MEDIAN LANGUAGE. Median was the language of the ancient Medes, who ruled over great parts of the Zagros mountains and Northwestern Iran (presumably in the form of a confederation of smaller city states, see MEDIA), before the Achaemenids. Since no Median texts survived for the period in question, it is not definitely clear, whether the Medes used Median in written form or some other language (e.g., Akkadian) for administrative or economic texts, annals, contracts, and correspondence. Therefore, Median is known only fragmentarily from evidence attested in various other languages, mainly proper names, but also some words, which are explicitly described as “Median”, are embedded in a Median context or are characterized by particular phonological features (see below). In the early period of Iranian studies this factual ignorance caused the arbitrary use of the term “Median” for various other languages, especially for Elamite (thus Oppert, 1879) and Avestan (qq.v.); but by now this case is closed.

On the basis of its phonological features, the Median language belongs to the Northwestern Iranian languages, but it is only one part of this group and should not be misunderstood as a collective term for the whole. Other Northwestern Iranian languages must have already existed in ancient times, as we can deduce from the attested names of tribes and countries and from the Middle Iranian evidence, which is seen in the tri- or bilingual Sasanian inscriptions and in the multilingual Manichaean texts of that time and which is confirmed by the modern Iranian data. Through that material it becomes clear, too, that there existed dialectological differences between (ancient) Median and (medieval) Parthian.

The Medes are, for the first time, mentioned in the Neo-Assyrian Empire on king Šalmaneser III’s so-called Black Obelisk (835 BCE), since under his reign (858-824 BCE) the Assyrians came into contact with Western Iranian peoples and the Medes became their most important opponents, who constantly rebelled. For two centuries Iranian names are mentioned in Assyrian inscriptions (esp. of Šamši-Adad V [823-811 BCE], Tiglath-pileser III [744-727 BCE], and Sargon II [721-705 BCE]), chiefly the names of minor princes (“city rulers”) and the toponyms belonging to them. Etymological interpretations of those anthroponyms often are not more than suggestions of some probability, but there are also secure examples like

Agnuparnu < *Agni-farnah- ‘Having glory by (god) Fire’,

Ašpabara < *Aspa-bāra– ‘Horseman’,

Bag(a)parna < *Baga-farnah– ‘Having glory by the gods’,

Bagdāti < *Baga-dāta– ‘Given by the gods’,

Daiukku < *Dahyuka– (from dahyu– ‘land, country’),

Kaštarītu < Xšaθrita (from xšaθra- ‘kingship, kingdom’; see below),

Mašdaiukku < *Mazdāyuka– (containing the theonym Mazdā-),

Mašdakku < *Mazdaka– (the same),

Mitraku < *Miθraka- (containing the theonym *Miθra-),

Satarpānu < *Xšaθra-pāna– ‘Protecting the kingship/kingdom’ (see below),

Šatašpa < *Satāspa– ‘Possessing hundreds of horses’ or

Šidirparna < *Čiθra-farnah- ‘Having glory by his lineage’ (for all details see Schmitt, 2009, s.vv.). The Median origins of these anthroponyms are evident from chronological and geographical reasons, but to retrieve reliable forms of Median names is only possible through the Achaemenid royal inscriptions of the end of the 6th century BCE.

The Babylonian and (so far as it is preserved) Aramaic versions of Darius I’s great Bisotun inscription (DB; q.v.) show consequently and without exception forms of names, that when containing some Old Persian special development, do not agree with the genuine Old Persian forms (which are reflected also in the Elamite text, however) and therefore must come from another Iranian dialect, for which obviously only the Median language can be considered. Those “Median” forms testify to Median influence on the Mesopotamian scribes’ tradition and must be seen in a wider context, because the same versions reflect also in other cases older Akkadian traditions and not the Old Persian equivalent (e.g., Bab. Ú-ra-áš-ṭu ‘Urartu’, Aram. ʾrrṭ ‘Ararat’ instead of OPers. Armina ‘Armenia’). Those Median forms are *Br̥ziya (Bab. Bar-zi-ia/iá, Aram. brzy) vs. OPers. Br̥diya; *Āθrina (Bab. At-ri-na-ʾ) vs. OPers. Āçina; *Čiθrantaxma (Bab. Ši-it-ra-an-taḫ-ma) vs. OPers. Ciçantaxma; *R̥tavarziya (Bab. Ar-ta-mar-zi-ia, Aram. ʾrtwrzy) vs. OPers. R̥tavardiya; *Suxra (Bab. Su-uḫ-ra-ʾ) vs. OPers. Θuxra, whereas Bab. Za-ʾ-tu-ʾ-a (from Med. *Zāt°) and OPers. Dātavahya are somewhat varying. The only example in some Xerxes inscription (XPa) is the name of the “Gate of All Nations”, OPers. Visa-dahyu– (XPa 12), which is given as Bab. Ú-ʾ-is-pi-da-a-ʾ-i (with initial Med. *Vispa-). From these parallels the following distinctive features of the Median language can be established:

