One Language, Two Grammars: the ‘Plight’ of Classical Armenian

by Robin Meyer (University of Oxford; Hon. Secretary for Student Associate Members)

Armenian is one of those Indo-European languages that very rarely gets much attention from students of historical linguistics or comparative philology; most frequently, it crops up only in discussions of the augment, laryngeals, and the Glottalic Theory. This, alas, is unlikely to change.
Yet, Armenian can serve as an interesting case study for a number of fields within linguistics, not least language contact and corpus linguistics. With these two topics in mind, allow me to introduce you to Armenian – albeit in extreme brevity –, and to illustrate one of its more curious traits: its two grammars.

Map_Armenia_BCE
Map of Armenia in the 2nd and 1st centuries BCE
An exceedingly short introduction: Iranian, Greek, and the Armenian language(s)

Armenian, attested in its Classical form (called գրաբար |grabar|) since the 5th century CE, is a language with a couple of twists. Until a ground-breaking paper by Heinrich Hübschmann (1875), Armenian was thought to belong to the Iranian language family. In fact, Armenian is most closely related to Greek – and even that not all that closely (Clackson 1994). For the most part, this relationship is not immediately obvious at the surface, particularly if compared to the similarities between, for instance, Vedic and Old Avestan, or Latin and Oscan.
The reason for its historical allocation to the Iranian family lies in the inordinate amount of Iranian loan words and calques, both lexical and phraseological, in Armenian. These are mostly taken from Parthian (North West Middle Iranian; Meillet 1911–12, Schmitt 1983). Less obviously, even certain Iranian syntactic structures and patterns have been replicated (Meyer 2013, 2016). These borrowings are, without doubt, owed to long-lasting contact between Armenian and Parthian speakers. Since the 5th century BCE, Armenia was under Iranian rule in one form or another: Achaemenid, Artaxiad, Arsacid Parthian, and later Sasanian Persian. For the most part, an Armenian king of Iranian origin ruled as primus inter pares among other Armenian and Iranian noble families. The history and ethnic composition of Armenia is, of course, far more complex than can be described in one sentence; excellent summaries can be found in Hovannisian (1997).

So far, so good. Plenty of other languages have undergone changes in contact with other languages (not least English owing to Norman French; Northern French because of German; and Latin under Greek and Semitic influence).
What makes Armenian different, at least from a historical linguistic point of view, is not its contact with Iranian; Armenian is, in fact, an important source for reconstructing Iranian lexical material not otherwise attested in Iranian languages (Nebenüberlieferung).
It is its indirect contact with Greek, that is the translation of the New Testament and a number of philosophical, theological, and grammatical treatises from their Greek original, that makes (certain kinds of) Armenian a little different. In these translations, dating back to the first and second half of the 5th century CE respectively, the Armenian translators quite consistently imitate Greek word order and idiom – but that’s not all.
In the latter, translations, many concepts are rendered in Armenian on a morpheme by morpheme basis. Gk. ἐν-εργ-εία “activity, agency”, for example, is translated as Armenian ներ-գործ-ութիւն |ner-gorc-ut‘iwn| “id.”.
Occasionally, the imitation of Greek goes so far as to introduce new grammatical categories. Although Armenian does not mark grammatical gender, some texts create gender-differentiated forms, e.g. for Armenian մի |mi| “one”, which occurs as եզ (m.), մի (f.), մու (n.) |ez, mi, mu| “id.” (Muradyan 2012:92).

The translations of philosophical, theological, and grammatical texts take it to such extremes that the Armenian version is almost unintelligible without prior knowledge of the Greek source, leading to the suggestion that they may have served as aides-mémoire rather than as translations proper.
In comparison, the New Testament translation is almost ‘tame’ – but at least one crucial issue remains. Grammars and linguistic studies of Classical Armenian have, until relatively recently, treated as one the ‘hellenophile’ version of the language used in the New Testament and that used in original texts (e.g. Koriwn’s Life of Maštoc‘, or the Epic Histories attributed to P‘awstos Buzandac‘i), and have frequently given preference to the biblical corpus.
Over all, this is of little consequence for phonology and morphology, which are indeed largely identical across the two corpora. As far as syntax is concerned, however, it does make a difference, e.g. in the realm of relative clauses (Meyer 2018).

Why corpora matter: counterfactual conditionals

As part of a larger project funded by the ANR in partnership with the College de France, colleagues and I have recently begun investigating the use of imperfective aspect in modal expressions in the languages of the Caucasus. My first goal was to find out whether the modal use of the imperfect in Armenian was uniform across these two corpora or not, and if not, whether any differences could be explained.

In Armenian, the imperfect tense is used for a number of purposes: as a narrative tense for background action, in an iterative or habitual function (1), to express unfulfillable wishes and metaphorical comparisons.

  1. այլ թագաւորն Վաղէս նեղէր զքրիսոնեիցն ժողովուրդն
    ayl t‘agaworn Vałēs nełēr zk‘ristoneic‘n žołovurdn
    “But Emperor Valens kept oppressing the Christian community …”

Interestingly, the imperfect is also used for the protasis (if-clause) and apodosis (main clause) in conditionals, both generalising and counterfactuals. As is maybe imaginable, this can be somewhat confusing, especially since (unlike Greek, for instance) there is no modal marker to distinguish the two uses. A differentiation is, accordingly, only possible from context. Compare examples (2) and (3) below.

