The language of the Sahname in historical and dialectal perspective ‘The Sahname is not only the most important of Persian epics and a monument of Iranian national tradition and culture, it is also of paramount importance for the history of the Persian language. Linguists generally prefer prose works when investigating a certain language and its grammar, because the language of any epic or poetic work may be affected by stylistic and metrical peculiarities. A living popular epic like the Sahname, however, should rank next to prose works in usefulness for linguistic purposes. Most Iranians know that the language of the Sahname belongs to a very early phase in the history of (New) Persian, earlier than that of “classical” poets such as Hafez and Sa‘ét. Although more educated Iranians are certainly able to read and understand the Sahname (henceforth Sn.), they do have problems understanding certain archaic words and constructions. This is no surprise given the long period of 1000 years that has elapsed since the Sn. was composed. Iranians are less aware that the language of the Sn. is not only archaic but does also show some dialectal peculiarities in comparison to modem Persian. Some words from the Sn. that seem to be archaic today, are in fact loans from other Iranian languages or local varieties of Persian. ‘The aim of this paper is to examine the historical and dialectal evolution of New Persian, and to determine the Sn’s place within this evolution. For the sake of analysis, it is necessary to separate the historical from the dialectal factors; later it will be seen that historical and dialectal factors sometimes act in combination and cannot be clearly separated from each other. The historical perspective With respect to the historical development of Persian, the first thing to emphasize is its continuous character. Chronologically, the language of the Sn. (10/11. cent. A. D.) would take approximately an intermediate position between the Middle Persian (Mp.) of the 3rd and modern New Persian (Np.) of the 20th century. The advent of Islam in Iran certainly caused a major cultural break in Iranian history and also had a heavy influence on the Persian language. But this influence was gradual, and even the intrusion of so many Arabic words into Persian was a gradual process that took centuries to reach its peak. The vocabulary of the Sn. still had many fewer Arabic loan words than that of later literary works. In many ways, the vocabulary of the Sn. resembles more that of Mp. than that of modem Np.,€.g.: Mp. Sshname modern Np. Words that are obsolete today: xi “thing” ( Mp) (obsolete) dar “here” (= Mp.) (obsolete) Words that changed their meaning: Sahr "country" (= Mp.) "city" veda "pure, holy" (= Mp.) “special” ‘These changes are well-known, as well as the fact that some phonological developments ee ee the Sn. eg. Mp. aba "with", abar"upon”> Sn. (a)ba, (a)bar> Np. (today) ba, bar Mp. andar “in”, hamé "always"> $n. (an)dar, (ha)mé > Np. dar, (verbal prefix) ml- Mp. 8&r “lion”, kor “blind” > Sn. (idem) > Np. tr, kar (majhill vowels)! | This applies to the Persian spoken in (most parts of) [ran only. In Afghanistan and Tajikistan, the ‘majhutl-vowels are still distinguished from the ma‘rdf ones to this day.Itis less generally known that the intermediate position of the language of the Sn. (between Mp. and 20th cent. Np.) can also be seen from most other areas of grammar. Two examples should serve to demonstrate this: The expression of reflexivity Mp. Sn. Class.