The place of Armenian in the Indo-European language family: the relationship with Greek and Indo-Iranian* (1)
The dialectal position of Armenian has attracted the attention of Armenologists ever since
Heinrich Hübschmann (1875/1877) proved that Armenian does not belong to the Iranian
group of Indo-European languages and should be treated as an independent branch of the
Indo-European family. Mainly under the influence of the centum / satm division, Armenian was considered to be in close relationship with the Aryan and Balto-Slavic languages for a long period, until Pedersen (first in 1906: 442), Meillet and others noted that the number of Greek-Armenian agreements is greater than the number of agreements between Armenian and any other Indo-European language. The relations between Armenian and Greek are sometimes regarded within a larger Balkan context including Phrygian, Thracian and Albanian.
Some scholars argued that there are a large number of similarities between Greek and
Armenian, which allow for the postulation of a common Graeco-Armenian language.1 It is
now clear, especially after Clackson’s (1994) thorough, albeit somewhat hypercritical treatment,that this case is not as strong as it is for Indo-Iranian and Balto-Slavic. The contact relations between Proto-Greek and Proto-Armenian may have been intense, but these similarities are considered insufficient to be viewed as evidence for discrete Proto-Graeco-Armenian.2
There are also connections between Armenian and Indo-Iranian on the one hand, and
between Armenian, Greek and Indo-Iranian on the other. Armenian is usually placed between Indo-Iranian to the east and Greek to the west, and on the northern side it might neighbour Balto-Slavic (and/or Germanic and others). The dialectal unity of Armenian, Greek, (Phrygian,)and Indo-Iranian is often discussed.3
After the well-known large-scale investigations of Porzig (1954) and Solta (1960), in the
period between 1970’s and 1990’s there have been made a number of attempts to study the relations between Indo-European branches by means of statistics: Tischler 1973, Davies / Ross 1977, Bird 1982 (updated 1993), Coleman 1992, etc. With respect to Armenian one has to mention especially the works of Jahukyan, 1980, 1983, and 1987: 86–222. For a critical account of these studies, see Clackson 1994: 7–8, 193–198. For a recent attempt to recover the first-order subgrouping of the Indo-European family using a new computational method, see Ringe /Warnow / Taylor 2002: 102-106.
As far as the morphological and lexical isoglosses are concerned, in these statistical investigations Greek and Indic mostly appear among the closest languages to Armenian. As an example,in the table below I present Jahukyan’s data on the first five language branches displaying the highest number of common features taken from his lists of 27 phonetic, 35 morphological,and 1400 lexical isoglosses.
Rather than discuss here the different theories of the relationships between Armenian and
other language branches and tackle every individual isogloss, which would require a copious monograph, I shall limit myself to a general outline of the most relevant issues regarding Greek and Indo-Iranian. After a short methodological outline (§ 1) and sections on phonological and morphological agreements (§§ 2–3), I shall turn to the main goal of this paper, the lexical material.
Applying the methodology outlined in § 1, I shall select the most illustrative examples from the lists that have been used before and will add some new material that has not been discussed in this context before. Additionally I present a number of new etymologies which are marked as HM.
In cases where I give no references, the relevant etymological material can be found in
HAB and Martirosyan 2010 s.v.
The methodological and thematic background of this paper largely coincides with that of
Clackson 1994, so I simply omit these discussions and refer the reader to this exemplary
A crucial methodological point of departure is that archaic features and independent developments are not significant for determining a close genetic relationship between two languages or dialects. Instead, one should rely on shared innovations from the outset. The drawback with this method is that there is often (if not always) the possibility of independent innovations yielding similar results. Nevertheless, the cumulative evidence decreases the likelihood of chance in such cases.
On the other hand, if a competing feature is present in a certain group of dialects that is
otherwise confirmed by a number of isoglosses, it should be taken into account even if we
cannot formally decide whether we are deling with an archaism or innovation. Thus Armenian,Greek, Phrygian and Indo-Iranian make use of the e-augment (§ 3.1) whereas other dialects do without it, and it is impossible to decide whether the presence or absence of an augment is to be regarded as dialectal innovations made in late Proto-Indo-European. Nevertheless,this is a significant isogloss, because a similar dialectal distribution is found, as we shall see, for a number of morphological and lexical variables.
In the case of, e.g., the genitive ending *os o,
however, some archaic traces are also found in other branches, for example Italic and Celtic (see § 3.2). This is reminiscent of e.g. IE *h2nēr,gen. *h2nr-ós ‘man’ that is basically represented by the dialect area under discussion (Armenian ayr, gen. an, Skt. nár,
Greek ἀνήρ, gen. ἀνδρός, Phryg. αναρ, cf. also Alb. njeri ‘human being, person’) but has also left some traces in Italic (Osc. ner)and Celtic (Mir. ner ‘boar’,
MWelsh ner ‘chief, hero’).4 A similar case is *h2erh3-uer/n- ‘arable land’. Such cases cannot be regarded as significant for the purpose of subgroupping or establishing areal contacts.
