Caucasian interface

A patchwork of languages

[see my presentation on The Semipermeable Barrier, Caucasus (Jena 2019)]

Four distinct and commonly acknowledged language families belong to the Caucasus and Transcaucasus region, one of which went extinct BCE:

  • Northeast Caucasian (or Nakh-Daghestanian)
  • Northwest Caucasian (or Abkhaz-Circassian)
  • Hurro-Urartian (extinct)
  • Kartvelian (or South Caucasian)

A common feature of the three contemporary language families is glottalization of consonants (Hewitt 2004: 32), which plays a central role in the Glottalic Theory for PIE (Gamkrelidze & Ivanov 1995: 5ff.). In addition, they all exhibit ergative case marking, which constitutes another typological argument for early PIE contact with the area (e.g. Beekes 1985: 172ff.). Within the fields of both linguistics and archaeology, the inclusion of a small handful of indirectly or scantily attested cultural languages into the Caucasian grouping has seemed straightforward (however difficult, if not impossible, to prove); these references are given below and may shed much-needed historical light on an otherwise complicated task. The central geographic location of the Caucasus makes it interesting in the development of PIE as the historic distribution of IE languages envelops the region completely, albeit with the later enrichment of Turkic and Mongolian. Additionally, and more pertinently for a discussion of PIE proper, the most prominent homeland hypotheses place the point of dispersal for the IE languages either north (the Pontic steppes) or south (Transcaucasia or the Anatolian highlands) of the region (§ 3). An important question also applies to the permeability of the region, the shortest route between Anatolia and the Pontic Steppe, yet the embedded linguistic patchwork seems to be old and inhospitable to foreign permanence, as suggested by the unique status of all three Caucasian families that seems mirrored genetically (Yunusbayev et al. 2011). The Armenian presence in the southern parts seems more a refuge from an earlier vast territory further west than its original point of dispersal (cf. Mallory 1989: 34), which is corroborated by its eclipse of the Urartian empire that appears to have had strong ties to the Caucasus (§ 2.1.3). Other foreign elements include Ossetic (an Iranian language) in the north and a few Turkic languages that all can be traced historically to Steppe migrations emanating from Central Asia (cf. Nichols 1997: 132f.). The trichotomy of the modern Caucasian languages is uncontroversial, while further attempts to unite some or all the families are complicated by their intricate and ancient relationship with mutual influence that obscure regular sound correspondences (cf. Deeters 1963: 41). The most pervasive and obvious unification is that of NW and NE Caucasian in a common North Caucasian (§ 2.1.4). The cultures historically confined to the Caucasus region are widely believed to have played a bigger role in the Fertile Crescent around the time of the incipient stages of PIE further north (cf. Roberts 1998: 72ff.), the last remnant of which may be linguistically corroborated by the ultimate disappearance of the Hurro-Urartian languages in the first millennium BCE.

Suggested literature:
Diakonoff & Starostin (1986) Hurro-Urartian as an Eastern Caucasian Language
Dumézil, Georges (1933) Grammaire Comparée des Langues Caucasienne du Nord
Hewitt, George (2004) Introduction to the Study of the Languages of the Caucasus
Nikolayev & Starostin (1994) North Caucasian Etymological Dictionary

Northeast Caucasian Languages (NE Caucasian) or Nakh-Daghestanian
The family consists of eight more or less independent branches, of which only Nakh and Lezgian have sufficiently varied internal dialectal variation to warrant linguistic sub-branching:

1. Dargi
2. Lak
3. Nakh (Bats, Ingush, and Chechen)
4. Khinalug
5. Lezgian
5a. Archi
5b. Samur (Tsakhur, Rutul, Kryts, Budukh, Udi, Lezgi, Agul, and Tabasaran)
6. Avar
7. Andian
8. Tsezian

Sometimes geographically also referred to simply as East Caucasian, the alternative term Nakh-Daghestanian is actually obsolete, but not very dissimilarly from how Indo-European fails to account for all of its internal constituents (cf. Armenian, Anatolian, Iranian, and Tocharian that are neither Indian nor European). The NE Caucasian languages are also occasionally divided into a Central Caucasian (Nakh) and a Northeast Caucasian proper language category, but these dichotomies are disputed and of secondary importance.

