Were the Mitanni Aryans really Indo-Aryans?

Mitanni (or Mittani) is a famous kingdom of the 15th-14th century BCE in northern Mesopotamia, famous especially because of the evident 'Aryan' (Indo-Iranian) nature of the names of its kings and deities. 

The most common theory is that they were more precisely Indo-Aryans, and it is repeated everywhere, although Kammenhüber maintained that the language is still Indo-Iranian and Diakonoff has proposed instead a connection with Dardo-Nuristani people. Mayrhofer himself in his article Indo-Iranisches Sprachgut aus Alalah said that the words used, except aika (see below) are not specific of Indic but are common Indo-Iranian. He remarked that even when they are not attested in Iranic, since the Old Iranian documents are limited, the 'only Indic' character of the word cannot be proved e silentio.
This alleged Indo-Aryan identity has been used to date the Ṛgveda, suggesting that it must be dated at the age of Mitanni or after it, since the Mitanni Indo-Aryan appears as more archaic than the Ṛgvedic language itself. So, let us see what the arguments of this identification are.

The most repeated reasons of the Indo-Aryan identity are s- instead of Iranic h- and aika- for 'one' (in aikawartanna, 'one turn' of the horse) instead of Iranic *aiwa- attested in Avestan and Old Persian. Now, Iranic h- from s- seems to be a late phenomenon, as Mayrhofer has shown from borrowings into Elamite, it is not earlier than the 8th/7th century BC (see here), at least in Western Iran. The change is attested in the Avesta, the date of which is not sure, but it can be after 1000 BCE (in any case it should not go beyond 1300 BCE), and the region is Central Asia, quite far from Mitanni. Old Persian is closer but is attested from the 6th century BCE. About Median we do not know much, but it is interesting that the capital of the Median empire, Ecbatana in Greek, Hamgmatāna- in Old Persian, can be referred as Sagbat/Sagbita in Assyrian texts (see here). Also the theonym Assara Mazaš (for Ahura Mazdā) in an Assyrian list of gods, dated 8th/7th century BC, has revealed a Median form with sibilant (see here). On the other hand, among the names from Nuzi we have Artaḫuma, which can be compared with a name in Elamite document, ir-da-u-ma, interpreted by Hinz 1975 as *ṛtā-humā 'through the right Order lucky', and he cites other names in Elamite or Akkadian starting with uma-, like u-ma-a’-pi-ri-a, interpreted as *humāfrya 'Phoenix-dear?' (because the Humā bird in Persian mythology is a sort of phoenix, associated with good fortune). The Persian term humā is derived from Middle Persian humāy and seems to be related with Avestan humāyā 'beneficial; blessed'. So, maybe Artaḫuma means 'blessed through Truth' and reveals a form with aspirated sibilant. It can be due to a different dialect, like Old Persian.    

As to aika-, it is curiously ignored how also in many Iranic languages there are terms for 'one' derived from that root (see here), for instance Middle Persian ēk, Zaza e:k (see here). Nuristani languages, besides Ashkun ach and a Waigali variant ēk, have forms of ev/ew, therefore the Nuristani connection does not seem so strong.
Also interesting is a remark from Raulwing's article on Kikkuli's treatise on horse training:

HERZFELD describes that in 1929, Riza Schah Pahlavi held horse races near the Afghan border of 3, 5 and maximum 9 rounds of 1.4 km (nearly a mile); as HERZFELD enthusiastically points out, “exactly as in Kikkuli’s and Haosravah’s times, and the horses were not tired at all”. 

So, it seems that in Iran the tradition of an odd number of rounds for horses until nine, as in Kikkuli's description, was still alive in the 20th century.

