The fun part of writing my cave-Thracian story is of course getting the chance to reconstruct the Thracian language. I poured through word lists, looked up cognates, shook, stirred, and here’s what crystallized.
Proto-Indo European > Thracian (examples) Notes
*bʰ > b (*bʰébʰrus > berbrus “beaver”, *bʰuǵ > Byzas name) p (*bʰel > pala, *bʰer- >por,per,pes,puis “son”, *bʰerHǵós > berza(s) “pine”) There might have been a dialect difference with b in the north (Dacian) and p in the south (Thracian).
*d > d (wódr̥ > Urda place, *déh₃-tis > Datos place) z (mendyos > Mezēnai “horsemen”, *deywós > ziu “Zeus”) st (*skeyd- ktístai “monks”) I think there was palatalization here, with dy/de/di becoming /dz/, which Greek scribes spelled as best they could. The other possibility is “ziu” was borrowed from Greek.
*dʰ > d (*dʰéh₁s > desas “god”, *dʰewbʰ- > Dapha place, *h₁rewdʰ- > Rhodope “place”)
*ē, > o (gʷḗn-h₂ > goni “woman”) e (*ǵʰēlo > zelâs “gold”, *h₃rḗǵs > rezas “king”) maybe a previous labialized consonant conditions the shift to o?
*e > a (*(s)kelh₁- kalsas “dry”, *bʰel > pala “mud”, *déh₃-tis > Datos place, *dʰewbʰ- > Dapha place, *h₂ekʷeh₂ > achel “water”) e (*kʷetwóres > ketri “four”, *dʰéh₁s > desas “god”, *ǵenh₁- > zenis “child”, *ǵʰwer- > zeri “beast”) ē (*h₁eǵ(oH) >ēsko “I”, mendyos > Mezēnai “horsemen”) o (*h₂ep- > ópē “river”) wouldn’t it great if Thracian had an /æ/ sound?? But a more likely possibility is the peculiarly Balkan “accented shwa” Bulgarian /ɤ~ɐ/, Albanian /ə/, Romanian /ə/.
*er > er (*bʰerHǵós > berza(s) “hill” *gʷʰer-mós Germe- “warm”) or, er, e, ui (*br- >por,per,pes,puis “son”) There may be some Persian influence on “son.”
*ew > o (*h₁rewdʰ- > Rhodope place) u (*kewH- > kuti “box”)
*ey > i (*skeyd- ktístai “monks”)
*ǵ > sk ((*h₁eǵ(oH) >ēsko “I”) z (*bʰerHǵós > berza(s) “hill”, *bʰuǵ > Byzas name, *ǵenh₁- > zenis “child”, *h₃rḗǵs > rezas “king”) This one is especially interesting because ǵ > z it’s an isogloss for “Eastern” Indo-European languages (the satem languages such as Balto-Slavic and Indo-Iranian). The word “ēsko,” though is from one one of the three known Thracian inscriptions, and was probably written by a Thracian, rather than Greek, scribe. I wonder, then, if <z> might not have been how Greeks tried to spell /ʃ/.
*ǵʰ > s ((*dʰé)ǵʰōm > semele “Gaia”, *ǵʰl̥h₃tóm > saldas “golden”) z (*ǵʰelh₃- zelta “gold”, *ǵʰēlo > zelâs “pink”) see above. I think it was /ʃ/.
*ǵʰw > z (*ǵʰwer- > zeri “beast”) Perhaps ǵʰwer > ʃver > ʒer?
*gʷʰ > g (*gʷʰer-mós Germe-“warm”) Fun to see those Germanic cognates, isn’t it?
*k > k (*kr̥snós > kersas “black”, *(s)kelh₁- kalsas “dry”, *kewH- > kuti “box”) *kr̥snós is interesting, because it only shows up in Eastern Indo-European languages.
*ḱ > s (*h₁éḱwos > esvas/aspios “horse”) see *ǵ. I think it was /ʃ/, and the *w became /v/, which gave Greek scribes trouble.
