Luc Vartan Baronian -- Armenian Linguistics


Research > Armenian Studies > Armenian Linguistics	Contact

The purpose of this page is twofold. First, I want to provide information on what Armenian linguistics is concerned with and how my research fits into the larger body of literature on the topic out there. Second, I also hope that non-linguists will find answers to frequently asked questions. This page contains the following three sections:

History and Dialectology of the Armenian Language
Synchronic Armenian Grammar
Frequently Asked Questions

History and Dialectology of the Armenian Language

The main issues in the history of the Armenian language have revolved around the origins question. Armenian is located in a crucial and strategic part of the world, at the crossroads of Europe and Asia and not too far from Africa, where humanity originated. Further, as an Indo-European (IE) language, Armenian is located within the general area that is judged most credible as the original location of that linguistic family (draw a circle from the Balkans to the Black Sea to Bactria to Mesopotamia and back to the Balkans). Could Armenia be the location of the original Proto-Indo-European language (PIE)? While I have not published on the topic, my own take on the issue is pretty standard: such a hypothesis is unlikely, given the fact that Armenian contains words for local flora and fauna not found in other IE languages. Of course, it's always possible that Armenia was the original site of PIE, but that the Armenian language came back to it after spending some centuries or millenia in Europe or elsewhere.

Armenians who are not linguists are often under the incorrect impression that there are only two possibilities for Armenian origins: either Armenian is the only IE language who stayed in place or Armenian is descendant from Phrygian, following Herodotous's claim that Armenians were Phrygian colonists. Yet, mainstream historical linguists are convinced by neither position. It is true that Armenian shares some features with Phrygian, but it shares even more features with Greek and Indic. In all cases, the features in question are not in sufficient numbers or of a type deemed convincing for establishing a common stage after PIE. The standard view is that Greek-Armenian-Indo-Iranian (plus Phrygian, Thracian and perhaps Albanian) forms a linguistic area where IE features spread at an early date, while the branches were probably already distinct, but still mutually intelligible.

When dealing with the origin of any language, it is important to remind ourselves that the issue is largely independent of the origins of the people. By most accounts, PIE was spoken about 7000 years ago. Before the colonial period, its descendant languages were spoken on an area that stretched from Ireland to Northern India. The genetic diversity in this vast area is immense and its unity goes back way before the earliest imaginable date for PIE linguistic unity. By any account, almost all speakers of IE languages today have to be largely the descendants of an ancient people that was once assimilated.

As far as the Armenian population is concerned, the studies stemming out of the new field of population genetics have all shown that the Armenian population is a very local population, with a minority of historical admixture from outside populations. In other words, even if the IE language came in prehistoric times from outside Armenia like its cultural spouse, the Christian religion, did in historic times, the prehistoric population of Armenia was not replaced by hordes of invading Indo-Europeans.

Concerning the dialect divisions between Eastern and Western Armenian, an issue arises as to where the linguistic border should be drawn. Western Armenian is associated with a voicing system whereby [dun] 'house' and [tun] 'you' correspond to Eastern [tun] and [du] respectively. Western Armenian is also associated with the [gë] present tense (gë sirem 'I love') as opposed to the [-um] present of Eastern Armenian (sirum em). Further, Western Armenian possesses an [-e(n)] ablative (Yerevan-en yegav 'He came from Erevan') that contrasts with Eastern Armenian's [-its] ablative (Yerevan-its yekav). Many dialects however display mixed patterns: for example, the Van dialect has the Western present, but the Eastern ablative and semi-Eastern voicing; the Gyumri dialect has a Western present and ablative, with a semi-Eastern voicing; the Sebastia dialect has semi-Western voicing with Western present and ablative. In a forthcoming paper, I argue that the dialects of the ancient minor kingdoms of Tsopk and Lesser Armenia were the true Western dialects, while the kingdom of Greater Armenia consisted solely of Eastern dialects. During the Ottoman period, the Western dialects gained social importance, and started exerting an influence on adjacent dialects, which gradually shifted their appearence to resemble Western dialects more and more.

