Frederik Kortlandt #306 (2016)
Phrygian between Greek and Armenian
In an earlier study (1988), I argued that we can think of Thracian as an early
dialect of Proto-Armenian. Both languages shared the devoicing of the PIE glottalic
stops *b, *d, *g to *p, *t, *k (cf. Dečev 1952: 7-9, Duridanov 1976: 101, 1988: 61, Brixhe &
Panayotou 1997: 200), e.g. Thracian Utus, Ουτως, Greek ὕδωρ. The alleged voiced
aspirates appear as plain voiced stops, e.g. γεντον ‘meat’ < *gwhento-, βρ(ο)υτος ‘beer’ <
*bhrūto- (cf. Georgiev 1977: 103). The voiceless stops remained unchanged, as is
especially clear from Pulpudeva ‘Plovdiv’, Greek Φιλιππόπολις (cf. Duridanov 1986,
Hamp 1988), also -πορις ‘son’ (with variants), Latin puer. The Thracians evidently did
not perceive the aspiration of the Greeks. Like Balto-Slavic and Daco-Albanian, the
Thraco-Armenian complex belongs to the satǝm languages.
Following Lejeune (1979), I assumed that Phrygian did not share the ThracoArmenian consonant shift. This view is still maintained by Brixhe (1997: 173, 2008:
74), Matzinger (2005: 378) and de Lamberterie (2013: 27f.). Their key example is podas,
which they identify with Greek acc.pl. πόδας ‘feet’. However, Lubotsky has pointed
out (2004: 232) that this suggestion does not yield an acceptable interpretation of the
sentence. Other pieces of evidence that Lejeune adduces (1979: 223f.) are bagun ‘gift’ <
*bhagom and the name benagonos, which he compares with Greek -γονος. As Lubotsky
points out, the former word may rather mean ‘idol’ and be a borrowing from Iranian,
or may even be identified with Greek acc.sg. πῆχυν ‘arm’ < *bhāǵhum, whereas the
latter may correspond to Greek -φονος or even -χθονος. On the other hand, Lubotsky
adduces the following words in order to show that the consonant shift operated in
Phrygian:
– acc.sg. Τιαν < *diēm, gen.sg. Τιος < *diwos, Gr. Ζεύς, Ζῆν, Διός < Διϝός.
– acc.sg. βεκος ‘bread’ < *bhH1ǵos, Gr. φώγω ‘bake’.
– nom.pl. κενα ‘generations’ < *ǵenH1esH2, Gr. γένεα.
– acc.sg. κ̣ναικαν ‘wife’ < *gwneH2iḱ-, Gr. γυναικ-.
– imp.mid. 3sg. lakedo ‘?devote’ < *slH2gw-, Gr. λαβεῖν ‘be taken’.
– pf.pt.mid. τετικμενος ‘cursed’ < *dediḱmH1nos, Gr. δικάζω ‘judge’.
– preverb τι ‘?through’ < *dis-, Gr. δια-.
I conclude that the Thraco-Armenian consonant shift affected Phrygian as
well and was a dialectal Indo-European innovation cutting through the division
between centum and satǝm languages, which may have arisen around the same time.
In view of its geographical distribution, I suspect that the devoicing of the glottalic
stops can be attributed to the influence of a Proto-Anatolian substratum because the
Anatolians passed through future Thracian and Phrygian territory on their way from
the Indo-European homeland north of the Black Sea to Anatolia. Indo-European
voiced stops originated only after the Anatolians and the Tocharians left the IndoEuropean homeland (cf. Kortlandt 2012). Contrary to de Lamberterie’s objection
(2013: 28), there is no reason to suppose that PIE *p, *t, *k became aspirated in order
to keep them distinct from *b, *d, *g when the latter were devoiced because
glottalization was probably preserved, as it was in Armenian (cf. Kortlandt 2003:
20-25, 126-128). The rise of aspirated stops in Armenian was a much more recent
development (cf. Kortlandt 2003: 28-31).