-Pers. Np. tan T xweS / xweStan > xweStan > — xweStan / xwad > xod (direct object) xwed > xweS> — xweS/xwad > xod (possessive) Modal verb constructions full infinitive > short inf. > short inf./ke+aorist> (ke +) subjunctive (eg. bayad raftan baad raft’ bayad raft/ bayad ke ravad _bayad (ke) beravad) ‘The case of reflexivity is a particularly convenient grammatical feature in examining the continuity in the evolution of Persian. Mp. xwéitan was gradually replaced by xwad (later xod) beginning only after the 10th century A.D. It is possible to trace in statistical terms exactly how this took place. Similarly, the evolution of the Np. modal constructions (from bayad raftan to bayad beravad) has developped only during the Np. period. The schemata above are simplified; as I have discussed both features in a forthcoming publication,? I will not go into detail here. Other features that exemplify historical-linguistic continuity include the word order of noun phrases (H = head, M = modifier [i.e., genitive or adjective attribute]): Mp. M-H or = H(t(g) M Sn. M-H or HiM Np. (today) only H-eM ‘The Ezafe construction, which is the only productive way to form a noun phrase today, was already wide-spread in Mp.? But the older reverse construction still occurred with about the same frequency in the Sn.4 ‘The constituent order of nominal compounds is connected to the word order of Noun Phrases. Please observe the following possessive compounds ("Bahuvrihis"): Mp. MH (weh-den "having the good religion") Sn. M-H (taham-tan “having a strong body” [epithet of Rostam]) Np. (today) M-H/H-M (siyah-post “black-skinned” / kalle-pak “empty-headed") 2A Linguist's Fresh View on ‘Classical Persian’, to be published in the Proceedings of the 4th “> European Conference of Iranian Studies (held in Paris, 6-10 September 1999) (In the series Cahiers de ‘Studia Tranica, Paris). 3 The word order of Mp. noun phrases has been treated by Mary Boyce, “The use of relative a in Western Middle Iranian”, Indo-Iranica. Mélanges présentés d Georg Morgenstierne d occasion de son soixante-dixiéme anniversaire, ed. G. Redard, Wiesbaden 1964, p. 28-47, and by Christopher J. Brunner, A syntax of Western Middle Iranian, Delmar 1977, p. 10ff. Both provide ample examples but not a comprehensive analysis of the various possible NP constructions of Mp. A quick view on a selection of Inscriptional and Manichacan Mp. texts of the 3rd century shows that in that period, the M-H and H-M_ constructions occurred with about the same frequency. ManMp. examples (adj. only): M-H wuzurg Sédth “great joy”, HM kar I wuzurg "great work", H-M bozegar wuzurg “great saviour 4 Examples: M-H ndrastda turunj “untipened lemon”, téra Sab “dark night", H-iM Sab-i tera “dark night", Bahram-i néw “the brave Bahram”. Cf. Np. today: H-e M Sab-e tartk “dark night". In Np. today, MH is possible only in poetic or archaizing usage.In both Mp. and the Sn., there were only head-final Bahuvrihis.5 They continue to be productive to this day, but recently a new Bahuvrihi type with reverse constituent order (the kalle-pak-type) has become very productive in Np (apparently starting from the substandard language). Both Ezafe constructions and nominal compounds show that the language of the Sn. shows a stronger affinity to Mp. than to modern Np. ‘Summing up the points made so far, the language of the Sn. represents a historical stage of Persian that appears to be in many respects - lexical, phonological, morphological, syntactical - closer to the Mp. of the 3rd century than to the Np. of the 20th. Nevertheless, it is undoubtedly justified to call the language of the Sn. "New" rather than "Middle Persian”. This is due mostly to extralinguistic factors, namely to the cultural shift and to the break in literary tradition that the Arab invasion brought about, and to the fact that the Sn. constitutes one of the first links in a new and continuous literary tradition, the Persian-Islamic one. The dialectal perspective The place of the language of the Sn., in the history of Persian, corresponds more or less to what one would expect from the outward chronological data. The question of dialectal features of the Sn., by contrast, is very complex and contains a number of unresolved problems. Any investigation into the dialectal characteristics of the Sn. should still start from Wolfgang Lentz’ pioneering study.” Lentz showed that many Np. words (especially those that occur in the Sn.) cannot be properly derived from Mp. or Old Persian (Op.), and therefore must be loans from some other Iranian language(s) or dialect(s). Lentz identified Parthian as the main source from which Persian had borrowed these words. He called the “Parthian” loan-words "northem” in the sense that West