(1) Med. sp vs. OPers. s < PIE. *ḱu̯:

(a) vispa– ‘all’ in Med. *Vispa-dahyu– ‘All-Nations-’ vs. OPers. Visa-dahyu– (see above) and Med. (OPers. loan) vispa-zana– ‘containing all races’ vs. OPers. *visa-dana– (in El. mi-iš-šá-da-na-; see below);

(b) aspa– ‘horse’ in Med. (OPers. loan) uv-aspa– ‘with good horses’ vs. OPers. uv-asa- and the proper names Aspa-canah– and Vištāspa-;

(c) *spaka– ‘dog’ in Gk. (“Med.”) spáka ‘bitch’ (according to Herodotus 1.110.1 “the Medes call the bitch spáka”) vs. OPers. *saka– (unattested; but cf. Mid.Pers. sg /sag/, NPers. sag ‘dog’);

(2) Med. s vs. OPers. θ < PIE. *ḱ:

only Med. *suxra– ‘red’ in *Suxra– (see above) vs. OPers. Θuxra– (this being the voiceless parallel to the voiced series in [3]);

(3) Med. z vs. OPers. d (written for [δ]) < PIE. *ǵ/*ǵ^h:

(a) *br̥zi– ‘high, sublime’ in *Br̥ziya– (see above) vs. OPers. Br̥diya-;

(b) *varz ‘to work’ in *R̥ta-varziya– (see above) vs. OPers. R̥ta-vardiya-;

(c) zana- ‘race’ in Med. (OPers. loan) vispa-zana– (see above) vs. OPers. *visa-dana- (only in El. mi-iš-šá-da-na-), which example with its two distinctive features is especially meaningful;

(4) Med. θr vs. OPers. ç < PIE. *tr:

(a) *čiθra– ‘origin, lineage’ in *Čiθran-taxma– (see above) vs. OPers. ciça– in Ciçantaxma-;

(b) *xšaθra– ‘kingship, kingdom’ in Xšaθr-ita– DBe 6 (see below) vs. OPers. xšaça-;

(c) *ātar-/āθr- ‘fire’ in *Āθr-ina– vs. OPers. Āç-ina- (see above).

On the other hand, the names of Medians mentioned in DB (Fravartiš, Xšaθrita, Uvaxštra, Taxmaspāda) and the toponyms (Sikayuvatiš, Māruš, Kunduruš, Hagmatāna ‘Ecbatana’) and choronyms (Nisāya, Kampanda, Ragā) concerning Media, where the major part of the rebellions described in DB had taken place, are not very revealing in this regard. Only Xšaθrita, a hypocoristic based on Med. *xšaθra-, the counterpart of OPers. xšaça– ‘kingship, kingdom’ (see above), is an additional confirmation of θr/ç; and perhaps the basic component of Sikayuvatiš may belong to OPers. θikā– ‘gravel’ (cf. s-/θ-), if the word- formation could be explained satisfactorily.