2
The development of the laryngeals in Phrygian was evidently the same as in
Greek (cf. Ligorio & Lubockij 2013: 186). Armenian shared the triple representation of
the laryngeals in word-initial position, e.g. *H1- in inn ‘nine’ < *H1neun, im ‘mine’ <
*H1mos, both with *i- < *e- before a nasal, further eluzanem ‘extract’, erek ‘evening’,
Greek ἐλεύσομαι, ἔρεβος, *H2- in ayr ‘man’, astł ‘star’, arew ‘sun’, Greek ἀνήρ, ἀστήρ,
Sanskrit ravis, *H3- in ołb ‘lament’, orb ‘orphan’, orjik‘ ‘testicles’, oskr ‘bone’, Greek
ὀλοφύρομαι, ὀρφανός, ὄρχις, ὀστέον (cf. Kortlandt 2003: 54-56, 75-78). Initial *H2e- and
*H3e- are reflected as ha- and ho-, respectively, e.g. haw ‘bird’, harawunk‘ ‘field’, hot
‘odor’, hoviw ‘shepherd’, Latin avis, arāre, odor, ovis, Greek ἄρουρα, ὄϊς (cf. Kortlandt
2003: 39-44, 54-56). It cannot be established if Thracian and Phrygian shared the latter
development because initial h- would not be reflected in the orthography of these
languages. Without taking the ablaut grades into account, de Lamberterie rejects the
rise of h- from *H- in Armenian (2013: 2965) on the basis of wrong etymologies,
positing *H2o- for hoviw, hołm ‘wind’, ost ‘branch’ and oskr. He also rejects the
reconstruction of *H3- for atamn ‘tooth’ and anun ‘name’, where pretonic *o- yielded
a- (cf. Meillet 1894: 153-157, Pedersen 1900: 100, Kortlandt 2003: 32), in spite of the
unambiguous evidence of Greek ὀδών, ὀδούς, ὄνομα, ὄνυμα, and especially νωδός
‘toothless’, νώνυμος ‘nameless’, which represent an archaic formation beside more
recent ἀνόδων, ἀνώνυμος. The word for ‘tooth’ must not be connected with Gr. ἔδομαι
‘will eat’ but with the root *H3ed- ‘bite, sting’ attested in Lith. úodas ‘gnat’, Gr. ὀδύνη
‘pain’, νώδυνος ‘painless’. The word for ‘name’ is also attested in Old Phrygian
onoman. Contrary to de Lamberterie’s statement (2013: 3374), I do not think that the
initial vowel of Albanian emër ‘name’ is a “voyelle prothétique”; as in Prussian
emmens (cf. Kortlandt 2009: 255) and Old Irish ainm, it reflects the vocalization of the
syllabic nasal. The word is a derivative of the root *H3neH3- of Gr. ὄνομαι ‘scorn’.
The inflection of nominal o-stems in Phrygian is virtually identical to that in
Greek and Armenian (cf. Brixhe 2008: 76, Ligorio & Lubockij 2013: 188, Kortlandt
2003: 45-51):
Greek
Phrygian
Armenian
-ος
-ον
-ου, -οιο
-ῳ
-οι
-os, -ος
-un, -ουν
-ovo, -ου
-oi, -ου
= dat.
= acc.
*-u
*-oyo
*-oye
*-e
-οι
-ους, -ονς
-ων
-οις
-οισι
-oi, -οι
?-oys
-ουν
?-oys, -ως
= dat.
*-uh, *-e
*-us
*-oc‘u
= gen.
*-su
SINGULAR
nom.
acc.
gen.
dat.
loc.
PLURAL
nom.
acc.
gen.
dat.
loc.
The singular endings represent nom. *-os, acc. *-om, gen. pronominal *-oso (Greek
and Phrygian, cf. *-eso in Germanic) and *-osio (Greek and Armenian, also in IndoIranian), dat. *-ōi (Arm. -oy- from the genitive), loc. *-oi, the plural endings nom.
pronominal *-oi (Arm. *-uh < *-ōs), acc. *-ons, gen. *-oom (cf. Kortlandt 2014b,
3
Arm. -c‘- < *-sk-), dat. *-ōis from the instrumental, loc. *-(oi)su (Greek -ι from the
singular).