Synchronic Armenian Grammar

Contrary to what non-linguists may think, most modern linguists working on Armenian do not deal with the historic (a.k.a. diachronic) aspects of the language, but with its synchronic aspects. In other words, most contemporary linguists at the PhD level or beyond work on the grammar of either Modern Eastern or Modern Western Armenian. Take a look for example at the following list of English-language PhD dissertations since 1990:

Tamrazian, Arminé. 1994. The syntax of Armenian: chains and the auxiliary. University of London.

Bert Vaux. 1994. Armenian Phonology. Harvard.
[Published in 1998 as The Phonology of Armenian, Oxford: Clarendon Press.]

Michelle Sigler. 1996. Specificity and Agreement in Standard Western Armenian. Massachusetts Institute of Technology.

Karine Megerdoomian. 2002. Beyond words and phrases: a unified theory of predicate composition. University of Southern California.
[Deals with Farsi and Eastern Armenian syntax.]

Luc Vartan Baronian. 2006. North of Phonology. Stanford.
[Chapter 4 deals with Western Armenian morphology.]

Hrach Martirosyan. 2008. Studies in Armenian Etymology with Special Emphasis on Dialects and Culture, Indo-European Heritage. Universiteit Leiden.
[Published in 2010 as Etymological Dictionary of the Armenian Inherited Lexicon, Leiden and Boston: Brill Academic Publishers.]

Hrayr Khanjian. 2013. (Negative) Concord and Head directionality in Western Armenian. Massachusetts Institute of Technology.

The discussions in these disserations are very formal and the non-specialist unfortunately does not gain much by reading them. Exceptions include the Martirosyan's historical dissertation (a dictionary), dialectal pronunciations in Vaux's dissertation and colloquial Western phrasings in Khanjian's dissertation, but the reader should not even expect to understand 90% of these two later dissertations without at minimum a BA in linguistics. Martirosyan's dictionary is accessible, although the non-Indo-Europeanist will find it difficult to understand the discussions about the possible derivations of words. The goal of the synchronic research, essentially, is to create or fine-tune an existing (generative) formalism to account for the facts of Armenian phonology, morphology and syntax. Armenian is interesting for this purpose because it was recently ranked as one of the most typologically odd languages: its features are rare or, rather, the combination of its known features is rare among the world's 6000 or so languages.

My dissertation and publications in this area of research are concerned with Western Armenian verbal morphology (conjugations), that is, how to build a set of formal rules and constraints that generate the correct set of Western Armenian verbal forms from a minimal (or optimal) set of memorized lexical entries. The most difficult issues lie in how to split the task between morphology proper and phonology, especially when such things as syllable count is relevant. I have also been working on the Armenian vowel alternation, although I have not yet published any work on this latter topic.

Frequently Asked Questions

Disclaimer: The following list of Q&A does not stem from my own research. They are simply summary statements about the state of knowledge in Armenian linguistics, both on the consensuses and areas of disagreement. I do not reference very basic facts that can be found in any general historical linguistics textbook such as Anttila[1], Campbell[2], Fortson IV[3], Hock[4], Mallory & Adams[5], Trask[6] or Watkins[7]. I only reference facts and hypotheses specific to Armenian that appear in few places. Of course, it's impossible to include every hypothesis that was ever proposed, but I tried to include as many credible alternatives as possible.