Geography and history
The present distribution is, as indicated by their epithet, centered in the northeast corner of the Caucasus on the shores of the Caspian Sea with the Nakh languages further inland. The earliest attestations of NE Caucasian languages are included in word lists assembled in the late 18th century (Hewitt 2004: 2)[Add. fragments of ancient Udi (or Albanian) has been found on Sinai dating from the middle of the first millennium AD (Alexidze 2007)]. There is no indication of significant migrations by NE Caucasian speakers either to or from its current location [Add. see Greppin for an eastern homeland (2008b: 135-36)], although Russian and Turkic encroachment continues to decimate the number of speakers, and archaeological continuation seems to indicate stability at least since the Bronze Age (Nichols 1997: 125). The internal diversification probably took place roughly contemporarily with incipient stages of the proto-Indo-Iranian linguistic community was present in its vicinity as is indicated by several borrowings (Dolgopolsky 1987: 18), while Nichols uses lexicographic data to pinpoint the spread, which she perceives to have happened after the domestication of sheep, cattle, and goats, but before horses became a stable word (1998: 225). Diakonoff & Starostin suggest that the Hurro-Urartian languages (§ 2.1.3) are to be considered a branch on the NE Caucasian language tree (1986), but the difficulties with which the comparison could be made, they argued, is like adducing genetic kinship between transcribed modern French with Sanskrit (1986: 98). Access to this potential treasure trove of ancient forms is thus hampered by the extremely complicated task of bridging of almost three millennia of silence and this suggested relation will consequently play a minor role in the loan etymologies.

Linguistic affinities with PIE
Relations with North East Caucasian are only inferred from the unspecified (North) Caucasian adstrate theories, most notably S. Starostin (2009), see below (§ 3.2.3).

Suggested literature:
Deeters & Solta (1963) Armenisch und kaukasische Sprachen
Diakonoff & Starostin (1986) Hurro-Urartian as an eastern Caucasian language
Hewitt, George (2004) Introduction to the Study of the Languages of the Caucasus

Northwest Caucasian Languages (NWC) or Abkhaz-Circassian
The family is rather small, consisting of only four extant languages (with dialects), divided into two subgroups, that teeter on a recently extinct intermediary.

  • Circassian branch:
    • Adyghe
    • Kabardian
  • Ubykh (extinct since 1992)
  • Abkhaz branch:
    • Abaza
    • Abkhaz

Geography and history
The NW Caucasian languages, true to their name, occupy the northwestern perimeter of the Caucasus mountains on the shores of the Black Sea and further inland. Like their eastern neighbors, their territory is continually dwindling under the cultural influence of Russian, despite the fact that Abkhaz remains a strong ethnic language in the breakaway republic of Abkhazia, whose de facto sovereignty from Georgia is militarily secured from Moscow. Word lists only began emanating from the area in the 17th century in the notebooks of travelers (Hewitt 2004: 1); the added fact that the NW Caucasian languages suffer from very late attestation and close geographic proximity severely inhibit internal reconstruction. Further, the root structure is mostly CV, allowing very little material for radical alignment (cf. Chirikba forthc.: 12f., Dolgopolsky 1989: 16). Consequently, the comparison with the reconstructed PIE that harks back at least 5.000 years is extremely difficult. Nichols considers NW Caucasian to have occupied roughly its current distribution for a considerable amount of time (1997: 125). Connections with Hattic (§ 2.5) have been proposed (Diakonoff & Starostin 1986: 2,97 and Dolgopolsky 1989: 14), possibly through North Caucasian proper (Kassian 2010: 320), but remain controversial (cf. Goedegebuure 2010: 949).

Linguistic affinities with PIE
Specifying the general trend of assigning Caucasian influence to some defining features of PIE, Bomhard suggests that PIE was in intense contacts with NW Caucasian in a period leading up to the split of the Anatolian branch (2015: 11), and has produced a list of 150 lexical items to substantiate the claim. Kortlandt similarly favors NW Caucasian adstrate as the defining factor in the genesis of PIE (2010: 6). The suggested North Caucasian family (§ 2.1.4) would demand clearer stratification of these proposals.

