There are remnants left of pre-Islamic poetry within western Middle Iranian languages: fragments of Manichean religious hymns, some poems preserved in the literature of Pahlavi, and poetical pieces in New Persian not following the rules of classical versification.




Pre-Islamic Iran certainly cultivated poetry. The Parthian minstrels (gōsān) enjoyed great fame (Boyce, 1957), and Ebn-al-Moqaffaʿ, in his famous statement about the languages of Iran, presents Parthian (pahlavi) as the language of “royal sessions,” i.e., probably court entertainments, while considering Middle Persian (pārsi) as the language of “mobads and scholars” (Lazard, 1971, pp. 361-62, 385). Yet there also existed a poetry in Middle Persian: tradition has retained the names of several poets and musicians of the Sasanian court, such as Bārbad, Bāmšād, and Nagisā. Of this production, which must have been brilliant, there are hardly any traces left, probably because it was purely oral. There are, however, remnants left of pre-Islamic poetry within western Middle Iranian languages. There are three kinds of these texts:

(1) Fragments of Manichean religious hymns, most of them in Parthian, but also in Middle Persian, belonging to the discoveries made in Central Asia in the early 20th century (see mainly Boyce, 1954, and also 1975, passim).

(2) Some poems preserved in the literature of Pahlavi. Benveniste (1930 and 1932) was the first to recognize that certain texts presented in manuscripts as prose are in fact pieces of poetry. He thus identified Draxt ī asūrīg, a poetic controversy between the goat and the palm tree, Ayādgār ī Zarērān, an epic piece later studied by Utas (1975), and the ‘Pahlavi apocalypse’ called Žāmāspnāmag. Henning (1950), for his part identified, among the Pahlavi Texts edited by J. M. Jamasp-Asana (2 vols., Bombay, 1897-1913), a poem of moral inspiration (dārom andarz-ē az dānāgān …). Tavadia (1950 and 1953) published another of the same kind (was raft hēm andar āwām …). A third one (xrad dāštār pānāg ī jān …) was simultaneously and independently published by Shaked (1970), who called it “A Hymn to Wisdom,” and by Tafazzoli (1972). To these pieces we may add “a rhymed ballad” published by Tavadia (1955). This text had previously been quoted (as prose) by Bailey (1943, p. 195); it was republished and commented by de Blois (2000), who shows that, although written in Pahlavi script, it is composed in a language which differs in dialect from that of Pahlavi literature and is, rather, to be identified with that of literature in Persian (i.e., New Persian): the text is not in traditional literary Pārsi, but in Dari.

(3) Poetical pieces in New Persian not following the rules of classical versification transmitted by works in Arabic or Persian. We find in Tāriḵ-e Sistān a hymn recited at the pyre of Karkoy (foroḵ-ta bādā rōš …), three little satirical verses which might have been composed in 60/680 at Basra by a certain Yazid b. Mofarreḡ (āb-ast-o nabiḏ-ast …), others addressed in 108/726 by the inhabitants of Balkh to a defeated general (az Ḵottalān āmaḏiya …), and in Ebn Ḵordāḏbeh a short piece deploring the ruin of Samarkand (Samarqand kand mand …); these four pieces, with their more or less problematic interpretation, are quoted in Ṣafā (1335/1956, pp. 144-45). To these can be added a couplet wrongly attributed to Abu Ḥafṣ Soḡdi (āhū-ye kūhī …; Lazard, 1970), and three verses discovered in Ebn Ḵordāḏbeh by Šafeʾi Kadkani (Qaysar māh mānaḏ …; Lazard, 1975, p. 605; Utas, 1994, p. 139). This very interesting piece is attributed to the famous Bārbad, a poet at the court of Ḵosrow Parviz, and its content indicates that it must actually have dated from the Sasanian period. We must also add another two poems of moral content preserved by the Toḥfat al-moluk (7th/13th century) composed in regular motaqāreb, but having rhymes that are not in accordance with Persian classical tradition (Lazard, 1974). On the other hand, we must not include two fragments in Manichean script pointed out by Henning (1962), for they do not differ in either language or verse form from Persian poetry of the 4th/10th century.

The Manichean manuscripts in Parthian and Middle Persian enable us to recognize the general form of the poems. The unit is generally the distich, the two verses of which are each divided into two hemistiches. There are no rhymes.

