Preliminary Remarks
Aramaic is a Semitic language with an exceptionally long documented history. It is first attested in written form in inscriptions datable to approximately 1000 BC and is still used as a spoken vernacular language by various minority communities in the Middle East. It had the status of an official lingua franca in the middle of the first millennium BC in the Persian Achaemenid empire, which extended from Egypt to India. In the first half of the first millennium AD it remained the main spoken language of the Levant and Mesopotamia, until the advent of Arabic in the region with the rise of Islam.
Spoken vernacular dialects of Aramaic, generally known as Neo-Aramaic dialects, have survived down to modern times in four subgroups (see Figure 1):
-
1. Central Neo-Aramaic (southeastern Turkey in the region of Ṭūr ʿAbdīn);
-
2. North-Eastern Neo-Aramaic (east of the Tigris);
-
3. Neo-Mandaic (southwestern Iran);
-
4. Western Neo-Aramaic (Syria).
The location of the neo-Aramaic subgroups in the Middle East.
Source: Based on Google Maps.
These subgroups are surviving islands of an original dialect continuum of Aramaic that is likely to have existed before the Islamic period. As remarked, the heartland of spoken Aramaic in antiquity was the Levant and Mesopotamia. The surviving Neo-Aramaic dialects spoken in Iran are, therefore, dialect islands on the eastern periphery of this original Aramaic heartland.
Subgroups of Neo-Aramaic in Iran and Their Location
The Neo-Aramaic dialects that have survived down to modern times in Iran belong to the eastern periphery of the North-Eastern Neo-Aramaic subgroup and to the Neo-Mandaic subgroup.
The dialect geography of Neo-Aramaic has undergone radical changes over the last hundred years due to a variety of upheavals and population displacements in the region that have resulted in the movement of a large number of the speakers of the dialects from the places where they have lived for many centuries.Footnote 1 As a consequence many of the dialects today are highly endangered. The following geographical description, therefore, relates to the situation that existed at the beginning of the twentieth century, before these major displacements.
North-Eastern Neo-Aramaic dialects
North-Eastern Neo-Aramaic (NENA)Footnote 2 is a highly diverse subgroup of over 150 dialects spoken by Christians and Jews in towns and villages east of the Tigris River in northern Iraq, southeastern Turkey and western Iran.Footnote 3
The Christian NENA dialects and the Jewish NENA dialects exhibit different geographical distributions in Iran. The Christian NENA dialects can be divided into two main subgroups according to their linguistic relationship, the Urmi subgroup and the Western Iran subgroup. The former is more diverse and contains a number of subdivisions. The latter consists only of the Christian NENA dialect of Sanandaj.
There were no dialects of Christian NENA spoken to the south of the Urmi plain throughout the area down to Sanandaj, where, as remarked, there was one isolated Christian NENA dialect.
The Christian dialects of the Urmi cluster exhibit a close relationship with the Christian NENA of the adjacent regions in southeastern Turkey and northeastern Iraq.Footnote 4 The Christian dialect of Sanandaj is closely related to the Christian dialect of Sulelmaniyya across the border in north-eastern IraqFootnote 5. It exhibits considerable structural differences from the dialects of the Urmi cluster.
Documentation of Christian NENA dialects of Iran has focused on the Urmi dialect and its varietiesFootnote 6 and on the Sanandaj dialectFootnote 7, with some short studies on the Salamas dialectFootnote 8.
The Jewish NENA dialects had a wider distribution in western Iran than the Christian NENA dialects. In the first half of the 20th century there were NENA-speaking Jewish communities in a variety of locations throughout the area extending from Salamas in the north in West Azerbaijan province to Kerend in the south in Kermanshah province.
In the region of Urmi in northwestern Iran there were several NENA-speaking Jewish communities, in particular in the towns of Urmi, Šəno (official name Ushnuye), Solduz (official name Naghade), Sablagh (now Mahabad),Footnote 9 and in the area of Salamas north of the Urmi plain.Footnote 10 The Jewish dialects of this area are closely related, but on structural grounds one can identify some subdivisions as presented above.
