I would like to take the opportunity offered in this congress to resume a problem which I had tackled some ten years ago, in a previous congress held in Paris and devoted to Kurdish Studies[1].

I had expressed a tentative theory about the place of Kurdish among the Iranian languages, within the framework of the historical grammar of that language[2]. However, my line of argument was not sufficiently documented. This is the reason why I wish to submit to this audience a new account of the problem, putting forward new and more detailed arguments for the solution which I am presently proposing.

The diachronic method examines the ways in which a language has made changes over some period of time. When used to establish the genetic affiliation of different languages, this method allows to determine the common ancestor, from which they all started off at some time in the past and those languages are to be considered as no more than regional varieties. They share a common origin and constitue a language family. We know for a long time that Kurdish, Zaza, Persian, Gilaki, Pashto, Ossetic, etc. belong to the same group, which is called among linguists, the Iranian family of languages.

But a serious problem may arise when scholars try to determine a classification of different languages inside a recognized family. Some languages are more closely related within a family and the linguists try to establish a subgrouping of different varieties. Inside each subgroup, it is also necessary to distinguish further groups, or branches, until we reach the level of the languages as they are actually spoken.

If we start from the lowest level, we can say that Sorani, Mukri, Kurmanji, and the Kurdish dialects, spoken from the Caucasus till the Turkmenistan, all derive from a common Kurdish language and constitute a subgroup of their own. But what is the relationship between the Kurdish subgroup and the Persian one ? Shall we consider that Persian and Kurdish are a common subgroup inside the Iranian family or two separate subgroups ? Is Kurdish derived from Persian or Persian from Kurdish ? And what can we say about the other Iranian languages ?

The historical approach has allowed linguists to distinguish, at the highest level, two great groups inside the Iranian family. The Eastern group, including Pashto, Munji, etc., the unity of which is still a disputed problem, will not retain us here, because Kurdish belongs undoubtedly to the Western group.

Unfortunately, the historical method has its own limits. In this particular case,  we lack information, for a long period of time, about the Western group, with the exception of the Persian branch. We have in our possesssion the Achaemenian Old Persian inscriptions, from the 6th till the 4th century BC. and documents in Middle Persian (Sasanian inscriptions, Mazdean Pahlavi and Manichean texts). All this provides us with a very good knowledge of the evolution of the Persian branch, but we do not know anything about other Western Iranian languages untill a very late time.

However, the historical approach allows us to draw a very important conclusion. Since a time as early as the 5th century BC., Persian had developped its own characteristics and had acquired a shape which made it different from the other Western Iranian languages. An example will be sufficient here. The old Indo-European voiced velar *g(h) was represented by d in Persian and by z in the other languages, as Persian dān- “to know” (Modern Persian mi-dān-am) ank Kurdish zān- (Sorani a-zān-im)[3], or the old personnal pronoun “I” (Latin ego) which is represented by Old Persian adam, but in Kurmanji, Tāleši, etc. by az (also the Eastern Pashto zə).

Unfortunately, as I said previously,  the historical method does not give us any possibility to elaborate a satisfactory classification of the Western Iranian languages. Nevertheless, we can draw our attention to a very important remark. As soon as the middle of the first millenium BC., Persian was distinct from the other Iranian languages. This is not at all surprising. Since the scattering of the primitive Iranian people, five or six centuries before, the course of time did favour the emergence of separate dialects inside the different Iranian tribes. Therefore, it is not highly speculative to assume that the languages we know in modern times had also acquired their own characteristics in a remote past. In other words, Kurdish, as well as Tāleši, Gilaki, etc., had begun the evolutionary process which led them eventually to the modern shape which we know currently.

If the historical approach is a failure to give us any certainty in relation to the classification of Western Iranian languages, we can go no further than to turn to typology.

Typology is a method which compares different languages according to their structural features. Comparison does not imply that those languages are genetically related. You could compare Chinese and French, as well as Kurdish and Japanese. There is a common agreement that typological similarities cannot be considered to be indicative of genetic relatedness. The stance that I take here on this issue is slightly different and less doctrinal, if I may use this word. Nothing cannot prevent us comparing, in the typological way, languages which share, as we know, a common ancestry, the subgrouping of which cannot be established by the way of the historical method. To put it otherwise : typological comparison may provide us with a clue which can help us to solve our problem in an historical view.

Before going any further, it is necessary to give a brief account of the most important members which constitute the group of  the Western Iranian Languages.

I shall call here “Medo-Caspian” a group of languages which are spoken in Eastern Turkey, in the Iranian Azerbaijan (the Ancient Media), on the shores of the Caspian See, in Khorassan, including some isolated languages in Afghanistan, and also in different parts of Central Iran. It is important to emphasize this point that the term “Medo-Caspian” is used here for the sake of convenience and does not imply at all that those languages constitute a homogeneous group.