As far as we can reconstruct the Phrygian verbal system, it is largely identical
to that of Greek and Armenian. It appears that there are reflexes of the present,
imperfect, aorist, perfect, future, subjunctive, optative, imperative, active and
transitive middle (cf. Sowa 2007, Ligorio & Lubockij 2013: 191f., Kortlandt 2003: 34-38,
107-116). Phrygian αδδακετ “corresponds to the aor. παρεθέμην of the Greek text,
which makes the preterite interpretation of this form very likely” (Lubotsky 1997: 127),
resulting in a meaning ‘has put’. This looks like a perfect derived from a k-aorist *dhēk,
like Latin fēcit (cf. Kortlandt 2016). This interpretation is supported by the 3rd pl.
perfect form δακαρεν, corresponding to Latin fēcērunt (cf. Ligorio & Lubockij 2013:
192). The 3rd pl. form *stār (Vedic ásthur) may be found in Phrygian ενσταρνα̣. Since
αδδακετορ and αββερετορ appear in the same contexts as αδδακετ and αββερετ, where
we also find αββερετοι, Ligorio & Lubockij assume that all these forms belong to the
middle paradigm (2013: 191). I find this highly unlikely. As Lubotsky points out
himself (1997: 12726), the most common forms are αδδακετ and αββερετορ while
αδδακετορ and αββερετ occur only in a few inscriptions. It is therefore probable that
the original formulae used either active αδδακετ or middle αββερετορ and later the
forms became mixed up. This suggests that αδδακετ was a perfect form ‘has put’
whereas αββερετορ and αββερετοι were middle forms of the imperfect ‘brought’ and
the present ‘brings’, respectively. While the present ending -τοι is the same as in
Greek, the imperfect form must be compared with the Armenian imperfect berēr <
*-etor, beriwr < *-etro (cf. Meillet 1936: 127, Kortlandt 2003: 37), like Latin -tur < *-toro
beside Oscan -ter < *-tro (Kortlandt 2007: 156). It appears that the original transitive
middle endings in *-(n)tr(o) were lost in Greek and Indo-Iranian, where the
intransitive endings 3rd sg. *-o and 3rd pl. *-ro were preserved in Vedic śáye, śére
‘lie(s)’, with added *-i.
Like Greek, Phrygian developed a transitive s-aorist, which is attested in 3rd sg.
εδαες ‘put’ < *dhēs-es and εσταες < ‘erected’ < *stās-es (cf. Lubotsky 1988: 17f.). I do not
think that the ending -es can be derived from another sigmatic aorist (thus Ligorio &
Lubockij 2013: 191, similarly Gorbachov 2005). More probably, it represents an enclitic
subject pronoun *es, like Tocharian A -ṣ, B -ṃ, Old Irish olsé ‘said he’ (cf. Kortlandt
2007: 114). The enclitic pronoun was also added to the root aorists ενεπαρκες
‘engraved’ < *pērk-es and εκανες ‘dug’ < *kēn-es, with a long root vowel from
monosyllabic lengthening (cf. Kortlandt 2015). Since edatoy < *dhētoi and estatoi <
*stātoi are evidently the middle counterparts of edaes and εσταες (Ligorio & Lubockij
2013: 191), they are original root aorists with a long root vowel and a generalized
primary ending. This is reminiscent of Tocharian, where the intransitive middle
endings spread to the transitive middle aorist (cf. Kortlandt 2014a: 84). As a result,
Phrygian has 3rd sg. -toi in the aorist versus -tor in the imperfect while the secondary
ending *-to was evidently lost. This is again reminiscent of Armenian, which has 3rd
sg. -r in the imperfect but not in the aorist. The form daΨet /dakset/ is strongly
reminiscent of the Latin future faxō ‘will do’ while the subjunctives egeseti and εγεσιτ
‘will hold’ < *seǵhes- beside με-βερετ < *bher- and με-τοτοσσειτι ‘will take away’ <
*-dH3(e)s- and ιστεικετ ‘will appear’ < *-deiḱ- suggest that there was no semantic
difference between primary and secondary endings here (as in the Indo-Iranian
subjunctive, cf. Beekes 1981). The zero grade of the root in τοτοσσειτι and the suffix
*-es- in εγεσιτ show that these forms must be compared with Oscan didest ‘will give’
4
and Umbrian ferest ‘will carry’ and represent original s-presents (cf. Pedersen 1921:
23), similarly Old Irish fo-lil ‘will support’ < *lilugst, subj. -bé ‘may be’ < *bhH3uest,
Lithuanian bùs ‘will be’, Tocharian tās- ‘put’ < *dhH1(e)s- (cf. Kortlandt 2007: 65-74).
