Is the Georgian language related to Basque, another European “outlier”?
The history of the Georgian language reveals some interesting patterns of cross-cultural interaction. Georgian can be traced back to a ancestral language— Proto-Kartvelian—that it shares with its close relatives: Mingrelian, Svan and Laz. Spoken in the second millennium BC, Proto-Kartvelian must have interacted closely with Proto-Indo-European, the ancestral tongue to most European languages, as well as those of Iran and northern India. This connection is indicated by the so-called ablaut patterns (like the English sing-sang-sung), which Proto-Kartvelian probably borrowed from Proto-Indo-European, alongside many specific words. The most notable among these loanwords is the reconstructed Proto-Kartvelian m.k.erd ‘breast’, which is said to be a cognate to the Indo-European kerd ‘heart’ (cf. the Latin cardio—and even the English heart).
While the connection of Georgian to Indo-European languages is solid, if distant, several scholars have searched for linkages to other languages, most notoriously Basque, a non-Indo-European “outlier” language in Europe. To this day, no proven connection has been demonstrated between Basque and any currently spoken languages; as a result, Basque remains a perfect isolate, an “orphan” language with no ties to any language family. But the idea that Basque might be related to some other languages, in particular Georgian and other languages of the Caucasus, has ignited a lot of interest among Vasconists (i.e. scholars of Basque) and Caucasianists alike.
The search for a connection between Caucasian languages and Basque dates to the work of Hugo Schuchardt in the early twentieth century. Schuchardt was chiefly interested in finding a North African connection for Basque, but in a 1913 paper he cited some parallels between Basque and the languages of the Caucasus. The Dutch linguist Christian Cornelius Uhlenbeck explored the same connection in a series of papers beginning in the early 1920s (e.g. “De la possibilité d’ une parenté entre le basque et les langues caucasiques”, published in the Revista Internacional de los Estudios Vascos) and continuing through 1940s. The Italian linguist Alberto Trombetti wrote an entire book in 1925 based on a long list of supposed Caucasian-Basque cognates. The Georgian linguist Nikolai Marr published several articles comparing the languages of the Caucasus and Basque, but Marr is now generally considered to have been more a myth-maker than a scientist (in a curious twist of fate, his teachings were declared anti-Marxist in an article published in the Soviet newspaper Pravda under the signature of Joseph Stalin himself, but this article was most likely inspired by the writings of Marr’s most energetic opponent, Arnold Chikobava). The French Indo-Europeanist (and Caucasianist) Georges Dumézil devoted a chapter of his 1933 book to putative cognates found among Basque and the Northern Caucasian languages. René Lafon, a French scholar of Basque, produced a long series of articles, published in 1930s through 1960s, arguing for a Basque-Caucasian link. This idea was further explored by the Norwegian Caucasianist Hans Vogt and the German linguist Karl Bouda, although they achieved very different results: Vogt’s conclusions were largely negative, whereas Bouda is perhaps the most enthusiastic of the proponents of a Caucasian-Basque link. Bouda’s papers from the late 1940s and early 1950s provide a most extensive list of cognates, including some 500 items.
The original inspiration for the Georgian- (or more generally Caucasian-) Basque link came from the existence of the ancient Kingdom of Iberia in the Caucasus (its self-designation was Kartli, hence “Kartvelian”), basically located in present-day eastern Georgia. Another supposed link involves the existence of the Basque place-name ending in –adze, similar to Georgian surnames ending in -dze or –adze (prevalent in western Georgia). (The most common surnames ending in -(a)dze are Kapanadze, Maisuradze and Giorgadze; one may also recall in this connection the names of the former Georgian president Eduard Shevardnadze and of the iconic ballet choreographer Giorgi Balanchivadze, better known as George Balanchine.) However, this suffix literally means ‘son’ in Georgian, so the connection to Basque toponyms is hardly substantiated.
