AZERBAIJAN vii. The Iranian Language of Azerbaijan



vii. The Iranian Language of Azerbaijan

Āḏarī (Ar. al-āḏarīya) was the Iranian language of Azerbaijan before the spread of the Turkish language, commonly called Azeri, in the region. The currency of Āḏarī in Azerbaijan during the first centuries of the Islamic period is attested by contemporary sources. The earliest reference to Āḏarī is the statement by Ebn al-Moqaffaʿ (d. 142/759), quoted by Ebn al-Nadīm (Fehrest, p. 13), to the effect that the language of Azerbaijan was Fahlawī (al-fahlawīya) “pertaining to Fahla,” and that Fahla was the region comprised of Isfahan, Ray, Hamadān, Māh Nahāvand, and Azerbaijan. A similar statement, on the authority of Ḥamza Eṣfahānī, and obviously deriving from the same source, occurs in Yāqūt’s Moʿjam al-boldān (III, p. 925, s.v. “Fahlaw”), and also in Ḵᵛārazmī’s Mafātīḥ al-ʿolūm (ed. van Vloten, pp. 116-17).

Next to Ebn al-Moqaffaʿ’s the oldest reference to Āḏarī, though no name is given the language, occurs in Balāḏorī’s Fotūḥ al-boldān (p. 328; cf. Qazvīnī, Bīst maqāla I, p. 145), composed in 255/869. He quotes the word ḥān, meaning “house” or “caravanserai” (Ar. ḥāʾer), as belonging to the “language of the people of Azerbaijan.” (This word shows the development in Āḏarī of Middle Iranian x to h, see below.) The oldest mention of the specific term Āḏarī occurs in Yaʿqūbī’s Ketāb al-boldān, composed in 276/891, p. 272; the population of Azerbaijan is described here as a mixture of Iranian Āḏarī (al-ʿajam al-āḏarīya) and old Jāvedānis (al-jāwedānīya al-qedam). By these terms he apparently means the Muslim Azerbaijanis and the Ḵorramdīnis or Jāvedānis, the followers of Jāvedān and Bābak, the neo-Mazdakite leaders who had held sway in Azerbaijan under al-Maʾmūn. It thus appears that the term Āḏarī was applied to both the population of Azerbaijan and their language.

The next testimony is the statement by Masʿūdī (d. 345/956) which points to the original unity of the language of the Iranians and its later differentiation into separate languages, such as Fahlawī, Darī, and Āḏarī—obviously the most prominent Iranian dialects in his estimation (Tanbīh, p. 78). Next we have the statement of Ebn Ḥawqal (d. ca. 981 /371 ) that “the language of the people of Azerbaijan and most of the people of Armenia (sic; he probably means the Iranian Armenia) is Iranian (al-fāresīya), which binds them together, while Arabic is also used among them; among those who speak al-fāresīya (here he seemingly means Persian, spoken by the elite of the urban population), there are few who do not understand Arabic; and some merchants and landowners are even adept in it” (p. 348). Despite the exaggeration concerning the spread of Iranian languages into Armenia and the currency of Arabic in Azerbaijan, the statement clearly attests to the fact that the language of Azerbaijan in the 4th/10th century was Iranian. Moqaddasī (d. late 4th/10th cent.) also affirms that the language of Azerbaijan was Iranian (al-ʿajamīya), saying that it was partly Darī and partly “convoluted (monqaleq)”; he means no doubt to distinguish between the administrative lingua franca, i.e., Darī Persian, and the local dialects (Aḥsan al-taqāsīm, p. 259). Further he says that the language of the Azerbaijanis “is not pretty . . . but their Persian is intelligible, and in articulation (fi’l-ḥorūf) it is similar to the Persian of Khorasan” (p. 378). Again he must mean Darī Persian, which then, as now, must have been current in the urban centers of Azerbaijan.

An anecdote preserved by Samʿānī (Ansāb, s.v. Tanūḵī) concerning Abū Zakarīyā Kāteb Tabrīzī (d. 502/1109) and his teacher Abu’l-ʿAlāʾ Maʿarrī refers again to the vernacular of Azerbaijan in the 5th/12th century. While Kāteb Tabrīzī was in Maʿarrat al-Noʿmān in Syria, he met a fellow-countryman and conversed with him in a language which Abu’l-ʿAlāʾ could not understand. When Abu’l-ʿAlāʾ asked him to identify the language, Kāteb told him it was the language of the people of Azerbaijan (read al-āḏarīya in the Hyderabad ed., III, p. 93; and al-aḏarbījīya [unpointed] in the Leiden ed.; cf. A. Kasravī, Āḏarī, p. 13 n. 1). The statement of Yāqūt (d. 626/1229) to the effect that “The people of Azerbaijan have a language which they call al-āḏarīya, and it is intelligible only to themselves” (Moʿjam al-boldān I, p. 172) makes it clear that Āḏarī was still current in Azerbaijan on the eve of the Mongol invasion.

From Zakarīyā b. Moḥammad Qazvīnī’s report in Āṯār al-belād, composed in 674/1275, that “no town has escaped being taken over by the Turks except Tabrīz” (Beirut ed., 1960, p. 339) one may infer that at least Tabrīz had remained aloof from the influence of Turkish until the time of Abaqa Ḥamdallāh Mostawfī writing in the 740/1340s calls the language of Marāḡa “modified Pahlavi” (pahlavī-e moḡayyar, as in Dabīrsīāqī’s reading, Nozhat al-qolūb, Tehran, 1336 Š./1957, p. 100; the reading pahlavī-e moʿarrab “arabicized Pahlavi” in Le Strange’s edition, p. 87, is not likely). Mostawfī also calls (ibid., p. 62) the language of Zanjān “straight Pahlavi” (pahlavī-e rāst) and the language of the Goštasfī province on the western side of the Caspian (i.e., north of the Persian Ṭāleš and south of Šīrvān) a Pahlavi close to the language of Gīlān (ibid., p. 92). By Pahlavi he, like Ebn al-Moqaffaʿ, obviously means in a general way the vernacular of northwestern and central Iran (an area coinciding with ancient Media). This language, however, was not, contrary to Marquart’s view (Markwart, Ērānšahr, p. 132 n. 5) the same as Parthian, as is evident from the written remains and surviving dialects of Āḏarī (see below).