The dialects of the other major cluster of Jewish NENA were spoken in various localities in an area that includes Sainqala, Bokan and Saqəz on its northern border, Sanandaj in the center, Bijar on the eastern border, and in the south Kerend and Qasr-e Širin. This cluster of dialects is remarkably uniform and only minor differences are found among the dialects of the aforementioned places where the dialects were spoken. The Jewish dialect of Sanandaj has been documented in detail.Footnote 11 Studies on other Jewish dialects of the cluster include those by Israeli on the dialect of Saqəz and by Hopkins on Kerend.Footnote 12
The Jewish NENA dialects of Iran are an eastern continuation of the subdivision of Jewish NENA dialects known as trans-Zab, the western boundary of which was the Great Zab river in Iraq.Footnote 13
Neo-Mandaic
Neo-Mandaic is spoken by the Mandaeans, who follow a religion that is a descendant of a pre-Islamic Gnostic sect. The traditional homeland of the Mandaean community is the south of Iraq and the adjacent Khuzestān province of southwest Iran. They are known in Iraq and Iran as “Sabians” (Arabic ṣābiʾūn, colloquial ṣubba), who are one of the “peoples of the book” (ʾahl al-kitāb) recognized in Islam.
Neo-Mandaic appears to have ceased to be the spoken language of the Mandaeans of Iraq by the beginning of the nineteenth century. There are references to a few speakers in Iraq in the twentieth century, but these seem to be of Iranian origin.Footnote 14
Up until the nineteenth century Neo-Mandaic was spoken in a variety of localities in the Khuzestān region. The Mandaeans subsequently came to be concentrated in Khorramshahr and Ahvāz, where two distinct varieties of the language survived until modern times.
Current Situation
The Christians from the Urmi region belong to the Assyrian Church of the East and the Chaldean Church, as well as a number of other denominations brought by missionaries in the nineteenth century.Footnote 15 There are still some Christians living in the Urmi region, almost exclusively now in the town of Urmi rather than the villages. A large proportion of the Christian community, however, has left the region. Some have settled in the large Iranian cities, mainly Tehran. In the nineteenth and early twentieth centuries many Christian speakers of the dialect of Urmi and other NENA dialects of northwestern Iran and the adjacent region moved to the Caucasus. These dialects, especially the Urmi dialect, is still spoken by communities in Georgia and Armenia, and elsewhere in the former Soviet Union.Footnote 16 A large proportion of speakers have now settled in the USA, in particular in California, and in Australia.
The Christians of Sanandaj belong to the Chaldean Church. In the nineteenth century several Christian families moved to Qazvīn, where their speech developed the distinctive trait of the realization of /w/ as /v/ under the influence of Persian.Footnote 17 In the middle of the twentieth century the Chaldean diocese of Sanandaj moved to Tehran and the Christian Neo-Aramaic speakers moved with it.
The vast majority of the NENA-speaking Jews left the region in the early 1950s and settled in the newly established state of Israel. In Kordestan and Kermanshah provinces a certain number remained until the Iranian Revolution in 1979, after which most of the remaining Jews left the region, the majority settling in Los Angeles in the USA, Israel or Europe. In the nineteenth century many Jewish NENA-speaking families from northwestern Iran, in particular from the region of Salamas, moved to Tbilisi in Georgia, and subsequently in 1950 were settled by Stalin in Almaty in Kazakhstan, where several speakers of the dialect can be found today.
The Mandaean community left Khorramshahr after it had been largely destroyed during the Iranian revolution in 1979 and in subsequent conflicts with Iraq. There are numerous Mandaeans living in the urban centers of Iraq and in communities that have settled outside the Middle East, especially in Sweden, Australia and the USA. The vast majority, however, do not speak neo-Mandaic. The number of competent speakers of the language is rapidly dwindling. Häberl estimated there to be around 100–200 elderly speakers, most of whom are living in Iran.Footnote 18 Neo-Mandaic, therefore, will inevitably become extinct within the next few years.