Iranian languages are said to be divided into two subgroups, northwestern and southwestem. A simplified tree diagram may serve to show the dialectal evolution of Wester Iranian: Proto-Iranian West-Iranian East-Iranian SW-Ir. NW-Ir. ca. 500 B.C. Old Persian ("Median") ca. 300 A.D. Mp. Parthian ‘ 900 A.D. - Np. Kurdish Gilakt TaleS1 etc. 5 With very few exceptions, e.g. tan-d(u)rust “healthy” in Mp. and Sn. © Please note that the syntax of compounds may differ from that of the NP’s in a given language. ‘Compounds may mirror syntactic rules of the earlier stages of a language, which have become “frozen” into lexical rules. 7 "Die Nordiranischen Elemente in der neupersischen Literatursprache bei Firdosi*. Zeitschrift fiir Indologie und [ranistik 4, 1926, p. 251-316.Although Parthian is certainly a good candidate for such a source language, I think “Parthian” as a blanket term covering most NW elements in Np. has been used too widely and uncritically since Lentz’ investigation. Cross-linguistic borrowing has been wide-spread among Iranian languages and dialects from earliest times. ‘The NW-Iranian loanwords that entered Persian may be classified into several chronological layers. I will propose a four-fold classification here, and illustrate each class with a selection of some typical examples: 1. NW loanwords (to Persian, attested in the Sn.) from the pre-Achaemenian period :§ buzurg “big” < Mp. wuzurg < Op. vaz(a)rka “id.” (NW because of -2-)? mihr “love; sun" -hr-) zar(t) "gold" -rr-. An altemative, inner-Persian explanation for the f- of this much discussed word hhas been given by P. O. Skjaervo, "Farnah-: mot méde en vieux-perse?", BSL 78, 1983, p. 241-259. ™ The “corrupt” ZorMp. writing dmyst’n (transcribed zmyst n) is usually considered a scribal error (see D. N. MacKenzie, “Notes on the Transcription of Pahlavi", BSOAS 30, 1967, p. 20, fn. 12). As there seem to be only three ZorMp. words with an initial z- that are regularly “miswritten” d- (besides zamestan: zamtg "soil", zaman “time"), the first two of which have replaced NW z- for SW d-, one may ask if ZorMp. dmyst’n and dmyg are really “corrupt”, or if they could not represent an early-Mp. linguistic reality where (as with "winter" in ManMp.) the NW loans with z-had not yet entered Mp. 12 ManMp. biehk is from *bigadka-. ZorMp. bizeik (cf. Pth. bizeik) is due to metathesis of i/2, the p- of Np. piziik is dissimilated from b-. aemury “bird” < Pth. mury, but Mp. (Z./M.) murw “id.” tuxm “seed, family” < Pth. tuxm, but Mp. (Z./M.) to(h)m “id.” durdy “lie” < Pth. durdy, but Mp. (Z./M.) dro(w) “id.” yoy "yoke" < NW (Pth.?) yoy, but Mp. (Z./M.) joy “id.” barzan “quarter” < NW (Pth.?) barzan, but Mp. (Z.) walan “id.” NW words that have (partly) been replaced by SW forms in later Np.:!3 burz "high" ~rr- took place in Persian.'® As for Sahr "country; (later:) city", Armenian axarh "world; 3 Besides the three NW words burz, pir and arj the SW equivalents bald, pusar and arz do also occur in the Sn. “4 Lentz had already called some of the loans “Median” and so detached them from the other ones chronologically (loc. cit., e.g. p. 260, 287), but he did not chronologically distinguish any of the other, loans. James Russell has suggested that some of these loanwords (¢.g., mihr) may indeed be not from - » Median but Avestan. This seems to be well possible for words from the religious sphere like mihr. However, since Avestan cannot be clearly distinguished from Median in terms of phonology (both: languages represent largely the same conservative stage as opposed to the Op. sound changes), this is difficult to prove or disprove. 1S Apparently, at the beginning of the Sasanian period (3rd cent. A.D.) not all regional variants of Mp. had been influenced by NW/Pth. to the same degree. 16 If, with Henning (“Mitteliranisch", Handbuch der Orientalistik, Erste Abt., IV. Bd.: franistik, 1. Abschnitt: Linguistik, ed. B. Spuler, Leiden-K6In 1958, p. 97) ManMp. indeed represents the form of Persian that was spoken at the court in the capital Ktesiphon. 