Of course, another strong argument for the Indo-Aryan identity is the mention of 'Vedic' gods in the treaty between the Hittite king Suppiluliuma and the Mitanni king Šattiwaza (here a translation): Mitra, Varuṇa, Indra and Nāsatya.  Let us see how is the reading of the cuneiform texts in the two versions of the treaty, one for Mitanni and one for the Hittites (from Fournet 2010):

DINGIR.MEŠMi-it-ra-aš-ši-il DINGIR.MEŠA-ru-na-aš-ši-il DINGIRIn-da-
ra DINGIR.MEŠNa-ša-at-ti-ya-an-na

DINGIR.MEŠMi-it-ra-aš-ši-il/-el DINGIR.MEŠÚ-ru-wa-na-aš-ši-il
DINGIRIn-tar DINGIR.MEŠNa-ša-a[t-ti-ya-a]n-na 

So, the name Mitra is clear, Indra is rather Indar, and Nāsatya is Našattiya with short a. These are minor differences but we cannot speak of exact identity with the Vedic forms. And all these three deities are present also in the Avesta: Mithra is one of the main yazata or worshipped deities, Indra and Nåŋhaiθiia had become important daēuuas ('pagan' gods, therefore demons), mentioned in the Vīdēvdād/Vendidad (see here). Also Mayrhofer, in the aforementioned article, spoke of 'gods of the Vedic and pre-Zoroastrian pantheon'.
We can also notice that before Mitra there is here the Sumerian sign DINGIR.MEŠ for a plurality of gods. The suffix -ššil according to the hypothesis of Fournet is a dual (as in Vedic mitrāvaruṇau), but there is no confirmation of this, and we would expect the same also for the Nāsatyas who are typically a divine couple. Diakonoff observes that -šši- is the morpheme for abstract words (e.g. šarrašši 'kingship'), and -l for the plural. He translates 'the mitraic gods' (belonging to sphere of the god of light). According to Giorgieri, the suffix -šše/i- indicates also concrete objects and nomina actionis (when added to verbal roots).
If we consider that mithra- in Avestan means "a contract, promise, agreement, friendship, commitment", and the god Mitra is clearly connected with agreements both in the Iranian world and in the Vedas, maybe we have to do here with the 'gods of agreements', which would be fine for a treaty. They can also be the spies of Mitra mentioned in the Mihr Yašt of the Avesta (see here).

And what about Aruna or Uruwana? Also these have the sign of plural gods, and the form is clearly not the same as Vedic Varuṇa: the identification has been done only because it is after Mitra, with whom Varuṇa is typically associated in the Ṛgveda. It is interesting that in Hittite texts, aruna- is the sea, and is often listed among treaty witnesses and conceived as a male deity in myths of Hurrian origin (see here). So, maybe the same deity (in the plural 'the gods of the sea(s)') is invoked here as witness? Or maybe it was confused with the Hittite Aruna for that reason. Now, also the Indian Varuṇa in post-Vedic tradition is the god of the sea, but this is not his main role in the Ṛgveda, and it would be strange to put it besides Mitra as such. Hittite aruna- has no clear etymology, but the Hittite etymological dictionary compares Vedic arṇa 'wave, flood, stream', arṇava 'wave, flood, foaming sea', arṇas 'wave, flood, stream, sea, ocean'. The Indo-European root should be *ar-, expressing motion (of waves).

In order to understand more, let us consider the equivalent (in the other list) Uruwana, which is quite different. Diakonoff denied the identification with Varuṇa and proposed to read Avestan urwa(n) 'soul', 'soul of the dead', so that Uruwanaššil would be 'gods of the sphere of the souls of the dead'.
Fournet, instead, sees in Uruwana a Hurrian version (since initial r- was not admitted in Hurrian) of *Ruwana < *(s)rouaHno, meaning 'the one in charge of flowing waters'. And Varuṇa, being the creator of the water-blocking Vṛtra, would be a metathesis of an older *Ruwana. This etymology is not convincing, but we can notice a parallel phenomenon in Hurrian urukmannu, a metallic part of the shield, if it really comes from rukma-, that in Vedic indicates a golden or silver plate (from the root ruc- 'to shine'). 
Now, in Avestan addition of u- is regular before ru- and rw- (see here), like av. uruuata 'rule, order, prescription' from *rwata, which is a metathesis of *wrata, since we can compare skr. vrata 'rule, religious vow'. In Avestan there is also uruuaiti 'verbal promise, treaty'.