These various testimonies, in spite of their being occasionally imprecise and uncritical, indicate that the population of Azerbaijan spoke a major Iranian language, termed Āḏarī after the name of the region. It formed a group with the dialects of Ray, Hamadān, and Isfahan and remained the prevalent language of Azerbaijan until the 8th/14th century and probably for some time thereafter.

The spread of Turkish in Azerbaijan.

The gradual weakening of Āḏarī began with the penetration of the Persian Azerbaijan by speakers of Turkish. The first of these entered the region in the time of Maḥmūd of Ḡazna (Ebn al-Aṯīr [repr.], IX, pp. 383ff.). But it was in the Saljuq period that Turkish tribes began to migrate to Azerbaijan in considerable numbers and settle there (A. Kasravī, Šahrīārān-e gomnām, Tehran, 1335 Š./1956, III, pp. 43ff., and idem, Āḏarī, pp. 18-25). The Turkic population continued to grow under the Ildegozid atabegs of Azerbaijan (531-622/1136-1225), but more particularly under the Mongol il-khans (654-750/1256-1349), the majority of whose soldiery was of Turkic stock and who made Azerbaijan their political center. The almost continuous warfare and turbulence which reigned in Azerbaijan for about 150 years, between the collapse of the Il-khanids and the rise of the Safavids, attracted yet more Turkic military elements to the area. In this period, under the Qara Qoyunlū and Āq Qoyunlū Turkmen (780-874/1378-1469 and 874-908/1469-1502 respectively), Āḏarī lost ground at a faster pace than before, so that even the Safavids, originally an Iranian-speaking clan (as evidenced by the quatrains of Shaikh Ṣafī-al-dīn, their eponymous ancestor, and by his biography), became Turkified and adopted Turkish as their vernacular.

The Safavid rule (905-1135/1499-1722), which was initially based on the support of Turkish tribes and the continued backing and influence of the Qezelbāš even after the regime had achieved a broader base, helped further the spread of Turkish at the detriment of Āḏarī, which receded and ceased to be used, at least in the major urban centers, and Turkish was gradually recognized as the language of Azerbaijan. Consequently the term Āḏarī, or more commonly Azeri, came to be applied by some Turkish authors and, following them, some Western orientalists, to the Turkish of Azerbaijan (see EI1-2, s.v. “Ādharī”).

Āḏarī survivals.

These are of three kinds: (1) words, phrases, poems, and scattered verses, recorded in various written sources; (2) the present-day dialects which continue Āḏarī, spoken mainly on the periphery of Azerbaijan to the south and southeast, but also in isolated pockets in the north and the center; and (3) vocabulary borrowed from Āḏarī into the Turkish of Azerbaijan. The credit for first bringing together a collection of Āḏarī survivals belongs to Aḥmad Kasravī (d. 1324 Š./1946; see Āḏarī yā zabān-e bāstān-e Āḏarbāygān, Tehran, 1304 Š./1925). He also sketched the Āḏarī background and a history of the gradual spread of Turkish in Azerbaijan. Although his linguistic observations and methods can not always be supported, his general conclusions were essentially valid and dispelled a widespread notion that no information was available on the original language of Azerbaijan beyond Turkish. (See the reflection of his research in İslâm Ansiklopedisi, s.v. “Âzerî,” where Âzerî-Fârisî lehcesi “Iranian Azeri dialect” is distinguished from Âzerî-Türk lehcesi “Turkish Azeri dialect”.) Later, other Āḏarī survivals were detected.