1. How do we know that Armenian is an Indo-European language?
Any two randomly selected languages share similarities. This can be due to chance or to universal tendencies. An example of the latter is the word for mother, which often contains an [m], probably because [m] is universally among the first sounds pronounced by babies and perhaps because of an ideophonic connection with the suckling sounds and gestures associated with nursing. Besides the requirement that the common words be in sufficient numbers, the major requirement for establishing a relationship between languages is that the correspondences be systematic: for example, English wasp, William, and war correspond to French guêpe, Guillaume and guerre ([w/g] correspondence), while English father, five and fire correspond to Greek pater, pente and pyr (f/p correspondence). When two languages share such systematic similarities, historical linguists search for an explanation. The two possibilities are that the languages were once in a situation of social contact (the reason why English borrowed Chinese words and vice versa) or that the two languages were once one and the same (for example, French, Italian and Spanish are all forms of Latin that evolved separately).

It is usually not a problem to distinguish common words due to a contact situation from common words due to an inherited common stage. For example, the words khanut 'shop' and shuka 'market' that Armenian shares with Syriac are obvious trade terms (hence, due to contact). However, Armenian shares words with Greek, English, Latin, Russian and Sanskrit that belong to a more intimate part of the lexicon, less prone to borrowing, which includes, but is not limited to kinship terms, small numbers and body parts. Further, Armenian shares grammatical prefixes and suffixes with Indo-European languages. If we think of a language as consisting of a machine (the grammar) that spits out its production (words and phrases), we can safely say that Armenian has an Indo-European machine and core production.

2. Is Armenian agglutinative? What does that mean and imply?
Armenian is somewhat agglutinative, with Modern Armenian being more so than Classical Armenian was. An agglutinative structure refers to the stringing of a root with a series of prefixes or suffixes. For example, English lov+er+s consists of the root lov(e) followed by the suffixes +er and +s. The more a language contains such structures, the more it is considered agglutinative. Besides being agglutinative, word structures can also be, among other possibilities, isolating or infectional. An isolating example from English is I will love, where each word carries a separate function (subject, future tense, verb) as opposed to I loved, where the past tense is more agglutinative. An inflectional example from English is sing/sang/sung where the tense is expressed by changing the internal vowel of the root.

Like English and like many other languages, Armenian shows agglutinative, isolating and inflectional structures: nav+er+u+n (boat+plural+dative+definite) 'of the boats' is agglutinative; siradz em 'I have loved' is isolating; mayr/mor 'mother/of mother' is inflectional.

It is important to note that the late 18th / early 19th century linguist and philosopher W. von Humboldt used such terms as isolating, agglutinative and inflectional to denote what he perceived were stages in the evolution of languages. In von Humboldt's mind, inflectional languages such as Sanskrit represented the summum of linguistic perfection. It was also thought by some that the use of such structures by different language groups were indicative of a special relationship. Both these ideas have since completely been discarded and the labels are now simply used as a convenient way to summarize a language's word structure, with no major theoretical consequence in mind.

It is possible that Armenian became more agglutinative under the influence of Turkish[8], but such a hypothesis is very difficult to prove, because the evidence is indirect. In any case, Armenian's pseudo-agglutinative status is not indicative of anything concerning its origins.

3. Was Armenia the site of the Indo-European Homeland?
In the current state of research, there are more linguistic arguments against this view than in favor of it. Armenian is definitely an Indo-European (IE) language and, by almost everybody's account, it is one that separated from the others very early on. Linguists argue that Proto-Indo-European (the mother language of Latin, Greek, Sanskrit, Hittite, Armenian and others) was most likely spoken circa 5000 B.C., shortly after the wheel was invented, but before the invention of bronze, because of the words we are able to safely reconstruct in this language. A plurality of anthropologists have linked the IE language dispersal to the kurgan cultural expansion from the steppes North of the Black Sea.[9] Others, arguing for an earlier date, link the dispersal to the spread of agriculture to Europe from (peninsular) Anatolia.[10] Armenia has been considered as a potential site within the Glottalic Theory, which has, however, failed to gain mainstream support among historical linguists. Other credible proposals include Bactria, the Balkans and Mesopotamia, all sites located not too far from Armenia, but not quite close enough.