Common heritage
Attempts have been made to demonstrate that PIE and the NW Caucasian languages share common heritage from a unique mutual prehistoric stage. Such a genetic connection was first suggested by Friedrich on the basis of typological features (1964: 208-9), and is now primarily championed by Colarusso, who dubbed the proposed super-family ‘Proto-Pontic’ (1981), but without wider acceptance. It may be noted that the Proto-Pontic hypothesis is at odds with the proposed North Caucasian language family (§ 2.1.4).

Suggested Literature:
Chirikba, Viacheslav (1996) Common Northwest Caucasian
Deeters & Solta (1963) Armenisch und kaukasische Sprachen
Kuipers, A.H. (1975) A Dictionary of Proto-Circassian Roots

Hurro-Urartian (HU)
The Hurro-Urartian language family consists of only two languages, Hurrian and Urartian, whose genetic affinity is well-established (cf. Wilhelm 2008a: 81 and Fournet & Bomhard 2010: 2ff.). Hurrian is the better attested language of the two constituents and was spoken in the northern Fertile Crescent (modern day northern Syria and Iraq and southern Turkey) from the latter parts of the 3rd millennium BCE through to the turn of the first millennium BCE. It was the language of the Mitanni empire known to Indo-Europeanists for its Indo-
Iranian superstrate and owes much of its attestation to the Hittite archives. Urartian was spoken from the Caucasian corner of the Caspian Sea to the Euphrates. Attestations run from c. 800 BCE to no later than 600 BCE and thus seem to continue in the vacuum left by the Hurrians, but not as a direct continuation hereof; the branches may have split around 2.000 BCE (Wilhelm 2008b: 105). The Urartian empire was likely displaced by the Armenians (Ajello 1998: 197ff., Solta 1963: 80ff.), which is evidenced by substrate words such as Armenian xnjor ‘apple’ from Hurrian ḫinzuri (Fortson 2010: 382 and, more elaborately, Greppin 2008). The archaeological record appears to link the civilization of the Hurrians and Urartu with the Caucasus, where it is first detectable in the fifth millennium BCE (Burney 1990), a notion corroborated by the Middle Eastern genetic component (Yunusbayev et al. 2011). Further connections with NE Caucasian have been proposed (§ 2.1.1 and § 2.1.4).

Suggested literature:
Diakonoff, Igor M. (1971) Hurrisch und Urartäisch
Diakonoff & Starostin (1986) Hurro-Urartian as an Eastern Caucasian Language
Fournet & Bomhard (2010) The Indo-European Elements in Hurrian
Speiser, E. A. (1941) Introduction to Hurrian

Excursus: North Caucasian
As mentioned above, various groupings of the Caucasian languages have been proposed, but the most pervasive theory remains North Caucasian that reconstructs the common ancestor of NW and NE Caucasian. Its strongest argument is by far the 1994 North Caucasian Etymological Dictionary by Nikolayev & Starostin, while profound skepticism can be found with Nichols (2003: 208). The question of their possible aboriginal unity gains saliency in the ability to reconstruct phonemes, morphemes, and, consequently, lexemes in greater depths of time and with greater (although still comparatively limited) geographic dispersal to check for secondary regional phenomena. If both families are significantly younger than PIE, it makes sense to consider the possibility of ancestral forms to inform the pertinent chronological stage in the region. Nikolayev & Starostin estimate Proto-North Caucasian to be at least 5.000 years old (1994: 60), which means that proto-forms existed no later than 3.000 BCE, and consequently comparable to the dates of PIE; at any rate, and per definition, significantly older than the established individual North Caucasian language families on their own. If the Hurro-Urartian connection with NE Caucasian holds true, a window opens into the more distant aspects of the makeup of the languages of the region that very likely were in contact with PIE. Additionally, the ergative HU *-s(ə) ending, which Diakonoff & Starostin compare to the NE Caucasian instrumental suffix *-s (1986: 75), and ultimately may prove to be the origin of the PIE nominative marker *-s. A truly significant informant of Caucasian language matters at the time of PIE may thus potentially be synthesized from all three proposed constituents, i.e. NW Caucasian, HU, and NE Caucasian (cf. Dolgopolsky 1989: 16). Ultimate primordial Sino-Caucasian affinities (cf. Kassian 2010: 321ff.) are way beyond the scope of the present thesis.

