ghəidr- in Lit gaidrùs “light (colour), clear”, Gk

phaidrόs “clear, bright, blazing”.

IE əu (<H2e) > Thr. au (av): bulgarian (Bg) Osogov(o) (a mountain) < Thr. *Asagav < IE

*Ok’o-ghəo-, compare the second element of the name to Got gawi “land, area”, pre-Greek gaia “land, area”.

IE ou > Thr. au: TN Pautalia < IE *pout-, cf. Old Prussian (OPru) HN Pauta < IE

*Poutā from the IE stem *peut-: *put- in Lit putà “foam”, Latvian putas “the same”.

IE ōu > Thr. o (maybe ō): TN Ost-aphos, Ostu-dizos, the first element being derived from

the IE *ōusto-s or *ōusta “mouth, estuary”, Lit úostas, uostà “estuary”, Lat ōstium “the same”.


In the word zetráia “pot” < IE *g’hetraā the IE diphthong eu is surprisingly substituted by e. In all likelyhood the word was not reproduced correctly by Pollux (2. century). Admittedly there is no sure example for the substitution of the IE eu, that is the case with the rest of the IE diphthonhs as well.


In the late Thracian we can assume a monophthongization of ai to i previous to consonants on the basis of the mountain name




Im- < Haimos (s. p. 30). The only example for a monophthongization au > o is the TN Potelense, a Latin development of *Potela < *Pautela (s. p. 39). However, because the same sound change is typical of the balkanic Latin from the 2. century AD onward (see for examples H. Mihăescu 1978, p. 185), it is difficult to decide whether the monophthongization of au to o previous to a stressed syllable (*Pautéla, *Potéla) took place in the late Thracian or was due to the development of the balkanic Latin.


The sonants


The IE sonants , , , have a twofold substitution in Thracian:

  1. IE , , > Thr. il, ir, in: TN Silta, Lit šitas “warm, comfortable” < IE *k’lto-; field name (FN) Zilmissus < IE *g’hm-is-(o)-, cf. Lit Žilmà, comprehensive (follstufig) Latvian zlme “verdant grass, a meadow”; TN Tirsai < IE *Tsoi, cf. OPru TN Tirs-kaymen, Lit tištis “thicket”; TN Dingion < IE *dhgh-, cf. Latvian dinga “plant; fructiferous land”; PN Brinkazis, Brinkainos < IE *bhrk-, cf. Lit brìnkti “to swell”. We have no evidence of IE > Thr. im.
  2. IE , , > Thr. ul (ol), ur (or), um (om), un (on): bólinthos “a bison” < IE

*bhn-ent-, cf. German Bulle < IE * bhn-ēn; póltyn “a wooden fort” (TN Poltym-bria) < IE *(s)p-tu-(n), cf. OIc spjald “a board” (*spel-to-s); burt-, burd- “a ford” in TN Burd-apa, Burd-enis, Burt-icom, Burtu-dizos < IE *bhd(h), cf. OBul (OCS) neprĕ-brьd-omъ “boundless, unlimitted” (literally “impossible to roam through” < Proto-Slavic *bьrd-); TN Kurpisos, IE root: kp, cf. Lit kupti “to grub”; brynchós “a guitar” < IE *bhrko-s, cf. Polish brzęk “clink, rattle”, Ukrainian br’ak, “clink, rattle”; Rumbo- (in TN Rumbo-dona) < IE *rb(h)-, cf. Latvian ruba “a waterfall, a shoot”, Lit rubas “an edge”; Sonkētēnē, a sobriquet of Hera, a derivation of TN *Sonkētā, cf. Lit sunkà “juice, sap, liquid” < IE *sk-. The fact that Thr. u was reproduced in Greek as o could be explained either by the openness of the Thracian vowel or by the absence of another option: the Greek υ already denoted ü or i.


There are three examples of IE sonants previous to a vowel: PN Skilas < IE *skьlo-s, cf. Lit skìlti (skiliù, skýliau) “to light a fire” and “to run/ride very fast”; TN




Keiris, Keir-para, Kēri-parōn (with Greek ει, η = i) < IE gьr(i)- in OI girí-, “a mountain”, Lit girià, žemait gìrė “a wood, a forest”; -thinas (-sinas) in PN Taru-thin(n)as, Taru-sinas, Taru-tinos < IE tьn-, a reduced grade of the IE root *ten- in Lat teneō, -ēre “to hold”.