Thus, I conclude that Greek, Phrygian and Thraco-Armenian reflect a single
Indo-European dialect area that was divided by two major isoglosses, viz. the
devoicing of the glottalic stops which separated Phrygian from Greek and the
satǝmization of the palatovelars which separated it from Thraco-Armenian. Phrygian
provides in several respects the missing link between Greek and Armenian. In
particular, the paradigms of the middle voice appear to have been more extensive than
what we find in the separate languages. The 3rd sg. endings are the following (cf.
Ligorio & Lubockij 2013: 191, Kortlandt 2003: 37f., 2007: 145f.):
present
imperfect
aorist
Greek
-ται, -τοι
-το
-το
Phrygian
-τοι
-τορ
-toy
Armenian
-y
-yr, -wr
-w
It appears that the original endings are best preserved in Armenian and that the
ending *-tor was lost in Greek (but not in Latin -tur and Old Irish -thar). The ablaut in
the Phrygian forms egeseti and με-τοτοσσειτι (Ligorio & Lubockij 2013: 192) reveals
their origin as athematic s-futures while the form daΨet < *dhēkset is clearly an aorist
subjunctive, like Armenian dic‘ē ‘put’ and tac‘ ‘gave’ (cf. Kortlandt 2003: 107-109). The
archaic character of the Phrygian language is corroborated by the Indo-Iranian and
Italo-Celtic evidence.
References
Beekes, Robert S.P. 1981. The subjunctive endings of Indo-Iranian. Indo-Iranian
Journal 23/1, 21-27.
Brixhe, Claude. 1997. Le phrygien. Langues indo-européennes (Paris: CNRS), 167-180.
Brixhe, Claude. 2008. Phrygian. The ancient languages of Asia Minor (Cambridge:
UP), 69-80.
Brixhe, Claude & Panayotou, Anna. 1997. Le thrace. Langues indo-européennes (Paris:
CNRS), 181-205.
Dečev, Dimitъr. 1952. Harakteristika na trakijskija ezik (Sofija: BAN).
Duridanov, Ivan. 1976. Ezikъt na trakite (Sofija: Nauka i izkustvo).
Duridanov, Ivan. 1986. Pulpudeva, Plovdiv, Plovdin. Balkansko Ezikoznanie 29/4,
25-34.
Duridanov, Ivan. 1988. Zur Frage der Lautverschiebung im Thrakischen. Balkansko
Ezikoznanie 31, 57-64.
Georgiev, Vladimir I. 1977. Trakite i tehnijat ezik (Sofija: BAN).
Gorbachov, Yaroslav. 2005. The origin of the Phrygian aorist of the type edaes.
Proceedings of the 16th annual UCLA Indo-European conference [=JIES monograph
no. 50], 191-217.
Hamp, Eric P. 1988. Плъп’дивъ > Плъвдивъ. Balkansko Ezikoznanie 31, 75-76.
Kortlandt, Frederik. 1988. The Thraco-Armenian consonant shift. Balkansko
Ezikoznanie 31, 71-74.
5
Kortlandt, Frederik. 2003. Armeniaca: Comparative notes (Ann Arbor: Caravan
Books).
Kortlandt, Frederik. 2007. Italo-Celtic origins and prehistoric development of the Irish
language (Amsterdam: Rodopi).
Kortlandt, Frederik. 2009. Baltica & Balto-Slavica (Amsterdam: Rodopi).
Kortlandt, Frederik. 2012. Proto-Indo-European glottalic stops: The evidence revisited.
<www.kortlandt.nl/publications/art282e.pdf>
Kortlandt, Frederik. 2014a. The Tocharian personal endings. Tocharian and IndoEuropean Studies 15, 79-86.