Linguistic affinities with PIE

Since Uhlenbeck (1901), a school of thought has explained the identity of the IE languages as an Indo-Uralic language that was marked by considerable influence of a Caucasian language, imbuing the incipient stages of PIE with traits of ergativity, additional places of articulation, and, seemingly later, gender. The dearth of comparative evidence for the linguistic communities in the Northern Caucasus (§ 2.1.1, § 2.1.2) postponed the more precise associations that have materialized in the most recent thirty years (e.g. S. Starostin 2009: 126f.). This analysis may be corroborated by Nichols’ typological dichotomy of northern and southern Eurasian languages (cf. § 1.5.2), where PIE exclusively falls within in the northern group with the marked exception of two particular features, viz. gender assignment and a high number of consonant articulation places (2007: 203f.). Although Nichols dismisses ergativity as a PIE trait (2007: 195), another southern feature, the otherwise asymmetric accusative system seems to indicate a period of influence from such a system (Uhlenbeck 1901; Beekes 1985: 172ff. & 2011: 214-216). Of particular lexical interest is the suggestion by Uhlenbeck that the IE languages consist of two distinct layers, “A und B,” one of which represents the structure of the language, and another that provided terms for “einzelne
Verwandtschaftsnamen, zahlreiche Körperteilnamen, Zahlwörter usw.” (1933: 397); such claims may now be more rigorously tested with the aid of an additional eighty years of research into the Caucasian languages.

Common heritage
Reversing the trajectories of the adstrate hypothesis, Trubetzkoy suggests that PIE developed from a Caucasian language under the influence of Uralic to become an independent language family (1939: 89).

Kartvelian or South Caucasian Languages
Four languages comprise this family, in which an early bifurcation separates Svan from the rest of the stock (cf. Klimov 1998: viii), thus:

  • Svan
  • Georgian-Zan:
    • Zan
      o Mingrelian
      o Laz
    • Georgian (attested already from Old Georgian in the 4th century CE)

Geography and history
The distribution of the Kartvelian languages is largely confined within the borders of modern day Georgia, with the gravity centered around Georgian as the country’s only official language and the three minor languages in the west. The Zan languages straddle the Black Sea, while Svan is confined to a handful of valleys in the western Caucasus mountains. The breakup of proto-Kartvelian is generally believed to have happened in the third millennium BCE (e.g. Smitherman 2012: 517), and different technological assemblages may aid the stratification (cf. Klimov 1998: ix). Nichols believes that Kartvelian is foreign in its present Caucasian distribution, originally “emanating from somewhere to the southeast of the Caspian Sea” (1997: 128), a hypothesis based on loan word trajectories and linguistic spread zones. This theory is worth questioning, as complete relocations of linguistic communities require more than circumstantial evidence, and is, indeed, contrary to the general assumption of a homeland equal to or in the vicinity of its current distribution (cf. Tuite 2008: 145, Dolgopolsky 1989: 10 fn.3, and Smitherman 2012: 517).
[ADD: A very recent discussion of the origins of Kartvelian and possible interface with PIE can be found on the blog]

Linguistic affinities with PIE
Based on a series of typological similarities, most notably their suggested glottalic mode of articulation of one of the PIE stop series (cf. § 1.4.6), Gamkrelidze & Ivanov suggest prolonged and intimate contacts with Kartvelian in the Transcaucasian region (1995: 768f.). Smitherman (2012) believes to have demonstrated clear evidence of similarly intense contacts, but is adamant in allowing the transfers to have occurred either in PIE proper or a slightly later, still undifferentiated branch. Kartvelian adstrate is suggested to have happened in Central Asia by Nichols (1997 and 1998, cf. § 3.1.2)

Common heritage
Kaiser & Shevroshkin (1986) assign strictly genetic affinity due to the semantic fields to which the purported lexical comparanda belong.

Suggested literature:
Fähnrich, Heinz (2007) Kartwelisches Etymologisches Wörterbuch
Klimov, Georgij A. (1969) Die kaukasischen Sprachen
— (1998) Etymological Dictionary of the Kartvelian Languages

Addendum: Hattic
Hattic, a language preserved in enigmatic cultic texts from the Hittite empire, continues to elude proper classification. The language is believed to have been spoken in central and northern Anatolia before the Anatolian languages, and, in particular, Hittite, gained supremacy in the area. Relations to NW Caucasian (§ 2.1.2) are possible, but remain unsettled (cf. Chirikba 1996: 407ff.)