The IE has resulted in Thracian in v, which was reproduced in Greek by υ, β, ου and in Latin by v and b. An incontestable example for the conservation of the Thr. v up to recent times in Bulgarian is the HN Veléka, that was inherited from the Thracians: it is a development of the root *el- in Lit velti “to wash”. Another example is the TN Βαίρος (= Vairos), used by Ptolemaios (2. century AD), cf. Lit vairùs, vairas “rotating, looping” < IE * o-ro-. Thr. v was reproduced as ου in the sobriquet of Dionissius Ουηρζεληνος, a derivation of the TN *Verzela, cf. Lit HN Vérža. In the first syllable of a word, previous to a vowel, v could disappear, apparently under Greek influence, e.g. TN Egerica, cf. Lit HN Veger, Latvian vedzere “a crick” < IE *eg-; TN Idē, Idakos < IE * idhu- “wood, timber” in old irish fid “a tree, wood; a forest”, OIc viđr “a tree”. In TN Chalá-stra the second element is derived of an older *strava, cf. Latvian stràva “to stream, a stream, a downpour”, then the -v- between two vowels has disappeared, presumably under Greek influence, after that a contraction of -aa- > -а- took place so that it came to -stra at the end. The Thracian v was not preserved after the dental d (<  IE dh), as in PN Desa-kenthos, where the first element Desa- < IE *dheso- “deity”, compare to the related word in Lithuanian dvãsas “a ghost, a spirit”. The conservation of Thr. v after the fricative consonants s and z is attested by the following examples:

1.      IE k’ > Thr. sv, (zv): in the sobriquet of the Heros Σουιτουληνος, Σουητουληνος (= Svitulēnos), a derivation of the TN *Svitula (s. p. 46); in the PN Εζβενις (= Ezvenis), Εσβενειος (= Esvenios), derived of the Thr. *esva-, ezva- “a horse” < IE *H1eko-.

2.      IE g’hv > Thr. zv in TN ρος Ζυακοζρερων (= *Zvakozrenōn) with first element being derived of IE * g’hәk-, cf. Lit žvãkė “light”.


In some Thracian dialects and possibly under Greek influence the v previous to the fricative consonants s and z has been dropped out. Compare for example the TN Zrynthos, Zrinthon (Ζήρυνθος, Ζήρινθον) from an older *Zvēr-unth- (-inth-) < IE *G’hēr-t- (s. p. 50); DN Sabazios, Sabadios (Σαβάζιος, Σαβάδιος) – from an older *Svabazios, *Svabadios < IE *sobhodhio-s.


In the word bria “a town, a city”, which was reproduced in Greek as βρία (Strabo) and in the same manner in many TN (*Αλααι-




βρια, *Βολβα-βρια, Μασκιοβρια, Μεσαμβρία, Πολτυμβρία etc.), the Greek β may have represented v or b. For good reasons, it is to assume that we have the phonem b in that case, that is: Thr. v (< IE ) previous to r altered to b. So the TN Mesemvrija (with this pronunciation in more recent times) is written by Herodotus (5. century BC) as Μεσαμ-βρίη, and in his time the Greek letter β denoted the sound b. The Slavs have adopted the name from the indigenous population with its original -b-, compare the evidence from the Slav sources: Nesбрь (Нєсбръ, from the 11. century), Nesębrь (Нєсбръ, AD 1337). In the PN Esbenus (Hesbenus) we have also a Thr. b < older v after an s.


The consonants


We can observe many changes in the Thracian system of the consonants, some of them are typical of other IE languages too.


The IE voiced aspirated consonants (the so called Mediae aspiratae) bh, dh, gh in Thracian have lost their aspiration and have merged with b, d and g (“und sind somit in reine stimmhafte Konsonanten b, d, g übergegangen”) .

IE bh > Thr. b: bólinthos “a bison” < IE *bhn-ent-, cf. German Bulle; TN Bérgē,

Bergúlē < IE *bhergho-, cf. OIc berg, OHG berg = German Berg; burt-, burd- “a ford” in TN Burd-apa, Burd-enis, Burt-icom, Burtu-dizos < IE *bhd(h)-.

IE dh > IE d: TN Dymē, Lit dmas- “dark, blackish brown (of cattle), Latvian dms “dark

brown”, OI dhūmá- “smoke” < IE *dhmo-s; -dizā, -dizos “a fortress, a fort” in TN Tyro-diza, Burtu-dizos, Tarpo-dizos < IE *dhig’hā (-o-s), OP didā, New Persian (NP) diz, dēz “a fortress”, Gk téichos “a wall”.