*kʷ > k (*kʷetwóres > ketri “four”) or ch (*h₂ekʷeh₂ > achel “water”) maybe /k/ became /x/ between vowels?
*ḱw > sv (*h₁éḱwos > esvas “horse”) zb (*ḱweytos > Zburulus name) see why I think /v/ was sometimes written <b>?
*l> l (*(s)kelh₁- > kalsas “dry”, *bʰel > pala “mud”, *ǵʰēlo > zelâs “wine”) a nice, stable sound.
*l̥ > al ( *ǵʰl̥h₃tóm > saldas “pink”)
*m > m ((*dʰé)ǵʰōm > semele “Gaia”, *gʷʰer-mós Germe- “warm”, mendyos > mezēnai “horsemen”, *srowmos > Strymon, Struma rivers) another nice stable sound.
*n > n (*ǵenh₁- > zenis “child”) 0 (*mendyos > Mezēnai “horseman”) I think this is palatalization and assimilation: *nde, ndi > ne, ni.
*ō > e ((*dʰé)ǵʰōm > semele “Gaia”)
*o > u (wódr̥ > udrēnas “aquatic”) â (*ǵʰēlo > zelâs/zelai “wine”) You can’t trust sounds at the ends of words.
*ow > y/u ( *srowmos > Strymon, Struma rivers) So <y> was probably written early enough that it still meant /u/.
*p > p (*h₂ep- > ópē “river”)
*r > r/rh (*h₁rewdʰ- > Rhodope mountain, *h₃rḗǵs > rezas “king”, *bʰébʰrus > berbrus “beaver”) I don’t think there was any functional difference between <r> and <rh>. Note the rhotization of previous accented vowel in beRbrus.
*r̥ > er (*kr̥snós > kersas “black”) rē (wódr̥ > udrēnas “aquatic”) ri (kr̥snós > Krisos place). The vowel wandering around the r makes me think this r stayed syllabic until very late. Perhaps the front vowel associated with it was /ɨ/ like in Romanian?
*s > s (*kr̥snós > Krisos place, *dʰéh₁s > desas “god”)
*sk > kt (*skeyd- ktístai “monks”) since this word was written by a Greek-speaker, I wonder if it might not actually have been /ʃ/. I’m less sure about this one than the other /ʃ/s, though.
*sr > str (*srowmos > Strymon, Struma rivers) For some reason, this sound shift makes me very happy.
*t > t (*kʷetwóres > ketri “four”, *déh₃-tis > Datos place)
*tw > t (*kʷetwóres > ketri four )
*u > u/y (*bʰuǵ >Buzas/Byzas/Byzantas name) meaning the settlement of Byzos that gave the Byzantine Empire its name might originally have been Thracian, meaning “young goat.” That’s cognate to English “buck,” by the way.
*w, *u̯ > 0 (wódr̥ > udrēnas “aquatic”) ph (*dʰewbʰ- > Dapha place) I think this means *wV> V and *wC> /v/, with *wb assimilating to just /v/.
Conclusions, it looks like Thracian was a Satem Language, but very close to the split between Satem and Kentum. My feeling is you have (Albanian+Illyrian) on one side, ((Greek+Phrygian)+(Armenian)) on the other, and between them ((Thracian+(Balto-Slavic))+(Indo-Iranian)). So when I make up Thracian words from PIE roots, I’ll try to use roots that survived in at least Greek, Albanian, and Lithuanian. But I’m sure that Thracian borrowed some words from Greek. It might have borrowed some words from some ancient Balto-Slavic or Persian languages, too.
I’m also assuming that Thracian was the substrate language that gives Albanian, Romanian, and Bulgarian their similar characters.
Now, with these sound shifts nailed down, I can take a stab at deriving some new Thracian words! But first, I’ll look back at those three Thracian inscriptions and see if I can translate them. Questions, comments, corrections, and encouragement welcome 🙂