1. Āḏarī in written sources. These include the following: (1) A sentence in “the language of Tabrīz” in Ḥamdallāh Mostawfī’s Nozhat al-qolūb (ed. Dabīrsīāqī, p. 98). (2) A sentence in the “Tabrīzī” language and two sentences attributed to Shaikh Ṣafī-al-dīn of Ardabīl, two double distichs (dobaytīs) probably by him, another dobaytī apparently in the language of Ardabīl, and one in the language of Ḵalḵāl, all of these in the Ṣafwat al-ṣafā of Ebn Bazzāz, a contemporary of Shaikh Ṣadr-al-dīn, the son of Shaikh Ṣafī-al-dīn, and therefore of the 8th/14th century (Bombay ed., 1329/1911, pp. 25, 107, 191, 220). (3) Eleven double dobaytīs by Shaikh Ṣafī-al-dīn, and therefore apparently in the language of Ardabīl, in the Selselat al-nasab-e Ṣafawīya of Shaikh Ḥosayn, a descendant of Shaikh Zāhed Gīlānī, the mentor (morād) of Shaikh Ṣafī-al-dīn (Berlin, 1343/1924-25, pp. 29-33). (4) A macaronic ḡazal by Homām Tabrīzī (d. 714/1314) in Persian and a local language which must be that of Tabrīz (see M. Moḥīṭ Ṭabāṭabāʾī, “Dar pīrāmūn-e zabān-e fārsī,” Majalla-ye āmūzeš o parvareš 8/ 10, 1317 Š./1938, p. 10; M. Ḥ. Adīb Ṭūsī, NDA Tabrīz 7/3, 1334 Š./1955, pp. 260-62). This specimen differs, however, from the sentence in Tabrīzī given by Ebn Bazzāz with respect to one important phonological feature: In Homām’s poem, the enclitic pronoun of the second person singular is -t, while in Ebn Bazzāz’s sentence it is -r (see below). (5) Two anonymous qaṣīdas in a manuscript written in 730/1329-30 and preserved in the Aya Sofia library in Istanbul (see Adīb Ṭūsī, ibid., 10/4, 1337 Š./1958, pp. 367-417); the dialect of these, judging from their phonology and some of the vocabulary which can be read with certainty appears to belong to the north-central Persian Azerbaijan, probably the Tabrīz-Marand region (see below). (6) One ḡazal and thirteen dobaytīs by Maḡrebī Tabrīzī (d. ca. 809/1406-07; see Adīb Ṭūsī, ibid., 8/12, 1335 Š./1956, pp. 121-27). (7) A text probably by Māmā ʿEṣmat, a mystical woman-poet of Tabrīz (d. 9th/15th cent.), which occurs in a manuscript, preserved in Turkey, concerning the shrines of saints in Tabrīz (see M. Nawwābī, ibid., 7/1, 1334 Š./1955, pp. 41-44; cf. Adīb Ṭūsī, “Fahlawīyāt-e Māmā ʿEṣmat wa Kašf-ī be-zabān-e āḏarī-eṣṭelāḥ-e rāžī yā šahrī,” NDA Tabrīz 8/3, 1335 Š./1957, pp. 242-57). (8) Three poems in the dialects of Ḵamsa and Qazvīn, quoted by Ḥamdallāh Mostawfī in Nozhat al-qolūb which, although not belonging to Azerbaijan in the narrow sense of the term, should be grouped with the other remnants of Āḏarī in accord with the classification of the modern Iranian dialects of the Qazvīn and Zanjān areas. These poems consist of a dobaytī by Abu’l-Majīd Bāygānī in the dialect of an environ of Qazvīn; two dobaytīs by Jūlāha of Abhar, apparently a contemporary of Mostawfī, in the dialect of Abhar, a town in Ḵamsa, and a fragment of nine dobaytīs, by a certain Uyanj or Utanj, in the dialect of Zanjān. The text of all three is extremely corrupt (E. G. Browne, JRAS, 1900, pp. 738-41). (9) Two dobaytīs by Kašfī, a ḡazal and seven dobaytīs by Maʿālī, five dobaytīs by Ādam, and seven by Ḵalīfa Ṣādeq from a jong (a manuscript of personal selections) found in Ṭāleš, and another jong from the Ḵalḵāl area (Kasravī, Āḏarī, 5th ed., pp. 57-61). Information is lacking concerning their authors and their dates of composition, but linguistically they are all close to the verses of Shaikh Ṣafī. (10) Ten words from the language of “Aḏarbāḏakān” in contrast to Persian, quoted in an old manuscript of Asadī Ṭūsī’s Loḡat-e fors in the Malek Library (no. 5839) (Ṣ. Kīā, “Kohnatarīn dastnevīs-e "Logat-e fors"-e Asadī Ṭūsī”, MDAT 3/3, 1335 Š./1956, pp. 4-5; idem, Ādarīgān: āgāhīhā-ī dar bāra-ye gūyeš-e āḏarī, Tehran, 1354 Š./1975). (11) Two short ḡazals, five lines each, by Badr Šīrvānī (Dīvān, ed. A. H. Rahimov, Moscow, 1985, pp. 665f.) in the language of “Kanār Āb,” in a local dialect of Šīrvān and possibly the mother tongue of the poet who was born in Šamāḵī. The language of these poems is almost identical to that of Shaikh Ṣafī-al-dīn’s dobaytīs (see below); notice čəman “my,” -r, the 2nd singular enclitic pronoun (read mehr-ər “your love,” cf. ḡam-ər “your sorrow”), “from,” “without,” kar-, the present stem of “to do,” vāč-, the present stem of “to say.”

It should be noted that the final section of Rūḥī Anārjānī’s 11th/17th-century Resāla, a literary miscellany, entitled “On the Terms and Phrases of Ladies, Grandees, and Dandies of Tabrīz” which has been assumed by a number of scholars to be in Āḏarī dialect (ʿAbbās Eqbāl, “Yak sanad-e mohemm dar bāb-e zabān-e āḏarī,” Yādgār 2/3, 1324 Š./1945, pp. 43-50; M. Moḡdam [Moqaddam], Iran Kūda 10, 1327 Š./1948, pp. 1-18; Saʿīd Nafīsī, ed., “Resāla-ye Rūḥī Anārjānī,” FIZ 2, 1333 Š./1954, pp. 329-72; Y. M. Nawwābī, NDA Tabrīz 9, 1336 Š./1957, pp. 221-32, 396-426; M. J. Maškūr, Naẓar-ī ba tārīḵ-e Āḏarbāyjān wa āṯār-e bāstānī wa jamʿīyatšenāsī-e ān, Anjoman-e Āṯār-e Mellī, Tehran, 1349 Š./1971, pp. 221ff.; M. Mortażawī, Zabān-e dīrīn-e Āḏarbāyjān, Tehran, 1360 Š./1981, p. 35), bears no relationship to Āḏarī, but as W. B. Henning ingeniously realized (“The Ancient Language of Azerbaijan,” TPS, 1954-55, p. 176 n. 5) refers to a vulgar form of New Persian, and actually attests to the continued currency of this language in Tabrīz even in the sixteenth century.

Of the written remains of Āḏarī, the dobaytīs of Shaikh Ṣafī-al-dīn are the most important: They are relatively old, their linguistic area and their author are known, and they are accompanied by a paraphrase in Persian which helps their understanding. Despite Ardabīl’s location at the eastern edge of Azerbaijan, in view of its significance both before and after the advent of Islam, its language must have been one of the more important dialects of Āḏarī. Before it fell into the hands of the Arabs, Ardabīl was the madīna, i.e., the metropolis, of Azerbaijan; it was the center of its fiscal administration and the seat of the Sasanian marzbān (Balāḏorī, Fotūḥ al-boldān, p. 325; Yāqūt, Moʿjam-al-boldān I, p. 197) and was confirmed as the capital of the region by Ašʿaṯ b. Qays during ʿAlī’s caliphate (Balāḏorī, Fotūḥ, p. 329). Some three centuries later Ebn Ḥawqal (Ṣūrat al-arż, p. 334) still mentions it as the center and the largest city of Azerbaijan (cf. Moqaddasī, Aḥsan al-taqāsīm, p. 375); Eṣṭaḵrī (Masālek, p. 181) refers to it as the largest city, the seat of the government (dār al-emāra), and the military encampment (moʿaskar) of the region (see further Qodāma b. Jaʿfar, Ketāb al-ḵarāj, p. 244 and Ebn Rosta, Aʿlāq, p. 106).