In Greek we have ρητός (<*wrētos) 'stated, covenanted, agreed', Cypriot ϝρήτᾱ 'agreement, treaty, law, pronunciation', Elean ϝράτρα 'verbal agreement, covenant'.
The concept of vrata is strictly connected with Varuṇa in the Ṛgveda. Jamison and Brereton, in their recent translation of the hymns, even state that his name is related to vratá “commandment” and therefore is the god of commandments. But if the root of vrata is that of speaking and declaring (cf. here), the root of Varuṇa is rather the same as the Sanskrit verb vṛ- 'to cover, screen, veil, conceal, hide, surround, obstruct; to ward off, check, keep back, prevent, hinder, restrain'. One could connect this root with the fact that the Vedic Varuṇa is the god who by his snares (pāśa) punishes and restrains those who transgress his commandments. But another interpretation is that he is the covering or all-encompassing Sky itself, a supreme celestial deity that observes everything, especially the deeds of men.
There are good parallels in Vedic varūtṛ 'one who wards off or protects, protector, defender, guardian deity', and Greek erymai 'to keep off, protect, save'. Varuṇa was the guardian of the cosmic order, called ṛta ('truth'). 
We can do another comparison with Avestan urvant 'seizing' (said of a bird), from the participle *wr-ant- according to Bartholomae: the original root can be different (PIE *wal), but the phonetic evolution would be the same. 
So, maybe the root *wṛ- had given a variant form *wrana that became *rwana and then Urwana. By the way, it is interesting that both *wrana and Urwana are close to Greek ouranos, the name of the sky and of the heavenly god that has not a clear etymology. 
Thus Uruwanaššil can refer to the 'gods who are guardians of verbal agreements or sacred commandments', or to the spies of Varuṇa that are often mentioned in the Ṛgveda.

But let us proceed with the numerous Indo-Iranian names that we find in Mitanni and in the Near East, among the rulers of various cities of Mesopotamia and Palestine, and the charioteers called in Hurrian maryanni. First of all, I would like to cite a name present in a list of names from Nuzi that is not cited in the list given by O' Callaghan in Aram Naharaim. It is given in the forms a-ri-ia, a-a-ri-ia, a-ri-i-ia. There are at least 11 persons with this name, two of them are charioteers. In Alalah, there are 24 a-ri-ia; one of them is also called a-ri-ia-an. Another is a maryanni with chariot, and another is brother of Irteya, a clearly Aryan name (see below).
The name clearly corresponds to Sanskrit ārya, Pāli ariya, Avestan airiia, Old Persian ariya. In Iranic use, it indicates the Iranic people itself, according to Herodotus 7.62 also the Medes were originally called Arioi. In India it was more a social and ethical term, 'noble, freeman' (see this post). We cannot say what was the meaning in Nuzi, but we can suppose that it meant 'member of the Aryan nobility'. A name A-ri-ia is found also among the Median city-lords in Assyrian sources (see here).
Now, I would like to remark some features of Indo-Iranian words in these names that are not Indic but rather closer to Iranic:

Arta-: many Mitanni Aryan names start with Arta- 'Truth', for instance Artadāma, Artaššumara, Artamanya. Now, arta- is typical of Old Persian and Parthian names, and it was found also in Median (see here). In Vedic, the form is Ṛta-, that does not give a single name in the Vedas: we find only the patronymic Ārtabhāga in BĀU 3.2.1.13 and matronymic ārtabhāgīputra in BĀU 6.5.2.

Moreover, there is a Mitanni name Artaya/Arteya (frequent in Nuzi and present in Alalah), that recalls the old name of the Persians, Artaioi, according to Herodotus 7.61. According to Hesychius, instead, artaioi means 'heroes' among the Persians. Parallels in Elamite documents are Irdaya, Irteya (Hinz 1975). The same form Irteya is very frequent in proper names from Alalah.