17 This is why Lentz, loc. cit., p. 278, calls it a "Median" loanword. '8 Cf. Pth. zarn “gold”. We are not able to date this sound change, but the occurrence of the same ‘change in East Iranian Khotanese could point to a relatively early date, when the Khotanese speakers were still in closer contact with Persians.country” shows that it may be a quite early Kulturwort from the political sphere.!9 These words may have been borrowed in pre-Parthian, if not late Achaemenian times.?° For the 3rd group of loans, Pth. influence is still possible. But for the larger last group, it is problematic. During the 3rd and 4th centuries A.D., Parthian lost its function as an official language of the Sasanian empire. Afterwards, its use was again restricted to its home province Parthia (in NE Iran).2! Some words show a mixed SW/NW form in Sn. and ZorMp., as opposed to a “pure” SW form in ManMp., e.g. Sn. durust “right” < ZorMp. drust, but ManMp. drist, Pth. drust; Sn. javed “eternal” < ZorMp. javed, but ManMp. jaydan, Pth. yavedan. These forms have been explained as "NW forms in SW pronounciation”.22 But it would also be conceivable that they derive from yet another dialect, lying somewhere between Mp. and Pth., and thus between SW and NW. There are several reports that provide useful information on the linguistic situation in Iran at the end of the Sasanian empire (7th cent. A.D.), the most well-known of these being that of the Arabic historian Ibn al-Mugaffa‘.2? It says that in NE Iran, Parthian (called Pahlavt by Ibn M.) was still spoken, but "Dart Persian” was also spoken there. Reports from the 10th century mention only "Dart Persian” as the language of NE Iran. It seems that towards the end of the Sasanian empire, Pth, was already declining and restricted to folkloric usage, and soon after the Arab invasion it became extinct as a living linguistic unit. It certainly lived on in local dialects for a while, but among the modern NW-Iranian dialects of today, there is no single one that could be called a descendant or even a close relative of Parthian. ‘The post-Sasanian loanwords in the Sn. must have entered Persian approximately during the 8th or 9th centuries. If they are from Pth., then "Parthian" must be understood here as a group of dialects, rather than as a unified language. It is probable that still at that time, there existed wide-spread and popular literary traditions in dialects, i.e. also in “Parthoid” dialects. Indeed, had these traditions not been wide-spread and popular, it would hardly be conceivable that they should have influenced Persian so heavily. It is possible that a strong Pth. literary tradition had developped, during the Arsacid and Sasanian reigns, that continued to be transmitted into the Islamic period, maybe into Ferdowsi's time. The two Mp. texts Draxt 1 Asdrtg and Ayddgar I Zaréran, that have preserved Pth. lexical elements, would then be traces of this tradition. I think, however, that lively literary traditions in other dialects probably coexisted and competed with the Pth. one. The local Fahlaviyyar dialect poems, which are attested in various regions of central and NW Iran from about the 13th century A.D. onwards, would be * later examples of these. i 19 It is the only Iranian loanword in Armenian (known to me) where the x- of an original Iranian *xi- is preserved (if with metathesis), cf. other words like Arm. inorh < Iranian *xnao8ra- “gratefulness”, ‘which must have been borrowed after the Iranian change *x-> 3-had taken place. The InscrPth. spelling Aitr (= Rahr/) seems to reflect a stage of Parthian before this change. 20 Even if the source language was Parthian, it was not the Parthian that we know from the 3rd cent. inscriptions and Manichaean texts. 21 Only in Central Asia was it still used as a missionary language of Manichaeism in later times. 22 Bg., Lentz, loc. cit., p. 286. 