2. Words borrowed from Āḏarī into Azeri Turkish. These include dardažar “ailing” and *kušn “field”, which occur in Shaikh Ṣafī’s dobaytīs (see Kasravī, Āḏarī, p. 41). Kārang (Jahān-e aḵlāq 4, 1956, pp. 84ff.) notes a number of Tati words used also in Azeri Turkish, e.g., dīm “face,” zamī “land, field,” olis, Azeri ulas “charcoal.” But to determine the full extent of such borrowings requires further research. Several authors, notably Adīb Ṭūsī (“Nomūna-ī čand az loḡat-e āḏarī,” NDA Tabrīz 814, 1335 Š./1957, pp. 310-49; 9/2, 3, 4, 1336 Š./1957, pp. 135-68, 242-60, 361-89; cf. M. Aržangī, ibid., 9/1, 2, pp. 73-108, 182-201; 10/1, 1337 Š./1958, pp. 81-93) have collected a large number of non-Turkish words used in the Azeri Turkish of the various parts of Azerbaijan (See Maškūr, op. cit., p. 263 for a count); but, ignoring proper linguistic criteria, they have taken them to be Āḏarī, whereas in fact, they are, by and large, Persian (or Arabic, borrowed through Persian), a fact which shows that Āḏarī, unlike Persian, has not affected the lexicon of Azeri Turkish significantly. The assumption of these researchers that the material in the last chapter of Rūḥī Anārjānī’s Resāla is Āḏarī (see above) has also tended to vitiate their conclusions. (For a listing of Azeri vocabulary see Y. M. Nawwābī, Zabān-e konūnī-e Āḏarbāyjān [Bibl.]; and Koichi Haneda and Ali Ganjelu, Tabrizi Vocabulary, An Azeri-Turkish Dialect in Iran, Studia Culturae Islamicae, no. 13, Tokyo, 1979.)

3. Present-day dialects or Āḏarī. Despite its continued decline over the centuries, Āḏarī has not died out and its descendants are found as modern dialects, mostly called Tati, sharing a wide range of phonological and grammatical features. Proceeding from north to south, these are: (1) The dialect of Kalāsūr and Ḵoynarūd, two villages of the Ḥasanow (Ḥasanābād) district of Ahar; (2) the dialect of Karīngān, a village of eastern Dīzmār in the Vazraqān district (baḵš) of Ahar sub-province (šahrestān); (3) the dialect of Galīnqaya, a village of the Harzand rural area (dehestān) in the district of Zonūz, Marand sub-province; (4) the Ḵalḵāli dialects spoken in the chief villages of the Šāhrūd baḵš (i.e., Askestān, Asbū, Derow, Kolūr, Šāl, Dīz, Karīn, Lerd, Kehel, Ṭahārom, Gelūzān, Gīlavān, and Gandomābād), in Karnaq, in the Ḵoreš-e Rostam baḵš, and in Kajal in the Kāḡaḏkonān baḵš of Ḵalḵāl; (5) the Tati dialects of the Upper Ṭārom (principally in the villages of Nowkīān, Sīāvarūd, Kalāsar, Hazārrūd, Jamābād, Bāklūr, Čarza, and Jeyšābād); (6) the Tati dialects of Rāmand and Zahrā, southwest and south of Qazvīn (i.e., the dialects of Tākestān, Čāl, Esfarvarīn, Ḵīāraj, Ḵᵛoznīn, Dānesfān, Ebrāhīmābād, and Sagzābād) which are close to the Tati of Ḵalḵāl and Ṭārom; (7) the dialects of Ṭāleš, from Allāhbaḵš Maḥalla and Šāndermīn on the border of Gīlān in the south to the Soviet Ṭāleš in the north, including the dialect of ʿAnbarān in the Namīn district of Ardabīl; all connected with the Tati dialects of Šāhrūd. This list does not necessarily exhaust the Āḏarī-speaking villages of Azerbaijan, and there may exist villages which the writer has not been able to visit, and where Tati is still understood (see A. A. Kārang, Tātī wa harzanī, pp. 27; he mentions a number of villages in Dīzmār and Ḥasanābad districts, including Arzīn, where the dialect was still understood in the 1940s; on the continued waning of Āḏarī, see below).

To the same group of dialects belong in a broad sense: (1) the dialect of Māsūla in the Fūmenāt district of Gīlān; (2) the language spoken in the Rūdbār of Gīlān (Raḥmatābād, Rostamābād, etc.), in the Rūdbār of Alamūt (Dekīn, Mūšqīn, Garmārūd, and Bolūkān), and in Alamūt (Moʿallem Kelāya, Estalbar, Gāzarḵān; Avānak, etc.); (3) the dialect of Ḵoʾīn and Safīdkamar in the Ījrūd of Zanjān, and a few villages in the Kūhpāya of Qazvīn (Zerejerd, Nowdeh, Asbemard, Ḥeṣār, etc.); (4) the dialect of Vafs, between Hamadān and Arāk. There are also a number of border dialects, such as the dialect of Ṭāleqān villages between Qazvīn and Karaj, and the dialects of Āmora and Āštīān, all much affected by Persian, that have close affinities with the group. In fact, the demarcation line between these dialects and their more northerly cognates cannot be sharply drawn. Kurdish, however, spoken in Mahābād in southwestern Azerbaijan and scattered in several other areas in the region, which some have supposed to be a descendant of Median, does not belong to this group and exhibits some clear differences with it. (See D. N. Mackenzie, “The Origins of Kurdish,” TPS, 1961, pp. 67-83.)