-aššura: in some names, we find this element, e.g. Be-ta-aš-šu-ra, Bi-ri-aš-šu-ra, Kal-ma-aš-šu-ra, Ša-i-ma-aš-šu-ra, Šú-na-aš-šú-ra (king of Kizzuwatna). The Assyrian national deity Aššur does not have the final -a, and is normally at the beginning of names. Moreover, Bi-ri-aš-šu-ra is clearly a Mitanni Aryan name (<*Priyāsura), and we have already seen that Assara is used in an Assyrian list of gods before Mazaš, showing a similar form with double sibilant.
However, it is possible that *asura here is not the god, but has the basic meaning of 'lord' applied to human beings, attested also in the Avesta. Bi-ri-aš-šu-ra, for instance, might mean 'dear lord'. The name Ša-i-ma-aš-šu-ra (variant Ša-mi-aš-šu-ra) from Nuzi is interesting, because the first element *šaima- can be compared with the Old Persian *çaima- 'superiority' found in the Elamite U-še-ma, interpreted as *hu-çēma- 'having a splendid superiority' (Tavernier 2007). The Indic form is śreman. So, Ša-i-ma-aš-šu-ra can be interpreted as 'lord of/with distinction'. O'Callaghan observes that he was owner of a horse, and he was father of *Biriazana, a clearly Aryan name that apparently has the word zana, the Avestan, Old Persian and Median equivalent of Sanskrit jana 'person, race, people' (priyajana in Sanskrit means 'dear person').

-atti: Biryatti (bi˗ri˗at˗ti, bi˗ri˗a˗at˗ti), corresponding to Indic priyātithi 'loving guests, hospitable'. The Avestan form is asti, and Mayrhofer reconstructs the Indo-Iranian form as *atHthi. The addition of -i- instead of the laryngeal or schwa between two consonants is typical of Indic, but here there is no trace, as in the other numerous compounds with this element found in Mitanni Aryan names (Aššuratti, Deuwatti, Indaratti, Mittaratti, etc.). It is interesting that the form is also not the same as in Avestan, but rather like the original Indo-Iranian, without aspiration.

Bard-, Birid-: Bartašwa (Nuzi), Biridašwa (Yanuamma) are two names that can be variant of the same one, that has been compared with Sanskrit Bṛhadaśva, meaning 'having high horses'. But bṛhad (from *bhṛǵh-ant) is very different from Birid-. On the other hand, in Avestan we have barǝz-, bǝrǝz(i)- 'high', and in Old Persian the name Bardiya, written B-r-di-i-y-, Bìr-ti-ia in an Elamite document (Tavernier, p.14) comparable with Biridiya found in Megiddo. In the Old Iranian names attested in non-Iranian texts (see here) we have a reconstructed element bṛdi-, including *Brdi-aspa-, "having a tall horse" (Elamite Bir-ti-is-ba, Middle Persian Burjasp).

-mašda: a name is bi-ir-ya-ma-aš-da, interpreted as corresponding to the Vedic Priyamedha (from *Priyamazdha). But we can see that the form mašda is closer to Avestan and Old Persian mazda 'wisdom'. Names with frya- (<*priya-) are common in Iranic, written pi-ri-ia in Babylonian sources (see here). In Khotanese, an Eastern Iranic language, we have even a form with voiced plosive, bria. It is also remarkable that, as noted by E. Hopkins, compounds with priya- are known in the Ṛgveda only in books VIII, I, IX and X, that are the latest ones: we can suspect an Iranic influence (cf. Talageri 2008, The Rigveda and the Avesta. The final evidence, pp.175-183).

mišta-: Mayrhofer connected mištannu 'bounty' with Vedic mīḍhá- “booty, price for fight”, but Avestan mižda, Middle Persian mizd 'wage, price' are closer.