Relationship between NENA and Neo-Mandaic
The two subgroups of Neo-Aramaic in Iran, NENA and Neo-Mandaic, exhibit a number of fundamental differences in phonology, morphology and syntax, the details of which cannot be given in this short overview article.Footnote 19 We shall mention just two examples of structural difference, one in phonology and one in the verbal system.
In the NENA dialects an original unvoiced pharyngeal consonant /ḥ/ [ħ] has shifted to a velar fricative /x/. This is a general feature of NENA and can be assumed to have existed in the proto-form of the subgroup. In Neo-Mandaic, on the other hand, an unvoiced pharyngeal has shifted to a larygeal /h/, e.g.:
In the NENA dialects there has been a restructuring of the verbal system of earlier Aramaic, whereby the perfective and imperfective finite verbal forms have been replaced by participles due to convergence with the model of Iranian languages.Footnote 20 Neo-Mandaic is more conservative than NENA in that it has retained the historical finite perfective form. This is illustrated below by the verb “to kill,” which has the historical root q-ṭ-l:
The NENA perfective form is historically a passive participle and is inflected with oblique agentive suffixesFootnote 21.
NENA and Neo-Mandaic are clearly two genetically distinct subgroups of Aramaic, which are now separated by considerable geographical space. It should be remarked, however, that the two subgroups share a number of lexical items that are not found in other Neo-Aramaic subgroups.Footnote 22 This applies in particular to the common NENA word məndi “thing,” which appears to be of Mandaic origin. In Neo-Mandaic mend-i means “a thing,” the final -i being the Iranian indefinite suffix. Without the suffix, Neo-Mandaic mendā means “thing.” The indefinite form mend-i has been transferred to NENA as a fossilized form (məndi), which is used for both indefinite and definite. Such lexical isoglosses between the two dialect groups may have spread at an earlier period by areal diffusion. This suggests that the two subgroups, which are now language islands, originally occupied a larger geographical area of Iran and were in contact.
Communal Dialect Splits in NENA
There are many differences in all levels of grammar and also in the lexicon between NENA dialects spoken by Christians and those spoken by Jews in the region, even in cases where the two communities live in the same location. This applies, for example, to the towns of Urmi and Sanandaj, where there were both Christian and Jewish communities. This is exemplified in (4), in which selected features from the Christian and Jewish dialects of Urmi (northwestern Iran) are compared:
The Christian and Jewish communities in Urmi were not separated from each other geographically or physically (there were no Jewish ghettos). There was, moreover, professional contact between them. The crucial factor that induced divergent linguistic development in Urmi and elsewhere in the region was the different group identities and social networks of the two communities.Footnote 25
The Impact of Languages in Contact
The contact languages of the region
The Neo-Aramaic dialects of Iran exhibit numerous aspects of historical development that have been induced by contact with other languages spoken in the region. The main languages or language groups with which the NENA dialects were in contact in western Iran are the following:
The main languages in contact with Neo-Mandaic in modern times are:
The influence was almost completely unidirectional, since there is no clear evidence of convergence of these languages with NENA or Neo-Mandaic, apart from a few NENA loanwords in Kurdish.Footnote 26
In some cases one can identify various historical layers of contact. In northwestern Iran, for example, the NENA dialects were influenced extensively by Kurdish and Azeri Turkish, especially in their lexicon. Nowadays most Christian speakers of NENA who still live in the area speak, in addition to NENA, Azeri, the vernacular of the Muslim population in this area, and Persian, the official language of Iran, but not Kurdish. The morphology of the Kurdish loanwords in the Christian Urmi NENA dialect indicates that they form an older historical layer of the lexicon than the many Azeri words. In general Kurdish loanwords have been adapted to the inflectional morphology of Neo-Aramaic but the loaned Azeri lexical items remain unadapted. This indicates that there must have been a more widespread knowledge of Kurdish in the NENA communities at an earlier period.Footnote 27
In what follows I shall present a few selected cases of features in the Neo-Aramaic dialects of Iran that have been induced by contact, concentrating for the most part on the NENA dialects.