The fact that these dialects are so relatively abundant and are spoken in contiguous areas over a vast territory confirms their being indigenous to these areas and speaks strongly against the possibility that they spread into Azerbaijan and its border regions from other areas. Their shared linguistic features place them in a well-defined group of North-West Iranian, with affinities with the Central dialects, spoken to the south and southeast of the Āḏarī language area. Āḏarī and the language termed Fahlawī in the medieval Islamic sources refer in fact to the northern and southern branches of the language spoken in the territory of ancient Media, broadly corresponding to their modern continuations, namely the Tati or Āḏarī dialects in central and western Iran (excluding Kurdish and Luri). On the analogy of New Persian one may call them New Median (see further below).

That only meager traces of the language spoken in the central regions of Azerbaijan have survived is only natural, since a language that comes under pressure from other languages disappears faster in the center than in the periphery. The fact that while there are some meager remains of Āḏarī from the north, the center, the east, and south of Azerbaijan, yet the western part of the province yields no comparable material, is no doubt due to the dominance in these regions, before the spread of Turkish, of other languages, such as Neo-Aramaic and Kurdish.

The process of the linguistic Turkification of Azerbaijan continues to this day, and even in the border areas the original dialects keep giving way to Turkish. In the course of his study of these dialects in the 1960s, the writer met a number of elderly people who could remember or had been told by their fathers or grandfathers that villages now speaking Turkish formerly spoke the Iranian dialect. In Ḥalab, a village in Ījrūd on the way from Zanjān to Bījār, he met in 1964 the last three men who still retained some shaky memory of their Tati, and in Galīnqaya there was in 1972 only one old man who could speak the native dialect fluently. (See also Kārang, Tātī wa harzanī, pp. 27-29; idem, “Ḵalḵālī,” Jahān-e aḵlāq 4, 1335 Š./1956, p. 83; Ḏokāʾ, Gūyeš-e Galīnqaya, p. 6.)

Linguistic features.

The absence of vocalization, the deficiencies of the Arabic alphabet in indicating the details of pronunciation, scribal errors, and the influence of classical Persian make the reading of the literary Āḏarī remains difficult. Nevertheless they reveal some genuine features of the phonology, grammar, and vocabulary of the language in which they are written. Here the features of two written remains are explored.

A. Shaikh Ṣafī-al-dīn’s dobaytīs. 1. Old Iranian intervocalic t > r. Examples: žir “life” (< *jit-, cf. Parthian jydg); the enclitic 2nd singular pronoun -(a)r (Pers. -[a]t); past tense forms: āmarim “I came” (< *āmat-), bori or beri “he was” (< *būt-), šoram or šeram “I went” (< *šut-), and žar “struck” (< jat-, Pers. zad) in dara žar “was pained” (Parthian drdjd; Henning, “Ancient Language,” p. 176 n. 4). The same sound change is found in two Tati dialects: Harzandi and the dialect of Kalāsūr and Ḵoynarūd; cf. Harzandi amārā “he came” (other examples: vör “wind” < *wāt-, kar “house” < *kat-, jörö-tan “stranger” < *(wi)yut-, Pers. jodā “separate”); Kalāsūri umarim “I came,” and šerim “I went” (other examples: vur “wind,” jeru “separate,” purez “autumn” < *pātēz [Pers. pāʾīz], zura “boy, son” < *zātak-). In other dialects, this change occurs only sporadically; cf., e.g., Kajali kerom “which” (< *katām-, Pers. kodām), and in the dialect of Derow in Ḵalḵāl šera “he went.” The enclitic pronoun of the 2nd singular is -r in Kajali and Šāhrūdi of Ḵalḵāl, also in Asālemi and Māsāli in the central and southern Iranian Ṭāleš area (but not in northern Ṭāleši or ʿAnbarāni). In the sentence in the dialect of Tabrīz recorded by Ebn Bazzāz as uttered by a contemporary of Shaikh Ṣafī-al-dīn, we find ḥarīf-ar žāta “your contender has come.” One can not measure the extent of this rule in the defunct dialect of Tabrīz by this instance alone, but note also the Iranian word därdäjär “sick, ailing” in Azeri Turkish, and the Azerbaijani placename Esparaḵūn, colloquial for Safīidaḵān, a village in Bostānābād, east of Tabrīz, probably “White spring,” with espara < *spētak- (Pers. safīd “white”). The change of intervocalic t to r is seen also in the so-called Tati, but actually (archaic) New Persian dialect of the Iranian-speaking Jews in the Apsheron peninsula and the northeast of the Azerbaijan S.S.R. The change, on the other hand, is not effected in the dialects of Ṭārom, Ḵoʾīn, Rāmand, and Alamūt areas to the south.

2. Old Iranian intervocalic č >j. Examples: riji “he pours,” (Av. raēca-), and navāji “you [sing.] do not say” (Parth. wʾc-). The same change is seen in the modern dialects of Šāhrūd, Kajal and Asālem: Šāhrūdi verijam “we flee,” vāje “he says;” Kajali mivrije “he flees;” and Asālemi bivrij “flee!” By contrast, in the dialects of Kalāsūr and Ḵoynarūd, Ṭāleš, Karīngān, and Harzand, č has become `: cf. Kalāsūri ruž “day,” namuž “prayer;” ʿAnbarāni ruža “fast,” nəmož “prayer;” Ṭāleši as spoken in the Soviet Union: tož “to rush, gallop,” bad-vož “defamer, slanderer;” Karīngāni vuž “say!;” Harzandi ruž “sun.”

3. A vowel phoneme /ö/ə/ is indicated by the variant spellings -w and -h: čw and čh, i.e., /čə/ “from” (< *hača, Pers. az); and ʾštw and ʾčth, i.e., /aštə/ or /ačtə/ “yours” (2nd sing., rendered by Pers. māl-e to, lit., “your property”). A similar phoneme is found in the modern dialects of Harzand, Ṭāleš, Kajal, and Šāhrūd (not in word-final position in Šāhrūdi).