-myašda: a name from the Amarna letters, zi-ir-dam-ia-aš-da, has been interpreted as zṛda-myazda 'one who makes an offering of the heart' (O'Callaghan). This form is very Iranic, in Avestan heart is zǝrǝd-, in Parthian zyrd (in Indic hṛd-). We can compare from the already cited list of Iranic names (Tavernier 2007): "*Zṛdiyavauš (Median) 'with a good heart'. Babylonian: Si-ri-di-a-muš."
The sound z in Akkadian cuneiform should be an affricate, while Iranic z is a voiced sibilant, but it is also considered a voiced sound (see here), therefore quite close. On the other hand, Babylonian s is considered a voiceless affricate. We can also wonder if we have to do here with a form with voiceless palatal, comparable with the Indo-European root *ḱrd-.  
There is also a curiously similar name *Irdamiasda, written in Elamite Ir-da-mi-ia-is-da, that has been interpreted as *Ṛta-myazda- "ritual banquet". In fact, Avestan miiazda is a sacred offering of food. In the Ṛgveda we have the form miyedha with the same meaning.

panza-: in the list of the number of rounds with the horses, we have pa-an-za-wa-ar-ta-an-na, clearly corresponding to Indo-Iranian pañca 'five'. If z is really a voiced sound, it is interesting that in many Iranic languages the form of the number five has also a voiced affricate or fricative: Persian panj, Parthian panǰ, Bactrian panzo, Pashto pinźë, Kurdish penj.

-ta(h)ma: the famous Mitanni king's name Artatama is written in the Amarna letters ar-ta-ta-a-ma(š) (see here). It has been interpreted through -dhāman as in the Vedic ṛtadhāman 'whose abode is truth or divine law', an epithet of Agni found twice in the late Vājasaneyi Saṃhitā. As a name it is used in the Viṣṇu Purāṇa for the 13th Manu. So, it is difficult to consider it as an ancient Indo-Iranian name, I would propose another interpretation, through an Old Persian *Artata(h)ma 'brave through Truth'. A corresponding Median name *Ṛtataxma is attested in the Elamite form Ir-da-tak-ma and similar. Moreover, many names with Old Persian tahma- are written tam-ma- in Elamite cuneiform.

-tarna: Kosyan has found in a cuneiform Hittite text datable to the kingdom of Suppiluliuma and related to the eastern kingdom of Išuwa, the name U-it-tar-na-, identifiable with the Iranian Vidarna, written mi-tar-na in Elamite. Kosyan quotes the translation 'support, security' by R. Kent and 'der auseinander reißt' ('that tears into pieces') by W. Hinz, from *wi-dṛna, participle from dar- 'to support' or dar- 'to tear apart'. The case is interesting not only because it suggests the existence of an Iranic name in the Hurrian kingdom of Išuwa, which had relations with Mitanni, but also because it is suitable to a comparison with the names of the Mitanni kings Šuttarna and Parattarna. Šuttarna was connected with a hypothetical skr. *sudharaṇa 'supporting well', but we see that the Iranic form that we find in Vidarna is closer. In Avestan we also have darǝna 'fortress, residence'.

wašanna: in Kikkuli's treatise, we find wa-ša-an-na-ša-ia translated "of the training area (for chariots)' and interpreted as coming from Indo-Iranian *waźhanasya. In Indic, this root *waǵh- becomes vah-, in Avestan vaz-, giving vaza- 'driving', vazō-raθa 'driving a chariot', vāza- 'draft animal'. There is also an Avestan name važāspa 'having draft horses', and the noun vāša- 'vehicle', but this must come from *wart-, according to the typical Avestan change rt>š that is not proper to Mitanni Aryan, considering the terms with -wartanna. As observed by Gamkrelidze and Ivanov (1995: 466): "It is interesting that several of the attested Mitannian Aryan words are semantically closer to their Iranian cognates than to the Indic ones: e.g. wasanna and Sogd. -wzn; -wartanna and Oss. aewwaerdyn." P.468: "Ossetic also preserves the ancient horse-breeding term aewwaerdyn 'horse training' (cf. above on the Mitannian Aryan cognate -wartanna)." P.627: "Mitannian Aryan wasanna 'racecourse', Sogd. 'nxr-wzn 'circle of the zodiac; path of stars' [...]"