The poetic texts in Pahlavi “prove to have the same underlying metrical principles” as the Manichean poems (Boyce, 1957, p. 40). Their form is not as evident as in the case of the latter, because they are presented like prose and, in addition, are often altered. We can nevertheless recognize distiches in the “Hymn to Wisdom,” Was raft hēm …, and the poem edited by de Blois (Kay bawād …). Henning (1950) divided the Dārom andarz-ē …into two hemistiches, as well as some passages of Draxt ī asūrīg (he refers to the distiches thus divided into four as strophes). Draxt ī asūrīg does not appear to be rhymed, but in other Pahlavi poems rhymes are not infrequent. In Dārom andarz-ē …, a rhyme in -ān appears in the first two hemistiches and, afterwards, at the end of each verse, that is, twice per distich. In Kay bawād …, a rhyme likewise in -ān appears in each verse. In “Hymn to Wisdom,” all the distiches, except for the first, end in a comparative in -tar. In Was raft hēm …, the first four distiches have a rhyme in -ag and the following four end in the word xrad. These rhymes differ from those of classical Persian poetry in that the same word or the same suffix is allowed to rhyme with itself.

Rhymes with this same structure and placement appear in almost all the pieces transmitted by Arabic or Persian texts. For example, in the hymn of the Karkoy pyre, all the (very short) verses rhyme together (in ōš). The same is true of the two poems quoted by Toḥfat al-molūk, although the metrics are the same as the classical model. For example, one begins: to az kār-e Keyḵ-osrow andāze gir / kohan gašte kār-e jahān tāze gīr/ jahān dār saḵ-t-o bas aš sost gir / be saḵ-ti-o sostī to yeksān begir …. On the other hand, the three verses attributed to Bārbad (Qaysar māhmānaḏ …) are not rhymed.

Overall, we may observe an evolution. The poetry of the Sasanian period was unrhymed. Later on, there developed in Pahlavi poetic forms containing the regular or non-regular repetition of the same syllable or the same word, either at the end of the verses or at the end of the distiches. These rhymed forms continued in popular poetry and also in Persian poems in classical prosody.

What is the basis for the structure of the Middle Persian and Parthian verse? Clearly it is not the short or long quantity of the syllables, as in classical Persian poetry. In his pioneering studies on Middle Persian poetry, Benveniste thought that the versification was syllabic. Henning (1950) refuted this thesis. He showed that the verses contained a variable number of syllables and that this number varied, for each poem, between precise limits. He thus reached the conclusion that the versification is based on stress.

Boyce, in the introduction to her edition of the Manichean hymn cycles, tried to clarify the rules by examining the (presumed) place of the accents of words in the verses of these poems. This meticulous examination did not lead to definition of the structures. The author concluded (1954, p. 47): “In fact few combinations of stressed and unstressed are excluded; and the number and fluidity of the resulting patterns make any general attempt at schematization of doubtful value.”

Lazard (1985) took up the question on a slightly different basis. He formed the hypothesis that two types of syllables are to be distinguished. Light syllables are all short syllables and those long syllables which belong to auxiliary words or words that are positionally weak (in prosthesis or anaptyxis). Heavy syllables are all other long syllables. He suggests that the stresses or beats (ictus) which define the rhythm of a verse always fall on heavy syllables. Analysis of the intact verses of the Manichean hymn cycles supports the following rules: (1) The stress often coincides with the accent on the word, but not always. We may suppose that the accent and the stress are characterized by different phonetic properties, for instance the accent by a higher pitch and the stress by an intensity or a length. (2) The basic rhythmic unit (foot) is formed by a heavy syllable that carries the stress, preceded by a variable number of heavy and light syllables; but for a given poem the number of heavy syllables per foot is limited. (3) Enclitics are treated in the same way as root words.

It appears that, in the two hymn cycles in Parthian published by Boyce, the verse has four stresses and the foot contains a maximum of three heavy syllables, including the one that bears the stress. The Manichean poem in Parthian M 10, published by Henning (1933) and also analyzed by Lazard (1985 and 2003), has four stresses per verse, but a maximum of two heavy syllables per foot. Lazard also applied these principles to some Pahlavi poems. In the “Hymn to Wisdom” (2001), and Was raft hēm … (2002), he found four stresses per verse with a maximum of two heavy syllables per foot. In the Draxt ī asūrīg (2003), there were four stresses per verse with a maximum of three heavy syllables per foot.

The rules of versification thus derived are still conjectural and do not explain everything. For example, they do not account for the purely statistical difference, pointed out by Boyce, between the average length of the verses in the two cycles of the Parthian hymns. We may suppose that the difference is connected with the music accompanying the poems.

Classical Persian versification is different from that of pre-Islamic Iran; it is based, as in Arabic versification, on the arrangement of short and long syllables. It was for a long time believed that this quantitative system was borrowed from Arabic poetry. This claim will have to be seriously qualified. In fact, although the principle is the same, the use that is made of it is not the same on both sides: the meters most used in Persian are rare or nonexistent in Arabic, and vice versa (Elwell-Sutton, 1975). This is not surprising, for, although any system of versification is more or less artificial, it is nevertheless conditioned by the phonological properties of the language. The phonology of Persian scarcely differs from that of Middle Persian (and Parthian), for from Western Middle Persian to New Persian the evolution is minimal; we merely need to take into account the massive borrowing of Arabic words, which increased the proportion of short vowels. Hence the natural rhythms of New Persian ought to be little different from those of Middle Persian (and Parthian). One is thus led to believe that there must be some lineal relationship between pre-Islamic Iranian meters and those of Persian poetry.