4. Old Iranian initial j > ž. Examples: žir “lile,” and žar “struck.” The same sound change is seen in the modern dialects of Kalāsūr and Ḵoynarūd: žan “woman,” žare “to hit,” žāte “to arrive”; Ṭāleši žen “woman,” žae “to hit”; Arazini žen and Kajali žan “woman,” bežana “strike!” The form žāta in Ebn Bazzāz’s sentence shows that this feature extended to the dialect of Tabrīz. In the dialects of Karīngān and Harzand, however, initial ž has become y: Karīngāni yan “woman” and “strike!,” yaz/yat- “to arrive,” and Harzandi yan “woman,” yare “to strike.”

5. Old Iranian x, xw > h in harda “he ate;” cf. sohrāb “rouge” in the manuscript of the Loḡat-e fors mentioned above (Kīā, p. 4). This development is regular in Kajali: (hardan “to eat,” hára “ass,” heriār “buyer,” howlig “sister”) but sporadic in the Šāhrūdi group: Šāli (h)ardan, cf. Gīlavāni ha “sister,” hezə “he wants” (Parth. wxāz-, wxāšt, but Pers. ḵᵛāh-, ḵᵛāst); but Šāli ḵri- “to buy,” ḵes/ḵel “to sleep,” etc. Cf. also Karīngāni hārdan “to eat,” haraši “sun” (Pers. ḵᵛoršīd): Harzandi horde “to eat,” höšn/höšt “to want,” hištan “self” (Pers. ḵᵛīštan); Kalāsūri horma “I ate,” hāmma “I read” (Pers. ḵᵛāndam); and in most Ṭāleši dialects: Asālemi hard-, ʿAnbarāni hāna bim “I was eating, used to eat,” and Northern Ṭāleši hova “sister“. But in Asālemi we find ženā-xāzī (Pers. ḵᵛastgārī), and in the dialect of Māsāl in southern Ṭāleš we find xa “sister,” xəšk “dry,” etc.

6. Old Iranian fr > hr in ahrā “tomorrow” (Pers. fardā < *fra-, cf. G. Lazard, La langue des plus anciens monuments de la prose persane, Paris, 1963, p. 145). In the modern dialects we find Kajali a(h)rā, Harzandi ohra (cf. also heraš/heröt “to sell” < *frawaxš-/frawaxt, Pers. forūš/forūḵt), Ḵīāraji of Rāmand ahrā, Šāli pašara “the day after tomorrow,” Šāndermīni and Māsāli pašerā, Tākestāni sarā “day after tomorrow,” Northern Ṭāleši havate “to sell,” hamue “to order” (< *framāt-, Pers. framūdan).

7. Oblique case/genitive in *-i (or so-called inverted eżāfa construction). This ending is written only in ōyān-i banda “the servant of the Lord” (dobaytī 11; on ōyān < Tk. oγan, see Henning, “The Ancient Language,” p. 176 n. 4; it is not a plural of oy “he,” as Kasravī thought) but may also be assumed in other cases, e.g., oyān(i) ḵāṣṣān “special friends of god,” čowgān(i) gur-im “I am the ball of the polo stick” (i.e., resigned to the divine will), and qodrat(i) zanjir-im “I am the chain of power” (dobaytī 3). Among modern dialects, Kalāsūri and Asālemi have accusative and genitive in -i, Ḵalḵāli in -e.

8. The personal pronouns have four forms:

  Direct Oblique Possessive Enclitic
1st az man -m
2nd te or eštö -r

This feature is shared by the dialects of Ḵalḵāl and Ṭāleš. For instance, the corresponding forms in the Šāli dialect of Šahrūd are:

  Direct Oblique Possessive Enclitic
1st az man čeman -m
2nd te te ešte -r

In Kajali the forms are:

  Direct Oblique Possessive Enclitic
1st az aman čəman -m
2nd əštə -r

and in Asālem:

  Direct Oblique Possessive Enclitic
1st az aman čəmən -m
2nd əštə -r

A similar scheme is found in the dialect of Čāl in Rāmand. In the rest of the Rāmand area, however, the oblique form is no longer used. The dialects of upper Ṭārom, e.g., Nowkīāni and Hazārrūdi, have a system of actually five pronominal forms (the pronouns for the direct object and the “logical direct object” in passive constructions are differentiated; see Yarshater, “The Tati Dialects of Ṭārom”). In Karīngāni and Harzandi the direct pronoun has been replaced by the originally oblique form, as in Persian.

9. The 2nd person singular ending is -i in the present indicative (riji “you pour,” navāji “you do not say”), but -š in the present subjunctive (mavāješ “you may not say”). A 2nd person singular ending -š is found in several Tati dialects. In Karīngāni, in particular, it is the common form; in Kalāsūr, it is found in the present indicative (bežareš “you strike”); in Šāhrūdi (Šāli and Kolūri), everywhere except the present indicative and the imperative (bešiš “you went,” age bevrijāš “if you should flee”); in Asālem, everywhere except in the imperative and the present subjunctive (biš “you were,” bebaš “be!”); in ʿAnbarāni, in the continuous past tense; and in Northern Ṭāleši throughout the verbal system. In Harzandi the ending -š does not occur.

10. A continuous present is made from the past stem if indeed, as it appears, the verbs in the fourth dobaytī are present tense, wrongly rendered by the past tense in the paraphrase of the Selselat al-nasab: be-koštim “I kill,” be-heštim “I let/leave,” and na-daštim “I am not harming” (on the last verb, see Henning, “The Ancient Language,” p. 176 n. 4). The same kind of formation is found in the dialects of Karīngān, Harzand, and Kalāsūr, Northern Ṭāleši, and in Asālemi, but not in the dialects of Southern Ṭāleši: Karīngāni heteine “I am sleeping” (cf. fesene “I sleep” < *xwafs-), Harzandi bāvāštān “he is carrying,” bo-hordān “he is eating,” Kalāsūri ba-durem “I am giving” (< *dāt-), be-žareš “you (sing.) are striking,” ba-šem “I am going,” Asālemi ba-vindiše “you (sing.) are seeing,” ba-bramastim “we are weeping.”