First of all, we find Zaza spoken by Kurdish people in Eastern Turkey, on a territory surrounded by Kurmanji and approximately limited by the towns of Erzincan, Erzurum, Mutki and Diyarbakir. Its old name “Dimli” has often been compared with “Daylam”, the ancient name of Gilān, and it is suggested that that language has its origins in the Caspian region. This is an attractive but still speculative hypothesis.

In Western Iran, on the territory of the Ancient Media, called nowadays Azerbaijan, we find the remnants of the Iranian language which was spoken there before the Turkish invasions. According to a suggestion of the Iranian scholar Ahmed Kasravi[4], it is convenient to call that language Āzari, which must not be mistaken for Turkish Azerbaijani. Āzari survives in isolated small areas : in NW Azerbaijan (Harzand, Dizmār), in NE (Xalxāl, Tārom), in the vicinity of Qazvin and some other regions.

In an eastward direction, along the western shores of the Caspian See. we find the Tāleši dialects, which are spoken in the Republic of Azerbaijan and in Iran, then the Gilaki dialects, in the city of Rasht and in all Gilān, and, lastly, Māzanderāni, till the border of Turkmenistan.

Slightly to the South of Māzanderān and at some distance East of Tehran, the town of Semnān has remarkably preserved its own language, the Semnāni, in spite of the constant pressure of Persian, spoken by the numerous travellers who take the road to Khorassan. Around Semnān, scattered in a desert region, is found a cluster of languages, Sorxei, Lāsgardi, Sangesari, etc. Those languages are still an unresolved problem. They are not related to each other and they cannot be considered as dialects of Semnāni. Their presence in the area could be explained as a consequence of the frequent invasions or forced migrations of populations, which  occured very often in that region.

In this respect, we still have to mention two isolated languages of Afghanistan. Parāči, spoken in some inaccessible valleys, in North of Kabul, and Ormuri, in South of Kabul and in a small region of Pakistan, Kanigram. In spite of their geographical situation, these two languages are not considered as part of the Eastern group of Iranian Langauges. They belong undoubtedly to the Western one.

If we come back to Eastern and Central Iran, we find a number of languages or clusters of dialects, often scattered in isolated areas.

First of all, a very interresting language, Gōrāni, spoken by some Kurdish tribes, in different small areas : Gawhāra (North of Kermanshah), Kandula (eastwards), Hawrāmān (West of Sanandaj), Bājalān (near Qasr-e Shirin), etc. This situation recalls that of Āzari and it induces us to suppose that Gorāni was once spoken on a larger and continuous territory. On the other hand, as it was the case with Zaza, Gōrāni has often been linked to the Caspian region (the name Gilān seems to be a frequent place name).

Other isolated languages are spoken in Iran : Tafreši dialects (near Hamadān and Sāve), various dialects of the Fārs province (including Sivandi), which have not yet been investigated enough, but they are not very important for our purpose, with the exception however of the Kaviri dialects, spoken in the Dašt-e Kavir, which offer very interresting features for our problem.

We shall end this review by Balōči, a very important language, spoken, in different and numerous dialects, on a considerable stretch of land, extending from South-Eastern Iran to South-Western Pakistan, as far as Karachi, with some isolated groups of Balōč, in Afghanistan and Turkmenistan. There is a general agreement that the Balōč tribes migrated into their present location, from the vicinity of the Caspian See, no later than in the late Sasanian times.

The main purpose of this paper is to show that all the languages of the so-called Medo-Caspian subgroup share common specific features in deep contrast with an other subgroup which is made up of Persian[5], Kurdish and the Kermanian dialects.

The Kermanian dialects constitute a very important cluster spread over a large area of Central Iran, limited by the towns of Kāšān, Ispahān, Kermān and Yazd. None of those dialects has ever gained the position of a literary language. A point of great importance for a linguist is that the Kermanian dialects seem to be at home in their present location. No trace of any substrate is detectable. This is an exceptional situation when one thinks that most of the Western Iranian languages have migrated in the course of history, with the valuable exception of Persian, which is attested in the Province of Fārs for a long time, as we have seen. This fact does not make easy the elaboration of a convincing theory about the classification of those languages.

 Actually, the large area where the Kermanian dialects are spoken seems to correspond to the Ancient Carmania which the Greek historian Herodotus locates on the North and East of the Fārs province, from which it is separated by moutains. Herodotus gives the strong impression that Persians and Carmanians are peoples who look very much alike. It is needless to say that the name of the old country survives till nowadays in the name of Kerman.

We can now proceed to our problem, beginning with the verbal system.

