Classification and origins
Armenian is an independent branch of the Indo-European languages. It is of interest to linguists for its distinctive phonological developments within that family. Armenian exhibits more satemization than centumization, although it is not classified as belonging to either of these subgroups. Some linguists tentatively conclude that Armenian, Greek (Phrygian) and Indo-Iranian were dialectally close to each other; within this hypothetical dialect group, Proto-Armenian was situated between Proto-Greek (centum subgroup) and Proto-Indo-Iranian (satem subgroup).
Armenia was a monolingual country by the 2nd century BC at the latest. Its language has a long literary history, with a 5th-century Bible translation as its oldest surviving text. Its vocabulary has historically been influenced by Western Middle Iranian languages, particularly Parthian, and to a lesser extent by Greek, Persian, and Syriac. There are two standardized modern literary forms, Eastern Armenian and Western Armenian, with which most contemporary dialects are mutually intelligible.
Although the Armenians were known to history much earlier (for example, they were mentioned in the 6th century BC Behistun Inscription and in Xenophon's 4th century BC history, The Anabasis), the oldest surviving Armenian-language text is the 5th century AD Bible translation of Mesrop Mashtots, who created the Armenian alphabet in 405, at which time it had 36 letters. He is also credited by some with the creation of the Caucasian Albanian alphabet. In The Anabasis, Xenophon describes many aspects of Armenian village life and hospitality in around 401 BC. He relates that the Armenian people spoke a language that to his ear sounded like the language of the Persians.
W. M. Austin (1942) concluded that there was an early contact between Armenian and Anatolian languages, based on what he considered common archaisms, such as the lack of a feminine gender and the absence of inherited long vowels. However, unlike shared innovations (or synapomorphies), the common retention of archaisms (or symplesiomorphy) is not considered conclusive evidence of a period of common isolated development.
In 1985, Soviet linguist Igor M. Diakonoff noted the presence in Classical Armenian of what he calls a "Caucasian substratum" identified by earlier scholars, consisting of loans from the Kartvelian and Northeast Caucasian languages. Noting that Hurro-Urartian-speaking peoples inhabited the Armenian homeland in the second millennium BC, Diakonov identifies in Armenian a Hurro-Urartian substratum of social, cultural, and animal and plant terms such as ałaxin "slave girl" ( ← Hurr. al(l)a(e)ḫḫenne), cov "sea" ( ← Urart. ṣûǝ "(inland) sea"), ułt "camel" ( ← Hurr. uḷtu), and xnjor "apple(tree)" ( ← Hurr. ḫinzuri). Some of the terms he gives admittedly have an Akkadian or Sumerian provenance, but he suggests they were borrowed through Hurrian or Urartian. Given that these borrowings do not undergo sound changes characteristic of the development of Armenian from Proto-Indo-European, he dates their borrowing to a time before the written record but after the Proto-Armenian language stage.
Loan words from Iranian languages, along with the other ancient accounts such as that of Xenophon above, initially led linguists to erroneously classify Armenian as an Iranian language. Scholars such as Paul de Lagarde and F. Müller believed that the similarities between the two languages meant that Iranian and Armenian were the same language. The distinctness of Armenian was recognized when philologist Heinrich Hübschmann (1875) used the comparative method to distinguish two layers of Iranian words from the older Armenian vocabulary. He showed that Armenian often had 2 morphemes for the one concept, and the non-Iranian components yielded a consistent PIE pattern distinct from Iranian, and also demonstrated that the inflectional morphology was different from that in Iranian languages.
The hypothesis that Greek is Armenian's closest living relative originates with Holger Pedersen (1924), who noted that the number of Greek-Armenian lexical cognates is greater than that of agreements between Armenian and any other Indo-European language. Antoine Meillet (1925, 1927) further investigated morphological and phonological agreement, postulating that the parent languages of Greek and Armenian were dialects in immediate geographical proximity in the Proto-Indo-European period. Meillet's hypothesis became popular in the wake of his Esquisse (1936). Georg Renatus Solta (1960) does not go as far as postulating a Proto-Graeco-Armenian stage, but he concludes that considering both the lexicon and morphology, Greek is clearly the dialect most closely related to Armenian. Eric P. Hamp (1976, 91) supports the Graeco-Armenian thesis, anticipating even a time "when we should speak of Helleno-Armenian" (meaning the postulate of a Graeco-Armenian proto-language). Armenian shares the augment, and a negator derived from the set phrase Proto-Indo-European language *ne h₂oyu kʷid ("never anything" or "always nothing"), and the representation of word-initial laryngeals by prothetic vowels, and other phonological and morphological peculiarities with Greek. Nevertheless, as Fortson (2004) comments, "by the time we reach our earliest Armenian records in the 5th century AD, the evidence of any such early kinship has been reduced to a few tantalizing pieces".
Graeco-(Armeno)-Aryan is a hypothetical clade within the Indo-European family, ancestral to the Greek language, the Armenian language, and the Indo-Iranian languages. Graeco-Aryan unity would have become divided into Proto-Greek and Proto-Indo-Iranian by the mid-third millennium BC. Conceivably, Proto-Armenian would have been located between Proto-Greek and Proto-Indo-Iranian, consistent with the fact that Armenian shares certain features only with Indo-Iranian (the satem change) but others only with Greek (s > h).
Graeco-Aryan has comparatively wide support among Indo-Europeanists for the Indo-European homeland to be located in the Armenian Highlands, the "Armenian hypothesis". Early and strong evidence was given by Euler's 1979 examination on shared features in Greek and Sanskrit nominal flection.
Used in tandem with the Graeco-Armenian hypothesis, the Armenian language would also be included under the label Aryano-Greco-Armenic, splitting into proto-Greek/Phrygian and "Armeno-Aryan" (ancestor of Armenian and Indo-Iranian).