IE gh > Thr. g: TN Bérgē, Bergúlē < IE *bhergho- “a mountain, a riverbank”; TN Dingion

< IE *dhgh-, cf. Latvian dinga “plant; fructiferous land”, OE dynge, OHG tunga “fertilisation”.


The IE voiced consonants (the so called Mediae) b, d, g have changed in Thracian to become their voiceless equivalents p, t, k (sound shift):

IE b > Thr. p: Skalpēnos, a sobriquet of Asclepius, a derivation of the TN




 *Skalpa < IE *Skolbā, cf. Lit HN Skab-upis, verb skabti (-biù, -biaũ) “to wash, to beat”; dinupula “a plant similar to the beetroot” < IE *k’un-ābōlā, cf. Lit šùn-obuolas “a wild apple tree”, crimean goth apel “an apple” < IE *bel-, *bl-

IE d > Thr. t: HN Utus < IE *Ūdo-s = *ūdo-s, OI ud-án „water“, Gk hýdos “the same”;

Taru- in PN Taru-thin(n)as < IE *doru- “a tree” in Gk dóry “a trunk, wood, a spear”, Hittite ta-ru “a tree, wood”.

IE g > Thr. k: TN Kíkones, singular Kíkōn < IE *gigō(n), cf. OE cwicu, En

quick; skárkē “small silver coins” < IE *skor-gā, cf. ONo skark “a noise”, OI kharjati “crunches”.


If one encounters within the borders of Thracia a geographical name, in which the IE voiced consonants (Mediae) are preserved, one can regard them as non-Thracian. Such is the case with the antique TN Salmydes(s)ós (Sophocles, Herodotus, Xenophon and others), that belonged to a coastal strip in Eastern Thrace, a town (nowadays Midia on the coast of the Black Sea) and maybe even a river. The basic form of the name is suested to have been *salm-udes-o- or *salm-udes- “salty water”, the first element being of the same root as Gk hálme “sea water, saltness, pepperiness”, the second as OI ud-án “water”, Gk hydos “water”; accordinly the IE d was preserved in the second part of the word –ydes(s)os. On the basis of this phonetical criterion we can judge that the name was not Thracian. The fact that in antique Thrace there were TN of non-Thracian origin is not astonishing. It is enough to look at the present-day toponimy of the country: along the Slav, we encounter here Romanian, Osman-Turkic, Greek as well as a smaller number of adopted TN from Thracian. On the Black Sea coast there are not only Slav but also old Greek names as Sozopol, Akhtopol (from Agatopolis), Gálata (a rocky cape near Varna) and so on.


The IE voiceless explosives (the so called Tenues) p, t, k have shifted in Thracian to their correspondent aspirated consonants (the so called Tenues aspiratae) p’, t’, k’ (ph, th, kh) (in Greek: ϕ, θ and χ):

IE p > IE p’ (ph): rhomphaia (ῥομϕαία) “a spear, a javeline” and later “a sword” < IE

*rump- in Lat rumpō, -ere “to break”; -aphos in TN




Ost-aphos (΄΄Οστ-αϕος), cf. OPru apus “a spring” < IE * p- “water, a river”.

IE t > Thr. t’ (th): -kenthos in PN Kenthos (Κενθος), Desa-kenthos (Δεσα-κενθος), Bithi-

centhus, Epta-kenthos (Επτα-κενθος) etc. < IE *kent-to-s “a child, descendant”, Lat re-cens “fresh, young, new”; TN Perinthos (Πέρινθος) < IE *Perto-s, OI párvata- “a mountain”, Hittite peruna- “a rock”; alonside Trausoi (Τραυσοί in Herodotus; a Thracian tribe in the southwestern Rhodopes) we encounter (in Livius) the form Thrausi too (with th < IE t, see above).

IE k > Thr. k’ (kh): brynchós (βρυνχός) “a guitar” < IE *bhrko-s, Polish brzęk “clink,

rattle”; Chala- in TN Chalástra (Χαλά-στρα) < IE *kālo- in OBul kalь “slime, dirt”; -sychis in PN Eptē-sychis (Επτη-συχις) < IE *sukis “a girl”, cf. Cymbrian hogen “a girl” < IE *sukā.