When an etymon is only found in two or three non-contiguous dialects, it may theoretically
represent an archaic PIE lexeme that has been lost elsewhere5 and is thus not significant
for our purpose. But when an etymon appears in a few dialects that can be regarded as contiguous at a certain stage, we should take it seriously even if the etymon has no PIE origin and cannot be thus treated as a shared innovation in the genetic sense. Two Indo-European dialects that were spoken in the same geographical area at a period shortly before and/or after the Indo-European dispersal could both develop shared innovations as a result of their interaction with neighbouring non-Indo-European languages.
After the Indo-European dispersal Proto-Armenian would have continued to come into
contact with genetically related Indo-European dialects. Simultaneously, it would certainly
also have been in contact with neighbouring non-Indo-European languages. A word can be of a substrate origin if it is characterized by: (1) limited geographical distribution; (2) unusual phonology and word formation; (3) characteristic semantics.
Theoretically, these two lexical corpora — lexical innovations on an inherited basis and
substrate words — can thus be placed within the same temporal and spatial framework. As
far as the relationship between Armenian and Greek is concerned, matters are particularly
complicated for two principle reasons: (1) it is often very difficult to know whether we are
dealing with an innovation or a substrate / cultural word, and (2) aside to lexical correspondences confined to only Armenian and Greek, there are also a large number of lexical agreements between Armenian, Greek and a few other European dialects. Many of these words belong to the semantic fields of the physical world, fauna, flora, agriculture and crafts. They may, therefore, shed some light on cultural and geographical characteristics of the environment where Proto-Armenian might have contacted Proto-Greek and other dialects in a late period around the time of the Indo-European dispersal. For these reasons, in section six, following the subsection on Armenian and Greek only, I also provide a few characteristic examples reflecting the broader areal context. I then present a brief discussion on the substrate (section 7) and add a summarizing table divided into semantic fields. Wherever a lexical agreement is likely to be an innovation rather than an isolated etymon, I mark it by shading.
2. Phonological isoglosses
2.1. A major and very complicated issue is the centum / satəm division which puts Greek
and Armenian on different sides of the line. Together with Indo-Iranian, Balto-Slavic and Albanian (the situation in Luwian is disputed), Armenian belongs to the satm group of languages which show palatalisation of the palatovelars and absence of a labial element in their reflexes of the labiovelars.6
2.2. Another phonological feature that unifies Indo-Iranian, Balto-Slavic and partly Armenian is the ruki/iurk-rule, the special development of *s after r, k, i, and u.7 It is interesting that Armenian shows a twofold development of *rsas reflected in t‘aršamim and t‘aamim ‘to wither’; the rš-reflex is in line with the ruki-development, whereas
– betrays an assimilation of *rs- to *rr,also seen in Greek. This issue can be placed within the framework of the development *s > h in Armenian, Greek, Iranian, Phrygian and Lycian (and also Brythonic Celtic).
Both developments may have resulted from a common change, although independent innovations are not unlikely either.8
2.3. The so-called “prothetic vowel”, viz. Gr. ἀ- (and ὀ) : Arm. a,and Gr. ἐ- : Arm. e- vs.
zero in other languages, is now interpreted as a vocalized reflex of the PIE initial laryngeal
followed by a consonant (see the table below). It is considered an important isogloss shared by Armenian and Greek, and possibly also Phrygian and Albanian.9 Clackson (1994: 36) notes that this may represent an areal feature since initial laryngeals might also have left vocalic reflexes in the Anatolian languages. However the latter statement seems to be uncertain.10
2.4. Vocative accent. Armenian manuscripts and dialects provide rich evidence for vocative
forms accented on the first syllable: háyrik ‘o father’, máyrik ‘o mother’, Kárapet, etc.; dial:
T‘iflis áxpr ‘o brother’, vúrt‘i ‘o son’; Loi órdi ‘o son’; Hamšen háyr-i ‘o father’, máyri ‘o
mother’; Akn hársnuk ‘o sister-in-law’, márik ‘o mummy’; Moks xrółper ‘o uncle’, t‘ágävur ‘o king’, Nor-Bayazet hárs- ‘o sister-in-law’, Hóomsim, Máyran, Márgarit, tnákolner “you whose house may be destroyed!”, even word combinations, such as úrban hars “you, dear sister-inlaw (to whom I may be sacrificed)”, Hóp‘sma Xat‘un ‘o you, Lady Hop‘sim’.
The vocative with initial accentuation may be considered an Indo-European inheritance.
In Vedic Sanskrit, the vocative, when accented, has the acute on the first syllable, e.g., voc. pítar vs. nom. pita. The same is found in Greek: ἄδελφε vs. ἀδελφός ‘brother’; δέσποτα vs.