As mentioned above, various scholars have proposed a number of putative cognates, but most of them have simply listed Basque words and morphemes that bear a vague resemblance to words and morphemes in one Caucasian language or another (including Georgian). Representative examples include: the Basque word etxe ‘house’ is matched with the word ča ‘hut’ in Lak (Northeast Caucasian); and the Basque ahari ‘ram’ is matched with words meaning ‘small lamb’ in three Northeast Caucasian languages (Chechen Eaxar, Ingush häxar and Batsbi axrab). Many of the proposed cognates require a leap of imagination to see the sound correspondences at all, as is the case with the Basque haragi ‘meat’ matched with the Circassian l« ‘meat’; the Basque hotz ‘cold’ matched with the Abkhaz sw ‘freeze’; Basque larri ‘anxious’ matched with Avar erize ‘be afraid,’ and so on. Among the specifically Georgian-Basque cognates that suffer from the same problem is the Georgian bza ‘box tree,’ matched with the Basque ezpel ‘box tree.’ These lists of putative cognates typically pair Basque words with words in some particular Caucasian language (or closely related languages). Yet, these works do not demonstrate a connection between Basque and the ancestral languages of Northwest, Northeast or South Caucasian languages, let alone a connection between Basque and the ancestral language of all Caucasian languages.
And since the search for a Caucasian-Basque link involves some forty Caucasian languages, it is not surprising that some resemblances can be found, if only by chance. Such a possibility is enhanced if one allows rather loose parallels in sound and meaning, as has indeed often been the case.
But the Caucasian-Basque hypothesis was fueled not only by questionable lexical cognates, but also by a number of typological similarities. Georgian, for example, shares with Basque its ergative case system, an elaborate scheme of verbal agreement, as well as such other characteristics as agglutinative morphology, Subject-Object-Verb order, vigesimal number system (base 20), and the presence of distributive numerals. A full discussion of such similarities would be too long and too technical to be included here, but some points of commonality do deserve discussion..
Let us briefly consider the ergative case system shared by Georgian and Basque. In English, word order encodes who did what to whom: John kissed Mary and Mary kissed John differ by who initiated the action and to whom it was applied, which we deduce solely from the word order). But many, if not most, languages use some overt morphological marking to encode the same thing. Noun case is one such type of marking: in a case language such as Latin or Russian, the equivalents of ‘John kissed Mary’ and ‘Mary kissed John’ differ not so much by word order but by the forms of the nouns ‘John’ and ‘Mary’. In scores of languages, Latin and Russian included, the subject of a sentence – here, the one who did the kissing – appears in one form, called “nominative”, whereas the object – here, the one who is kissed – appears in another form, called “accusative”. Since “who did what to whom” can be deduced from the forms of the nouns, word order in many case languages is much freer than it is in English.
However, not all languages that rely on case use the same nominative-accusative model employed by Latin and Russian. Both Georgian and Basque follow a different model, known as “ergative-absolutive” (after the names of the two main cases). In this model, the subject appears in the so-called “ergative” case – but only if an object is also present. If there is no object, as in for example ‘John left’, the subject appears in a different form called “absolutive”. This latter form is also used for objects.
Here are some illustrative examples from Georgian. In the sentence (1a), which has both a subject and an object (it is thus called a “transitive sentence”), the subject bich’ ‘boy’ is marked with the ergative suffix -ma, and the object dzaghl ‘dog’ is marked with the absolutive suffix -i. In the sentence (1b), which has a subject but no object (and is thus an “intransitive sentence”), the subject is not marked by the ergative suffix -ma, but rather by the absolutive suffix -i. In effect, the object in (1a) and the subject in (1b) are in the same form.
bich’-ma dzaghl-i bagh-shi da-mal-a.
boy-ERG dog-ABS garden-DAT-in hid.AOR
‘The boy hid the dog in the garden.’
dzaghl-i bagh-shi da-i-mal-a.
dog-ABS garden-DAT-in hid.AOR
‘The dog hid in the garden.’
The same thing is true of Basque: the ergative suffix -k attaches to subjects but only if an object is also present (as in (2a)); if there is no object, the subject appears in the absolutive case, which is marked by the absence of the ergative suffix -k.
ehiztari-a-k otso-a harrapatu du.
hunter-DEF-ERG wolf-DEF.ABS caught has
‘The hunter has caught the wolf.’
otso-a etorri da
wolf-DEF.ABS arrived is
‘The wolf has arrived.’
But the ergative case system is not the only typological similarity between Georgian and Basque. Both of these languages require its verbs to agree not only with subjects (as in the English I play vs. He plays), but also with objects. Both the ergative case system described above and the pluripersonal agreement (i.e. agreeing with both subjects and objects) appear exotic to a speaker of any major European language, so it is tempting for us to look for a familial connection between Georgian and Basque.