Because of the comparatively weak aspiration of p’, t’, k’ in Thracian, relative to Greek (so Detchev), these consonants are often reproduced in Greek as π, τ, κ and Latin p, t, c respectively. E.g. we often encounter, alongside -phara in TN Breiero-phara, -para (< IE *(s)porā) as a second element of TNs: Agata-para, Bessa-para, Brento-para, Drusi-para etc. Alongside -αϕος in TNΟστ-αϕος -apa is also to be found, in TN Burd-apa and in the sobriquet of the nymphs in an inscription from the same place: Βουρδ-απ-ηναι; Along -kenthos, -centhus in PNs -centus occurs as well: Dias-centus, Disa-centus, Epta-centus,

-kentios in PN Μεστι-κεντιος; along -sychis (s. above) also -sykis in PN Ρυμη-συκις,

-sykos in PN Δεντυ-συκος, Επτε-συκος, -sucu in PN Dentu-sucu.


The above discussed changes in the three groups of IE consonants bh dh gh, b d g and

p t k, known in linguistics as a “sound shift”, have not all happened at the same time but in different periods; otherwise one should have expect to find only one group of consonants (ph th kh). The sequence of the mentioned phonetic processes can specified as follows: first, the IE Tenues p, t, k have shifted to p’, t’, k’ (ph, th, kh); after the process had been completed and the phonetic law had stopped to function, the second shift began (the change b, d, g > p, t, k). Only after the end of this shift




the third one took place – bh, dh, gh > b, d, g. The chronology can be illustrated schematically:






T          =         Tenues (p, t, k)

M         =         Mediae (b, d, g)

MA      =         Mediae aspiratae (bh, dh, gh)

TA        =         Tenues aspiratae (p’, t’, k’)




The IE palatal consonants k’, g’, g’h, which (according to V. Georgiev) have emerged from velar k, g, gh previous to front vowels and j, are represented in Thracian by the spirants s, þ / z, đ:

IE k’ > Thr. s (þ): Asamus < IE *ak’mo- “a stone, rocky”, OI aśman- “a stone; sky”, Av

asman- “the same”, Pelasgian asámynthos “a bath tub”, Lit akmuõ, eñs “a stone”; TN Pusinón, Lit pušýnas “a pinewood”, pušìs “a pine tree”, Gk péukē “a pine tree” < IE *peuk’-: *puk’-.

IE g’ > Thr. z (đ): HN Arzos < IE *Arg’os, cf. Gk argós “white”; -zenis in PN

Dia-zenis < IE *g’enis, cf. Gk Διογένης, old Lat genō, -ere “to create, to bear”, OI jánati “creates, gives birth”; Dentu- (with D- = đ) in PN Dentu-pēs, Dentu-brisa, Dentu-sucu, etc. < IE *g’ēnt-(i)- in Lat gēns, -ntis “a family, a tribe”, Av zantu- “a district”.

IE g’h > Thr. z (đ): zetráia “a pot” < IE *g’hetraā with the root IE *g’heu- “to pour”,

cf. Gk chýtra “a crock”; -zelmis “an offshoot, a plant” in PN Abru-zelmis, Dia-zelmis and others, cf. Lit želmuõ, -eñs “an offshoot, a plant” < IE *g’hel-.


The IE labiovelar sounds k, g , gh in Thracian have lost their labial element and have become pure velar sounds, k’ (kh), k, g:

IE k > Thr. k’ (kh), in Gk χ: HN Achelōos (Άχελος), Achelōn (Άχελν) < IE *əkel

“water”, cf. Lit HN Akẽl, Lat aqua “water” < *əka. Because of the weak aspiration of the Thracian k, it was written in Greek with χ: Ketri- in PN Ketri-poris (Κετρι-πορις) < IE * ketr- “four”, Gk tetra- “four”, Lit keturì “four”, Latvian četri, OBul četyre “four”.

IE g > Thr. k: TN Kellai, -cella in TN Syra-cella < IE *gelnā in OHG quella, German

Quelle “a spring, a source”; TN Kabylē < IE *Gəbhulā, En quab < IE *gəbh-; TN Kalíndoia (Kalindia), cf. OPru FN Galynde, Lit gãlas “an end, a boundary (of a field, forest, etc.)”, Latvian gals “area, region” < IE *golo-s.

IE gh > Thr. g: génton “meat” < IE *ghento- in OI hatá- “beaten, cut”; PN Gaidrēs,

< IE *g həidr- in Lit gaidrùs “light (colour), clear”, Gk phaidrόs “clear, bright, blazing”.