δεσπότης ‘master (of the house), lord’; πάτερ vs. πατήρ ‘father’, etc.; in modern Iranian languages:
in Persian, the stress is on the initial syllable of the vocative noun or phrase. In Kurdish
Awroman, when no vocative particle is present the stress is brought forward to the first
syllable of a noun.11 This isogloss is highly hypothetical.
(TO BE CONTINUED)
1 Note the term Helleno-Armenian in Hamp 1979: 4–5; 1983: 6; 1992: 58.
2 For such a moderate approach, see Meillet 1936: 9, 141–143. For more literature and different discussions of this and related issues, see Meillet 1922/1950; Pedersen 1924; Bonfante 1937, 1981; Adjarian 1937, AčaHLPatm 1,1940: 23–99; Makaev 1967; Schmitt 1972–74: 34–40, 64–67; 1975: 27; Širokov 1977, 1980; de Lamberterie 1978–79;
1992: 236–239; Jahukyan 1980; Wyatt 1982; Hamp 1983a; Beekes 2003: 152–153; Ringe / Warnow / Taylor 2002: 102–106. See especially the monographs: Arutjunjan 1983; Clackson 1994. For the relationship of Armenian with Balkan (and Asia Minor) languages, see Jahukyan 1970; 1987: 296–306; Diakonoff 1984: 103–104, 110–112, 120–121, 18417,
188–190; Holst 2009: 49ff; Kortlandt 2010: 4–6, 31–32, 78; de Lamberterie 2012 and 2013; individually: Phrygian (Pedersen 1925: 44–49; Haas 1939; Bonfante 1946; Jahukyan 1968; Orel 1993; Clackson 2008: 124), Albanian (Pedersen 1900; Kortlandt 1980, 1986; Rusakov 1984), Thracian (Kortlandt 2003: 83–87; Beekes 2003: 153). For Urartian names of Balkan origin, see Petrosyan 2002: 179–182 and 2005 (with literature).
3 For various views and discussions, see Meillet 1896: 149–155; 1936: 142; Pedersen 1924: 224–225 = 1982: 307–308; Specht 1935: 29–30, 102–103; 1939: 8, 12–14; AčaHLPatm 1, 1940: 85–86; Thieme 1954: 582–590; Porzig 1954:162–164; Birwé 1956: 6; Solta 1960: 459ff; Schmitt 1967: 259–260; Makaev 1967: 453–455, 461; Xač‘aturova 1973, 1979;Pisani 1979: 210; Euler 1979: 18–23; Jahukyan 1980: 4; Wyatt 1982: 27; Gamkrelidze/ Ivanov 1984, 1: 417–418 = 1995:365; de Lamberterie 1986; Mańczak 1987; Pisowicz 1987; Schmidt 1987; Hamp 1992; Lehmann 1993: 19; Clackson 1994: 201–202; 2008: 124; Mallory/Adams 1997: 29; 2006: 78–79, 109–110, 455; Stempel 2000; Ritter 2006; Schmitt 2007: 22–23; Fortson 2010: 203, 383.
4 On this etymon see de Vaan 2008: 406–407; Matasović 2009: 289; Martirosyan 2010: 61–62; Beekes 2010, 1:103–104.
5 For the problem of determining a PIE word, see e.g. Mallory/Adams 2006: 107–110.
6 See Pedersen 1925: 7, 44–47; Allen 1978; Shields 1981; Gamkrelidze/ Ivanov 1984, 1: 417 = 1995: 365; Clackson 1994: 54–55. It is remarkable that Luwian preserves the original labialized reflexes of labiovelars (for relevant literature,see Szemerényi 1996: 611).
7 See Martirosyan 2010: 709–710; Beekes 2011: 30, 126–127, 137. I do not share the view (see Olsen 2011: 26–27 with lit.) on the final r as a ruki-development in Armenian.
8 For references and a discussion of these two issues see Clackson 1994: 54, 21086; for *s > h, cf. Szemerényi 1985; Schmidt 1988: 602. Note that the change s > h in Lycian and Brythonic Celtic is certainly not a common archaism with Greek, Armenian, and Iranian according to received opinion on the matter (cf. already Meillet 1896:151 on Celtic).
9 For literature and a discussion, see Martirosyan 2010: 714–716; de Lamberterie 2013: 29–34. See also Schmidt 1988: 602. For Phrygian, see Ligorio / Lubotsky forthc. (section 4.3); for Albanian, see Demiraj 1994.
10 For a discussion, see Kloekhorst 2006 and his manuscript monograph on Hittite accentuation, notably the section “Words containing aC()”.
SOURCE Journal of Language Relationship • Вопросы языкового родства • 10 (2013) • Pp. 85—137 • © Martirosyan H., 2013