11. Vocabulary. Note asra “tear” (cf. Šāhrūdi asərk, Asālemi, Māsāli, and ʿAnbarāni asərg, Harzandi ösör, Karīngāni aster; cf. also ásra [fem.] in the dialects of Rāmand and ars in the Persian dictionaries) and ahra “tomorrow” (see above, no. 6). The question whether -a in asra is a feminine marker (as it is in Rāmandi) and whether Āḏarī of Ardabīl distinguished grammatical gender, can not be determined on the basis of the material at hand. Its affinities lie mostly with modern dialects which do not have the category of gender (see below).

It can be seen from the foregoing that the language of the dobaytīs is not identical with any one modern descendant of Āḏarī. Its greatest affinity seems to be on the one hand with the Tati dialects of Kalāsūr and Ḵoynarūd to the northwest (t > r, j > ž, 2nd singular -š, continuous present from the past stem), and on the other with the dialects of the central Ṭāleš area to the east (j > ž, four-fold personal pronoun, 2nd singular -š, continuous present from the past stem), and Ḵalḵāli (t > r in some instances, j > ž in Kajali, four-fold personal pronoun). This agrees well with Ardabīl’s geographical position. By contrast, the dialects of Harzand and Karīngān, the Āstārā region, and of Soviet Ṭāleš to the north that B. V. Miller (Talyshskiĭ yazyk, Moscow, 1953, pp. 253ff.) for lack of information about Tati and southern Ṭāleši dialects thought were closest to Āḏarī, are relatively remoter. (Northern Ṭāleši is characterized by the dropping or greatly reducing of unstressed syllables, t does not become r, the enclitic pronouns are -ə and -əon for 2nd singular and plural, respectively.)

Another conclusion that can be drawn from these comparisons is that Ṭāleši should not be grouped with the Caspian dialects, as is commonly done on the basis of their geographical location, but rather with the Tati dialects of Azerbaijan, particularly Šāhrūdi.

B. The Istanbul qaṣīdas. The phonology and vocabulary of the language attested in this poem link it with the area of Tabrīz and Marand. Note the following features.

1. Old Iranian ā > ū in āžūr “free” (Pers. āzād), dūr “hold!” (Pers. dār), gūn “soul” (Parth. and Mid Pers. gyān, NPers. jān), *huzdan “to ask, want” (Pers. ḵᵛāstan), pūydūr “permanent” (Pers. pāydār), and vad-nehūd “bad-natured” (Pers. bad-nehād).

2. Old Iranian intervocalic t > r in āžūr, -r “you” (Pers. -t), zūnar “he knows” (< *zān-, Pers. dānad), and žaran “to strike” (< *jat-, Pers. zadan).

3. Old Iranian intervocalic č > j in jeman “my own” (< Old Iranian hača-).

4. Old Iranian x, xw > h in harda “eaten” (Pers. ḵᵛorda), *hūzdan “to ask, want”; cf. hošk “dry” (< Old Iranian *huška).

5. Vocabulary. Note gūn “soul,” *karend “they do, make” (Parth. kar-), sag “stone” (Pers. sang), and vūn “blood” (Av. vohunī, Pers. ḵūn).

The position of Āḏarī among the Iranian languages.

It is obvious that the language of as broad an area as Azerbaijan could not have been uniform throughout and must have exhibited a variety of local dialects. The statement by Moqaddasī (Aḥsan al-taqāsīm, p. 375) to the effect that seventy dialects were spoken in the region of Ardabīl, despite its gross exaggeration, has to be taken to refer to the variety of its local subdialects. On the other hand, the fact that the language of the entire Azerbaijan has been called Āḏarī in the early sources and placed alongside Darī and Pahlavi implies that the dialects of the region were similar enough to be called by a single name.

Azerbaijan and the “Jebāl” of the medieval geographers, that is, the mountainous west-central part of the Iranian plateau, coincide geographically with ancient Media and was inhabited by Median tribes in ancient times. Although no independent written document in ancient Median has yet come to light, its fundamental phonological features are known from the Median words and names which occur in Old Persian inscriptions and, less frequently, in Greek (e.g., IE. ĝ, and ĝh < Med(ian) z, OPers. d; IE. kṷ > Med. sp, OPers. s; IE. tr and tl > Med. θr, OPers. ç; see Kent, Old Persian, secs. 8-9; M. Mayrhofer, Die Rekonstruktion des Medischen, Anz. d. Österreichischen Akad. d. Wiss., Phil.-hist. Kl., 1968, 1, Vienna; G. L. Windfuhr, “Isoglosses: A Sketch on Persians and Parthians, Kurds and Medes,” in Monumentum H. S. Nyberg II, Acta Iranica 5, Tehran and Liège, 1975, pp. 457-72). All these features are characteristic also of Āḏarī and its modern relatives. Thus there are no linguistic arguments against the derivation of Āḏarī from Median, which is based upon compelling geographical and historical evidence (see below), and such a conclusion can in no way be invalidated by the fact that the phonological peculiarities of Median are found, by and large, in all northwestern branches of Iranian, including Parthian, or by the fact that it has not been possible to find exclusive Median isoglosses (see P. O. Skjærvø, BSL 78, pp. 244-51). It will be noted that Āḏarī differs from Parthian in some important respects, e.g. “came” is from *ā(g)mata- (as in Persian) against Parthian āγad < *āgata-; Parthian has a suffix -īft and the eżāfa čē both unknown in Āḏarī.