Selected contact-induced features
Enrichment of NENA sound system
In the sound system of consonants in Christian NENA dialects in northwestern Iran there was a high degree of convergence with the sound systems of languages in contact. In Christian Urmi,Footnote 28 for example, an additional series of unaspirated stops has developed through convergence with the phoneme inventories of Eastern Armenian and Kurmanji Kurdish. In earlier Aramaic, and Semitic in general, unvoiced stops are aspirated. In (6) the unaspirated phonemes are distinguished with a circumflex diacritic above or below the letter):
Armenian has a triad of velar stops (aspirated, unaspirated and voiced k, k̭, g) but no uvular stops. In order to converge with the Armenian type system the historical uvular *q of Christian Urmi was fronted to converge with an unaspirated k̭ and as a result there was a push-chain effect whereby the NENA velars *k and *g were in turn moved forward to a palatal articulation /c/ [cʰ] ∼ [ʧʰ], /č̭/ [c] ∼ [ʧ] and /ɟ/ [ɟ] ∼ [ʤ]. The realization of the existing affricates /č/, /č̭/ and /j/, in their turn, tended to be pushed further forward in some varieties to [ʦʰ], [ʦ] and [ʣ]. A similar push-chain fronting of the dorsal consonants and affricates in order to converge with an Armenian type of system without an uvular is attested in other languages of the region of northwestern Iran, notably in the Azeri dialects of this area, known as Western Azeri.Footnote 29 Such a process occurs in languages in what Stilo calls the ‘transition area’ between the language areas of Transcaucasia and northern Iran. Each of these language areas contains languages of diverse genetic groups but they share similar phonological systems, which are typically represented by those of Armenian (Transcaucasia) and Persian (northern Iran). In Azeri, unlike in Armenian, however, there was only a binary system of stops (voiced and unvoiced) without unaspirated unvoiced counterparts, which is a feature shared with Persian. The original uvular in Azeri in Iran was voiced [ɢ] and so this moved forward to the voiced velar [g] and the other dorsals moved forward.Footnote 30 Christian Urmi NENA has converged with both the Azeri and the Armenian type of systems, in that it has unaspirated stops and has undergone a push-chain fronting of the dorsals and affricates to eliminate an uvular, thus:
The consonant system of the Jewish NENA dialect of Urmi corresponds more closely to that of Persian than to those of Armenian of Transcaucasia or Western Azeri,Footnote 31 in that it has retained the uvular /q/ and the velars /k/ and /g/ have not been palatalized. As in Persian, /q/ tends to be voiced and fricativized, especially in the intervocalic position.Footnote 32
It is relevant to note that Garbell, in her study of the impact of Kurdish on Jewish Urmi, concluded that the dialect had been influenced by the Kurdish dialects spoken to the south of Urmi in Iran and northeastern Iraq, rather than those spoken to the north.Footnote 33 It is the northern dialects of Kurdish that exhibit the Armenian type of opposition between aspirated and unaspirated stops. The Jewish NENA of the Urmi region, therefore, is likely to have originated from the south whereas the Christian NENA originated from the northern region.
Zagros d
In the Jewish NENA dialects of western Iran the interdental consonants *θ and *ð have developed into a lateral /l/,Footnote 34 e.g.:
An intermediate stage of development appears to have been *θ > d, *ð > d, whereby both interdentals became a voiced stop d. This intermediate stage is attested in some NENA dialects of northwest Iran, e.g.:
The lateral /l/ would therefore have been the outcome of a lenition of the stop /d/.