  1. թէ առնէր ոք հանգիստ աղքատացն մեծ պատիժս կրէր ի թագաւորէն
    t‘ē aṙnēr ok‘ hangist ałk‘atac‘n mec patižs krēr i t‘agaworēn
    “If (=whenever) anyone gave solace to the poor, he received a severe punishment from the king.”
  2. սա թէ մարգարէ ոք էր, ապա գիտէր
    sa t‘ē margarē ok‘ ēr, apa gitēr
    “If he really were a prophet, he would know …”

One of the standard grammars (Jensen 1959) notes that present counterfactuals always employ imperfects, while in past counterfactuals, a pluperfect may be used, but often protasis and/or apodosis simply employ imperfects.
Upon investigating the distribution of the usage of imperfect, pluperfect and other tenses in counterfactual conditional clauses in both the Gospels and a selection of five early Original texts, it turns out that this statement requires some relativisation:

  • in the Gospels and the Original corpus, present counterfactuals behave according to Jensen’s notion.
  • there are 4 occurrences of past counterfactuals in the Gospels
    • 3 using pluperfects only
    • 1 using imperfects only
  • there are 22 occurrences of past counterfactuals in the Original corpus
    • 1 using pluperfects only
    • 15 using non-pluperfects in the apodosis
    • 6 using imperfects only

Although the evidence is extremely limited, it may be speculated that in the Gospels, the tenses used in present and past counterfactuals reflect a differentiation made in the Greek from which they were translated: imperfects are used in both languages for present counterfactuals; for past counterfactuals, Greek uses aorists which Armenian renders as pluperfects.
One clue that makes this speculation somewhat palatable is the one occurrence listed above where this rule seems not to hold true. For in this instance, it turns out that Greek uses  a verb incapable of forming an aorist, with the result that the imperfect has to stand in for the Greek aorist, which is faithfully imitated in the Armenian translation (Mt. 23:30).

The distribution in the Original corpus is rather more complex, and as yet no ready explanation  has been found. The most promising possible explanation, namely that the choice of imperfect or pluperfect should rely on aspect rather than any other consideration, explains most but not all occurrences; other explanations fare even worse.

Two grammars

While more research is evidently needed, the usage of the tenses in Armenian counterfactuals illustrates, on a very small scale, why the composition of corpora is important, particularly if there are known differences in textual history.
With its various extraneous influences, Classical Armenian might very well be said to have (at least) two grammars: that of the Original texts, mostly uninfluenced by Greek; and that of the translation literature. In practical terms, this rarely leads to unbridgeable differences; for linguistic research, however, it is vital to tease the two apart.
One of the next big projects in Armenian historical linguistics therefore ought to be a detailed synchronic grammar (à la Kühner-Gerth) of Original texts; for the moment, though, we have to rely on smaller case studies of individual grammatical patterns.


For a somewhat fuller account of the above discussion, have a look at my recent paper at the Imperfective Modalities in Caucasian Languages (IMMOCAL) workshop at the Historical Linguistics of the Caucasus conference at the EPHE in Paris.

I’m grateful for any thoughts, suggestions, or questions. Leave a comment below, or get in touch via Twitter (@rbnmyr).


References:

Clackson, James P. (1994) The Linguistic Relationship between Armenian and Greek, Oxford/Cambridge, MA: Blackwell.

Hovannisian, Richard G. (ed., 1997) The Armenian People from Ancient to Modern Times, 2 vols., New York: St Martin’s Press.

Hübschmann, Heinrich (1875) “Ueber die stellung des armenischen im kreise der indogermanischen sprachen,” Zeitschrift für vergleichende Sprachforschung auf dem Gebiete der Indogermanischen Sprachen 23 (1), 5–49.

Jensen, H. (1959) Altarmenische Grammatik, Heidelberg: Carl Winter.

Meillet, Antoine (1911–12) “Sur les mots iraniens emptruntés par l’arménien,” Melanges de la Société Linguistique de Paris 17, 252–250.

Meyer, Robin (2013) “Armeno-Iranian Structural Interaction: The Case of Parthian wxd, Armenian ink‘n,” Iran and the Caucasus 17 (4), 401–425.

——— (2016) “Morphosyntactic Alignment, Pattern Replication, and the Classical Armenian Periphrastic Perfect,” in S. W. Jamison et al. (eds), Proceedings of the 26th Annual UCLA Indo-European Conference : Los Angeles, October 24th and 25th, 2014, Bremen: Hempen, 117–133.

——— (2018) “Syntactical Peculiarities of Relative Clauses in the Armenian New Testament,” Revue des Études Arméniennes 38.

Muradyan, Gohar (2012) Grecisms in ancient Armenian, Leuven: Peeters.

Schmitt, Rüdiger (1983) “Iranisches Lehngut im Armenischen,” Revue des Études Arméniennes 17, 73–112.

Do you have a comment?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s