The IE spirant s remained unchanged in Thracian in every position, which could be seen from a lot of TNs: Sérmē, Síngos, Sartē, Ostaphos, Kypsela,




Antisara, Kurpisos, Asamus, Atlas, Iuras and others. IN the late Thracian this sound began to acquire voiceness in some positions (e.g. previous to a vowel or v, between vowels and after r it began to alter to z). Examples: zalmós “fur” from an older *salmos, cf. OPru salmis “a helm”, Lit šálmas “the same”, OI śárman- “a screen, a shield, a cover”, IE root *k’el- “to envelop, to hold, to rescue”; PN Ezbenis (or Ezvenis, in Gk Εζβενις), Hezbenus with a z, alongside Esbenus, Esbeneios (or Esveneios, Εσβενειος), a derivation of Thr. esva, ezva “a horse” < IE H1ek’u̯o-, cf. OI áśva-h “a horse”, Av aspa-, Lat equus; TN Pizos (Πιζος) from an older *Pisos, cf. Latvian pīsa “a groundless morass, where only small birches and spruces flourish” < IE *pīdsā, Gk pīsea (plural) “wet places, meadows”; Diza- in PN Diza-zelmis, Diza-pēs, Diza-por, Diza-tralis (with z) alongside Disa-centus (and Disa-kenthos) from an older Desa- < IE *dheso- “a deity”, cf. Lit dvāsas “a ghost, a spirit”; PN Kerza f., Cerzula m. from an older *Kersa, *Kersula, cf. Thr. PN Kersēs, Kersos, Lit kéršas “covered with white and black spots, piebald” < IE *k()ər(ə)so-s “black”.


The IE sonorant sounds r, l, m, n have not experienced a change in Thracian.


The voiceless obstruents, no matter of what origin, can become voiced in the neighbourship of a sonorant (assimilation):

  1. IE dr > Thr. tr > dr: PN Gaidrēs < IE *ghəidr- in Lit gaidrùs “light (colour), clear (weather)”, Gk phaidrόs “clear, bright, blazing”, from IE *ghēi, which must have resulted in *gaitr- in Thracian; TrN Odrysai < *Udrusai, from IE *udrā, that is with a dr instead of the expected tr < IE dr. On the other hand, we have zetráia “pot” < IE *g’heutraā, that is with a preserved tr (or thr).
  2. IE rd > Thr. rt > rd: Burd- in TN Burd-apa, Burd-enis alongside Burt- in TN Burt-icom, Burtu-dizos < IE *bhd(h) “a ford”; Bg HN Arda (of a Thracian origin), in addition FN campus Ardiensis (4. century) alongside HN Artánēs, Artiskós (Artakēs, Artakos), cf. OI árdati “flows”, Gk ardō “to water”; -thurdos in DN Zbel-thurdos (Zber-thurdos, Zber-turdus) < IE *(s)tdo-, cf. OHG sturzen, German stürzen “to fall”.
  3. IE br > Thr. pr > br: Skombros (a mountain) < IE *Skbros, cf. Lit kumbrỹs, kubris “a small hill, a mound”.
  4. IE nt > Thr. nt(h) > nd: TN Brendice alongside Brentice and Brento-para < IE *bhrento- “a deer”, cf. Messapian bréndon “a deer”. It is possible that



this change occurred only in late Thracian because there are examples of a preserved nth: bólinthos “a bison”, TN Périnthos, Zērynthos.

  1. IE tn > Thr. t(h)n > dn: midne “village”, as in in its Latvian parallel *mĩtne “residence, dwelling, accommodation” < older *mītinē or *mītenē < IE *meit(h)-: *mt(h)-.
  2. IE ng, nk > Thr. nk > ng: TN Linos, Lit TN Lìngės < IE *lg-; HN Angítēs, TN Anissós < IE *ank-, *ang- “to bend” in OI añcati “bends, arches”, anká-h “a bend, a crook”


The sound composition d might have resulted in Thracian in a fricative sound ž (from an older ), which was reproduced in Gk as ζ and in Lat as z, because those two languages do not have fricatives of the type of ž, š. An example of such a rendering of IE d  is the HN Tonzos (Τόνζος), which the Slavs of the upper reaches of the river have adopted under the form of OBul Tža (Bg Tъža); nowadays the river’s name in Bulgaria is Tundža, possibly a Turkicised form instead of *Tunža. Apparently the latter form was borrowed after the 9.-10. century, as in Slav (Bulgarian) the phonetic law of the open syllables (an effect of which was the emergence of a nasal from the combination of a vowel and a nasal consonant previous to a consonant) lost its impact. Compare to the Byzantine Τονζα (Theophanus, Chronicle, 8. century). The fricative -ž- in Bg Tъža can be explained by the assumption, that it was present in Thracian name as well, thus the HN in Thracian sounding something like *Tunža (or -as), almost the same way as it is pronounced today.