The main argument against considering that Armenia is the original site of the IE languages is that Armenian almost certainly contains a substrate that other IE languages lack. Obvious words include those for 'camel', 'apple', 'grape/vine', 'plum', 'quince', 'mint', 'pomegranate', 'tin'. Many of these words are from the local but now extinct Hurro-Urartian family.[11] It is a typical case of a language coming in from outside adopting the local names for flora and fauna. Had these words originated in Armenian, we would expect to find them in the other Indo-European languages, but we don't.

4. Was the land of Ḫayaša-Azzi populated by (Proto-)Armenians?
Apart from a few personal names, we have no trace of the language that was spoken in Ḫayaša-Azzi, so the short answer is we don't know. The word Ḫayaša does resemble the Armenian endonym Hay to some extent and it was located in the Northwestern part of the traditional Armenian linguistic domain, but the Azzi part of the name is mysterious. Furthermore, the original Hittite inscriptions could actually read something closer to Khayasha, but it is possible that this discrepency was an adaptation of the alleged Armenian Hay. The -ša part may have been a Hittite suffix, not present in the language of Ḫayaša.[12]

5. What is the etymology of the endonym Hay 'Armenian'?
We don't know, but there are many proposals. We know that PIE initial *p became [h] in Armenian (as in *ph2ter becoming hayr 'father'). Therefore, one of the strategies has been to look for PIE words beginning with *p. One of them is PIE *potis 'lord', but the vowel [o] is problematic. Others have proposed a link to Paeonia and Paeonians (name of a Balkanic people implicated in the Trojan War).

The Armenian consonant [h] can also originate more directly from a laryngeal consonant. For example, the word hazar '1000' was borrowed from an Iranian language as such. Or, to take a native example, Armenian hav 'hen' is descendant of PIE *h2éu-i-. Therefore, Hay could be derived from Hatti, which was originally the name of an ancient Anatolian people speaking a linguistic isolate. The name was later adopted by Hittites on the same territory. If the Armenian language came in from the West, it could have adopted and adapted the name as well when it crossed Hittite lands. Alternatively, Jahukyan derives it from PIE *h2ei-e/os- 'metal/copper/iron', along with the toponym Ḫayaša. In support of this latter hypothesis, Martirosyan notes that the name of the people later known in the same area by the Greeks as Chalybes has a very similar etymology: 'hardened iron, steel'.[12]

Yet another proposal is Petrosyan's Eti-uni (name of a local people contemporary with Urartians), but, here, the initial consonant is lacking altogether.

6. What is the relationship between Armenian, Phrygian and Greek?
There are three possibilities, presented here from the least conservative to the most conservative view. The first argues that the Phrygian language became Armenian when Phrygian colonizers settled in Armenia, probably assimilating and merging with local Urartians. This theory finds its historical justification in Herodotous who claims that Armenians originated in Phrygia and, hence, earlier in the Balkans, where Phrygians are said to have come from. Strabo, writing later, but from a closer contact, points to Thrace as a point of origin. The mainstream linguistic position rejects this filiation. It is in fact not at all obvious how one could derive Armenian from Phrygian. Granted, our knowledge of Phrygian is limited and the native Armenian vocabulary is small in comparison to most IE languages. Nevertheless, we have many Armenian/Phrygian pairs and there are plausibility issues in transforming one into the other. For example, PIE *meh2ter 'mother' became Phygian matar. It is not obvious how one would then get Armenian mayr, an intermediate stage closer to Greek or Latin mater would work better. A knock-down argument against Armenian being a descendant of Phrygian is that the latter has merged the PIE distinction *d/*dh into [d], while Classical Armenian preserves it as [t]/[d].[13] Once a phonological distinction has been lost in a language, it is logically impossible to recreate it in exactly the same words. The only way to salvage the filiation would be to hypothesize that Phrygian orthography did not correspond exactly to Phrygian pronunciation and that PIE *dh was preserved.