The Neo-Aramaic dialects that have this feature were spoken in areas where in Iranian and Turkic languages a /d/ following a vowel or sonorant undergoes lenition, known as “Zagros d,” resulting in it being realized as an approximant or a sonorant. This has a variety of outcomes across the languages. In Hawrami dialects it is realized as an interdental approximant, a semivowel /y/ or a lateral /l/.Footnote 35 In Bakhtiari /d/ after a word-internal vowel or glide is realized as an interdental approximant or in some dialects as a sonorant r.Footnote 36 In some Turkic varieties in the Zagros area post-vocalic d is weakened to the sonorant /r/ or sometimes /y/ in combination with front vowels.Footnote 37
The development of /d/ to lateral /l/ in the Jewish NENA dialects can be regarded as a lenition due to the perceptual matching of post-vocalic NENA /d/ with the outcomes of weakened Zagros d in the contact languages. The area in which the Zagros d feature is attested in the Jewish NENA extends into the dialects of northeastern Iraq, as is the case with Sorani Kurdish.Footnote 38
It is significant that the lenition of /d/ with outcomes characteristic of Zagros d is a feature that is found only in the Jewish NENA dialects of the region and not the Christian NENA dialects. This may reflect that the Christian dialects have historically had less exposure to contact languages with this feature. Other phonological features of Christian dialects of the Urmi cluster, as remarked above, point to a northern origin, whereas the phonology of Jewish Urmi NENA corresponds more to that of languages to the south of the Urmi region.
The lack of a type of lenition corresponding to Zagros d in the Christian NENA dialect of Sanandaj may reflect a less intense contact with languages of the region than is the case with Jewish NENA. The Christian dialect exhibits a lesser degree of convergence with the local Kurdish dialect than the Jewish NENA dialect of Sanandaj also in other features (see below). This is reflected, furthermore, by the fact that there is a higher degree of Kurdish loanwords in the Jewish NENA of Sanandaj than in the Christian dialect.
Demonstrative pronouns
The Christian NENA dialects of the Urmi region have a system of demonstrative pronouns that express three types of deixis, namely (i) near to speaker, (ii) far from both speaker and hearer, and (iii) near to hearer and anaphoric, e.g.:
A similar three-way system of demonstratives is found in Eastern Armenian,Footnote 40 in some Kurmanji Kurdish dialects,Footnote 41 and in NENA dialects and Arabic spoken in the Kurmanji Kurdish area of northern Iraq and southeastern Turkey.Footnote 42 There are also formal similarities between the demonstratives of Christian Urmi NENA and those of the contact languages.
By contrast, the Jewish NENA dialects of the Urmi region and other NENA dialects south of Urmi in western Iran have a two-way system of demonstrative pronouns expressing (i) near deixis and (ii) far deixis and anaphoric, e.g.:
This corresponds to demonstrative systems in Iranian languages spoken in western Iran south of Urmi, such as Sorani Kurdish, Hawrami and Bakhtiari, as well as Persian and Turkic varieties of Iran. There is also some degree of convergence of the form of NENA demonstratives to that of the other languages of this area. The Jewish NENA singular far deixis demonstrative pronoun o “that” resembles its counterpart in Turkic o. It is also relevant to note that the gender distinction of this demonstrative has broken down in the Jewish NENA dialects. This may also have been induced by contact with Turkic, which has no gender distinctions.
The NENA near deixis pronoun ʾiya, which is attested in C. Sanandaj and also in some NENA dialects in Iraq, closely resembles the discontinuous near demonstrative ī … a of Hawrami, e.g. ī hanār = a “prox.dem.adj pomegranate = dem.m” (“this pomegranate”).Footnote 44 These formal resemblances are not the result of loaning of morphological matter into the NENA dialects but rather the result of the shaping of the native morphological material in imitation of corresponding forms in languages in contact.