Otherwise a number of words attested in the Old Persian royal inscriptions, but in a form differing from the normal Old Persian phonological development, are attributed to Median (first by Hübschmann, p. 116, but systematically only by Meillet and Benveniste, pp. 7-13, secs. 8-16), chiefly because the Persians since Cyrus II (q.v.), having cast off the yoke of Median rule, regarded themselves as heirs to the Medes. Nonetheless Media and the Medes took a prominent position in the Persian Empire, and obviously the Median language, too, as a supplier of loanwords. The reason for choosing such forms is a tendency to archaizing, which becomes evident most strikingly, when the Old Persian text exhibits a foreign, i.e., the Median form (e.g., DNa 10f. vispa-zana– ‘containing all races’, with sp and z), whereas the genuine Old Persian form is found in the Elamite version (DNa El. 8 mi-iš-šá-da-na- = OPers. *visa-dana-, with s and d). Those “Medisms” concern certain typical fields like kingship, military or jurisdiction (cf. Mayrhofer, p. 20). But of crucial significance is the fact that each single attribution of some word or name at any rate is starting from the distinctive features expressly attested as “Median” and has firm grounds in the historical-geographical context, and it must become probable that it is bound in some old or archaic tradition (and is on no account an innovation). That phenomenon is most clearly seen in cases, where both forms, the genuine Old Persian one and the foreign one, occur side by side, as in the example just quoted (OPers. visa– vs. Med. vispa– “all”), in asa– ‘horse’ (DB I 87) vs. aspa– (only in an older compound and in proper names) or in the satrap’s title (OPers. xšaça-pāvan- lit. ‘protecting the kingship/kingdom’ vs. non-Pers. *xšaθra-pā– in Gk. satrápēs, etc. or *xšaθra-pāna– in Bab. aḫ-šá-ad/da-ra-pa-nu, Aram. ḥštrpnʾ, etc., and as a personal name already at the end of the 8th cent. BCE in Ass. Sa-tar-pa/ba-nu; see Schmitt, 1976). Such doublets are reflected also in the Elamite renderings of Iranian terms and names attested in the Persepolis Administrative Archives (q.v.). From the hard core of such corresponding phonetic pairs then one may proceed even further by docking other forms with them (For the method to be applied see Schmitt, 1984, pp. 188-93; 2003, passim).

“Non-Persian,” however, does not mean automatically “Median,” since the same phonetic developments are also found in other Iranian languages, and scholars did not always agree on the extent of Median influence on Old Persian. Particularly restrictive in that regard was Ilya Gershevitch (1964), who for many doublets (esp. s/θ, z/d) reckoned with internal Persian variants and acknowledged as Median features only sp (< PIE. *ḱu̯) and a few others, that are disputed, however, conversely by most other scholars (e.g., f- in farnah- ‘glory’; see below). But often the Median hypothesis has proved to be wrong, because some (mostly etymological) pre-conditions were false or not all relevant criteria were observed. For example, the OPers. title, xšāyaθiya– ‘king’, often was regarded as a loanword, because –θiy– erroneously was based on PIr. *-ti̯-, which regularly gives OPers. –šiy-. But since –θiy– just as well may be derived from PIr. *-θiy– and since the explanation of xšāyaθiya– as a vr̥ddhi-formation with suffix –iya– on the basis of *xšay-aθa- ‘kingship’ is morphologically more plausible, there is no need to think of Median origin here (see Schmitt, 2014, pp. 286f.).

Likewise, the sound change of PIr. *hu̯ > *f, that for a long time had been considered as typically Median, cannot be ascertained as a characteristic trait of that language. The only example for it is the stem OIr. *farnah– ‘glory, happiness, etc.’ (vs. YAv. x^var^ənah-), that is richly attested in proper names and for Median already from the 8th century BCE (first Šataparnû in 737 BCE, Bag(a)parna in 714 BCE, and Parnua in 711 BCE; see Schmitt, 2009, s.vv.). This fact alone, however, is not sufficient to regard it as Median (or belonging to some Median dialect), since the anthroponomastic evidence seems to be rather pan-Iranian than especially Median. On the contrary the initial *f– most likely can be explained through phonetic dissimilation as proposed by P. O. Skjærvø (1983). Moreover, by denying the Median character of that sound change *hu̯ > *f one removes a great difficulty, because the name of the Median king (OPers.) Uvaxštra (which in Greek is Kyaxárēs [see below], that clearly is based on PIr. *Hu̯-) should not have existed with this initial.

Apart from the identification mark Gk. spáka ‘bitch’ scarcely any reliable information about the Median language may be obtained from the Greek sources, mainly for chronological reasons, since such sources date only from the time, when the Medes’ political significance had been surpassed by that of the Persians. Besides, Greek authors do not exactly distinguish between “Median” and “Persian”, and of course any knowledge of Iranian dialects is not to be expected anyway. Therefore only the names of the Median kings may be considered that are mentioned in genealogical sequence by Herodotus and Ctesias (qq.v.) or occasionally by Xenophon (q.v.). In favor of Herodotus’ list, which is diametrically opposed to the one of Ctesias, one can observe that the four names quoted by him may be connected with forms attested in Oriental sources entirely independent of him: Dēïókēs = Ass. Daiukku (the name of a Mannaean governor at Sargon II’s time; see above); Phraórtēs = OPers. Fravartiš (the name of a Median rebel against Darius I from Uvaxštra’s family, who adopted the name Xšaθrita [see above]); Kyaxárēs = OPers. Uvaxštra (from whose family two rebels claimed to be descended); and Astyágēs = Bab. Ištumegu (the name of the Median king, whom Cyrus II defeated decisively).