The second possibility is that Armenian, Phrygian, Greek and perhaps Albanian together form a branch of IE that we could label Balkanic. This would mean that these languages were once one and the same (they underwent a common stage between PIE and their current form). Indeed, there are some reflexes of PIE in Greek, Armenian and Phrygian especially that are strikingly similar.[13] One reason this possibility is not considered proven by most historical linguists is that the nature of most of these common reflexes is lexical, which means it can be due to an intensive and ancient contact, rather than the result of a common stage. Another reason is that Armenian also shares an important set of common ancient reflexes of PIE with Indo-Iranian. It is thus not obvious to decide whether we should group Armenian along with an unestablished Balkanic or Armeno-Indo-Iranian (or neither).

The third possibility is the dialect continuum hypothesis, which appears to be most likely to most specialists. Under this view, Armenian separated early on from the other IE languages, but remained in close contact with Greek, Phrygian and Indo-Iranian. At the time, these languages must have been separate, but still mutually intelligible, such that it was possible for them to undergo common changes. Armenian underwent some innovative changes with Greek and Phrygian, but others with Indo-Iranian.[14] For a modern analogy, think of how American English and British English still undergo common changes through contact, even though they separated four centuries ago (an example of this is the quotation marker like in sentences originating in the 20h century such as ...and he was, like, "I didn't do it").

Under either the second or third view, Herodotous's claim could be interpreted as saying that Armenian speakers tagged along the Phrygians, with whom they shared some cultural affinities (Herodotous mentions clothing and arms) but were distinct from them linguistically.

7. What is the relationship between Armenian and Indo-Iranian?
Early Indo-Europeanists thought that Armenian was a divergent form of an Iranian language. German linguist H. Hübschmann convincingly demonstrated that Armenian is not Iranian, but an independent branch of Indo-European that has undergone massive borrowings from Iranian into its lexicon (Parthian and Persian, mainly).[15] At the core of Armenian, there remain several hundreds of words (perhaps up to a thousand) designating body parts, kinship, small numbers and other common vocabulary that have undergone regular sound changes distinct from the ones that are common to Indo-Iranian languages. Nevertheless, there remains a set of common features between Armenian and Indo-Iranian that, if they are due to contact, must be very old (before the two branches were too differenciated). Particularly striking are the common features of Armenian and Indic that are lacking in Iranian.[14]

8. What is the relationship between Armenian(s) and Aramaic/Syriac/Arameans (or their names)?
They belong to separate linguistic families: Armenian is an Indo-European language and Aramaic was a Semitic language. There are many ancient loanwords in Armenian that came from Aramaic/Syriac and their civilisation played an important role in the Christianization process of Armenia.

The words Armenia(n) predate Christianization. It is possible that the words Armenia(n) and Aramaic are somehow connected, but we do not know.

One of the Syriac-language churches shares the religious doctrines of the Armenian church. The population DNA studies have shown that Armenians and Assyrians (whose traditional language is Syriac), but not Maronites, are extremely close genetically. At present, it is difficult to judge whether this is due to a common origin or to a high level of intermarriage between the two communities.

9. What is the relationship between Armenian and Hittite?
Other than geographic proximity and a handful of common features that could be due to early contact or archaism, the two languages don't share anything beyond their common IE heritage. American linguist W. Austin once hesitantly proposed (his words) that Armenian could be considered part of the Anatolian branch of IE, along with Hittite and especially Lycian[16], but his arguments were convincingly rejected.[17]

Note, however, that Armenian and Anatolian do not share the words used by other IE languages for wheels. It is possible then that these two branches split from the rest before the invention of the wheel. Such a hypothesis, if proven, would confirm Anatolian and Armenian as very archaic and would push back in time the date at which PIE was spoken. Note also that, based on indirect evidence, Petrosyan has suggested that an Eastern Anatolian language may have existed in Armenia in ancient times.[18]