The marking of definiteness
The Sorani Kurdish definite marking nominal suffix -aka has spread to the NENA dialects spoken south of Urmi in the Sorani-speaking area, the most northerly dialect where this is documented being the dialect of Šəno (Ushnuye). In the NENA dialects it has the form -ăke, which corresponds to the oblique form of the Kurdish definite article (-ăkay) rather than the nominative form (-ăka), e.g. Jewish Sanandaj bela “house,” belăke “the house”; Christian Sanandaj besa “house,” besăke “the house.”Footnote 45 The suffix is found also in other languages of the region such as Turkic varieties,Footnote 46 BakhtiariFootnote 47 and Hawrami.Footnote 48
Nominal genitive annexation
In earlier Aramaic the annexation of a noun as a genitive attribute to another noun is expressed by attaching the attributive particle d- to the dependent noun (glossed att below), e.g.:
In most NENA dialects of the region, the attributive particle has become an affix on the head noun, e.g.
This position of the NENA attributive particle replicates the placement of the ezafe clitic on the head noun in parallel constructions in Kurdish dialects in the north of the region, e.g.:
In the Jewish NENA dialects in southern Kordestan province and in Kermanshah province, however, the attributive particle is omitted in this construction, e.g.:
This is a replication of the construction in the Kurdish dialect of Sanandaj and in Southern Kurdish, in which the ezafe particle is omitted, e.g.:
In Neo-Mandaic, where the main contact languages are Arabic and Persian, a genitive relationship between nouns continues to be expressed by the Aramaic attributive particle d-, e.g.:
Alignment
Ergativity is found in the Jewish NENA dialects of Kordestan and Kermanshah provinces in clauses with past perfective and present perfect verbal forms, with distinct inflectional endings marking the subject of transitive and intransitive verbs, e.g.:
With the exception of the Jewish dialect of Sulemaniyya and Ḥalabja across the border in Iraq,Footnote 50 in other NENA dialects throughout the NENA area the subject of transitive and intransitive past perfective and perfect verbs have the same inflectional markers.Footnote 51
The Sorani Kurdish dialects of western Iran and Hawrami exhibit ergative alignment in clauses with past perfective and present perfect verbal forms.Footnote 52 It is likely that the ergativity of the Jewish dialects of western Iran, therefore, arose by convergence with the neighboring ergative Iranian languages. The Christian NENA dialect of Sanandaj has, as in other features, not converged so closely with the Iranian contact languages and retains the general NENA practice of expressing the subject of transitive and intransitive past perfective verbs with the same inflectional markers,Footnote 53 e.g.:
Persian and Bakhtiari have accusative alignment.Footnote 54 Neo-Mandaic is likewise accusative in its alignment. This is due to the fact that it has been in contact with Iranian languages with accusative alignment, such as Persian and Bakhtiari, and also with Arabic, rather than ergative Iranian languages.
Concluding Remarks
This brief overview of the Neo-Aramaic dialects of Iran has focused in particular on the factors that have brought about divergence between the dialects and their linguistic change.
It has been shown that the social factor of communal identity has played a fundamental role in the divergence of the NENA dialects spoken by Jews and those spoken by Christians. This factor, indeed, was no doubt operative in the maintenance of clear distinctness between NENA and Neo-Mandaic, which is spoken by Mandaeans. NENA and Neo-Mandaic are genetically different subgroups of Neo-Aramaic, but there is evidence that they may have been in geographical contact at an earlier period.
Many features of change in the Neo-Aramaic dialects have been brought about by convergence with patterns in the Iranian and Turkic languages of the region. Most of the discussion has concentrated on such contact-induced changes in the NENA dialects. Different changes have occurred in the NENA dialects in accordance with different structures in the contact languages of particular areas of western Iran. Study of these contact-induced changes provides evidence for the geographical origin of the various dialects. It has been shown, for example, that the Jewish NENA dialects of the Urmi region show greater convergence with contact languages spoken to the south of the Urmi region. The dialects of the Christian Urmi NENA cluster, by contrast, exhibit greater convergence with contact languages in the northern area of western Iran.
It has also been shown that the Jewish dialects have in some features converged more closely with structures in the contact languages than is the case with the Christian dialects of the same geographical area. The main case study for this is the divergent development of the Jewish and the Christian NENA dialects of Sanandaj.