Yama: Yama and Yamiuta 'helped by Yama?' (dynast of Guddashuna) are names attested in the Amarna letters, a Yamibanda 'servant or follower of Yama?' was prince of Taanach. In Alalah we have one Yaman (Ia-ma-an). A name Yama(ka) is found also in the cuneiform documents of the Achaemenid period. About Yamibanda, bandaka in Old Persian means 'vassal, follower', in modern Persian banda means servant, and among the Iranic names, we have *Bagabanda 'who serves or supports Baga', written Ba-ka-ban-da in Elamite. The form Yami- might derive from a reduction of the last vowel (the opposite of Avestan Yima, cf. also Persian Jam, Jamshīd), while a feminine Yamī like the Vedic sister of Yama seems unlikely. A cult of Yama as god is not only in India as king of the dead, but also among the Kalash (in the past also the Nuristanis), for whom Imra/Imro (Yama-rāja) is the creator god.

Yašdata: this is the name of a prince from Palestine, written ia-aš-da-ta in Amarna letters. O'Callaghan cites an intepretation from Avestan *yaza-dāta 'given by the sacrifice', but this is not attested. However, a comparable name is found among the Iranic names, written Ia-iš-da-da, interpreted by Hinz and Koch as an Old Persian *yazdāta, and by Tavernier as an extension with the suffix -āta from yašta 'consecrated', a name apparently attested in various forms in Elamite documents. Another similar word is Avestan yaoždāta 'purified', although the first element comes from *yauš 'health, vital force', so we would expect rather *yaušdata.

-zana: among Nuzi names, there is Aššuzana that has been derived from Indo-Iranian *ašwa-canas, that has no attestation in Indic, while there is an Iranic Aspačanā (nom.) 'delighting in horses', attested in Elamite document as Aš-ba-za-na.

Besides the names, there are also the epithets for horses from Nuzi pinkara-nnu and paritta-nnu, that have been compared with Vedic piṅgala 'reddish-brown, tawny, yellow' and palita 'grey': it is interesting that in both cases instead of Indic l we have an r, which is the rule in Old Iranian languages.
Finally, we can also remark that there is no trace of Indo-Aryan aspirated consonants in Mitanni Aryan words, although a way of writing it in cuneiform script could be that found, for instance, in the Hittite name Tudḫaliya.

Thus, we can see that not only many elements have a form closer to Old Persian, Median or Avestan, but also parallels in the Old Persian names attested in cuneiform documents. Forms like bard/birid- and -tama suggest affinity with Old Persian, while others like zird- with Median, Parthian and Avestan. We can suppose that at least two different dialects were present among the Mitanni Aryans, possibly a stage of Persian and Median, being Medes and Persians the most ancient Iranic peoples mentioned in Assyrian sources from the 9th century BC and acting in Western Iran in historical times. However, it is also possible that the people of Mitanni became the Matieni or Matiani mentioned by Herodotus and other ancient historians, placed close to lake Urmia, called 'lacus Matianus' in Latin, in Northwest Iran. The name could be a simple metathesis of the ancient form Maitani or Maiteni found in Nuzi, and we can suppose that the Mitanni Aryans came from that region (where the important site of Hasanlu was found) and a part of them always remained there, also after the end of the Mitanni kingdom. As Diakonoff says (see here): "The term Mittanne, older Maitene, is the name of the dynasty; the country itself was called Hanigalbat. The dynastic name is perhaps connected with Herodotus' Matieni, a tribe living somewhere in the neighbourhood of Lake Urmiyah—a likely place for the origin of the dynasty, and as a place of contacts between Hurrians and Indo-Iranians." In this case, they would be a particular Aryan tribe speaking an archaic dialect with specific affinities with other Iranian languages.

 
Giacomo Benedetti

24/5/2017 (with later additions)