23 See G. Lazard's comprehensive study “Pahlavi, Parsi, Dari: Les langues de I'lran d'aprés Ibn al- Mugaffa*, Iran and Islam (FS Minorsky), Edinburgh 1971, p. 361-391.One should also ask if the NW loanwords in the Sn. may be attributed to dialects/languages other than Pth. It is difficult to answer this question because we have no information about the predecessors of the modern NW-Iranian languages and dialects (such as Kurdish, Baluchi, or Gtlakt). On the basis of the present data, one can at least not exclude the possibility that the loanwords originated from Pth. or Parthoid dialects, but they could also be from the dialects of the Semnain region or from some central plateau dialects. The word for "son” makes it only possible to exclude a number of dialects like TaleS1 and Zazaki, as possible sources for the NW elements: Ph on Semnant —Natanzi (CD) Gilakt. Talest Zaz. “son” pubr —par/pusar_ pir por poser z0e kag ‘There must still have been Parthoid dialects around the time the Sn. was written, which looked quite different from the predecessors of modem dialects like Semnant and Gilaki This is already clear from the fact that most NW dialects of today have preserved a system of at least 2 cases, while Pth. had given up all case distinctions.2 A glimpse of what these dialects looked like may be gained from a dialect that was spoken by the religious group of the Hurdfis in Gorgan, in the 14th century, of which we have some knowledge. In this dialect, the verb "to come” was ds- as in Pth. (< *a-i-sa-, with inchoative -sa-), while in all NW dialects of today it is ay- or the like (< *d-i-, e.g. Semn -y- Bal. ay-, Kd. (h)e-), So far, I have only discussed lexical issues. I will now in passing raise the question (that was not asked by Lentz) how the grammar of the Persian in the Sn. relates to that of other Persian texts from the same period. Some grammatical characteristics of the Sn., as compared to ENp. (= Early New Persian) prose works from about 950-1050 A.D., would include:?7 ENp Sn. 1. Personal suffixes attached to prepositions seldom only with bar, pes 2. mar without -r& seldom (not as case prefix) never 3. Certain prepositions: - baz) as "back to” often never - ba(6) as comitative "with" often seldom - zt "towards" only in mss. from E. Iran occurs 4. Present perfect of type kardastam only in mss. from NE. Iran occurs 24 Although Gtlakt shows a SW (loan) form here, one should not exclude it as a possible source < language, because it may still have been “more NW” in the 89. centuries. 25 Like with the word for “son”, Tal. and Zaz. have another etymon for “daughter”: kine resp. keyn 26 In most contemporary NW dialects, there is a masc. obl. ending -1 < Olr. *-ahya. The Mp. forms duxtfduxtar (Ret JObL.) “daughter” may be reflected in Sn. duxt(ar) (= Ret. = Obl.) “id.", where the oblique ending -ar had lost its case function but had not yet been fully generalized (as it would be in later Np.). Alternatively, the forms without -ar could show an influence from Pth. duxt (Ret/Obl). 0: from dialects like Natanzt dot (for the Mp. oblique ending -ar after relation terms, see N. Sims- Williams, "Notes on Manichaean Middle Persian Morphology”, Stir 10, 1981, p. 165-176, and P.O Skjaervo, "Case in Inscriptional Middle Persian, Inscriptional Parthian and the Pahlavi Psalter”, Stl 12, 1983, p. 47-62; 151-181 27 The list is based on Lazard’s La langue des plus anciens monuments de la prose persane, Paris 3I cannot go into detail here, but the schema makes it clear that the Sn. shows some grammatical peculiarities as compared with prose texts from the same period, and at the same time it shares a number of features with mss. that were written in NE-Iran (Transoxania). Persian was not yet dialectally standardized at that time, and many mss. showed dialectal peculiarities, to varying degrees, of the regions where they were written or copied. The Present Perfect I would like to take only one point here, No 4, and discuss it in greater detail, because it leads to a more general point. Studying the present perfect in a text sample from the Sn. of about 700 verses, I realized that the two types of present perfect (karde'am vs. karda'stam) are not distributed homogeneously. The dialectal present perfect in -ast ‘occurs only in the direct speech parts of the text: Direct speech Framework story simple with -ast simple with -ast 1 Sg. rafta'am (4x) -- : = 2Sg. raftat (4x) —_raftastt (1x) - = 3Sg. rafta (15x)29 raftast (9x) rafta (~ 15x) — ‘The fact that forms like kardast occur in direct speech only, may have an important methodological implication. I think that the framework story and the direct speech parts of any literary work should generally be studied seperately at first, because they represent two different kinds of text which may have different authors and different ways of transmission. This is obvious in modem Persian belles lettres, where the author often chooses colloquial registers when one of the acting persons speaks. In the case of the Sn., the raftast-forms of the direct speech parts may also represent a colloquial register or give a dialectal nuance.» If there were indeed minstrels reciting the Sn., they might have tried to recite the direct speech parts in a more lively manner, attributing an individual way of speaking to some of the more prominent acting persons.>! ‘One must certainly try to verify such a hypothesis with the aid of other texts. So far, I have had time to do so with only one other text, NezAmo'l-Molk's Siyasatname. Having read ca. 20 pages (the passage that relates Mazdak's revolt), I indeed made a remarkable discovery. The present perfect (rafte ast) occurs from time to time, but only in direct speech parts, and largely irrespective of aspect, while in the framework story the preterite (raft) is used throughout, as the narrative time. : 28 The greater part of Rostam and Sohrab, vol. 2, pp. 117-173 of Dj. Khaleghi-Motlagh's new edition (Costa Mesa 1990). 29 The characteristic 3 Sg. present perfect rafta, without the copula ast, occurs seldom in ENp. prose texts. It may be a stylistic or metric feature of the Sn. 30 After preparing this paper, I came across the article by A. N. Boldyrev, "Perfekt II v novopersidskom. literaturnom jazyke", Izvestija Akademija Nauk Sojuza SSR (Otdelenije literatury i jazyka, 1946, tom V, vyp. 6), in which the author already saw that in the Sn., raftast-forms occur only in the direct speech parts, Boldyrev favors the colloquial explanation for it. 31 Alternatively, there may have been raftast-forms also in the framework story, but have been replaced later by copyists. A copyist may “emend” the framework story according to what he feels is “correct language”, but leave the direct speech passages (which he feels are spoken by another person) untouched.‘This seems to indicate that in 1th century A.D. Persian, the present perfect was the “general” past tense of everyday speech, while the preterite had the same function for more formal speech. This would not exclude the presence of an aspectual distinction, but the question whether the speech act was a formal one (from the narrator) or an informal one (from an acting person), may have been more important. Applying this to the Sn., it could mean that the usage of present perfect forms in the framework story of Rostam and Sohrab argues for this framework story to be of spoken origin, as well. Ferdowsl, then, would not have taken it from a written source. I am aware, of course, that this can only be one argument for such a hypothesis, which must be fleshed out with all possible historical and philological evidence. ‘Summary By way of summary and conclusion, I would like to emphasize the following points, Firstly, I tried to show that the language of the Sn. should be seen as one instance in a continuous historical development from Mp. to Np. Secondly, I tried to discuss the “Parthian” dialectal elements in Np. in a new light. The historical viewpoint that I added to the dialectal issue, the stratification of the loanwords into 4 chronological layers, showed that we should be more cautious when we use the term “Parthian” in this context. Then, I looked at the dialectal relation of the Sn. to other Persian texts of the time. By so doing, I came across an interesting methodological distinction between the direct speech and framework story parts. Although I cannot claim to have solved any of the vexing problems attached to the text and transmission of the Sn., I hope I have raised some interesting questions, and shown that linguistics may be able to contribute to their solution.