10. What is the relationship between Armenian and Urartian?
The relationship is strictly one of contact. Urartian is currently not considered to be related to the Indo-European family to which Armenian belongs, although there have been proposals of a deeper common origin. Urartian is attested earlier than Armenian, but on the same territory. Urartian's only confirmed relative was Hurrian (also extinct), although a relation to modern Northeast Caucasian languages has been suggested by I. M. Diakonoff and S. Starostin (the jury is still out). A handful of words are shared between Armenian and Urartian, but they are undoubtedly due to contact.[11] In some case, it is possible that the words went from Armenian to Urartian (as opposed to the other direction generally assumed), which would indicate a cohabitation of the two languages long before the Armenian language was itself attested. Claims to the effect that the Urartian language is at the origin if either Armenian or Turkish are due to nationalist propaganda and have no scientific basis.

11. What is the relationship between Armenian and Udi?
The relationship is strictly one of contact between two unrelated languages. Armenian is an IE language, while Udi is a Northeast Caucasian language. It has been suggested, but not proven, that Urartian may have been related to Northeast Caucasian languages. The Udi language was once spoken on a territory called Utik and adjacent to Artsakh (Nagorno-Karabakh) and the Udi people are religiously affiliated with the Armenian Church. The contact must have been quite intensive, for Udi has even borrowed the Classical Armenian complementizer t'e.

12. What is the relationship between Armenian and Ethiopian?
There is no relationship whatsoever. Armenian is an Indo-European language (along with Greek, English, Latin, Russian, Sanskrit and others) and Ethiopian languages are Semitic, hence Afro-Asiatic (along with Arabic, Berber, ancient Egyptian, Hausa, Hebrew and others). Armenians and Ethiopians are known to have traded since ancient times (their Christian doctrine is even quite similar) and there are still Ethiopian Armenians today, but there is no general linguistic trace of this. People often point to the similar look of the Armenian and Ethiopian alphabets. The Ethiopian alphabet is older. At best, an influence in style one way or the other is possible, but such a cosmetic feature has no bearing on the structures of each respective language.

13. What is the relationship between Armenian and Basque?
At best, it is possible that Basque is related to one of the substrate languages of Armenian. This was the programatic hypothesis of J. Karst (his words), who is better known for his earlier excellent grammar of Middle (Cilician) Armenian[19], but who also thought that Armenians were anciently the speakers of a language resembling Basque and who were assimilated by Indo-European speakers.[20] Basque is affiliated with no language family known to us. Nevertheless, there are words that are similar in Basque and Armenian.

An alternative explanation to Karst's hypothesis is that both Basque and Armenian picked up substrate words from now extinct European/Mediterranean languages. In favor of this hypothesis, many of the words common to Armenian and Basque are also found at random in European and Caucasian languages. There is every reason to think that 7000 years ago, Europe looked very different linguistically than it does now and that non IE languages were dominant on that continent. That words for flora and fauna especially were picked up randomly by newcomer languages (such as the IE ones) from these ancient languages is almost a certitude, but it remains difficult to prove which words exactly, precisely because it was done on a piece-by-piece basis.

The idea that the Basque language is a form of Armenian (or vice versa) is complete non-sense based on a misinterpretation of Karst's work and on hypotheses that predate the Comparative Method in linguistics.

14. What is the relationship between Armenian and Sumerian?
No direct relation. Sumerian is a linguistic isolate and the oldest language attested, linked to the very beginings of human civilisation. Recent work however has identified some very plausible Indo-European words in Sumerian.[21] These words are not Armenian by a long shot, but it is an open question as to how Sumerian acquired them (if they are indeed IE). One possibility is that the IE homeland was in Mesopotamia and Sumerian acquired these words through immediate contact. Another possibility is that the words were acquired via trading with nations to the North who had previously acquired these words via trading with IE speakers. Note that under either scenario, it is possible that, rather than going from IE to Sumerian, the words in question went from Sumerian to IE (hard to tell).