Kurdish, Persian and the Kermanian dialects have similar formations of the verbal conjugation. Two features are here important. First, tenses are formed regularly from two stems : present and past. The present stem is the continuation of the ancient present : Old Persian kunāmiy “I do” and the past stem is based on the old perfect participle : karta- “done” :

             Persian             kard                kon-

             Abyānei           kard                kar-, etc.

Most of the Medo-Caspian dialects have a very different present tense formation. They use mostly a composed expression made of the ancient present participle kunant- “doing” and the verb “to be”. This is quite similar to the English “I am doing” (but usually with the meaning “I do”). The affix -nt- changes into –nd-, or more often is reduced by assimilation to ­–nn-, -n-. This formation has both habitual and actual present meaning, in contrast with the subjunctive which has no affix and describes a single action :

In Zaza, the present tense is formed with an affix –an- added to the root : barm- “to cry” : pr. st. barm-an- ; ras- “to arrive” : ras-an- ; kiš- “to kill” : kiš-an- ; kar- “to do” : kan­- (< *kar-an-), etc. The subjunctive lacks any affix and is formed with the bare root and the common Iranian prefix bi-.

The Āzari dialects follow the same pattern, but the old participial affix is sometimes better preserved as in Harzandi kond-en “I do”, in contrast with subj. kon-im “that I do”. Some dialects have lost that formation, but preserved it semantically in creating a periphrasis, having the meaning “I am at doing” : Keringāni mun-an be-kārde, Šāli az kerā paj-əm “I am cooking” (kerā is a particle expressing continuity), Kajali az mišim kore raz “I am going to the garden”. The Tāleši (nothern dialects) has regularly developped such a formation which has become the regular paradigm of the present : az karde-da-m “I do”, which must not be confused with the past kard-əm-e “I did”.

Gilaki seems to have lost this formation, probabily under pressure of Persian, while Māzanderāni has partially preserved the affix of the present participle, but it has merged at times with the personal endings : nevis-eni “you write”, contrasting with subj. be-nevis-i “that you write”, nevis-en “he writes” and be-nevis-e “that he write”. Šamerzādi, a very conservative Māzanderāni dialect, spoken outside Māzanderān, not far from Semnān has pors-ammam, pors-anni “I, you ask”, opposite subj. ba-pors-am, ba-pors-i “that I, you ask”.

Semnāni does not keep any trace of the affix, but, as one can expect, the surrounding dialects, isolated as they are in the desert, are more conservative. Sorxei has pors-enda, pors-ende “you, he ask/s”, but subj. pors, pors-ey “that you, he ask” ; also in Lāsgerdi. Sangesari has the same : pors-endi, pors-ende “I, you ask”, but subj. por-si, pors-e “that I, you ask”.

Parāči has remarkably retained the old present participle : ān-em xartön, tū-ē xartön, ö-ē xartön “I am, you are, he is eating”, but Ormuri has lost it. Instead, it developped a periphrasis : bu xra “he eats, is eating”, with a particle bu which recalls a similar construction in some Āzari dialects.

Gōrāni has no affix in the present indicative, which has probably been reshaped under the influence of both Persian and Kurdish, but the imperfect has surprisingly special personal endings which could be explained as tiny remnants of the old participle affix : karene, kareni, karenme, karende “I was, you were, we were, you were doing”.

Balōči has also lost the participle formation of the present, but it has on the contrary developped progressive expressions, as a continuative ravagā un “I am going” and an iterative ravān un “I am (still) going”.

Persian, Kurdish and the Kermanian dialects have followed a very different way. As we can see in Middle Persian, the old indicative present kunāmiy, became kunam, kunom, kunēm, all three free forms with the same meaning, as a general present “I do”, as a progessive present “I am doing”, as a future “I shall do”, and also as a subjunctive “that I do”. This is the so-called aorist, which soon was felt unable to describe the process of the action. A future-subjunctive was then formed with the loose particle be, well spread among almost all the Iranian languages.

The durative aspect was first expressed by an adverb, in Middle Persian hamē, Persian hami, the meaning of which is “always, all the time”. In the Persian of the time of Fedowsi, this is still a free particle : hami kār konam, kār konam hami, etc. “I work [I make work]”, but in classical Persian kār hami konam became usual and the particle was prefixed to the verb : (kār) mi-konam.

The prefix appears also in a reduced form a- (sometimes also e-), with lost of the dental : Suleymāni, Sūrči, some Southern Kurdish dialects a-ka-m “I do”. The same in Tafreši : Vafsi a-vesin-om “I sit”, Āštiāni e-š-om “I go”, etc. The same appears very often in the Kermanian dialects : Abyānei e-kar-ān, Abuzeydābādi o-kor-ō “I do” (with e- > o- by vocalic assimilation, 2d sg. e-kar-ē “you do”). The prefix may also be completely lost as in many Southern Kurdish dialects and in Kermanian : Nāini kir-i “I do”, or even be displaced : Gazi ker-ān-e, Sedei ker-ōn-e, etc.