Likewise, the fact that the Āḏarī group of dialects shares a few isoglosses with some geographically and linguistically distant dialects in southeastern Iran, namely Lāri and Baškardi, which, like Persian belong to the South-Western Iranian dialects does not affect our conclusion with regard to the derivation and provenience of Āḏarī. The isoglosses shared with Lāri are the 2nd singular ending -š and the continuous present from the past stem; cf. Lāri ačedāeš “you are going,” čedeš “you went” (A. Eqtedārī, Farhang-e lārestānī, Tehran, 1334 Š./1955, p. 269); the isoglosses shared with Baškardi are: t > r in North Baškardi (e.g., zar- “to strike”) and the continuous present based on the past stem (e.g., North Baškardi akerdénom, South Baškardi bekert(en)om “I am doing,” see G. Morgenstierne in HO I, iv, 1: Linguistik, Leiden, 1958, p. 178). There is no need for assuming any special historico-geographical connection between the Āḏarī group and Lāri and Baškardi to explain these isoglosses. Indeed, since Āḏarī is phonetically a typical North-Western dialect but Lāri and Baškardi typical South-Western dialects, such an assumption would create more problems for historical Iranian linguistics than it would solve. In the case of other Iranian languages and dialects, too, we occasionally find isoglosses crossing other, fundamental, isoglosses and spanning large distances. One typical case is that of Sogdian and Old Persian (see Henning, Mitteliranisch, p. 108).

Historically, Media was divided into Greater Media, which was the area where today the Central dialects are spoken, and Lesser Media or Azerbaijan. Doubtless it is this geographical division which is reflected in the linguistic distinction between al-āḏarīya and al-fahlawīya of our medieval sources. (The fact that while there are some meager remains of Āḏarī from the north, the center, the east, and the south of Azerbaijan, yet the western part of the province yields no comparable material, is no doubt due to the dominance in these regions, before the spread of Turkish, of other languages, such as Neo-Aramaic and Kurdish.) Since there is no historical evidence that the population of the Median territories was ever dislocated on a significant scale, or that its language was superceded by any other language than Persian (in the urban centers) and Turkish (in Azerbaijan), the conclusion is inevitable that the affiliated Iranian dialects spoken in Azerbaijan, Ḵamsa, Qazvīn, Ṭāleš, Hamadān, Nahāvand, Ḵᵛānsār, Kāšān, Isfahan, and Semnān, to mention only the chief regions, can be none other than the descendants of the Old Median language, today divided roughly into a northern, Āḏarī, group and a southern, “Fahlawī” or “Central” group of dialects.



Given in the text. The dialect materials referred to in the article, except for the Ṭāleši of the Soviet Union, Arazīni, Baškardi, and Lāri, were collected by the author between 1955-72.

See also M. Qazvīnī’s review of Kasravī, Āḏarī, repr. in Bīst maqāla, Tehran, 1332 Š./1953, I, pp. 178-86.

On the modern dialects see ʿA. Kārang’s pioneering treatise on the dialects of Karīngān and Galīnqaya, Tātī wa harzanī, du lahja az zabān-e bāstān-e Āḏarbāyjān, Tabrīz, 1333 Š./1954.

Y. Ḏokāʾ, Karīngānī, Tehran, 1332 Š./1954.

Idem, Gūyeš-e Galīnqaya, "harzandī," Tehran, 1336 Š./1957.

J. Matini, “Daqīqī, zabān-e darī wa lahja-ye āḏarī,” MDAM 11/4, 1354 Š./1975, pp. 559-75.

M. Mortażawī, “Nokta-ī čand az zabān-e harzanī,” NDA Tabrīz 6/3, 1333 Š./1954, pp. 304-14.

Idem, Feʿl dar zabān-e harzanī, Tabrīz, 1342 Š./1963.

Y. M. Nawwābī, Zabān-e kunūnī-e Āḏarbāyjān, Tabrīz, 1334 Š./1955 (published earlier as a series of articles in NDA Tabrīz 5 and 6, 1332-33 Š./1953-54).

E. Yarshater, “The Tati Dialect of Shāhrud (Khalkhāl),” BSOAS 22, 1959, pp. 52-68.

Idem “The Tati Dialect of Kajal,” BSOAS 23, 1960, pp. 257-68.

Idem, “The Tati Dialects of Rāmand,” in A Locust’s Leg. Studies in Honour of S. H. Taqizadeh, ed. W. B. Henning and E. Yarshater, London, 1962, pp. 240-45.

Idem, “Marāḡīān-e Alamūt wa Rūdbār wa zabān-e ānhā,” Majalla-ye Īrānšenāšī 1, 1346 Š./1967.

Idem, A Grammar of Southern Tati Dialects (Median Dialect Studies I), The Hague and Paris, 1969.

Idem, “The Tati Dialects of Ṭārom,” in W. B. Henning Memorial Volume, ed. M. Boyce and I. Gershevitch, London. 1970, pp. 451-67.

M. Mortażawī provides a listing of the Persian articles on topics related to Āḏarī in Zabān-e dīrīn-e Āḏarbāyjān, pp. 56ff.; of interest is a paper he entitled “Bīst vāža-ye āḏarī dar ḥawāšī-e nosḵa-ye ḵaṭṭī-e Ketāb al-bolḡa” (Twenty Āḏarī words on the margin of the MS. of the K. al-bolḡa) read by M. Mīnovī at the sixth conference of Iranian studies (1974?), but apparently not yet published. On Median and the “Median” dialects see also A. Meillet, Grammaire du vieux perse, 2nd ed. by E. Benveniste, Paris, 1931, p. 7, par. 8; I. Gershevitch, “Dialect Variation in Early Persian,” TPS, 1964 [1965], pp. 1-29; P. O. Skjærvø, “Farnah: mot mède en vieux perse?” BSL 79, 1984, pp. 241-59.

On the dialectology of Middle Iranian see also W. Lentz, “Die nordiranischen Elemente in der neupersischen Literatursprache bei Firdosi,” ZII 4, 1926, pp. 252-316, and P. Tedesco, “Dialektologie der westiranischen Turfantexte,” Monde oriental 15, 1921, pp. 184-258.

Search terms:

 آذربایجان،زبان ایرانی آذربایجان azarbayjan,zaban e irani azarbayjan azarbaayjan,zaban e irani azarbayjan  azarbaijan,zaban e irani azarbayjan


(E. Yarshater)

Originally Published: December 15, 1988

Last Updated: August 18, 2011

This article is available in print.
Vol. III, Fasc. 3, pp. 238-245