This idea was first expressed by Benveniste (1932, p. 293): “The originality of the Persians as regards poetic technique consisted of subjecting the syllabic Iranian meter to Arabic quantitative prosody.” If we replace “syllabic” by “accentual,” this formula preserves its value. This idea has also been expressed by Utas (1994, p. 140): “The origin of many of the New Persian meters must be sought in earlier Iranian rhythmic structures that were formally adapted to a quantitative structure.” Grunebaum (1955, p. 18), for his part, has suggested that certain meters used by Arab poets, “the ramal, the mutaqârib, and perhaps the kḫafîf, may be considered as adaptations of Persian (Pahlavi) meters to Arabic linguistic conditions.”

Lazard has tried to clarify the relationships. He first discovered the origin of the robāʾi meter in the accented distich āhu-ye kuhi … (1970). Analyzing the epic Baluchi meter, which has 7 to 10 syllables and contains four stresses and a final cadence <–′ v  –′ >, he deemed it related to the motaqāreb; he then made the assumption that this Baluchi meter was the continuation of the Middle Persian epic meter (1994). Using the same type of analysis with meters of the Pahlavi poems “Hymn to Wisdom” (2001), Was raft hem … (2002), and Draxt ī asūrīg (2003), and with the Manichean poem in Parthian M 10 (2003), he found grounds for connecting the first one with the catalectic motaqāreb, a narrative or didactic meter; the second one with one kind of hazaj, a narrative meter; the third one with various Persian meters based on a four-syllable foot; and the fourth one with the acatalectic motaqāreb, a lyrical verse.

The method is as follows: The accentual meter is coded in a “structural formula” containing all the syllables, required or optional (the latter shown in parentheses), and indicating the stresses. Then one seeks among the usual meters of classical Persian poetry one which has the same number of syllables and in which the long syllables correspond to the stressed syllables of the accentual meter. For example, the meter of the “Hymn to wisdom” is represented by the following formula (where the capital letters symbolize the syllables bearing the stress): (x) (x) X (x) x X (x) (x) X x X, with the condition that all optional syllables are never simultaneously present or absent; additionally, as mentioned above, there are no more than two heavy syllables in a foot. It appears that this formula superimposes itself exactly on the structure of the narrative motaqāreb; moreover, the stress falls on those syllables where, in the motaqāreb meter, the word accents are most frequently placed (Rypka, 1936, pp. 200-201; cf. Lazard, 1994, p. 87). Not only do the stresses of the accentual meter correspond to some of the long syllables of the quantitative meter, but also all short syllables of the quantitative meter correspond to some of the optional syllables of the accentual meter.

The correspondences are as follows:

“Hymn to wisdom” (x) (x) X (x) x X (x) (x) X x X
motaqāreb narrative v - - v - - v - - v -
Parthian poem M 10 (x) x X (x) x X    (x) x X (x) x X
motaqāreb lyric v - - v - -     v - - v - -
Was raft hem (x) X (x) (x) X (x) X (x) (x) X
hazaj - - v v - v - v - -
Draḵt ī asūrīg (x) (x) x X (x) (x) x X     (x) (x) x X (x) (x) x X
ramal v v - - v v - -    v v - - v v - -
ramal - v - - - v - -    - v - - - v - -
hajaz v - - - v - - -     v - - - v - - -
Ahu-ye kuhi … x X x x X     x X x x X x x X
robāʾi - - v v -     v - v - - v v -

Lineal relationships between pre-Islamic versification and Persian versification have not been strictly proven. However, considering that all the above-mentioned conditions about the number and nature of the syllables on both sides are fulfilled, it may be thought that the correspondences observed are not altogether illusory. It is true that the quantitative meter is always longer than the corresponding accentual meter. But we know that New Persian has more short syllables than Middle Persian. Moreover, the short syllables of the classical Persian meter are often merely the third mora of an overlong syllable, that is, actually a phantom syllable.


Harold W. Bailey, Zoroastrian Problems in the Ninth-Century Books, Oxford, 1943.

Emile Benveniste, “Le texte du Draxt asūrīk et la vérification pehlevie,” JA 216, 1930, pp. 193-225.

Idem, “Le Mémorial de Zarēr, poème pehlevi mazdéen,” JA 220, 1932, pp. 245-93.

Idem, “Une apocalypse pehlevie: le Žāmāsp Nāmak” RHR 53, 1932, pp. 337-80.