But one particular quirk throws the relationship in doubt. While both Georgian and Basque employ the ergative case model and pluripersonal agreement, only Basque follows the ergative model for its agreement, selecting the same agreement morphemes for intransitive subjects (i.e. subjects that occur with no object in sight) and for objects, whereas agreement with transitive subjects (i.e. subjects that occur with an object in the same sentence) is marked by a different set of morphemes. In contrast, Georgian follows the nominative-accusative model for its agreement: agreement with subjects is marked by one set of morphemes, regardless of whether an object appears in the same sentence or not, while agreement with objects is done by another set of morphemes. In other words, for the purposes of selecting a case suffix, Georgian treats subjects differently depending on whether the object is present or not, whereas for purposes of agreement on the verb, all subjects are created equal. Basque, on the other hand, relies on the ergative-absolutive model more extensively. Thus, the apparent typological resemblance between Georgian and Basque – their ergative case system and the pluripersonal agreement – does not hold up on a closer inspection.
Although the common presence of non-Indo-European grammatical properties (together with putative cognates) has been enough to persuade some linguists that there must be a connection between Georgian and Basque, this is a dangerous assumption, as typological resemblances have rarely been useful in proving common descent from an ancestral language. As a result, most linguists today – including most notably R.L. Trask, Luis Michelena, V.A. Chirikba – reject the idea of a Georgian- (and more generally, Caucasian-) Basque link.
Moreover, genetic studies to date have not found any link between the Basques and the peoples of the Caucasus. For example, Bertorelle et al. (1995) conclude that “a genetic test of this hypothesis, based on classical markers [i.e. blood groups and protein electromorphs], did not show any particular genetic link between Caucasians and Basques”. This conclusion is further supported by the work of Nasidze and Salamatina (1996), who found that “Georgians seem to be genetically differentiated in relation to European populations, again as shown by classical markers”. Nor was mitochondrial DNA any help, as Comas et al. (2000), who studied a 360-base-pair stretch in HVR I of the mtDNA control region, concluded that “the putative linguistic relationship between Caucasian groups and the Basques, another outlier population within Europe for classical genetic markers, is not detected by the analysis of mtDNA sequences”.
The negative findings of modern scholarship have not, however, prevented certain schools of populist nationalism from propounding an “Iberian” theory of pre-history based on the supposed kinship of Basque and Georgian. A website called “Iberia Forever”, dedicated to the “spiritual mission of Georgia”, claims that “According to the latest studies of modern Kartvelologists (Jan Braun, and others), the view is gaining ground on Basque being a fourth Kartvelian language”. Another Georgian site follows the discredited work of Nikolai Marr in arguing that “the Georgians and the Basques (in Spain) are the sole survivors [of the proto-Iberians], though the extinct Etruscans in Italy may have belonged to a kindred family”. Such claims are taken much further in an imaginative YouTube video modestly entitled “Iberian Heritage: Let the Truth Speak for Itself” (see the image to the left). Such ideas occasionally filter into the mainstream media. A 2003 Washington Times article, for example, begins by noting that
“It may come as a surprise that the Georgians of the former Soviet Union and the Basques of ancient Iberia, now Spain and Portugal, have a common ancestry, but early Greeks and Romans called those Georgians and Basques Iberians.”
The search for common ancestry among seemingly disparate ethnic groups is understandable, and is beneficial insofar as it draws people to the study of history, geography, linguistics, and genetics. But it is the responsibility of scholars to occasionally throw cold water on fond dreams that do not withstand scrutiny.
Bertorelle G, Bertranpetit J, Calafell F, Nasidze I, & Barbujani G. (1995) “Do Basque- and Caucasian-speaking populations share non-Indoeuropean ancestors?” European Journal of Human Genetics 3:256–263.
Comas D, Calafell F, Bendukidze N, Fañanás L, & Bertranpetit J (2000) “Georgian and Kurd mtDNA Sequence Analysis Shows a Lack of Correlation Between Languages and Female Genetic Lineages” American Journal of Physical Anthropology 112:5–16.
Nasidze IS, Salamatina NV. (1996) “Genetic characteristics of the Georgian population”. Gene Geogr 10:105–112.
« Genetic clues to the Ossetian past
The linguistic and genetic mosaic of the... »