Also two toponyms, attested in Greek and Latin, have been discussed in this connection, viz. Gk. Aspadána ‘Isfahan’, later Parth. ʿspʾhʾn /Ispāhān/, and Lat. Ganzaca (= Gk. Gázaka) in Media Atropatene (related to Parth. gzn/gazn/, Mid.Pers. gnz /ganz/, Arm. loan ganj ‘treasure’, etc.). The conclusions drawn from them by A. Perikhanian (q.v., 1966, p. 25: prothetic vowel a-) and W.B. Henning (q.v., 1963, pp. 196f.: metathesis *zn > nz) respectively, however, find no support in the older material.

Studies like those by Walther Hinz (q.v.) or Jan Tavernier which attempt to put together the Old Iranian words and names attested in foreign languages (Elamite, Babylonian, Aramaic, etc.), sometimes mark the reconstructions as “Median”, usually only on the basis of phonological features as discussed above. It goes without saying that such attributions to Median (instead of Northwestern Iranian or else) are not always properly well-founded.

Therefore, in short, only a few phonological features of the ancient Median language can be established as well as some words, in which they take shape. Since no texts have been preserved in the Median language, it is not possible, however, to distinguish particular morphological or syntactic features, which could reveal differences from other Iranian languages.

Middle Iranian or modern Iranian idioms descended from the ancient Median language are not attested and cannot be established with certainty. The modern Northwestern Iranian dialects or dialect groups like Āḏari/Azeri (q.v., see Azarbaijan vii.), Tāti and others and part of the Central Iranian dialects in any case belong to the group of the Northwestern Iranian languages. The same has been claimed for Kurdish, Baluchi (qq.v.), and Sivandi, too, without success though, because it is not possible to derive them directly from the ancient Median language by unambiguous characteristic phonological or other developments. In any case it is not sufficient that they show features like *z, *sp or *θr (see above), since these are not exclusively Median, but Northwestern Iranian altogether. One should add that any information on Median dialects in ancient times is lacking. So we must confine ourselves to the remark that all those Northwestern Iranian idioms spoken on the territory, which in Achaemenid or earlier times was called Media, may be termed “Median” only in a geographical, but not in a linguistic sense.

A third Western Middle-Iranian dialect (apart from Middle Persian and Parthian) seems to be required by forms characterized by particular phonological features and attested indirectly by loanwords and names, especially in Armenian. This dialect often is called “Middle Median” ([MMed.] or Atropatenian). The most distinctive developments are OIr. postvocalic *d > Parth. δ, Mid.Pers. y, but MMed. h (e.g., OIr. *spāda– ‘army’ > Parth. ʿspʾd /ispāδ/ [also in Arm. spar-a-pet ‘commander’], Mid.Pers. *spāy [Arm. loan spay], MMed. *spāh [Mid.Pers. ʿspʾh /ispāh/, Arm. loan spah, but not Parth.!]) and OIr. *fr– > Parth., Mid.Pers. fr-, but MMed. hr– (e.g., Parth. fryštg /frēštag/ ‘messenger’, Mid.Pers. frystg /frēstag/, MMed. hr– in Mid.Pers. loan hrystg /hrēstag/, Arm. loan hreštak, but not Parth.!). Of some importance for assuming such Middle Median influence on Armenian are for geographical as well as chronological reasons the Iranian elements (names as zryhr /Zarēhr/) in Aramaic inscriptions found in Armenia and belonging to the reign of king Artašēs I (ARTAXIAS I; 189-160 BCE), when the Parthians (from whom the Armenians borrowed the major part of their vocabulary) were not yet in the northwest of Iran and had not yet any contact with the Armenians (cf. Henning, 1963; Perikhanian, 1966; Perikhanyan, 1993, pp. 4-23). But because any particular feature of the ancient Median language is lacking in that material, those pre-Parthian borrowings of the Armenian language also cannot be ascribed to Median with certainty.