The t- prefix is also the rule in all Kurdish and Kermanian dialects, but its use is here restricted to some verbs beginning with a vowel and this prefix seems to be incorporated to the stem. Kurmanji az tīnim “I bring”, Wārmāwa min tēm “I come”, min tērim “I bring”. In Zakho, Gulli, etc. az tēm “I come” is the regular form. In Southern Kurdish dialects, we find similarly min tyām “I come”, min tyārim “I bring”, etc. In the Kermanian dialects this form is much more spread, even in dialects which use otherwise the reduced form a- of the prefix, or no prefix at all : Qohrudi a-t-ömödūn, Ardestāni a-t-ōmoyō, Nāini t-omi  “I was coming”, Abuzeydābādi o-t-ō, Ardestāni t-orō  “I come”, Abuzeydābādi a-t-ārō, Abyānei e-t-orān, Tāri a-t-ārō, Nāini t-āri “I bring”, and many other verbs.

We find also a stressed deictic suffix which is shared by the three members of our subgroup and has a demonstrative force, or is identical to a definite article, mostly in its referential function. Persian has such a suffix, the use of which is however restricted to the spoken language : pesar-é “the (previously mentioned) son”. In the same way, Sorāni has kur-á, which is preferably used with the demonstrative : aw kur-á. The Kermanian dialects have exactly the same stressed suffix (also –e, -o) : Qohrudi yene-y- á “the woman”, Abuzeydābādi pe-y-á “the father”, Tāri pǖr-á, Varzenei por-ó  “the son”, etc.

As for the adjective, it takes often a suffix, which is used only in the attributive position. The origin of this suffix is not known. In Āzari dialects, we find Keringāni čuk-a mert “good man”, Hazārrudi sor-a xalā “red dress”, Čāli kov-a qāter “grey mule” and in Tāleši, Lenkorāni yol-a do “great tree”, Māsulei isbi-a asb “white horse”, Māsāli xās-a kina “pretty girl”. In Gilaki we have rāst-ə gəb “true word” and in Māzanderāni xor-ε kitob “good book”. In the same way, Semnāni pāk-a šiša “clean bottle” and also, but without any suffix, Sorxei esbi asm, Sangesari esbi asb “white horse”. No suffix either in Parāči : činö γus “small house”, nērök ösp “male horse, stallion”, and in Ormuri spēw yāsp “white horse”. In Balōči however, the suffix is regularly used : mazan-ēn asp “big horse”, ispēt-eⁿ jāmag “white shirt”.

                        1          *mana pitā                 “mine father (where “mine” is the old genetive)

                        2          *hača mana pitā        “of me father” (with redundant preposition)

                        3          *manahya pitā           “of-mine father” (with double genitive)

The ezāfe construction is also found in Zaza and in Gōrāni, two languages which we have provisionally connected to the Medo-Caspian subgroup. We could suppose that the ezāfe construction has been borrowed from Persian or Kurdish, but this is unlikely. Moreover, Gōrāni has a specific adjectival ezāfe which does not exist elsewhere. Therefore, we are obliged to consider here a new division in the so called Medo-Caspian subgroup.

We have seen that Persian and Kermanian offer the same line of evolution, although they are not related genetically. This is not at all surprising. Linguists know for a long time that neighbour languages have influence on each other. But the similar convergence in Kurdish is amazing, since this language is spoken today at a great distance from both Persian and Kermanian. There is only one way to explain this : when Kurdish acquired its modern shape, it was spoken in the vicinity of the other two, probably on a territory roughly limited by South Media, North Persia and East Carmania.

In his description of the country East of Carmania, the Greek historian writes that it was occupied by three great tribes : the Achaemenidae, the Magi and the Pateischoreis. The first two tribes are the Persians and the Medes. The third name is attested in an inscription of Darius I, on his tomb at Naqš-e Rostam. as the tribal name Pātišuvariš [Pātišxwariš] of Gaubaruva, a high dignitary. This is undoubtedly an Iranian name, even if its etymology is still disputed. Strabo gives also the names of two subtribes : the Amardoi (or Mardoi), also a good Iranian name, and the Kurtioi. We can, at least, conclude that the latter is the name of an Iranian speaking tribe, maybe that of the ancestors of the Kurds.

Before this, this part of the Caspian shores could have been occupied by the Zaza. This position would explain its connection with Gilān, the former name of which was Daylam. From a linguistic point of view, it explains also some features which Zaza shares with the languages of the region, but also the ezâfe contruction which it has in common with his ancient southern neighbour, Kurdish.