Compositions of the consonants m, l and the semivowel are represented in Thracian as follows:

IE m > Thr. ml: Zyml-yzdriēnos (a sobriquet of Asclepius) with ml < m , forms without

an epenthetic -l- are to be encountered too: Zym-ydrēnos, Zym-yzrēnos; compare to the form with a metathesis ml > lm: Zylm-ydrēnos, Zylm-yzdrēnos.

 IE l > Thr. ll: Kersullos (a sobriquet of Zeus) from an older *Kersulos, a derivation of a

TN *Kersula; Raimullus (a sobriquet of Apollo) from an older *Raimulos, a derivation of a TN *Raimula.


Changes in a series of consonant compositions are to be observed:

1.      IE d-t > t-t > Thr. s-t: HN Nestos < IE *Ned-tos < *ned in OI nádati “clinks, roars, hisses”, nad “a river”, OIr nes “a river”.




2.      IE kt > Thr. tt > t(th), reproduced in Gk as θ: the mountain name Athōs, Athōn < IE *Aktō-s, *Aktō(n), cf. Gk akt “a steep bank”; Bg Atija (a rocky peninsula, westward of Sozopol on the Black Sea) from a Thr. *Athija, attested in a Graecised form as Anthium, Antheia < IE *Aktiā, cf. Gk akt. This sound change must have occurred later than the previous one, at a time when the dissimilation of IE d-t (t-t) to st was already a fact.

3.      IE ts, ds > Thr. s (z): TN Calsus < IE *(s)kolt-so-s, cf. Latvian kàlst

(-stu, -tu) “become dry”, an extention to kàltêt “to dehydrate, to dry, to cure”, kaltuonis “dried tree”; TN Pizos (Πιζος) from an older *Pisos, cf. Latvian pīsa “a groundless morass, where only small birches and spruces flourish” < IE *pīdsā, Gk psea (plural) “wet places, meadows” < IE *pīdses, root: IE *pīd-, cf. Gk pîdax “a spring, a source”, OIc fit “a meadow”.

            4.   IE sr > Thr. str: HN Strmn (Στρμών) < IE *Srū-mōn, TN Strýmē, root:

IE *sreu- - srú - in Lit srti (srūvù, dial. srūnù) “to fill with water, to stream, to flow”; -stra in TN (originally a HN) Chalá-stra < IE *sroā, cf. Lit sràva “a flow”, Latvian stràva “to stream, a stream, a downpour”.

5.   IE ln > Thr. l(l): TN Kellai, -cella in the TN Syra-cella < IE *gelnā in OHG 

      quella, German Quelle “a spring, a source”.

6.   IE mt > Thr. nt: Tyntēnoi, the name of the inhabitants of a TN *Tynta,

cf. Lit tuñtas “a band, a troop, a mass” from an older *tumtas, a derivation of the verb tuméti “to coaulate”, tumùs “viscous”.


The results of the research on the phonetic system of the Thracian can illustrated in the following table:




a, ə, o a
e e
i i
u u
ā ā
ē, ē- ē (later > ī), -
ī ī
ō ō ( > ū > > ī)
ū, ū- ū ( > > ī), -







ai, əi ai (later ī)
ei ei ( > ī ?)
ou, əu au (av)
, , , il, ir, (im ?), in
ul (ol), ur (or), um (om), un (on)
bh, dh, gh b, d, g
b, d, g p, t, k
p, t, k ph, th, kh
k’ s, þ
g’, g’h z, d (= đ)
k k’ (kh)
g k
gh g
s- s- (later z-)
-s- s- (later z-)
r, l, m, n r, l, m, n



The combinatory sound shifts in Thracian are reflected in the following table:




sr str
dt (tt) st
r vr > br (r ?)
dh d
k’ sv, sb, zb (zv)
gh zv, z
kt th





ts (ds) s (z)
l ll
ln ll (l)
m ml
mt nt
d ž
dr tr > dr
tr thr, tr
rd rt, rd
br pr > br
nt nth, nd
tn t(h)n > dn
ng, nk nk > ng