Kortlandt, Frederik. 2014b. Reconstructing Balto-Slavic and Indo-European. Baltistica
49/1, 5-13.
Kortlandt, Frederik. 2015. Sigmatic and asigmatic long vowel preterit forms. Journal of
Indo-European Studies 43, 236-242.
Kortlandt, Frederik. 2016. The Indo-European k-aorist.
<www.kortlandt.nl/publications/art302e.pdf>
de Lamberterie, Charles. 2013. Grec, phrygien, arménien: des anciens aux modernes.
Journal des Savants 2013/1, 3-69.
Lejeune, Michel. 1979. Regards sur les sonores i.e. en vieux phrygien. Florilegium
Anatolicum [Fs. Laroche] (Paris: Boccard), 219-224.
Ligorio, Orsat & Lubockij, Aleksandr. 2013. Frigijskij jazyk. Jazyki Mira: Reliktovye
indoevropejskie jazyki Perednej i Central’noj Azii (Moskva: Academia), 180-195.
Lubotsky, Alexander. 1988. The Old Phrygian Areyastis-inscription. Kadmos 27/1,
9-26.
Lubotsky, Alexander. 1997. New Phrygian inscription no. 48: Palaeographic and
linguistic comments. Frigi e Frigio: Atti del 1° simposio internazionale (Roma:
Consiglio Nazionale delle Ricerche), 115-130.
Lubotsky, Alexander. 2004. The Phrygian Zeus and the problem of the
“Lautverschiebung”. Historische Sprachforschung 117, 229-237.
Matzinger, Joachim. 2005. Phrygisch und Armenisch. Sprachkontakt und
Sprachwandel (Wiesbaden: Reichert), 375-394.
Meillet, Antoine. 1894. Notes arméniennes. Mémoires de la Société de Linguistique de
Paris 8, 153-165.
Meillet, Antoine. 1936. Esquisse d’une grammaire comparée de l’arménien classique
(Vienne: Mékhitharistes).
Pedersen, Holger. 1900. Wieviel laute gab es im Indogermanischen? Zeitschrift für
vergleichende Sprachforschung 36, 74-110.
Pedersen, Holger. 1921. Les formes sigmatiques du verbe latin et le problème du futur
indo-européen (København: Høst & Søn).
Sowa, Wojciech. 2007. Anmerkungen zum Verbalsystem des Phrygischen.
Indogermanische Forschungen 112, 69-95.
Summary
We can think of Thracian as an early dialect of Proto-Armenian. Both
languages shared the devoicing of the PIE glottalic stops *b, *d, *g to *p, *t, *k. The
alleged voiced aspirates appear as plain voiced stops. The voiceless stops remained
unchanged. The Thracians evidently did not perceive the aspiration of the Greeks.
The Thraco-Armenian consonant shift affected Phrygian as well and was a dialectal
6
Indo-European innovation cutting through the division between centum and satǝm
languages. In view of its geographical distribution, I suspect that the devoicing of the
glottalic stops can be attributed to the influence of a Proto-Anatolian substratum
because the Anatolians passed through future Thracian and Phrygian territory on
their way from the Indo-European homeland north of the Black Sea to Anatolia. The
development of the laryngeals in Phrygian was evidently the same as in Greek.
The inflection of nominal o-stems in Phrygian is virtually identical to that in
Greek and Armenian. As far as we can reconstruct the Phrygian verbal system, it is
largely identical to that of Greek and Armenian. It appears that there are reflexes of
the present, imperfect, aorist, perfect, future, subjunctive, optative, imperative, active
and transitive middle. I conclude that Greek, Phrygian and Thraco-Armenian reflect a
single Indo-European dialect area that was divided by two major isoglosses, viz. the
devoicing of the glottalic stops which separated Phrygian from Greek and the
satǝmization of the palatovelars which separated it from Thraco-Armenian. Phrygian
provides in several respects the missing link between Greek and Armenian. In
particular, the paradigms of the middle voice appear to have been more extensive than
what we find in the separate languages. The archaic character of the Phrygian
language is corroborated by the Indo-Iranian and Italo-Celtic evidence.