François de Blois, “A Persian Poem Lamenting the Arab Conquest,” in Hillebrand, ed., Studies in Honour of C. E. Bosworth, Leiden, 2000, vol. 2, pp. 82-95.

Mary Boyce, The Manichaean Hymn Cycles in Parthian, London, 1954.

Idem, “The Parthian gōsān and Iranian minstrel tradition,” JRAS,1957, pp. 45-52.

Idem, A Reader in Manichaean Middle Persian and Parthian, Leiden, 1975.

L. P. Elwell-Sutton, “The Foundations of Persian Prosody and Metrics,” Iran 13, 1975, pp. 75-97.

Gustav E. von Grunebaum, Kritik und Dichtkunst, Studien zur arabischen Literaturgeschichte, Wiesbaden 1955.

Walter Bruno Henning, “Geburt und Entsendung des manichäischen Urmenschen,” Nachrichten der Gesellschaft der Wissenschaften zu Göttingen,1933, pp. 306-18.

Idem, “The Disintegration of the Avestic Studies,” TPS,1942, pp. 40-56.

Idem, “A Pahlavi Poem,” BSOAS 13, 1950, pp. 641-48.

Idem, “Persian Poetical Manuscripts from the Time of Rūdakī,” in A Locust’s Leg. Studies in Honour of S. H. Taqizadeh, London, 1962, pp. 89-104.

Gilbert Lazard, “Ahu-ye kuhi: le chamois d’Abu Hafs de Sogdiane et les origines du robâi,” in W.B Henning Memorial Volume, London, 1970, pp. 238-44.

Idem, “Pahlavi, pârsi, dari: les langues de l’Iran d’après Ibn al-Muqaffa’,” in Iran and Islam in Memory of the Late V. Minorsky, Edinburgh, 1971, pp. 361-91.

Idem, “Deux poèmes persans de tradition pehlevie,” in Ph. Gignoux and A. Tafazzoli, eds., Mémorial Jean de Menasce, Leuven, 1974, pp. 433-40.

Idem, “The Rise of the New Persian language,” in R. N. Frye, ed., The Cambridge History of Iran IV, Cambridge, 1975, pp. 595-632.

Idem, “La métrique de la poésie parthe,” in Papers in Honour of Professor Mary Boyce, Leiden, 1985, II, pp. 371-99.

Idem, “Le mètre épique baloutchi et les origines du motaqâreb,” in L. Johanson and B. Utas, eds., Arabic Prosody and its Applications in Muslim Poetry, Stockholm, 1994, pp. 81-90.

Idem, “La versification d’un poème pehlevi,” in A. A. Sadeghi, ed., Tafazzoli Memorial Volume, Tehran, 2001, pp. 39-47.

Idem, “Encore la versification pehlevie,” Jerusalem Studies in Arabic and Islam 26 (= Studies in Honour of Shaul Shaked), Jerusalem, 2002, pp. 130-39.

Idem, “Le mètre du Draxt asûrîg “ in Orientalia suecana 51-52, 2002-03, pp. 327-36.

Idem, “La versification d’un hymne manichéen en parthe,” in C. G. Cereti et al., Religious Themes and Texts of Pre-Islamic Iran and Central Asia. Studies in honour of Professor Gherardo Gnoli, Wiesbaden, 2003, pp. 223-30.

J. Rypka, “La métrique du mutaqarib épique persan,” Travaux du cercle linguistique de Prague 6, 1936, pp. 192-207.

Ḏ. Ṣafā, Tāriḵ--e adabiyāt dar Irān I, Tehran, 1956.

Shaul Shaked, “Specimens of Middle Persian Verse,” in W.B. Henning Memorial Volume, London, 1970, pp. 395-405.

Ahmad Tafazzoli, “Andarz ī Wehzād Farrox Peroz containing a Pahlavi Poem in Praise of Wisdom,” Studia Iranica 1, 1972, pp. 207-17.

J. C. Tavadia, “A Didactic Poem in Pahlavi,” in M.P. Kharegat Memorial Volume, Bombay, 1953, I, pp. 271-75; and idem, Indo-Iranian Studies, Santiniketan, 1950, pp. 86-95.

Idem, “A Rhymed Ballad in Pahlavi,” JRAS,1955, pp. 29-36.

Bo Utas, “On the Composition of the Ayyātkār i Zarērān,” in Monumentum H. S. Nyberg, Leiden, 1975, II, pp. 399-418.

Idem, “Arabic and Iranian Elements in New Persian Prosody,” in L. Johanson and B. Utas, eds., Arabic Prosody and its Applications in Muslim Poetry, Stockholm, 1994, pp. 129-41.

(Gilbert Lazard)

Originally Published: November 15, 2006

Last Updated: November 15, 2006